Beruflich Dokumente
Kultur Dokumente
02 Multifactorial
The relational expression of multifactorial and the gamma function is shown here.
For instance, in case of double factorial,
/
1 1 1
(2n -1)!! = 2n n + / = 2n n +
2 2 2
= 2 (n +1) = 2 n !
2 2
(2n)!! = 2n n + / n n
2 2
These are used to express the half-integration of a integer-power function later.
Moreover, we obtain the following Maclaurin expansions as by-products.
2 =
x
f x + x + x + x +
0!! 2!! 4!! 6!!
f (0) 0 f '(0) 1 f " (0) 2 f 3 (0) 3
( )
f =
x
x + x + x + x +
3 0!!! 3!!! 6!!! 9!!!
Next, multinomial thorem and generalized multinomial thorem are shown as follows.
Theorem 3.3.1
For real numbers x1 , x2 , , xm and a natural number n , the following expressions hold.
r r r
n r1 rm -2 n r1 rm-2 rm-2-rm-1 rm-1
x1+ x2++xm = Σ Σ Σ x2 xm-1
n n -r1 r1-r2
x1 xm
r1=0 r2=0 rm -1=0 1 2 m-1
(1.1)
r1+ r2+ + rm-1
r + r + + r
n n n n rm-2 + rm-1
= Σ Σ Σ
r1=0 r2=0 rm -1=0 1 2 m-1 r2+ + rm-1 rm-1
n -r1- - rm-1 r
x1 x21 x32 xmm-1
r r
Theorem 3.4.1
The following expressions hold for real numbers and x1 , x2 , , xm s.t.x1 x2+ x3++ xm .
-1-
r r r
r1 r m -2 r1 rm-2 -r1 r1-r2 r -rm-1 rm-1
x1+ x2++xm = Σ Σ Σ xm-1
m-2
. x1 x2 xm
r1=0 r2=0 rm -1=0 1 2 m-1
(1.1)
r1+ r2+ + rm-1
r + r + + r
rm-2 + rm-1
= Σ Σ Σ
r1=0 r2=0 rm -1=0 1 2 m-1 r2+ + rm-1 rm-1
-r1- - rm-1 r1 r2 r
x1 x2 x3 xmm-1
(1.2)
Where, x 1 x 2+ x 3++ x m is allowed at >0 .
Higher order calculus of the product of many functions and super order calculus of the product of many functions are
expressed just like these later.
What should be paid attention here are the following property of generalized binomial coefficients.
r
n n
Σ
r=0
nC r = Σ
r=0
That is, once generalized binomial coefficients was used, the upper limit of should be . Therefore, when n is
a natural number and p is a real number, the following holds in most cases.
r f(p,x)
n p
n p Σ nC r f(n,x) Σ
r=0 r=0
When the original coefficient is not binomial coefficient e.g. 1 ,
r f(p,x)
n p
n p Σ nC r f(n,x) Σ
r=0 r=0
r f(p,x)
p p
Although Σ
r=0
1f(p,x) is difficult, Σ
r=0
is satisfactory. What enables super calculus in this text
04 Higher Integral
f(x)dx
x x x x x
f(x)dx 1 , f(x)dx 2 , , n
a1 a2 a1 an a1
And these are called higher integral with variable lower limits . On the other hand,
-2-
Theorem 4.1.3
r =0, 1, , n
<r> <r+1>
Let f be continuous functions defined on a closed interval I and f be the arbitrary
ar , x I .
<r>
primitive function of f . Then the following expression holds for
x x n -1 x x
f(x)dx n = f (x) - Σ f -r a n-r
<n > <n >
dx r
an a1 r=0 an a n- r+ 1
f(x)dx
x x
<n > n -1
<n -r> (x -a)r
n
=f (x) - Σf (a)
a a r=0 r!
x x x Left: Lineal 3rd order integral
3
sin xdx = cosx
3 2 1 Right: Lineal 3rd order primitive
2 2 2
x x x
3x2 Left: Collateral 3rd order integral
sin xdx = cosx + -1 Right: Collateral 3rd order primitive
0 0 0 2
for r =1, 2, , n
<r>
Furthermore, from Theorem 4.1.3, we see that ar must be all zeros of f in order for the
higher integral of f (x ) to be lineal.
(x -t)
x x 1 x
f(x)dx n =
n-1
f(t)dt
a a (n) a
Reimann-Liouville Integral of the right side is important. However, in the higher integral, since the left side itself has
an operation functions, the right side is not indispensable.
By the way, replacing the left side in Theorem 4.1.3 for Reimann-Liouville Integral and shifting the index by -n ,
we obtain the following.
(x -t)
(x -a)r
n -1
(r) 1 x
n-1 (n )
f
<0>
(x) = Σf (a) + f (t)dt
r=0 r! (n) a
(n)
This is the Taylor expansion of f (x ) around a . And Reimann-Liouville Integral of f is the remainder term called
Bernoulli form
x dx
x x 1 m!
m 2 m+2 m+2
= x = x
0 0 (m +1)(m +2) (m +2)!
-3-
Then, when is a positive number, it is as follows.
!
x dx
x x
2
= x +2
0 0 ( +2)!
Here, ! can be expressed by gamma function (1+) . Thus
(1+ ) +2
x dx
x x
2
= x
0 0 (1+ +2)
By such an easy calculation, we obtain the following expressions for elementary functions. Where, , denote the
ceiling function and the floor function respectively.
e dx = (1) e
x x
x n n x
n
2
x x x
sin xdx = sin x - n
n 2 1
2 2 2
n
x
x x
cosxdx n = cos x -
(n -1) 1 0 2
2 2 2
e x - (-1)ne
x x x -x
sinh xdx n =
n i 2 i 1 i 2
2 2 2
e x + (-1)ne -x
x x x
cosh xdx = n
(n -1) i 1 i 0 i 2
2 2 2
x x tan -1x n /2
tan x dx = -1
Σ (-1)k nC n-2k x n-2k
n
0 0 n ! k=0
log1+ x 2 n /2
+ Σ (-1)k nC n+1-2k x n+1-2k
2n ! k=1
1 n /2
- Σ (-1)r nC n+1-2r (1+n)- (2r)x n+1-2r
n! r=1
x x xn tan -1x n /2
(-1)k nC n-2k x n-2k
n! Σ
-1 n -1
cot x dx = cot x -
0 0 n! k =1
-4-
n-2r
x x n/2 x
sin x dx = Σ
-1 n
2 sin -1x
an a1 r=0 (2r)!! (n -2r)!
n-2r+1
n /2 n -2r+1 (s -1)!!2 x
+ Σ Σ (-1) n -2r+1C s
s
2 (n -2r +1)!
1-x 2
r=1 s=0 (s +2r -1)!!
Where, a1 = i 1.508879 , a2 = 0 , a3 = -i 0.475883 , a4 = 0 ,
n-2r
x x xn n/2 x
cos x dx = cos x - Σ
-1 -1 n
2
-1
sin x
an a1 n! r=1 (2r)!! (n -2r)!
n-2r+1
n /2 n -2r+1
s (s -1)!!2 x
- Σ Σ (-1) n -2r+1C s 2 (n -2r +1)!
1-x 2
r=1 s=0 (s +2r -1)!!
Where, a1 = 1 , a2 = 0 , a3 = ? , a4 = 0 ,
n-2r-1
x x xn (n -1)/2 (2r -1)!! x
-1
sec xdx = n
sec x - Σ
-1
logx + x 2-1
1 1 n ! r=0 (2r)!!( 2r +1)! (n -2r -1)!
n-2r
s n -2rC s
(s -1)!!2 x
n /2 n -2r
+Σ Σ (-1) 2 x 2-1
r=1 s=0 2r +s (s +2r -1)!! (n -2r)!
n-2r-1
x x xn (n -1)/2 (2r -1)!! x
csc x dx = csc x + Σ
-1
-1 n
logx + x 2-1
an a1 n! r=0 (2r)!!(2r +1)! (n -2r -1)!
n-2r
n /2 n -2r Cs
n -2r (s -1)!!2 x
-Σ Σ (-1)s x 2-1
r=1 s=0 2r +s (s +2r -1)!!2 (n -2r)!
Where, a1 , a2 , , an are all complex numbers.
x x tanh -1x n /2 n-2k log1- x 2 n /2 n+1-2k
n
Σ nC n-2k x Σ nC n+1-2k x
-1
tanh x dx = +
0 0 n ! k =0 2n ! k =1
1 n /2
nC n+1-2r (1+ n )- (2r)x
n+1-2r
n! Σ
-
r=1
x x xn n tanh -1x n /2 n-2k
Σ nC n-2k x
-1 -1
coth x dx = coth x +
0 0 n! n ! k=1
log1-x 2 n /2 n+1-2k
+
2n ! k=1 Σ nC n+1-2k x
1 n /2
nC n+1-2r (1+ n )- (2r)x
n+1-2r
n! Σ
- |x|<1
r=1
(-1)r x n-2r
x x n/2
sinh x dx = Σ
-1 n
2 sinh -1x
an a1 r=0 (2r)!! (n -2r)!
n-2r+1
n /2 n -2r+1
r+s (s -1)!!2 x
+Σ Σ (-1) n -2r+1C s 2 (n -2r +1)!
1+x 2
r=1 s=0 (s +2r -1)!!
Where, a1 = 1.508879 , a2 = 0 , a3 = -0.475883 , a4 = 0 ,
n-2r
x x n/2 x
cosh -1x dx n = Σ 2 cosh -1x
1 1 r=0 (2r)!! (n -2r)!
-5-
n-2r+1
n /2 n -2r+1
s (s -1)!!2 x
-Σ Σ (-1) n -2r+1C s 2 (n -2r +1)! x
2
-1
r=1 s=0 (s +2r -1)!!
n-2r-1
x x xn (n -1)/2 (2r -1)!! x
sech x dx = sech x + Σ
-1
-1 n
sin -1x
an a1 n! r=0 (2r)!!(2r +1)! (n -2r -1)!
n-2r
n /2 n -2r n -2r Cs
(s -1)!!2 x
+Σ
r=1
Σ (-1)s
s=0 2r +s (s +2r -1)!!2 (n -2r)!
1- x 2
x x xn (n -1)/2 (-1)r(2r -1)!! x
csch -1x dx n = csch -1x + Σ sinh -1x
an a1 n! r=0 (2r)!!(2r +1)! (n -2r -1)!
n-2r
r+s n -2rC s (s -1)!!2 x
n /2 n -2r
+Σ Σ (-1) 2 x 2+1
r=1 s=0 2r +s (s +2r -1)!! (n -2r)!
Where, a1 = 0 , a2 = 0.6079 , a3 = 0 , a4 = 1.5539 ,
(6) Termwise Higher Integral and Taylor series of higher primitive function
Theorem 4.6.1
r =0, 1, , n
<r> <r +1>
Let f be continuous functions defined on [a , b ] and f be the arbitrary primitive function
<r>
off . At this time, if f (x ) can be expanded to the Taylor series around a, the following expressions hold for
x [a , b] .
<n > n -1
<n -r> (x -a)r (r) (x -a)n+ r
<n -r> (x -a)r
f (x) =Σf (a) +Σ f (a) =Σf (a)
r=0 r! r=0 (n + r)! r=0 r!
<n>
This expression shows that the Taylor series of f consists of the constant-of-integration polynomial and the
termwise higher integral of f (x ). The following can be said from this.
(1) A termwise higher integral with a fixed lower limit is collateral generally.
(2) It is the following case that a termwise higher integral with a fixed lower limit is lineal.
(a)=0 for r =1, 2, ,n (a) 0 , for at least one s 0
<r> (s)
f & f
For example,
x x (x -a)n+ r
e dx x n
a -
= Σ ea is collateral higher integral.
a a r=0 (n + r)!
x x (2r)!
tan -1x dx n = Σ(-1)r x
n+2r+ 1
is lineal higher integral.
0 0 r=0 ( n +2r +1)!
Next, the following is obtained from the Taylor series of the higher integral of log x .
n-1
1 (-1) n -1
H
n 1
Σ
r=1 r(r +1)(r + n )
=
n! Σ
r=0
(-1)rnC r Hn-r n =Σ
k =1 k
(-1)r-1 2n 1 n -1
Σ
r=1 r(r +1)(r + n)
=
n!
log 2 -H n +
n! Σ
r=0
nC r Hn-r
n -1 (-1)n-1
Σ
r=0
(-1) nC r Hn-r r
=
n
-6-
(1- x)p
1 p
dx = 2plog 2- (1+p )-+Σ (1+r)+
0 1+ x r=1 r
Example
1 1 1 1 1
+ + + + =
1234 2345 3456 4567 18
1 1 1 1 4 8
- + - +- = log 2 -
1234 2345 3456 4567 3 9
x x (-1)n-1 e -2k xn
x
tanh x dx n = Σ(-1)
k-1
+
0 0 2n-1 k =1 kn n!
n -1 (k +1) x n-1- k
- Σ(-1)k x >0
k=0 2k (n -1- k)!
n
(-1)k
x x
sec x dx = 2Σ
n
n cos (2k +1)x -
0 0 k=0 (2k +1) 2
n /2 2 (2k) n-2k
+ Σ (-1)k-1 x |x|<
k=1 (n -2k)! 2
-7-
)x
x x e -(2k +1
sech x dx n = (-1)n2Σ(-1)k x >0
k=0 (2k +1)n
(3) Riemann Odd Zeta & Dirichlet Odd Eta
Comparing Taylor series and Fourier series, we obtain Riemann Odd Zeta and Dirichlet Odd Eta. For example,
2n+1
Σ
2 -1B2k
2k
(-1)n 22n
(2n +1) = 2k+2n+1 - log 2
22n-1 k=1 2k (2k +2n +1) ! (2n +1) !
(-1)n 22n n -1 2n+1-2k 22k -1
- Σ(-1)k (2k +1)
22n-1 k =1 (2n +1-2k) ! 22k
2n+1
(2n)!
B2k
1 2n +1 1
(2n +1) = (-1)n log - Σ -Σ 2k+2n+1
k=1 k k=1 2k (2k +2n +1) !
1 n -1 2n-2k+1
(2k +1)
Σ
+ (-1)n (-1)k-1
k=1 (2n -2k +1) !
2n-1 n-1- j
e -k n -2
j-1 2 -1 1
(n) = n-1 Σ (-1) n
+ Σ(-1) n-1- j
k-1
(n -j)
2 -1 k=1 k j=1 2 j!
(-2)n-1 2 -1B2k
2k
1 1 log 2
- n-1 Σ - +
2 -1 k=1 2k (2k + n -1)! 2 n! (n -1)!
(-1)n-1 E2k 2k+2n
(2n -2k) 2k
2 2
n
(2n) = Σ +Σ(-1) k-1
2 k=0 (2k +2n ) ! k=1 (2k) !
x x (2k)!
tan -1x dx n = Σ(-1)k x 2k+
n+1
|x|< 1
0 0 k=0 (2k + n +1)!
xn
x x (2k)!
cot -1x dx n = - Σ(-1)k x 2k+
n+1
0< x 1
0 0 2 n! k=0 (2k + n +1)!
sin
x x 2
(2k -1)!! n+1
-1
x dx = Σ
n
x 2k+ |x|< 1 collateral
0 0 k=0 (2k + n +1)!
xn
x x 2
(2k -1)!!
cos -1x dx n = -Σ x 2k+
n+1
|x|< 1 collateral
0 0 2 n! k=0 (2k + n +1)!
tanh
x x (2k)! n+1
-1
x dx n = Σ x 2k+ |x|< 1
0 0 k=0 (2k + n +1)!
sinh
x x 2
k (2k -1)!! n+1
-1
x dx = Σ(-1)
n
x 2k+ |x|< 1 collateral
0 0 k =0 (2k + n +1)!
xn 2
x x 2 n 1 (2k -1)!!
-1
sech x dx = n
log +Σ -Σ x 2k+n
0 0 n! x j=1 j k=1 2k (2k +n)!
0< x <1 collateral
-8-
xn 2
x x 2 n 1 (2k -1)!!
-1
csch x dx = n
log +Σ -Σ(-1)k x
2k+n
0 0 n! x j=1 j k=1 2k (2k +n)!
0< x <1 collateral
(2) By-products
1 (1- x)n n /2
nC n+1-2r (1+ n )- (2r)
n-1
0 1- x 2
dx = 2 log 2 - Σr=1
1 (1- x)n n /2 log 2 n /2
0 1+ x 2 dx =
4 Σk=0
(-1)k
nC n-2k +
2 Σk=1
(-1)k nC n+1-2k
n /2
- Σ (-1)r nC n+1-2r (1+n)- (2r)
r=1
n-1
2
(2k)! 1 n /2
Σ nC n+1-2r (1+ n )- (2r)
n! Σ
log 2 - =
k=0 (2k + n +1)! n! r=1
(2k)! n /2 log 2 n /2
4 n! Σ 2n! Σ
k k
Σ (-1) = ( -1 ) nC n-2k + (-1)k nC n+1-2k
k=0 (2k +n +1)! k=0 k =1
1 n /2
- Σ (-1)r nC n+1-2r (1+ n)- (2r)
!
n r=1
2n -1C n-1
2
(2k -1)!! n /2 1
Σ = -Σ 2
k=0 (2k + n +1)! (2n)!! k=1 (2k -1)!! (n -2k +1)!
(2k -1)!! 1 n 1 n /2 1 1
Σ 2k (2k +n)!
=
n!
log 2+Σ
j=1 j
+Σ 2 (n -2r)!
k=1 2r(2r -1)!!
r=1
-9-
(1) Definitions and Notations
<p>
f (x) denotes the non-integer order primitive function of f (x ). And we call this Super Primitive Function of
f(x). Since there is a super primitive function innumerably, which f p (x) means follows the definition at that time.
< >
We call it Super Integral to integrate a function f with respect to an independent variable x from a (0) to a (p )
continuously. And it is described as follows.
x x
x x
p
f(x)dx = f(x)dx dx
a(p ) a(0) a(p ) a(0)
And
when a(k)= a for all k [0 , p ] , we call it super integral with a fixed lower limit ,
when a(k) a for some k [0 , p ] , we call it super integral with variable lower limits .
x x p -1 x x
<p > <p >
f(x)dx p = f (x) - Σ f -r a(p -r) dx r
a(p ) a(0) r=0 a(p ) a(p -r+1)
Especially, when a(r)= a for all k [0 , p ] ,
x x p -1 (x -a)r
f(x)dx n = f
<p >
(x) - Σf <p -r>
(a)
a a r=0 (1+ r)
And when Constant-of-integration Function is 0, we call the left side Lineal Super Integral . and we call f <p>(x)
Lineal Super Primitive Function.
Oppositely, when Constant-of-integration Function is not 0, we call the left side Collateral Super Integral and we call
the right side Collateral Super Primitive Function.
For example,
p
x
x Left: Lineal p th order integral
p
sin xdx = sin x -
p 0 2 Right: Lineal p th order primitive
2 2
x x (-1)r Left: Collateral p th order integral
sin xdx p =Σ x 2r+1+
p
0 0 r=0 (2r +2+ p) Right: Collateral p th order primitive
- 10 -
Let
(x -t)p-1
f(t) = t , g(x -t) =
(p)
Then
f(t)g(x -t)dt
<p > 1 x x
x t (x -t)
p-1
= dt =
(p) 0 0
(f g )(x) . Then we take the Laplace transform of (f g)(x)
We find out that this is a convolution
(f g)(x) F(s)G(s)
( +1)
f(x) = x
= F(s)
s +1
x
p-1
1 (p) 1
g(x) = = = G(s)
(p) (p) s p sp
( +1) 1 ( +1) ( +p +1)
F(s)G(s) = =
s +1 s
p ( +p +1) s +p+1
( +1) a+p
Finally, taking the inverse Laplace transform, we obtain x .
( +p +1)
Because the technique of Fractional Integral is difficult like this, it is more difficult to obtain the super primitive
function of log x by this technique.
Above all, the problem is that Fractional Integral cannot treat the lineal non-integer order integral such as sin x.
Because Riemann-Liouville Integral cannot be applied to the integral with variable lower limets.
On the other hand, in Super Integral that I advocates, first of all, we obtain the following higher integral.
(1+ ) +n
x
x x
dx n = x ( 0)
0 0 1+ + n
And replacing the index of the integration operator n with a real number p , we obtain the following very easily.
(1+ ) +p
x
x x
dx p = x ( 0)
0 0 1+ + p
Furthermore, performing the higher integral with variable lower limits and replacing the index of the integration operator
with a real number, we can obtain the super integral such as sin x easily.
e dx = (1) e
x x
x p p x
log x - (1+ p) -
log xdx = (1+p)
x x
p
xp
0 0
p
x
2
x
sin xdx = sin x - p
p 0
2 2
p
x
x
cosxdx p = cos x -
(p -1) -1 2
2 2
- 11 -
x x e x - (-1)pe -x
sinh xdx p
=
p i 0 i 2
2 2
e x + (-1)pe -x
x x
cosh xdx p
=
(p -1) i -1 i 2
2 2
tan -1x
x x p
tan -1x dx p =
Σ (-1)k x
p-2k
0 0 (1+ p ) k =0 p -2 k
log1+ x 2
p p+1-2k
+ Σ (-1)k x
2(1+ p) k=1 p +1-2k
1 p
(1+ p )- (2r)x
p+1-2r
-
(1+ p) r=1 Σ (-1)r
p +1-2r
p -1
x x x tan x p
cot x dx =
-1 p -1
cot x - Σ (-1)k x
p-2k
0 0 (1+ p) (1+ p) k=1 p -2k
log1+ x 2
p p+1-2k
-
2(1+ p) k=1 Σ (-1 )k
p +1-2k
x
1 p
(1+ p )- (2r)x
p+1-2r
+ Σ
(1+ p) r=1
(-1)r
p +1-2r
tanh -1x
x x p
tanh -1x dx p = Σ x
p-2k
0 0 (1+ p ) k=0 -2
p k
log1- x 2
p p+1-2k
+ Σ
2(1+ p) k=1 p +1-2k
x
1 p
(1+ p )- (2r)x
p+1-2r
- Σ
(1+ p) r=1 p +1-2r
(7) By-products
(-1)k (-1)k (1+ n , 1/m)
k k
n n (2n )!! n n
Σ
k=0 2k +1
=
(2n +1)!!
, Σ
k=0 mk +1
=
m
q+p
(-1)r r
r r
q -1 q <p > x (-1) q
(q , p) = Σ x
(p) Σ
, =
r=0 p+r r=0 p + r
- 12 -
22k 22k -1B2k
tan xdx
x x
p-1
p
=Σ x 2k+ 0< x <
0 0 k =1 2k (2k + p) 2
22k 22k -1B2k
tanh xdx
x x
p
=Σ x 2k+
p-1
0< x <
0 0 k =1 2k (2k + p) 2
secxdx
x x E2k
p
=Σ x 2k+p 0< x < collateral
0 0 k =0 (2k +p +1) 2
sech xdx
x x
p
E2k
=Σ x 2k+p 0< x < collateral
0 0 k =0 (2k +p +1) 2
22k 22k -1B2k
cotxdx
x x 2k+ p-1
p
= -Σ x- < x<
k=1 2k (2k + p) 2 2
2 2
E2k
x 2k+p
x
cscxdx p =Σ x- < x< collateral
k=0 (2k +p +1) 2 2
2 2
)x
x x e -(2k +1
csch x dx = (-1) 2Σ p
p
p
x>0
k=0 (2k +1)
)x
x x e -(2k +1
sech x dx p = (-1)p2Σ(-1)k x>0
k=0 (2k +1)p
x x (2k)!
tan -1xdx p = Σ(-1)k x 2k+p+1 0< x <1
0 0 k =0 (2 k + p +2 )
xp
x x (2k)!
cot -1xdx p = -Σ(-1)k x 2k+p+1 0< x <1
0 0 2 (1+ p) k=0 (2k + p +2)
2 (2k -1)!!
-1 p
= log + (1+ p) + -Σ x 2k+p
0 0 (1+ p) x k=1 2k (2 k + p +1)
0< x <1 collateral
xp
x x 2
2
k (2k -1)!!
-1
csch xdx = log + (1+ p)+ -Σ(-1)
p
x 2k+p
0 0 (1+ p) x k=1 2k (2k + p +1)
0< x <1 collateral
09 Higher Derivative
- 13 -
of f(x)..
Moreover, we call it Higher Differentiation to differentiate a function f with respect to an independent variable x
repeatedly. And it is described as follows.
n
d d d d d d
f(x) = f(x) : n pieces
dx n dx dx dx dx dx
r =0, 1, , n
(r)
When f are continuous functions on a closed interval I and are the r th derivative functions of f,
the following expression holds for x I .
dn
( ) = f n (x)
( )
nf x
dx
This theorem guarantees that only the lineal exists in the higher differentiation.
n
(sin x) n
()
= sin x +
2
n
(cosx) n
( )
= cos x +
2
(n ) e x - (-1)-ne -x
(sinh x) =
2
e x + (-1)-ne -x
(cosh x) n =
( )
2
etc.
- 14 -
(n ) (n -1)! n /2
n+1-2r
tan x = (-1)n nΣ
(-1)rnC n+1-2r x
-1
x +1
2 r=1
(n ) (n -1)! n /2
cot x
-1
= (-1)n-1 nΣ
(-1)rnC n+1-2r x n+1-2r
x +1
2 r=1
etc.
etc.
(6) By-Products
n /2 n
Σ (-1)k nC n+1-2k = -2n/2 sin
k=1 4
k +(n -1)! (n -1)!
Σ
k=0
(-1)k
k!
=
2n
Example
-5C 4 + 5C 2 - 5C 0 =4
1
12 - 23 + 34 - 45 +- =
4
- 15 -
(1) Taylor Series
(n )
22k 22k -1B2k 2k- n-1
(tanh x) = Σ x |x|<
n +1 2k(2k - n -1)! 2
k=
2
n! 22k B2k
(cot x)
(n )
= (-1) n
n+1
-Σ x 2k-n-1 0< x <
x n +1
2 2 - -1
k( k n ) !
k=
2
n! 22k B2k
(coth x)
(n )
= (-1) n
n+1
-Σ x 2k-n-1 0< x <
x n +1 2k(2k -n -1)!
k=
2
n! 22k -2B2k
(csch x)
(n )
= (-1) n
n+1
+ Σ x 2k-
n-1
0< x <
x n +1
2 2 -
k( k n )-1 !
k=
2
E2k
(sec x) n =
( )
Σ
n +1 (2k -n)!
x 2k-n |x|<
2
k=
2
E2k
(sech x) n =
( )
Σ
n +1 (2k -n)!
x 2k-n |x|<
2
k=
2
4
(1-2n) = Σ
24n-1 k =n 2k(2k -2n)!
(2n -1)! + Σ
(-1)n-1 2n 22k B2k 2k
4
4
(1-2n) =
24n-1 k=n 2k(2k -2n)!
22n -1B2n
(1-2n) =
2n
- 16 -
(-1)n 22k 22k -1B2k 2k-2n-1
4
(-2n) = Σ
22n+1 k=n +1 2k(2k -2n -1)!
(-1)n 2n 22k B2k 2k
4
4
(-2n) = (2n)! - Σ
22n+1 k=n +1 2k(2k -2n -1)!
E2n
(-2n) =
2
(4) Other by-products
1n 2n 3n 4n B2k
+ + + + = n! + (-1)n Σ
e1 e2 e3 e4 k=
n +1
2k(2k -n -1)!
2
2 -1B2k
n n n n 2k
1 2 3 4
n-1
- + - +- = (-1) Σ
e1 e2 e3 e4 k=
n +1 2k (2k -n -1)!
2
1 n
3 n
5 n
7 n
n! (-1)n 22k -2B2k
2 k =Σ
+ + + + = -
e1 e3 e5 e7 2 n +1 2k(2k - n -1)!
2
n n n n n E2k
1 3 5 7 (-1)
- + - +- = Σ
e1 e3 e5 e7 2 k=
n +1 (2k -n )!
2
p p p p
1 2 3 4
1
+ 2
+ 3
+ + (1+ p ) p >0
e e e e4
1p 3p 5p 7p (1+ p)
1
+ 3
+ 5
+ 7
+ p >0
e e e e 2
22k -2B2k 2k
(2n -1)! = Σ
k=n 2k(2k -2n )! 2
2k
2 B2k 2k
(2n)! = Σ
k= n +1 2 2
k k n
( -2 -1)! 2
(n ) (2k)! n
cot x Σ (-1)k
-1
=- x 2k+1- |x| <1
k=
n -1
(2k +1- n)!
2
2
(n ) (2k -1)!!
sin x Σ
-1
= x 2k+1-n |x| <1
k=
n -1
(2k +1- n)!
2
- 17 -
2
(n ) (2k -1)!!
cos x Σ
-1
=- x 2k+1-n |x| <1
k=
n -1
(2k +1- n)!
2
(2k -1)!! (2k +n)! -2k-n-1
(n )
csc x
-1
= (-1)nΣ x |x| > 1
k=0 (2k)!! (2k +1)!
-1 (n ) n-1 (2k -1)!! (2k +n )! -2k-n-1
sec x = (-1) Σ x |x| > 1
k =0 (2k)!! (2k +1)!
(n ) (2k)! n
tanh x Σ
-1
= x 2k+1- |x| <1
n -1
k=
(2k +1- n)!
2
(n ) (2k + n)! -2k-1- n
coth x = (-1)(n)Σ
-1
x |x| > 1
k=0 (2k +1)!
2
(n ) (2k -1)!!
sinh x Σ (-1)k
-1
= x 2k+1-n |x| <1
n -1
k=
(2k +1- n)!
2
(n ) (n -1)! n-1 (2k + n -1)! -2k-n
cosh
-1
x = (-1)n-1 + (-1) Σ 2 x x >1
xn k=1 (2k)!!
2
(n -1)!
(n ) k (2k -1)!!
csch x - Σ (-1)
-1 n
= (-1) n x 2k-n 0< x <1
x n 2k (2k -n)!
k=
2
2
(n ) (n -1)! (2k -1)!!
sech x -Σ
-1 n
= (-1) n x 2k-n 0< x <1
x n 2k (2k - n )!
k=
2
(4) By-Products
(2k)! (n -1)! n /2
Σ (-1)k = (-1)n Σ (-1)k nC n+1-2k
k=
n -1
(2k +1- n)! 2n k =1
2
Example
1
12 - 34 + 56 - 78 +- =-
2
d d d d
f(x) = f(x) : p pieces
dx p dx dx dx
- 18 -
function of f . And let a (r ) be a continuous function on the closed interval [0 , p ].
Then the following expression holds for a(r ), x I .
dp d p p -1 (r) x x
= (p )
f(x) f (x) + Σ f a(p -r) dx r
dx p dx p r=0 a(p ) a(p -r)
Especially, when a(r)= a for all k [0 , p ] ,
dp (p ) d p p -1 (r) (x -a)r
f(x) = f (x) + Σ f (a )
(1+ r)
dx p dx p r=0
when a = -
dp x Left: Lineal p -th order differenciation
pe
= ex
dx Right: Lineal p -th order derivative
As seen from this example, the lineal and the collateral exist in the super differentiation unlike the higher differenciation
(x -t)
dp 1 dn x
n-p-1
f(x) = f(t)dt n = ceil(p)
dx p (n -p) dx n a
Since the left side has lost the operating function, Reimann-Liouville Differintegral of the right side is very important.
=
1 1
(x -t)
1 x
2 1 d 1-
2
-1
(log x) log t dt
(1-1/2) dx 1 0
1
(x -t)
1 d x -
2
= log t dt
dx 0
x
-2
1 d x -t
= -4 x tanh -1 x -t (log t -2)
dx x 0
Long calculation continues.
- 19 -
1
1 d log x +2 log 2 - 2
= 4 x (log 2-1) +2 x log x = x
dx
Because the technique of Fractional Derivative is difficult like this, it is unknown in whether the case of p =1/3 is
calculable in this way.
Also, it is the same as the case of super integral that Riemann-Liouville Differintegral cannot be applied to the non-
integer order derivative such as sin x .
On the other hand, in Super Derivative that I advocates, first of all, we obtain the following higher derivative.
log x - (1+p)- p
log xdx
x x
p
= x
0 0 (1+ p)
Since differentiation is the reverse operation of integration, reversing the sign of the indexof the integration operator,
we obtain the following immediately.
log x - (1-p)- -p
(log x) p =
( )
x
(1- p)
Substituting p = 1/2 for this,
log x - 1/2- x - 2
1 1
(log x) 2 =
1/2
e x - (-1)-pe -x
(sinh x) p =
( )
2
- 20 -
e x + (-1)-pe -x
(p )
(cosh x) =
2
tan -1x
-1 (p )
-p
tan x =
(1-p) Σ
(-1)k x -p-2k
k=0 -p -2k
log 1+ x
2
-p
+ Σ (-1)k x -p+1-2k
2(1-p) k=1 -p +1-2k
1 -p
Σ (-1)r (1- p )- (2r)x
-p+1-2r
-
(1-p) r=1 -p +1-2r
tan -1x
-1 (p )
-p
cot x = - Σ (-1)k x -p-2k
(1-p) k=0 -p -2k
log1+ x
2
-p
- Σ (-1)k x
-p+1-2k
2(1-p) k=1 -p +1-2k
1 -p
Σ (-1)r (1- p )- (2r)x
-p+1-2r
+
(1-p) r=1 -p +1-2r
(7) By-Products
(-1)k (1+ n , -1/m)
k = -
n n
Σ
k=0 mk +(m -1) m(mn +m -1)
B() is Beta function
q +1- p (-1)r q
q (p )
x = Σ
(1-p) r=0 r +1-p r
x q-p
q -p (-1)r q -1
(q , -p) = -
p Σr=0 r +1- p r B() is Beta function
E2k
(sec x) p
( )
Σ
2k-p
= x 0< x < collateral
p +1 (2k -p +1) 2
k=
2
E2k
(sech x) p =
( )
Σ p +1 (2k -p +1)
x 2k-p 0< x <
2
collateral
k=
2
- 21 -
22k 22k -1B2k 2k- p-1
(cot x)
(p )
=- Σ x- < x<
k=
p +1
2k (2k - p) 2 2
2
E2k 2k-p
p +1 (2k -p +1)
x-
(csc x)
(p )
= Σ < x<
2 2
k=
2
(2k +1)p
(csch x) (p )
= (-1) 2Σ -p
(2k +1)x x >0
k=0 e
(2k +1)p
(sech x)(p) = (-1)-p2Σ(-1)k ( )x x >0
e 2k +1 k=0
(2k)!
-1 (p ) p
tan x = Σ (-1)
k
x 2k+1- 0< x < 1
k=
p -1
(2 k +2- p )
2
(p ) x -p (2k)! 2k+1- p
cot x
-1
= - Σ (-1)k x 0< x < 1
2 (1-p) k=
p -1
(2k +2- p)
2
2
(p ) (2k -1)!!
sin x Σ
-1
= x 2k+1-p 0< x < 1 collateral
p -1 (2k +2- p )
k=
2
(p ) x -p (2k -1)!!
2
cos x Σ
-1
= - x 2k+1-p 0< x < 1 collateral
2 (1-p) p -1 (2k +2- p )
k=
2
(p )
(2k)! p
tanh x = Σ
-1
x 2k+1- 0< x < 1
p -1 (2k +2- p )
k=
2
2
(p ) k (2k -1)!! p
sinh x Σ
-1
= (-1) x 2k+1- 0< x < 1 collateral
k=
p -1
(2k +2- p)
2
x -p 2
- Σ 2k (2k -p+1) x
(p ) 2 (2k -1)!!
sech
-1
x = log + (1- p)+ 2k-p
(1- p) x k=
p
2
(p ) x 2 (2k -1)!!
csch
-1
x = log + (1- p)+ - Σ (-1)k x 2k-p
(1- p) x k=
p
2k (2k - p+1)
2
et
x x cos t x sin t x 1
Ei(x) = dt , Ci(x)= dt , Si(x)= dt , li(x)= dt
- t t 0 t 0 log t
x x 1 n -1
n-1- r
Ei(x)dx n = x nEi(x)- e xΣr! x
- - n ! r=0
- 22 -
x x 1
(n-1)/2
Ci(x)dx n = Ci(x)x n - sin x Σ (-1)r(2r)! x
n-1-2r
n ! r=0
(n-2)/2
+ cosx Σ (-1)r(2r +1)! x n-2-2r n 2
r=0
x x 1
(n-1)/2
Si(x)dx n = Si(x)x n + cosx Σ (-1)r(2r)! x
n-1-2r
0 0 n! r=0
(n-2)/2
+ sin x Σ (-1)r(2r +1)! x n-2-2r
r=0
n-1-2r
(n-1)/2
rx
- Σ
r=0
(-1)
(2r +1)(n -1-2r)!
n n r n- r
n r Ei(rlog x)
0 0 r=1
x x 1 p
li(x)dx p =
Σ (-1)r p- r
x Ei(r +1)log x x 0
0 0 (1+ p) r=0 r
1 -p
x 0
r
Σ
(p )
li(x) = (-1)r x -p- Ei(r +1)log x
(1- p) r=0 r
x x 1 p
log|log x|dx p = x plog|logx| +Σ(-1)r
p-r
x Ei(rlog x)
0 0 (1+ p) r=1 r
1 -p Ei(rlog x)
(log|log x|)(p) = p log|logx| + Σ(-1)r x 0
x (1-p) r=1 r xr
x dx x 1- k 2x 2
F(x, k) = , E(x, k) = dx
0 1- x 1- k x
2 2 2 0 1- x 2
x dx
(x,c,k) =
1+ cx 1- x 1- k x
0 2 2 2 2
In this chapter, we expand these to a double series or a triple series and calculate the arc length of an ellipse and
a lemniscate using thiese. Next, we calculate these term by term.
2r +1 r - s s
(-1) -1/2 1/2 r
E(x, k) = ΣΣ k x 2s 2r+1
r=0 s=0
r
c
s-t t
(-1) -1/2 -1/2
r s
r-s
(x,c,k) = ΣΣΣ k 2t
x 2r+1
2r +1 r=0 s=0 t=0
- 23 -
(2) Termwise Higher Calculus
(-1)r(2r)!
r-s s k x
x x r -1/2 -1/2
F(x, k)dx =ΣΣ n 2s 2r+n+1
0 0 r=0 s=0 (2r +n +1)!
r
(2r +n +1)! r - s s
x x (-1) (2r)! -1/2 1/2
r
E (x, k)dx = Σ Σ n
k x 2s 2r+n+1
0 0 r=0 s=0
r
ΣΣΣ (2r +n +1)! s - t t k
x x (-1) (2r)! -1/2 -1/2
r s
( x,c,k)dx = c n r-s 2t
x 2r+n+1
0 0 r=0 s=0 t=0
dn r (-1)r(2r)!
r-s k
-1/2 -1/2 2s 2r- n+1
F(x, k) = Σ Σ x
dx n r=
n -1 s=0 (2r - n +1)!
s
2
n r
r-s s k
d
r (-1) (2r)! -1/2 1/2
E(x, k) = Σ Σ 2s 2r-n+1
x
dx n r=
n -1 s=0 (2r - n +1)!
2
n s (-1)r(2r)!
-1/2
k
d r -1/2
r-s
(x,c,k) = Σ Σ Σ c 2t
x 2r-n+1
dx n r=
n -1 s=0 t=0 (2r - n +1)!
s-t t
2
dp r (-1)r(2r +1)
r-s k
-1/2 -1/2
Σ Σ (2r -p +2)
2s 2r-p+1
F x, k
( ) = x
dx p r=
p -1 s=0
s
2
p r (-1)r(2r +1)
r-s s k
d -1/2 1/2
E(x, k) = Σ Σ 2s 2r-p+1
x
dx p r=
p -1 s=0
(2r -p +2)
2
p s (-1)r(2r +1)
-1/2
k
d r -1/2
r-s
(x,c,k) = Σ Σ Σ c 2t
x 2r-p+1
dx p r=
p -1 s=0 t=0
(2r - n +2) s -t t
2
a a r
x x m -1 -n
f g dx n = Σ
<0> (0)
f
<n + r> (r)
g
r=0
n 1
s f a a
-n + r
n -1 m -1 x x
- ΣΣ <n -r+ s>
an-r
(s)
ga dx r
r=0 s=0 n-r
n n- r+1
- 24 -
a a
n -1 r -1 r -1 x x
t s m +n-1-r + tC m-1 f a
m+ n -r+s (m+ s)
+ (-1)m ΣΣΣ C ga dx r
r =1 s =0 t=s n- r n- r
n n- r+1
a a f
m
(-1) n -1 n -1C k x x
<m+ k > (m+ k )
dx n
(n , m) Σ
+ g
k=0 m + k 1
n
when f
<r>
(a) = 0 (r =1, 2, ,m+n -1) or g (s)(a ) = 0 (s =0, 1, ,m+n -2) ,
x x
p q n
(ax +b) (cx +d) dx
b b
- -
a a
x x
(ax +b)p (cx +d)mdx n
b b
- -
a a
x x
n 1 n x
log xdx = log x log x -Σ
2
0 0 k=1 k n!
(r)
(1+ n +r)
-n
n+r 1
+ Σ(-1)r-1 x
n
log x -Σ
r=1 r k =1 k
x n -1 C r (n -r + , -x)x r
x x
n -1
x
e x dx = n
Σ +R
- - (-x) r=0 (n -1)!
- 25 -
x 0
0
R= n -1 n -1 C r (n -r + )x r
2i sin Σ x >0
r=0 (n -1)!
(1+ m) m- r
r
x x m -n
e x x mdx n = e x Σ x
- - r=0 (1+ m - r)
s! x r xr
= e log |x| +Σ Σ
x x n -2 n -2-r n -1
e x log x dx n x
- Ei(x)Σ
- - r=0 s=0 (r + s +1)! r=0 r!
n
n
4 4
x x
e sin xdx = sin e x
sin x - n x
- -
n
e cosxdx = sin 4 e cos x - 4
x x n
x n x
- -
x x (-1)n n -1
e x dx = -x
Σ (-1)rn -1C r (n -r + , x)x r
n
(n -1 ) ! r=0
(1+ m) m- r
n
x x (-1) m -n
e -x x mdx n = Σ ( -1) r
x
e x r=0 r (1+ m - r)
s!(-x)r
x x n -2 n -2-r
e log x dx = (-1)n e -x log|x| + Σ
-x n
Σ
r=0 s=0 (r + s +1)!
n -1 (-x)r
- (-1)n Ei(-x)Σ
r=0 r!
n
e
n
x x
sin xdx n = (-1)n sin
-x
e -xsin x +
4 4
n
e
n
x x
-x
cosxdx = (-1)n sin
n
e -xcos x +
4 4
(3) By-Products
n /2 (-1)k -n-1
(n +1)
1
Σ
k=0 2k +1 n
C 2k =
n +1
sin
4
sin
4
(n +1)
n /2 (-1)k -n-1
1
Σ
k=1 2k
nC 2k-1 =
n +1
sin
4
cos
4
-1
Example
C0
8 C2
8 C4
8 C6
8 C8
8 16
- + - + =
1 3 5 7 9 9
C1
8 C3
8 C5
8 C7
8 5
- + - + =
2 4 6 8 3
- 26 -
time, if there is a number a such that
f (a) = 0 r [0 , m + p] g (a ) = 0 s [0 , m +p -1] ,
<r> s()
or
then the following expression holds.
r f
x x m -1 -p
f g dx p = Σ
<p + r> (r)
g + Rmp
a a r=0
k f
(-1)m 1 p -1 x x
<m+ k > (m+ k )
(p , m) Σ
Rmp = g dx p
k=0 m + k a a
x x
p q s
(ax +b ) (cx +d) dx
b b
- -
a a
(1/a)s+ r (1+p)(1+q) (ax +b)p+s+ r
r
-s
=Σ
r=0 (1/c)r (1+p +s + r)(1+q -r) (cx +d)r-q
x x
(ax +b )p (cx +d)mdx s
b b
- -
a a
(1/a)s+ r (1+p)(1+m) (ax +b)p+s+ r
r
m -s
=Σ
r=0 (1/c)r (1+p +s + r)(1+m -r) (cx +d)r-m
log x - (1+ p+r)- (1+ ) +p
r
x x -p
x log xdx p = Σ x
0 0 r=0 (1+p+ r) (1+ -r)
-p log x - (1+p+ r)- (1+ m)
x log xdx r (1+p+r) (1+ m -r) x
x x m
m+ p
=Σ m p
0 0 r=0
(1+ m) (p + r)
Σ r (1+ m - r) x sin x - 2
x x-p m
x sin x dx = m p m- r
ap a0 r=0
(1+ m) (p + r)
Σ r (1+ m - r) x cosx - 2
x x-p m
x cosx dx = m p m- r
ap a0 r=0
(1+ m) m- r e x -(-1)p+ re -x
a a r
x x m -p
x m sinh x dx p = Σ x
p 0
r=0 ( 1+ m - r ) 2
(1+ m) m- r e x +(-1)p+ re -x
aa x r
x x
m p
m -p
cosh x dx = Σ x
p 0
r=0 ( 1+ m - r ) 2
- 27 -
p
p
x x
-
-
e xsin xdx p = sin 4 e xsin x -
4
p
p
x x
-
-
x
e cosxdx = sin p
4 e xcos x -
4
(-1)p
x x
p -1
-x
e x dx = Σ p
(-1)r (p -r + , x)x r
( p r=0
) r
(1+ m) m- r
p
x x (-1) m -p
e -x x mdx p = Σ (-1) r
x
e x r=0 r (1+ m - r)
p
p
x x
e sin xdx = (-1)p sin
p
-x
e -xsin x +
4 4
p
e
p
x x
-x p p
cosxdx = (-1) sin e -xcos x +
4 4
(3) By-Products
(-1)k -p-1
(p +1)
p
1
Σ
k=0 2k +1 2k
=
p +1
sin
4
sin
4
(p +1)
= p +1 sin 4
(-1)k p
1 -p-1
Σ
k=1 2k 2k -1
cos
4
-1
r f
(n ) n n (n -r)
(x)g r (x)
()
f(x)g (x) =Σ
r=0
- 28 -
(1+ ) -r (n -r)
r
n
{n } n
x sin x =Σ x sin x +
r=0 (1+ -r) 2
(1+ ) -r (n -r)
r
n
{n } n
x cosx =Σ x cos x +
r=0 (1+ -r) 2
(1+m) m-r (n -r)
r
n
{n } m
x sin x =Σ
m
x sin x +
r=0 (1+m -r) 2
(1+m) m-r (n -r)
r
n
{n } m
x cosx =Σ
m
x cos x +
r=0 (1+m -r) 2
(1+ ) x -(n -r) -x
r (1+ -r)
n (n ) e -(-1)
n e - r
x sinh x = Σ x
r=0 2
(1+ ) x -(n -r) -x
r (1+ -r)
n (n ) e +(-1)
n e - r
x cosh x = Σ x
r=0 2
(1+ m) x -(n -r) -x
r (1+m -r)
m
n (n ) e -(-1)
m e m- r
x sinh x = Σ x
r=0 2
(1+m) x -(n -r) -x
r (1+m -r)
m
n (n ) e +(-1)
m e m- r
x cosh x = Σ x
r=0 2
n-1
r (n - r)(r)
(-1)(n ) n n -1
log x =
2
2(n)log x -Σ n
x r=1
(1+ )
r (1+ -r) x
x (n )
n x
n
- r
e x = e Σ
r=0
(1+m)
r (1+m -r) x
x m (n )
n x
m
m- r
e x = e Σ
r=0
n (r)
r x
(n ) n
x x x r-1
e log x = e log x + e Σ(-1) r
r=1
n
-n
x (n )
e sin x = sin e xsin x +
4 4
n -n
= sin e cos x +
4
x (n ) x
e cosx
4
e x - (-1)-re -x
r
(n ) n n
e sinh x = e xΣ
x
r=0 2
x -r -x
r 2
n e + (-1) e
(n ) n
e cosh x = e Σ
x x
r=0
(1+ )
r (1+ -r) x
(n )
n n
-(n -r) - r
e x = e Σ(-1)
-x -x
r=0
(1+ m)
r (1+m -r) x
(n ) n m
e x = e Σ(-1) -(n -r)
-x m -x m- r
r=0
n (r) -n
log x - Σ
x
(-1)(n ) n
e log x =
-x
x r
e r r=1
- 29 -
-n
n
(n )
e sin x
-x
= -sin e -xsin x -
4 4
-n
n
(n )
e cosx
-x
= -sin e -xcos x -
4 4
e x - (-1)-re -x
r
(n ) n n
e sinh x
-x
=e -x
Σ (-1) -n+ r
r=0 2
x -r -x
r 2
(n ) n n e + (-1) e
e cosh x
-x
= e -xΣ(-1)-n+ r
r=0
n
(n )
sin x
2
= -2n-1cos 2x +
2
n
2 (n )
n-1
cos x = 2 cos 2x +
2
n 3n n
(n ) 3
sin x
3
= sin x + - sin 3x +
4 2 4 2
n n n
= cos x +
2 2
(n ) 3 3
cos x
3
+ cos 3x +
4 4
etc.
(3) By-Products
n n
n n
n /2 (n -1)/2
r r
Σ (-1) nC 2r Σ (-1) nC 2r+1 = 2 sin
2 2
= 2 cos ,
r=0 4 r=0 4
n /2 (n -1)/2
n-1
Σ C 2r = 2
n , Σ nC 2r+1 = 2n-1
r=0 r=0
Theorem 19.1.1
r =0, 1, 2, , let f -p+ r be arbitrary
< >
Let B(x,y ) be the beta function and p be a positive number. And for
(r)
primitive function of f (x )and g be the r th order derivative function of g (x ). Then the following expressions hold
r
(p ) m -1 p (p -r)
(x)g r (x) + Rmp
()
f(x)g (x) =Σ f
r=0
(-1)m 1
-p -1 <m+ k > (p )
f (x)g (m+ k )(x)
(-p , m) Σ
Rmp =
k=0 m + k k
Especially, when n =0, 1, 2,
r f
(n ) n n (n -r)
(x)g r (x)
()
f(x)g (x) =Σ ( Leibniz )
r=0
- 30 -
integer. Furthermore, if = -1, -2, -3, , it shall read as follows.
(1+ ) (- + r)
(-1)-r
1+ - r (- )
(1/a)-s+ r (1+p)(1+q) (ax +b)p-s+r
r
(s) s
(ax +b ) (cx +d) =Σ
p q
r=0 (1/c)r (1+p -s + r)(1+q -r) (cx +d)r-q
-s+ r
(1+p)(1+ m) (ax +b)p-s+ r
r (1/c)
p m (s)
s (1/a)
m
(ax +b ) (cx +d) =Σ
r=0
r (1+p -s + r)(1+ m -r) (cx +d)r-m
log x - (1-p + r)- (1+ ) - p
r
(p ) p
x log x =Σ x
r=0 (1- p +r) (1+ -r)
log x - (1-p+ r)- (1+ m) m- p
log x = Σ
m (p ) p m
x x
r r=0 (1- p+r) (1+ m -r)
(1+ m) (p -r)
sin x =Σ
p
x sin x +
(p ) m
m m- r
x
r (1+ m -r)
r=0 2
(1+ m) (p -r)
cosx =Σ
p
x cos x +
(p ) m
m m- r
x
r (1+ m -r)
r=0 2
(1+ m) x -p+ r -x
sinh x = Σ
p
{p } m e -(-1) e m- r
x
m
x
r (1+ m - r)
r=0 2
(1+ m) x -p+ r -x
cosh x = Σ
m {p }p m e +(-1) e m- r
x x
r (1+ m - r)
r=0 2
(p ) log x log x - 1-p -
x -p
log x
2
=
1- p
p log x - (1- p +r)-(r)
- x -p Σ(-1)r
r=1 r 1- p + r
(1+ ) - r
(p ) m -1 p
e x = e Σ
x x
x + Rmp
r=0 r 1+ - r
(-1)m 1 (1+ )
-p -1 x - m-k
(p )
(-p , m) Σ
Rmp = e x
k=0 m + k k 1+ - m -k
(1+m) m- r
x m (p ) x
m p
e x = e Σ x
r=0 r (1+ m - r)
-p x p
(p )
e sin x = sin
x
e sin x +
4 4
-p
p
x (p )
e cosx = sin e xcos x +
4 4
x
p e - (-1) e -x -r
(p )
x x
e sinh x = e Σ
r=0 r 2
p e x + (-1)-re -x
(p )
x x
e cosh x = eΣ
r=0 r 2
(-1)-p m -1 (1+ ) - r
-x (p )
e x =
e x Σ
r=0
( -1 ) r p
r 1+ - r
x + Rmp
- 31 -
x
(p)
(-1)k (1+) - m-k
1 -p -1
Rmp =
(-p , m) Σ
k =0 m+ k k (1+ - m -k ) ex
(-1)-p (1+ m) m- r
r
m (p )
m p
e x
r
Σ(-1)
-x
= x
e x r=0 1+ m - r
-p
p
4
(p )
e sin x
-x
= (-1)-p sin -x
e sin x -
4
-p
p
cos x -
4 4
(p )
e cosx
-x
= (-1)-p sin e
-x
e x - (-1)-re -x
r
(p ) -x -p+ r
p
e sinh x Σ
-x
=e (-1)
r=0 2
x -r -x
r 2
(p ) -x -p+ r
p e + (-1) e
e cosh x
-x
=e Σ (-1)
r=0
p
(p )
sin x
2
= -2p-1cos 2x +
2
p
2 (p )
p-1
cos x = 2 cos 2x +
2
p 3p p
3 (p ) 3
sin x = sin x + - sin 3x +
4 2 4 2
p p
p
3 (p ) 3 3
cos x = cos x + + cos 3x +
4 2 4 2
etc.
(3) By-Products
p p
p p
k
p
k
p
Σ Σ
2 2
(-1) = 2 cos , (-1) = 2 sin
k=0 4
2k k =0 2k +1 4
r = 2
p p p
Σ = 2p-1 , Σ = 2p-1 , Σ p
p >-1
r=0 2r r=0 2r +1 r=0
Theorem 20.1.1
When fk
(r)
denotes the r th order derivative function of fk (x) (k =1, 2, ,) ,
r r r
n r1 r-2 n r1 r-2
f1
n - r1 r1-r2
f -1
r
f1 f2 f =Σ Σ Σ
(n )
f2
r1=0 r2=0 r-1=0 1 2 -1
Theorem 20.1.2
When f
(r)
denotes the r th order derivativefunction of f (x ) and is a natural number,
r r r
n r1 r-2 n r1 r-2
f
n - r1 r1 -r2
f -1
(n ) r
f (x) =Σ Σ Σ f
r1=0 r2=0 r-1=0 1 2 -1
- 32 -
Example
(1+ ) s
r s
n r
n r
x e sin x
x (n ) r
=Σ Σ x -n+ e x sin x +
r =0 s =0 1+ - n + r 2
n-1 (n ) 3(n -1)! n 3log x
n-1 n!
log x = (-1) Σ
3 2
log x + (-1 )
xn x n r =1 (n -r)r
(-1)n-1 n-1 r -1 n!
+ n Σ Σ
r =2 s =1 (n -r)(r -s)s
x
Theorem 20.2.1
be the r th order derivative function of fk (x ) k =1, 2, , ,
(r )
fk r
<>
Let fk be the arbitrary r th order primitive
function of fk (x ) , m, n are natural numbers and B(n , m )be the beta function. If there is a number a such that
r r r
x x m -1 r1 r-2 -n r1 r-2
f1
n + r1 r1 -r2
f -1
r
f1 f2 f dx n =Σ Σ Σ f2
a a r1=0 r2=0 r-1=0 1 2 -1
+ Rmn
k k
(-1)m n-1 m + k1 k2 n -1C k1 k-2
k -2 m + k1 k2
Rmn = Σ Σ Σ Σ
(n , m) k 1=0 k 2=0 k 3=0 k -1=0 m + k1 k2 3 -1
f
x x
m+ k 1 m+k 1 -k 2 k 2-k 3
f -1dx n
k
1 f2 f3
a a
Theorem 20.2.2
(r )
f (x ), fk r
<>
Let f be the r th order derivative function of be the arbitrary r th order primitive function of f (x ),
m, n are natural numbers and B(n , m)be the beta function. At this time , if there is a number a such that
r r r f
x x m -1 r1 r-2 -n r1 r-2 n + r1 r1 -r2
f -1 + Rmn
r
f dx =Σ Σ Σ
n
f
a a r1=0 r2=0 r-1=0 1 2 -1
- 33 -
k k
(-1)m n-1 m + k1 k2 n -1C k1 k-2
k -2 m + k1 k2
Rmn = Σ Σ Σ Σ
(n , m) k 1=0 k 2=0 k 3=0 k -1=0 m + k1 k2 3 -1
f
x x
m+ k 1 m+k 1-k 2 k 2-k 3
f -1dx n
k
f f
a a
Example
r s (1+ + n + r ) x
(1+) s
x x m -1 r -n r
x e x sin x dx n = Σ Σ +R
+n+ r
e x sin x + n
m
- - r =0 s =0 2
(1+)
s
C m+ r
m
m+ r s
x x
(-1) n -1 m + r n -1 r +m+ r x
Rmn = Σ Σ x e sin x + dx n
(n , m) (1+ + m + r )
r =0 s =0 - - 2
log x - (1+n ) - n
log x dx
x x
3 n
= x (log x)
2
0 0 (1+n )
-n log x - (1+n + r ) -
- 2x n log xΣ(-1)r (r )
r=1 r (1+n + r )
r log x - (1+n + r ) -
r s (1+n + r ) (r -s ) (s)
n -n
r-1
+ x ΣΣ(-1) r
r=2 s=1
n
x x x Cr
cos (2m -2r +1) x -
1 m 2m +1
cos
2m+1
x dx n = Σ
(n-1) 1 0 22m r=0 (2m -2r +1) n 2
2 2 2
(-1)m-r2m +1C r n
x x x
n 1 m
Σ
2m+1
sin x dx = 2m n
sin (2m -2r +1) x -
n 2 1 2 r=0 (2m -2r +1) 2
2 2 2
n
x x Cm Cm x x
1 m -1 2m 2m
cos
2m
x dx n = Σ n
cos (2m -2r)x - + dx n
an a1 22m-1 r=0 (2m -2r ) 2 22m an a1
(-1)m-r2mC m n
dx
x x Cm x x
1 m -1 2m
sin
2m
x dx n = Σ n
cos (2m -2r)x - + n
Cm n
cos (2m-2r)x -
m -1 2m
Σ n
=0 k =1, 2, ,n
r=0 (2m -2r ) 2
(-1)m-r2mC m n
= 0
m -1
Σ n
cos (2m -2r)x - k =1, 2, ,n
r=0 (2m -2r) 2
cos x dx = 2 Σ r -2r
x 1 1
sin ( -2r) x
0 r=0
sin x dx = 2 Σ r -2r sin ( -2r)x - 2
x 1 1
r=0
2
x 1 m -1 2m Cr 2m Cm
2m-1 Σ
cos 2m x dx = sin {(2m -2r)x } + x
0 2 r=0 2m -2r 22m
- 34 -
sin
Cr Cm
x 1 m -1 2m 2m
m-r
2m-1 Σ
2m
x dx = (-1) sin(2m -2r)x + x-
2 r=0 2m -2r 22m 2
2
Theorem 21.1.1
p, r are positive numbers, m be a natural number, fk r be the r th order derivative function of
()
Let
fk (x) k =1, 2, , , fk r be arbitrary r th order primitive function of fk (x) and B(p , m ) be
<>
r r r
x x m -1 r1 r -2 -p r1 r-2
f1
p + r1 r1-r2
f -1
r
f1 f2 f dx =Σ Σ Σp
f2 + Rmp
a a r1=0 r2=0 r-1=0 1 2 -1
k k
m k-2
k
(-1) m + k1 k2 k -2
1 p -1 m + k1 k2
Rmp = Σ Σ Σ Σ
(p , m) k 1=0 k 2=0 k 3=0 k -1=0 m + k1 1 k2 3 -1
f
x x
m+ k 1 m+k 1-k 2 k 2-k 3
f -1dx p
k
1 f2 f3
a a
Theorem 21.1.2
(r)
Let p, r are positive numbers, m be a natural number, f be the r th order derivative function of f (x ),
<r>
f be arbitrary r th order primitive function of f (x ) and B(p , m ) be the beta function. At this time,
if there is a number a such that
(a ) = 0 r [0 , m + p] or f (s)(a ) = 0 s [0 , m +p -1]
f
<r>
r r r
x x m -1 r1 r-2 -p r1 r-2
f
p + r1 r1 -r2
f dx p =Σ Σ Σ f -1 + Rmp
r
f
a a r1=0 r2=0 r-1=0 1 2 -1
m + k k k k k
(-1) 1m p -1 m +k k
m + k1 k2 k k -2 1 2 -2
Rmp = Σ Σ Σ Σ
(p , m) k 1=0 k 2=0 k 3=0 k -1=0 1 1 2 3 -1
f
x x
m+ k 1 m+k 1-k 2 k 1-k 2
f -1dx p
k
f f
a a
Example
x r s (1+ + p + r ) x
(1+) s
x x m -1 r -p r
+R
+p+ r
e x sin x dx p = Σ Σ e x sin x + p
m
- - r =0 s =0 2
(1+)
s
m
C
p -1 m+r x x
p (-1) m +r s +m+ r x
R = Σ Σ x e sin x + dx p
m
(p , m) r m+ r (1+ + m + r )
r =0 s =0 - - 2
log x - (1+p ) -
log x dx = (1+p
x x
3 p p 2
x (log x)
0 0 )
- 35 -
log x - (1+p + r ) -
r (1+p + r ) (r )
-p
- 2x p log xΣ(-1)r
r=1
r log x - (1+p + r ) -
r s (1+p + r ) (r -s ) (s)
p -p
r-1
+ x ΣΣ(-1) r
r=2 s=1
p
(-1)m-r2m +1C r
x x 1 m
p
Σ
2m+1
sin x dx = p sin
(2m -2r +1)x -
a(p ) a(0) 22m r=0 (2m -2r +1) 2
a(s) s [0 , p ] are zeros of lineal super primitives of cos
2m+1 2m+1
Where, x or sin x .
Theorem 21.2.1
p, r are positive numbers, m be a natural number, fk r be the r th order derivative function of
()
Let
fk (x) k =1, 2, , , fk r be arbitrary r th order primitive function of fk (x) and B(p , m ) be
<>
r r r
m -1 r1 r-2 p r1 r-2
f1
p - r1 r1-r2
f -1
r
f1 f2 f =Σ Σ Σ
(p ) p
+R f2 m
r1=0 r2=0 r-1=0 1 2 -1
(-p , m) Σ Σ Σ Σ m + k k k k k
(-1) m
1 -p -1 m +k k
m + k1 k2 k k -2 1 2 -2
Rmp =
k 1=0 k 2=0 k 3=0 k -1=0 1 1 2 3 -1
(p )
f1
m+ k 1 m+k 1-k 2 k 2 -k 3 k -1
f2 f3 f
Theorem 21.2.2
(r)
Let p, r are positive numbers, m be a natural number, f be the r th order derivative function of f (x ),
<r>
f be arbitrary
r th order primitive function of f (x ) and B(p , m ) be the beta function. At this time,
if there is a number a such that
r r r
m -1 r1 r-2 p r1 r-2
f
p + r1 r1 -r2
(p )
f -1 + Rmp
r
f = ΣΣ Σ f
r1=0 r2=0 r-1=0 1 2 -1
m + k k k k k
(-1) m
1 -p -1 m +k
m + k1 k2 k k k -2 1 2 -2
Rmp = Σ Σ Σ Σ
(-p , m) k 1=0 k 2=0 k 3=0 k -1=0 1 1 2 3 -1
(p )
f f -1
m+ k 1 m+k 1-k 2 k 2-k 3 k
f f
- 36 -
cos m x , sin m x
Super Derivatives of
p
m (p ) 1 m /2
cos x = m-1 Σ mC r (m -2r) cos (m -2r)x +
p
2 r=0 2
p
m (p ) 1 m /2
sin x = m-1 Σ mC r (m -2r) cos (m -2r) x -
p
+
2 r=0 2 2
(n =1, 2, 3, )
(n ) (n ) (n )
g = g ( f) , fn = f (x)
j1 j2 jn
n! f1 f2 fn
Bn,rf1 , f2 , , fn =Σ
j1! j2! jn! 1! 2! n!
Next, the algorithm that generates Bn,rf1 , f2 , , fn without an omission is shown. And the derivatives up to
the 8 th order of z = g {f (x )} are calculated using this.
Even if the above algorithm is used, it is not so easy to obtain Bn,rf1 , f2 , , fn .
However, when the core function f (x )is the 1st degree, it becomes remarkably easy.
Formula 22.1.4
When g
(n)
=g
(n)
( f ) , fn = f
(n)
(x) (n =1, 2, 3, ), if f (x ) is the 1st degree, the higher derivative function
with respect to x of the composition g {f (x )} is expressed as follows.
(n ) (n ) n
gf(x) = g f1
And, this can be easily enhanced to the super-differentiation in this case. That is, the following expression holds
for p > 0 . This is the grounds for which we have used "Linear form" since "12 Super Deribative" as a fait accompli .
(p ) (p ) p
gf(x) = g f1
Next, trying the higher differentiation of some compositions, we obtain the following formulas.
n
f (x) (n )
e Σ
f (x)
=e Bn,rf1 , f2 , , fn
r=1
(x) (n ) (x) n
e -f = e -f Σ (-1)rBn,rf1 , f2 , , fn
r=1
(n ) n 1 r r
ge
x
= ΣS(n,r)g (r)e rx , S(n,r) = r Σ(-1)s (r -s)
n
r=1 ! s=0 s
- 37 -
(n ) n
ge -x = (-1)nΣS(n,r)g (r)e -rx
r=1
n
(n ) r-1
log f(x) = Σ(-1) (r -1)!Bn,rf1 , f2 , , fnf -r
r=1
(n ) n
(r) 0! 1! n-1 (n -1)!
g (log x) = Σ g Bn,r , - 2 , , (-1)
r=1 x x xn
dn 1 1 n
n (z)
= Σ (-1)k Bn,k 0(z) , 1(z) , , n-1(z)
dz ( z) k=1
Formula 23.1.1
<n > (k )
Let f = f(x), g be the lineal higher primitive function of g (f ) and h , h are the functions of f such that
dx 1 d kh
k = 1, 2, 3,
(k )
h= = (1) , h =
df f df k
Let Sk , Mk , rk are the polynomials such taht
r r
m k -1 -1 rk1 rk1 rk2 rk 2 rk k-1 rk k -1
Sk = Σ r Σ Σ Σ k =1, 2, ,n
rk1=0 k1 r =0k2 k2 r =0k3 k3 r =0
kk rkk
x x R
h
n+ Rn1 Rn1 -Rn2 -Rnn Rnn
gf(x )dx
n
= S1S2Sn g h n n-1
h + Rmn 1
a a
g
f f f
g
f f f
hdf hdf
1+ R11+m2 R11-R22+m2 R22
+ S1M2 h h df
fa fa fa
+S S M g
f f
2+ R21 + m3 R21 -R32+m3 R32-R33 R33
1 2 3 h h h df hdf
fa fa
- 38 -
g
f
n-1+ Rn-1 1+ mn Rn-1 1-Rn2+mn Rn2-Rn3 Rnn
+ S1S2Sn-1Mn h h h df
fa
r g
x m 1-1 -1
1+ r11 r11
gf(x)dx =Σ h + Rm11
a r11=0 11
= (-1) g
f
m1 m1 m1
Rm11 h df
fa
r r
r21
r
x x m 1-1 -1 m 2-1 r21 -1
g
2+ r11 +r21 r11 +r21 -r22 r22
gf(x)dx = Σ ΣΣ
2
h h
a a r11=0 11 r =0 r =0
21 22 21 22
+ Rm21
r
m2
r
m 1-1 -1 m2 f 1+ r11+ m2 r11-r22+ m2 r22
g
m2
Rm21 =Σ (-1) Σ h h df
r11=0 11 r =022 22 fa
r r r r r
r r r
r
x x x -1
m 1-1 -1 -1 m 2-1 r21 21 m 3-1 r31 r32 31 32
gf(x)dx
3
=Σ Σ Σ Σ Σ Σ
a a a r11=0 11 r21=0 r22=0 21 22 r31=0 r32=0 r33=0 31 32 33
g
3+ r11+ r21+ r31 r11+ r21+ r31-r22-r32 r22+ r32-r33 r33
h h h + Rm31
r r
m3
r Σ Σ r r
m 1-1 -1 m 2-1 r21 -1 r21 m3
m3 r32 r32
Rm31 = Σ (-1) ΣΣ
r11=0 11 r21=0 r22=0 21 22 r32=0 r33=0 32 33
g
f
2+ r11+ r21+m3 r11+r21-r22-r32+m3 r22 +r32-r33 r33
h h h df
fa
r g
m2
r
m 1-1 -1 m2 f f
m2 1+ r11+ m2 r11-r22+ m2 r22
+Σ (-1) Σ h h df hdf
r11=0 11 r =0
22 22 fa fa
+ (-1) g
f f f
m1 m1 m1
h df hdf hdf
fa fa fa
Next, trying the higher integration of some compositions, we obtain the following formulas
n+ R
(1+)
xn
n1
x x
1
x - c dx = x - c S S S n
1-
c
c n
1 2 n
1+ + n + Rn1 x
(1/ ) (1/ ) (1/ )
1/ - Rn1+Rn2 1/ - Rn n-1 +Rnn 1/ - Rn n
n n
Σ (1+) + n+ Σ rk
(log x) dx
x x rk n
= x nΣΣΣ(-1)
rk
n k =1
Π k n (log x) k =1
1+ + n +Σrk
0 0 r1=0r2=0 rn=0 k=1
k=1
(1+ ) n x
n-k
+ (-1) k-1
Σ(-1) nC k
n! k=1 k
n
n
n +Σrk
(log x) n
x x Σ rk n r n-1+Σ rk
1 k =1
dx n = -x nΣ Σ Σ(-1) k =1 Πk k n
k =1
log x
0 0 r1=0 r2=0 rn=0 k=1
n +Σr k
k =1
- 39 -
1 n -1
n-1 r
+ (x -1) log|log x| -Σ(-1)n-rn -1C r x log(n -r)
(n -1)! r=0
n
1+n +Σr
0 0 r1=0 r2=0 rn=0 k=1
k
k =1
1 n
n!
log|log x| - (x -1) +Σ(-1) C ( + logr ) x
n r n-r
+ n r
r=1
Formula 23.1.1 is so complicated and is not applicable to any composition. The 1st, the inverse function of the core
function must be known. The 2nd, the higher primitive function of the enclosing function g (f ) must have the property
such as lim g <n>( f )= c . Considering these, there are not many compositions that Formula 23.1.1 is applicable.
n
However, when the core function f (x ) is the 1st degree, it becomes remarkably easy.
Formula 23.1.2
When f (x )= cx +d ,
x x n f f
1
n
gf(x)dx = g( f )df n
a a c fa fa
And, this can be easily enhanced to the super integral in this case. That is, the following expression holds for p > 0.
This is the grounds for which we have used "Linear form" since " 07 Super Integral " as a fait accompli.
1. The series of the higher integral of a function f (x ) results in one integral of e x , sin x, cos x, sinh x, cosh x ,
2. The series of the higher integral of a function f (x ) is expanded into the series of the higher integrals of e x , etc.
f(x)dx
x x
These are introduced and proved in this chapter. In this summary, we assume lim
m a
m
= 0. Also,
a
fa n f (x)x=a
() (n)
we describe the n th order differential coefficient at a with
f(x)e dx
x x x
Σ f(x)dx r = ex -x
r=1 a a a
= e f(x)e dx
x x x
Σ(-1)
r-1
f(x)dx r -x x
r=1 a a a
a f(x)dx 2
f(x)e dx + e f(x)e dx
x
x 1 x x
Σ x x
2r-1 -x -x
= e
r=1 a a a
a f(x)dx 2
f(x)e dx - e f(x)e dx
x
x 1 x x
Σ x x
2r -x -x
= e
r=1 a a a
- 40 -
f(x)sin xdx
x x x x
Σ(-1)
r-1
f(x)dx 2r = sin x f(x)cosxdx - cosx
r=1 a a a a
Example
e dx +-
x x x x x x x x
e xdx e xdx + e xdx x
3 5 7
a a a a a a a a
x dx + x dx x dx +
x x x x x x x x
x dx
2 2 2 4 2 6 2 8
0 0 0 0 0 0 0 0
a f(x)dx e dx
s*r
()
x x (x -a) x x
Σ
r=0 a
r
=Σ
r=0
fa r Σ
s=0 (s + r) !
= Σ fa r
r=0
()
a a
x-a r
2s+1+ r
a sinh(x -a)dx
() (x -a)
x x x x
Σ
r=1 a
f(x)dx 2r-1 = Σ fa r Σ
r=0 s=0 (2s +1+ r) !
= Σfa r
()
r=0 a a
r
sin(x -a)dx
2s+1+
() (x -a) () x
x x x
Σ(-1)r-1
r=1 a a
f(x)dx
2r-1
= Σfa r Σ(-1)s
r=0 s=0 (2s +1+ r) !
= Σ fa r
r=0 a a
r
2s+ r
(-1)r f(x)dx = Σfa Σ(-1)s cos(x -a)dx
x x (x -a) x x
Σ
r=0 a a
2r
r=0
(r)
s=0 (2s + r) !
= Σfa r
r=0
()
a a
r
Example
s+r
x x x x x x (x -1)
log x + log x dx + log x dx 2 + log x dx 3 + = Σ(-1)r-1(r -1)!Σ
1 1 1 1 1 1 r=1 s=0 (s + r) !
sin x dx
x x x x x x
sin x dx 5 sin x dx 61
2 4 6
= sin x + +
0 0 0 0 0 0
- 41 -
3r
2s 2s+ r
(x -1) (r -1)! (x -1)
= Σ +Σ
4 s=0 (2s) ! r=1 2 r/2
sin
4 Σ s=0 (2s + r) !
cosh(x -1)dx
1 x 1 x x
2
= cosh(x -1) + cosh(x -1) dx - +
4 2 1 2 1 1
f(x)e dx
x x x x x
Σr f(x)dx r = ex f(x)e -xdx + e x -x 2
r=1 a a a a a
Example
e
x x x x x x x x x x x
dx 4 dx + = e x e dx + e x
2 3 4 -x -x 2
1 dx +2 dx +3 dx
a a a a a a a a a a a
= e x-a(x -a)
= e log x e dx - e log x e dx
x x x
-x x -x x 2
0 a a
=e
-x
x Ei(x)- + e -x
-1
f(x)dx
x x
In this summary, we introduce the formulas in the case of lim c m m
=0 (c > 0) .
m a a
f(x)e
x x x
Σc r f(x)dx r = ce cx -cx
dx
r=1 a a a
f(x)e
m x x x
Σ(-1)r-1c r
r=1 a a
f(x)dx r = ce -cx
a
cx
dx
f(x)e dx
x x c x x
Σc 2r-1
r=1 a a
f(x)dx 2r-1 =
2
e cx
a
f(x)e -cxdx + e -cx
a
cx
c f(x)dx
2
f(x)e dx - e f(x)e dx
x x x x
c
Σ
r=1
2r
a a
2r
= e cx
a
-cx -cx
a
cx
(-1) r-1c f(x)dx = c sin cx f(x)cos cx dx - c cos cx f(x)sin cx dx
x x x x
Σ
r=1
2r
a a
2r
a a
Example
e dx
x x x x x x x x
c
1
e x dx + c
2
e x dx + c
2 3
e x dx c
3 4
x 4
+
a a a a a a a a
ee ce cx e
(1-c)x (1-c)a
cx
x
x -cx -e
= ce dx =
a 1-c
- 42 -
e dx
x x x x x x x x
c
1
e xdx c
3
e xdx + c
3 5
e xdx c
5 7
x 7
+-
a a a a a a a a
e sin cx dx
x x
= c cos cx e xcos cx dx + c sin cx x
a a
x a
ce ce
= 2
- 2
[cos{c(x -a )}- csin{c(x -a )}]
1+ c 1+ c
x dx +
x x x x x x x x
x dx c x dx c
2 2 2 4 2 4 6 2 6 8 2 8
c x dx + c
0 0 0 0 0 0 0 0
Example
dx +
x x x x x x x x
dx 4c
1 2 2 3 3 4 4
1c dx + 2c dx + 3c
a a a a a a a a
e dx + ce e
x x x
= ce cx cx
dx = ce c(x-a (x - a)
-cx -cx 2 )
a a a
s+r
x x c s(x -a)
Σc r
r ()
f(x)dx r = Σ f (a)Σ
r=0 a a r=0 s=0 (s + r)!
f(x)dx
2s+1
x x c (x -a)
Σ
r=1
c 2r-1
a a
2r-1
= Σ f (a)Σ
r=0
(r)
s=0 (2s +1+ r) !
2s+1+ r
f(x)dx
2s+1
x x c (x -a)
Σ
r=1
(-1) r-1 2r-1
c
a a
2r-1
= Σ f (a)Σ(-1)
r=0
(r)
s=0
s
(2s +1+ r) !
Example
log x dx
x x x x x x
1 2 2 3 3
log x + c log x dx + c log x dx + c +
1 1 1 1 1 1
s+r
r-1 c s(x -1)
= Σ(-1) (r -1)!Σ
r=1 s=0 (s + r)!
- 43 -
secx dx c secx dx secx dx secx dx
x x x x x x x x
-
1 3 3 5 5 7 7
c +c -c
0 0 0 0 0 0 0 0
2s+1 2s+1+ 2r
c x
= ΣE2rΣ(-1)s E2r : Euler Number
r=0 s=0 (2s +1+ 2r) !
tan
x x x x x x
tan x dx c tan x dx c x dx +
-1 2 -1 2 4 -1 4 6 -1 6
tan x + c
1 1 1 1 1 1
3r
2s 2s 2s 2s+ r
c (x -1) (r -1)! c (x -1)
=
4Σ
s=0 (2s) !
+Σ
r=1 2r/2
sin 4 Σ
s=0 (2s + r) !
n (log k)r (log n)r+1
r = nlim
Σ -
k=1 k r +1
Then, the following higher calculus and super calculus are obtain. These are all lineal.
1 (z -1)
+ Σ(-1)r r + -Σ s (log 2) r-s
2 r=0 r +1 s=0 s (r+n )!
log(z -1) -(p) - 0
<p>(z) = (z -1)
p-1
2(p )
log r+12 r-1 r r+p
1 (z -1)
+ Σ(-1)r r + -Σ s (log 2) r-s
2 r=0 r +1 s=0 s (1+ r + p )
-n
(-1) n !
(n)(z) =
2(z -1) n+1
log r+12 r-1 r r-n
1 (z -1)
+ Σ(-1) r +
r
-Σ s (log 2) r-s
2 r=0 r +1 s=0 s (1+ r - n)
log(z -1) -(-p) - 0
p (z) =
( )
(z -1)-p-1
2(-p)
log r+12 r-1 r (z -1)r-p
(1+ r - p)
1
+ Σ(-1) r +
r
-Σ s (log 2) r-s
2 r=0 r +1 s=0 s
- 44 -
26.3 Higher and Super Calculus of Dirichlet Eta Function
zn
-n
log 2r -1
<n >
(z) = + (-1)nΣ(-1)r-1 Re(z) > 0
n! r=2 2r -1
z
-n
zn k (-1)r-1 k
log (2r -1)
(z) =
<n>
+ (-1) ΣΣ
-n
n! k=2 r=2 2k+1 r (2r -1)
z
zp k (-1)r-1 k -p
log (2r -1)
<p>(z) = + e p iΣΣ
(1+p) k=2 r=2 2k+1 r (2r -1)
z
- 45 -
Formula 26.4.2h ( Higher Derivative )
-n n
z r-1 log (2r -1)
(n)(z) = + (-1) Σ(-1)
-n
Re(z) > 0
(1-n) r=2 (2r -1)
z
r-1
z -p k log p (2r -1)
-p i
k (-1)
(z) =
(p )
(1-p)
+e Σ Σ k+1 r (2 -1)z
k =2 r=2 2 r
2011.04.16
2012.10.28 Renewal
2016.01.05 Updated
2018.04.04 Added Chap.25
2018.06.11 Added Chap.26
Kano Kono
Alien's Mathematics
- 46 -
1 Gamma Function & Digamma Function
Calculation.
Deform (z -1)! as follws
Since z is already good in anything besides 0, -1, -2, , replacing function z -1! with z ,
we obtain the desired expression
z -1
1 1 z
(z)= Π 1+ 1+ (2.1)
z n =1 n n
Calculation.
Further deform (1.1) as follows
z
n !n
lim
n z(z +1)(z +2)(z + n )
1 123 n 2 3 n
= lim
n z (z +1)(z +2)(z + n) 1 2 n -1
z
1 123 n 2 3 n +1 n +1
= lim 1
n z (z +1)(z +2)(z + n) 1 2 n n
-1-
z
1 1 1
1+ 1+ 1+
1 1 2 n
= lim
n z z z z
1+ 1+ 1+
1 2 n
z -1
1 1 z
= Π
z n =1
1+
n
1+
n
e
(z) = -t z-1
t dt Re(z)>0 (3.1)
0
Proof.
n
t
n t
n(z) = 1- z-1
dt (3.2)
0 n
Here
n
t
lim 1- = e -t
n n
so
e
n
t
n t z-1 -t z-1
lim 1- dt = t dt (3.3)
n 0 n 0
(1-s)
n
n t 1 1
1- t z-1 dt = (1-s)n n z-1 s z-1 n ds = n z n
s z-1 ds
0 n 0 0
1
n zn
(1-s)
sz 1
n-1
=n z
(1-s)n + s z ds
z 0 z 0
1
n zn n zn (n -1)
(1-s)
s z+1 1
= (1-s)n-1 + n-2
s z+1 ds
z z +1 0 z(z +1) 0
n zn(n -1)321 n z n !
s
1
z+n-1
= ds =
z(z +1)(z +2)(z +n -1) 0 z(z +1)(z +2)(z +n)
Hence
n
n t n! nz
lim 1- t z-1 dt = lim
n 0 n n z(z +1)(z +2)(z +n)
Thus, from (1.1),(3.3),(3.4) we obtain
-2-
e
(z) = -t z-1
t dt Re(z)>0 (3.1)
0
e
1 z -
= e z z Π 1+ n
(4.1)
(z) n =1 n
Calculation.
We employ a reciprocal of Gauss expression (1.1). Then
n
1 1 1
1+ + ++ z z z
2 3
e
e z z - z -
= z 1+ e -z 1+ e 2 1+ n
elog n z 1 2 n
e
z z - z -
=e 1+ e -z
1+ e 2 1+ n
1 2 n
z
1 1 1 z z
e
1+ + ++ - log n z z z - z -
lim e 2 3 n
1+ e -z 1+ e 2 1+ n
n 1 2 n
z
z z - 1 1 1
=e zΠ 1+ e n
where = lim 1+
n
+ ++ - log n
n =1 n 2 3 n
z
e
1 z -
= e z z Π 1+ n
(z) n =1 n
(z + n)
= z(z +1)(z +2)(z + n -1) n is a natural number (5.4)
(z)
(z)
= (z -1)(z -2)(z - n) n is a natural number (5.4')
(z - n)
(-z) (1+z +n)
= (-1)-n n is a nonnegative integer (5.5)
(-z -n) (1+ z)
(z +n)
lim =1 (5.6)
n (n )n z
-3-
Proof.
From Gauss expression (1.1)
Next,
1 1 n +1
k =1 e
n 1 - 1 1 1 - -
Π 1+ k
= 1+ 1+ 1+ e -1e 2 e n
k 1 2 n
n +1 -1+ 2 + + n (n +1) Log n -1+ 2 + + n
1 1 1 1
2 3
= e = e e
1 2 n n
(n +1) -1+ 2 + + n - Log n
1 1
= e
n
1 1
n =1
1 - n 1 -
Π 1+ e n
= lim Π 1+ e k = e -
n n k =1 k
Substitute this for Weierstrass expression (4.1) as follows.
1
e
1 1 -
= e 1 1 Π 1+ n
= e e - = 1
(1) n =1 n
Hence (1)=1 , and from (5.1) (2)=1(1)=1 i.e. (5.2) follows.
Substitute positive integer n for (5.1) one by one as follows.
(-z)
= (-z -1)(-z -2)(-z -n )
(-z -n)
(1+z +n)
= (-1)-n(z +1)(z +2)(z +n) = (-1)-n
(1+z)
This equation holds also for n = 0. Moreover, this equation holds obviously for the positive integer z .
Then,this quation includs a part of Singular Point Formulas ( Later 1.3 ).
From (5.4)
-4-
1
= z(z +1)(z +2) (z +n -1)(z +n) ( z )
(z +n)
(z +n) z(z +1)(z +2) (z +n -1)(z +n) ( z )
lim = lim
n (n -1)! n z n (n -1)! n z(z +n)
z(z +1)(z +2) (z +n -1)(z +n) n
= lim (z)
n n !n z z +n
1
= 1(z) = 1
(z)
By substituting n -1! = n for this, (5.6) follows.
n + =
1 147(3n -2) 1
3 3 n
3
2 258(3n -1) 2
n+ =
3 3n 3
1 159(4n -3) 1
n+ =
4 4n 4
3 3711(4n -1) 3
n+ = n
4 4 4
(z)(-z) = - , (z)(1-z) =
z sin z sin z
z
1 1
+z -z = , (1+z)(1-z) =
2 2 cos z sin z
22z-1
1
(2z) = (z) z+
2
33z-1/2
1 2
(3z) = (z) z+ z+
2 3 3
Proof.: Omitted.
-5-
'(z)
n+1
dz = log(n !) (7.2)
1 (z)
Calculation.
Just do logarithm integral calculus obediently, as follows:
'(z)
x+1
dz = [log(z)]x+1 = log (x +1) - log(x)
x (z) x
'(z)
n+1
dz = [log(z)]n+1 = log (n +1) - log(1)
1 (z) 1
= log(n !) - log 1
= log(n !)
Calculation.
'
'(z) "(z)(z) - '(z)2 "(z) '(z)
2
(z)
=
(z)2
=
(z)
-
(z)
From this
' 2
"(z) '(z) '(z)
(z)
= (z) + (z)
On the other hand, from next section (2.8 Trigamma Function)
'(z) '
d 1
= (z) = Σ
(z) dz n =0 (z + n )2
(1) The 1
(0) =
(1) =1 , (2) =1
-6-
2 2
1 -1!! 3 1!! 1
= = , = =
2 0 2 1 2
= =
5 3!! 3 7 5!! 15
= , =
2 2 4 2 2 2 8 3
=
2
9 7!! 105 11 9!! 945
= , = =
2 2 16 4
2 325
= 2.678938 = 1.354118
1 2
,
3 3
= 3.625600 = 1.225417
1 3
,
4 4
(2) The 2
22
1 21 3 4
- = - = -2 , - = =
2 1!! 2 3!! 3
24
5 23 8 7 16
- =- =- , - = =
2 5!! 15 2 7!! 105
26
9 25 32 11
- =- =- , - =
2 9!! 945 2 11!!
-7-
1.2 Digamma Function
e -t e -zt
(z)= - dt (integral expression) (1.0')
0 t 1-e -t
In addition, the derivative more than 2nd order are callecd Trigamma ,Tetragamma, Pentagamma, etc.,
and generaly, n th order derivative is called Polygamma Function.
These are dnoted like z , z , ,
(0) (1) (2) (n)
z , z including the digamma, and named
generally with Psi Function.
1 1 1 n 1
(z) = - - +Σ - = lim log n -Σ (2.1)
z n =1 n n + z n k =0 k + z
1 1 1
(z +1) = (z) + = - + Σ - (2.3)
z n =1 n n+ z
n -1 1 n 1
(z +n) = (z) + Σ , (z -n) = (z) - Σ (2.4)
k =0 z + k k =1 z - k
n 1
(1+n) = - + Σ (2.5)
k =1 k
1
= - - 2 log 2 (2.6)
2
1 n -1 1
n = - - 2 log 2 + 2Σ (2.7)
2 k =0 2k+ 1
d 1
( 1 ) (z) = (z) = Σ (Trigamma Function) (2.8)
dz n =0 (z + n )2
where = 0.577215664901532860606512090082402431042
Proof
Let invers Weierstrass expression in the previous section (1.1.4 ) as follows.
z
n
- z
(z) = e z Π
-1
e n
n =1 n +z
And differentiate the logarithm of this with respect to z, then
-8-
'(z) e -rz z -2 e z/n / n n /(n + z)
2
= - - z - -1 + Σ z/n - Σ
(z) e z n =1 e n =1 n /(n + z)
1 1 1
= - - +Σ -
z n =1 n n+ z
Hence the first half of (2.1) follows.
Next,
1 1 1 1 1
- - +Σ - = - + Σ -Σ
z n =1 n n+ z n =1 n n =0 n + z
n 1 1 1
= - lim Σ - log n + Σ -Σ
n k =1 k n =1 n n =0 n + z
n 1
= lim log n - Σ
n k =0 k + z
1 1 1 1 1 1 1
(1) = - - + - + - + - + = -
1 1 2 2 3 3 4
1 1 1 1 1 1 1 1 1
(2) = - - + - + - + - + - +
2 1 3 2 4 3 5 4 6
= 1 -
1 1 1 1 1 1
= 1 - - + + + + + + +
2 3 4 2 3 4
Thus (2.2) follows.
Differentiate both side of (z +1) = z z(the previous (5.1)) with respect to z , then
'(z +1) = z ' z+ z
Divide both side of this by (z +1) = z z , then
1 1
(z +1) - (z) = , (z) - (z -1) =
z z -1
1 1
(z +2) - (z +1) = , (z -1) - (z -2) =
z +1 z -2
1 1
(z +n) - (z +n -1) = , (z -n +1) - (z -n) =
z +n -1 z -n
Add these on each other, then
-9-
n -1 1 n 1
(z +n) - (z) = Σ , (z) - (z -n) = Σ
k =0 z + k k =1 z - k
1 1 2 1 1
= - + Σ -Σ = - + 2Σ - 2Σ
2 k =1 k k =0 2k +1 k =1 2k k =0 2k +1
(-1)k-1
= - - 2Σ = - - 2 log 2
k =1 k
Hence (2.6) follows.
Substitute z = 1/2 n for (2.1) ; then
1 1 1
+ n = - + Σ - Σ
2 k =1 k k =0 1
k+ +n
2
1 1 n -1 1 n -1 1
= - + 2Σ - 2Σ - 2Σ + 2Σ
k =1 2k k =0 2k +1+2n k =0 2k +1 k =0 2k +1
1 1 n -1 1
= - + 2Σ - 2Σ + 2Σ
k =1 2k k =0 2k +1 k =0 2k +1
n 1
= - - 2 log 2 + 2Σ
k =1 2k -1
1 1 1 1 1
- n = - +Σ -Σ = - +2Σ -2Σ
2 k =1 k k =0 1 k =1 2k k =0 2k +1-2n
k+ -n
2
1 n -1 1 1
= - + 2Σ - 2Σ - 2Σ
k =1 2k k =0 2k +1-2n k =n 2k +1-2n
1 n 1 1
= - + 2Σ + 2Σ - 2Σ
k =1 2k k =1 2k -1 k =1 2k +1
n 1
= - - 2 log 2 + 2Σ
k =1 2k -1
d d 1 1 1
(z) = - - + Σ -
dz dz z n =1 n n + z
1 1 1
= 2 +Σ = Σ
z n =1 (n +z)2 n =0 (n + z)2
- 10 -
1.2.3 Spcial values of the Digamma Function
From properties of the digamma function (1.2.2 ), the following special values are obtained.
(1) The 1
(0)= -
(1)= - , (2) = - + 1
3 11
(3)= - + , (4) = - +
2 6
5 29
(5)= - + , (6) = - +
4 20
97 739
(7)= - + , (8) = - +
60 420
(2) The 2
1
= - - 2 log 2
2
3 1
= - = - - 2 log 2 + 2
2 2
5 3 4
= - = - - 2 log 2 + 2
2 2 3
7 5 23
= - = - - 2 log 2 + 2
2 2 15
9 7 176
= - = - - 2 log 2 + 2
2 2 105
11 9 563
= - = - - 2 log 2 + 2
2 2 315
13 11 6508
= - = - - 2 log 2 + 2
2 2 3465
- 11 -
1.3 Singular Point Formulas
Formula 1.3.1
When (z), (z), m(z) denote gamma function, digamma function and polygamma function
respectively, following expressions hold for n =0, 1, 2, 3, .
(0)
= (-1)n(1+n) = (-1)n n! (1.1)
(-n)
(-n) m!
= (-1)m-n m =0, 1, 2, 3, (1.2)
(-m) n!
(-n)
= (-1)n+1n! (1.3)
(-n)
(-n)
=1 (1.4)
-(n +1)
m(-n)
dm
=1 m(z)= m (z) (1.5)
m-(n +1) dz
Proof
From Gauss expression ( 1.1.1 ), the following expression holds.
When z = -1, -2 , this expression becomes the indeterminate form, and the value is not decided.
Now assume z -1 , -2 , , then z +1 , z +2 , , z + n become nonzero all.
(z +1) nz z
= lim = lim =z (1.0)
(z) n z +n +1 n z 1
+1+
n n
Here substitute z =0 for this; then
- 12 -
Replacing k with n gives (1.1). And using this
1 1 1 1
log n - + + ++
(z) z z +1 z +2 z+n
= lim
(z) n n! n z
z(z +1)(z +2)(z +n)
Assum z 0 and multiply z + k k =0, 1, 2, by numerator and denominator; then
z+k z+k z+k z+k z+k
(z + k)log n - + ++ +1+ ++
(z) z z +1 z + k -1 z + k +1 z+n
= lim
(z) n n! n z
0 0 0 0 0
0log n - + ++ +1+ ++
(-k) -k - k +1 -1 1 n
= lim
(-k) n n! n -k
(-k)(-2)(-1)112(n -k)
(-k)(-2)(-1)112(n -k)
= lim -
n
n! n -k
Calculate this in detail ;
When z -k k =0
(-k) n 0n!
= lim - =1
(-k) n n!
When z -k k =1, 2, 3,
(-k) k+1 nk
= lim (-1) k !
(-k) n (n -k +1)(n -k +2)(n -k +k)
1
= lim (-1)k+1k !
n k 1 k 2 k k
1- + 1- + 1- +
n n n n n n
- 13 -
= (-1)k+1k !
Hence Replacing k with n gives (1.3).
Next using (1.2),(1.3)
(-n) (-n) (-n) -(n +1)
=
-(n +1) (-n) -(n +1) -(n +1)
1
= (-1)n+1n!-(n +1) n+2
=1
(-1) (n +1)!
Thus we obtain (1.4).
Finally,
z
1 1 1
m(z) = (-1)m+1m! m+1
+ + +
(z +1)m+1 (z +2)m+1
From this
1 1 1 1
+ + + +
m-(z +1) ( z )
- -1 m+1
( z)
- m+1
( z )
- +1 m+1
( z )
- +2 m+1
=
m(-z) 1 1 1
+ + +
(-z)m+1 (-z +1)m+1 (-z +2)m+1
Assuming z 0 , denominator is also non-zero and positive. Then
1
m-(z +1) (-z -1)m+1
=1+
m(-z) 1 1 1
+ + + +
(-z)m+1 (-z +1)m+1 (-z + n)m+1
1 1
Let z n then
m+1
. Therefor the summation of terms after
m+1
-z +n -z +n +1
converges to a finite value. m >0 . Consequently, second term in the right hand side converges to 0.
Thus
m-(n +1) m-(z +1)
= lim =1
m(-n) zn m(-z)
i.e. (1.5) was poroved.
Examples
The results actually calculated with computational software are as follows.
- 14 -
Significance of these formulas
z = 0, -1, -2 are singular points (pole of order 1). The value of these functions is . Nevertheless,
the arbitrary ratio of z,z on these points is removable singular point. esides, all those ratios reduce
to the integer or the reciprocal number of the integer. formulas (1.1) (1.4) mentioned above insists on this.
For exsample
(-3) (-3) (0) -7!
= = = 840
(-7) (0) (-7) -3!
(-3) (-3) (-7) -7! 1 1
= = =
(-7) (-7) (-7) -3! (-1)87! 6
(-8) (-8) (-8) -5!
= = (-1)98! = 120
(-5) (-8) (-5) 8!
etc.
These are phenomena which is peculiar to the gamma function and the digamma function and do not occur
in the polygamma functon more than the trigamma function.
For example, if we adopt the ratio 1z/0z of the trigamma functkion and the digamma function,
z =0, -1, -3, are all poles of order 1 of this function (ratio) , and generaly, if we adopt the ratio
mz /nz m >n , z =0, -1, -2, -3, become poles of order m -n . Functions more than
the trigamma are right different from Functions less than the digamma in a dimension.
What I can say in polygamma functions more than the trigamma is only that the ratio of polygamma functions
2003.12.19
K. Kono
Alien's Mathematics
- 15 -
2 Multifactorial
Definition 2.1.1
When m, n denote a natural number,
m !! n (n -2)(n -4) 531 m =2n -1 (11)
Example.
2!! = 2, 4!! = 24, 6!! = 246,
1!! = 1, 3!! = 13, 5!! = 135,
0!! = (-1)!! = 1
Formula 2.1.2
When n, and (z) denote a natural number, a circular constant and a gamma function respectively,
the following expressions hold.
n! = n!!(n -1)!! (2)
/
1 1 1
(2n -1)!! = 2n n + / = 2n n + (31)
2 2 2
= 2 (n +1) = 2 n !
2 2
(2n)!! = 2n n + / n n
(30)
2 2
(-2n)!! =, 0!! = 1 (4)
2
-(2n +1)!! = (-1)-n , (-1)!! = 1 (5)
(2n -1)!!
Proof
First, calculate (11) (10), as follows.
(2n)!!(2n -1)!! = (2n -1)(2n -3)(2n -5) 531
2n (2n -2)(2n -4) 642 = (2n)! (2)
Replacing 2n with n, we obtain (2).
1 1 1
1+ =
2 2 2
=
1 3 3 3 3 1 1
2+ = 1+ =
2 2 2 2 2 2 2
=
1 5 5 5 5 3 1 1
3+ = 1+ =
2 2 2 2 2 2 2 2
-1-
135(2n -1)
1 1 (2n -1)!! 1
n+ n
= = n
2 2 2 2 2
1 1 1
(2n -1)!! = 2 n + / = 2n n + /
n
(31)
2 2 2
In a similar way, we obtain (30).
Substitute n =0 for (30) ; then
0!! = 20(1) = 1
Replace n with -n in (30) ; then
2 (2n)!
= (-1)-n = (2n -1)!! (5)
(2n -1)!! 2nn !
Example 1
1 1 15
13!! = (27-1)!! = 27 7+ / = 27 /
2 2 2
= 1281871.25430543/1.77245385 = 135135.0000
Example 2
2 2 2
(-3)!! = (-1)1 = -2 , (-5)!! = (-1)2 =
1!! 3!! 3
Formula 2.1.3
f (0) 0 f '(0) 1 f "(0) 2 f (3)(0) 3
x
f = x + x + x + x + (6)
2 0!! 2!! 4!! 6!!
-2-
Calculation
f '(0) 1 f " (0) 2 f (3)(0) 3
f(x) = f(0)+ x + x + x +
1! 2! 3!
' " (3)
f (0) 0 f (0) 1 1 f (0) 2 2 f (0) 3 3
= 0 2x + 1 2x + 2 2x + 3 2x +
2 0! 2 1! 2 2! 2 3!
f (0) f '(0) f "(0) f (3)(0)
= (2x)0+ (2x)1+ (2x)2+ (2x)3+
0!! 2!! 4!! 6!!
Replacing x with x/2 , we obtain (6).
x 0! 1! 2 2! 3 3! 4
log 1 = x- x x - x
2 2!! 4!! 6!! 8!!
2+ x 0! 2! 3 4! 5 6! 7
log = x+ x + x + x +
2- x 2!! 6!! 10!! 14!!
1 2+ xi 0! 2! 3 4! 5 6! 7
log = x- x + x - x +
2i 2- xi 2!! 6!! 10!! 14!!
-3-
x 1 2+ xi
tan -1 = log
2 2i 2- xi
x (-1)!! 1 3 1!! 1 5 3!! 1 7
1 x 2dx = x x - x x -
0 2!! 3 4!! 5 6!! 7
x dx 1!! 1 3 3!! 1 5 5!! 1 7
=x x + x x +
0 1 x 2 2!! 3 4!! 5 6!! 7
-4-
1 0! 2! 4! 6!
tan -1 = - + - +
2 2!! 6!! 10!! 14!!
1 0! 2! 4! 6!
tanh -1 = + + + +
2 2!! 6!! 10!! 14!!
(-1)!! 1!! 3!! 5!!
2 =1+ - + - +
2!! 4!! 6!! 8!!
1 1!! 3!! 5!! 7!!
=1- + - + -
2 2!! 4!! 6!! 8!!
(-1)!! 1!! 3!! 5!!
1 = + + + +
2!! 4!! 6!! 8!!
-5-
2.2 Triple factorial
Definition 2.2.1
When m, n denote a natural number,
m !!! m (m -3)(m -6) 741 m =3n -2 (12)
Example.
1!!! = 1, 4!!! = 14, 7!!! = 147,
2!!! = 2, 5!!! = 25, 8!!! = 258,
3!!! = 3, 6!!! = 36, 9!!! = 369,
0!!! = (-1)!!! = (-2)!!! =1
Formula 2.2.2
When n and (z) denote a natural number and a gamma function respectively,
n! = n!!!(n -1)!!!(n -2)!!! (2)
1 1
(3n -2)!!! = 3n n + / (32)
3 3
2 2
(3n -1)!!! = 3n n + / (31)
3 3
3 3
(3n)!!! = 3n n + / = 3n(n +1) = 3nn ! (30)
3 3
(-3n)!!! = , 0!!! = 1 (4)
(-1)!!!(-2)!!! = 1 (5)
3
-(3n +1)!!!-(3n +2)!!! = (6)
(3n -1)!!!(3n -2)!!!
Proof
First, calculate (12) (11) (10), as follows.
(3n -2)!!!(3n -1)!!!(3n)!!!
= (3n -2)(3n -5)(3n -8) 741
(3n -1)(3n -4)(3n -7) 852
3n (3n -3)(3n -6) 963 = (3n)!
Replacing 2n with n, we obtain (2).
-6-
Next, since (1+ z) = z z ,
1 1 1
1+ =
3 3 3
1 4 4 4 4 1 1
2+ = 1+ = =
3 3 3 3 3 3 3
1 7 7 7 7 4 1 1
3+ = 1+ = =
3 3 3 3 3 3 3 3
1 147(3n -2) 1 (3n -2)!!! 1
n+ = n
= n
3 3 3 3 3
1 1
(3n -2)!!! = 3n n + / (32)
3 3
In a similar way, we obtain (31),(30).
Substitute n =0 for (30) ; then
0!!! = 3 (1) = 1
0
-7-
Example 1
2 2 20 2
17!!! = (36-1)!!! = 36 6+ / = 36 /
3 3 3 3
= 729389.03492617/1.35411794 = 209439.9998
Example 2
3 3
(-4)!!!(-5)!!! = =
1!!!2!!! 2
Formula 2.2.3
f (0) 0 f '(0) 1 f " (0) 2 f 3 (0) 3
()
x
f = x + x + x + x + (7)
3 0!!! 3!!! 6!!! 9!!!
Calculation
f '(0) 1 f " (0) 2 f (3)(0) 3
f(x) = f(0)+ x + x + x +
1! 2! 3!
' " 3 ()
f (0) 0 f (0) 1 1 f (0) 2 2 f (0) 3 3
= 0 3x + 1 3 x + 2 3 x + 3 3 x +
3 0! 3 1! 3 2! 3 3!
' " (3)
f (0) 0 f (0) 1 f (0) 2 f (0)
= (3x) + (3x) + (3x) + (3x)3+
0!!! 3!!! 6!!! 9!!!
Replacing x with x/3 , we obtain (7).
-8-
(4) Expansions by the triple factorial of the logarithmic function
x 0! 1! 2 2! 3 3! 4
log 1 = x- x x - x
3 3!!! 6!!! 9!!! 12!!!
3+ x 0! 2! 3 4! 5 6! 7
log = x+ x + x + x +
3- x 3!!! 9!!! 15!!! 21!!!
1 3+ xi 0! 2! 3 4! 5 6! 7
log = x- x + x - x +
2i 3- xi 3!!! 9!!! 15!!! 21!!!
x3 (-1)!!! 1 4 2!!! 1 7 5!!! 1 10
1 x 3dx = x x - x x -
0 3!!! 4 6!!! 7 9!!! 10
x dx 1!!! 1 4 4!!! 1 7 7!!! 1 10
3
=x x + x x +
0 1 x 3 3!!! 4 6!!! 7 9!!! 10
1
1 1 1 1 1
e 3
= + + + + +
0!!! 3!!! 6!!! 9!!! 12!!!
1
- 1 1 1 1 1
e 3
= - + - + -
0!!! 3!!! 6!!! 9!!! 12!!!
1 1 1 1 1 1
cos = - + - + -
3 0!!! 6!!! 12!!! 18!!! 24!!!
1 1 1 1 1 1
sin = - + - + -
3 3!!! 9!!! 15!!! 21!!! 27!!!
-9-
1 1 1 1 1 1
cosh = + + + + +
3 0!!! 6!!! 12!!! 18!!! 24!!!
1 1 1 1 1 1
sinh = + + + + +
3 3!!! 9!!! 15!!! 21!!! 27!!!
4 0! 1! 2! 3!
log = - + - +
3 3!!! 6!!! 9!!! 12!!!
2 0! 1! 2! 3!
-log = + + + +
3 3!!! 6!!! 9!!! 12!!!
0! 2! 4! 6!
log 2 = + + + +
3!!! 9!!! 15!!! 21!!!
1 0! 2! 4! 6!
tan -1 = - + - +
3 3!!! 9!!! 15!!! 21!!!
1 0! 2! 4! 6!
tanh -1 = + + + +
3 3!!! 9!!! 15!!! 21!!!
(-1)!!! 2!!! 5!!! 8!!!
3
2 = 1+ - + - +
3!!! 6!!! 9!!! 12!!!
1 1!!! 4!!! 7!!! 10!!!
3
= 1- + - + -
2 3!!! 6!!! 9!!! 12!!!
(-1)!!! 2!!! 5!!! 8!!!
1 = + + + +
3!!! 6!!! 9!!! 12!!!
- 10 -
2.3 Multi factorial
Definition 2.3.1
When m, n denote natural number and !!! (k pieces) !k ,
m!k n (n -k)(n -2k) (1+ k)1 m =kn -(k -1) (1k-1)
n (n -k)(n -2k) (k -1)+k(k -1) m =kn -1 (11)
Example
1!4 = 1, 5!4 = 15, 9!4 = 15 9, 13!4 = 15 913,
2!4 = 2, 6!4 = 26, 10!4 = 2610, 14!4 = 261014,
3!4 = 3, 7!4 = 37, 11!4 = 3711, 15!4 = 371115,
4!4 = 4, 8!4 = 48, 12!4 = 4812, 16!4 = 4812, 16,
0!4 = (-1)!4 = (-2)!4 = (-3)!4 = 1
Formula 2.3.2
When k , n and (z) denote a natural number and a gamma function respectively,
n! = n!k (n-1)!k (n-2)!k n-(k -1)!k (2)
k -s k -s
(kn -s)!k = k n n + / , s =1, 2, k -1 (3s )
k k
k k
(kn)!k = k n n + / = k n(n +1) = k nn! (30)
k k
(-kn)!k = , 0!k = 1 (4)
Proof
First, calculate (1k-1) (11) (10), as follows.
- 11 -
(kn -1) kn -(k +1) kn -(2k +1) k +(k -1) (k -1)
kn (kn -k) (kn -2k) (k + k) k
= (kn )!
Replacing 2n with n, we obtain (2).
Next, since (1+ z) = z z ,
k-s k-s k-s
1+ =
k k k
=
k-s 2k - s 2k - s 2k - s 2k - s k - s k-s
2+ = 1+ =
k k k k k k k
=
k-s 3k - s 3k - s 2k - s k - s k-s
3+ = 1+
k k k k k k
(nk -s)(2k - s)(k - s) (kn -s)!k
k-s k-s k-s
n+ = =
k kn k kn k
k-s k-s
(kn -s)!k = k n n + / (3s )
k k
Especially substituting s =0 for (3s ) , we obtain (30)
Substitute n =0 for (30) ; then
0!k = k 0(1) = 1
Replace n with -n in (30) ; then
- 12 -
-(k -1)n (n -1)! k n -(k -1)n kn(n -1)! k
n
= (-1) = (-1)
(kn -1)! (kn)!
kn! k n
= (-1)-(k -1)n { from (2) }
(kn)!k (kn-1)!k (kn-2)!k kn-(k -1)!k
-(k -1)n kn! k n
= (-1) { from (30) }
k nn!(kn-1)!k (kn-2)!k kn-(k -1)!k
k
= (-1)-(k -1)n (6)
(kn-1)!k (kn-2)!k kn-(k -1)!k
Example 1
5-3 5-3 32 2
27!5 = (56-3)!5 = 56 6+ / = 56 /
5 5 5 5
= 15625240.83377994/2.21815954 = 1696464.0025
Example 2
(-1)-(5-1)5 5
(-6)!5(-7)!5(-8)!5(-9)!5 = =
1!52!53!54!5 24
Formula 2.3.3
f (0) 0 f (1)(0) 1 f (2)(0) 2 f (3)(0) 3
x
f = x + x + x + x + (7)
k 0!k k !k (2k)!k (3k)!k
Calculation
f (1)(0) 1 f (2)(0) 2 f (3)(0) 3
f(x) = f(0)+ x + x + x +
1! 2! 3!
f (0) 0 f (1)(0) 1 1 f (2)(0) 2 2 f (3)(0) 3 3
= 0 kx + 1 k x + 2 k x + 3 k x +
k 0! k 1! k 2! k 3!
(1) (2) (3)
f (0) 0 f (0) 1 f (0) 2 f (0)
= (kx) + (kx) + (kx) + (kx)3+
0!k k !k (2k)!k (3k)!k
Replacing x with x /k , we obtain (7).
- 13 -
(2) Expansions by the multifactorial of the trigonometric function
x x0 x2 x4 x6 x8
cos = - + - + -
n 0!n (2n)!n (4n)!n (6n)!n (8n)!n
x x1 x3 x5 x7 x9
sin = - + - + -
n n !n (3n)!n (5n)!n (7n)!n (9n)!n
x 0! 1! 2! 3!
log 1 = x- x2 x3 - x4
n n !n (2n)!n (3n)!n (4n)!n
n +x 0! 2! 4! 6!
log = x+ x3 + x5 + x7 +
n -x n !n (3n)!n (5n)!n (7n)!n
1 n + xi 0! 2! 4! 6!
log = x- x3 + x5 - x7 +
2i n - xi n !n (3n)!n (5n)!n (7n)!n
- 14 -
1!n x n+1 (n +1)!n x 2n+1 (2n +1)!n x 3n+1
x dx
= x + (3n)! 3 +1 +
0
n
1 x n n !n n +1 (2n )!n 2n +1 n n
1 0! 1! 2! 3!
log 1 = - -
n n !n (2n)!n (3n)!n (4n)!n
n +1 0! 2! 4! 6!
log = + + + + (n >1)
n -1 n !n (3n)!n (5n)!n (7n)!n
1 0! 2! 4! 6!
tan -1 = - + - +
n n !n (3n)!n (5n)!n (7n)!n
1 0! 2! 4! 6!
tanh -1 = + + + +
n n !n (3n)!n (5n)!n (7n)!n
(-1)!n (n -1)!n (2n -1)!n (3n -1)!n
n
2 =1+ - + - + (n >1)
n !n (2n)!n (3n)!n (4n)!n
1 1!n (n +1)!n (2n +1)!n (3n +1)!n
n
= 1 - + - + -
2 n !n (2n)!n (3n)!n (4n)!n
(-1)!n (n -1)!n (2n -1)!n (3n -1)!n
1 = + + + + (n >1)
n !n (2n)!n (3n)!n (4n)!n
- 15 -
1 1 1 1 1
sin 1 =
- + - + -
1! 3! 5! 7! 9!
1 1 1 1 1
cosh 1 = + + + + +
0! 2! 4! 6! 8!
1 1 1 1 1
sinh 1 = + + + + +
1! 3! 5! 7! 9!
1 1 1
log 2 =1- + - +
2 3 4
1 1 1
tan -11 = 1 - + - +
3 5 7
1 1 1
tanh -11 = 1 + + + + =
3 5 7
1
=1-1+1-1+1-
2
Formula 2.3.4
When n , k are two or more natural numbers and a is an integer such that |a|< n ,
a +(k -1)n!n
a
= (when a >0) (8)
n k nk
aa +(k -1)n!n
= (when a <0) (8')
nk
Calculation
When a >0 ,
a(a +1n)a +(k -1)n
a a a a
= +1 + k -1 =
n k n n n nk
a +(k -1)n!n
=
nk
When a < 0 , since a !n = 1, a +1n >0, , a +(k -1) n > 0 , then
a = a a !n (k =1)
a(a +1n) = a (a +1n)!n (k =2)
a(a +1n)(a +2n) = a (a +2n)!n (k =3)
hence by inducton
- 16 -
Example
1 1471+3(k -1) 1+3(k -1)!!!
= k
=
3 k 3 3k
-1 -125-1+3(k -1) -1-1+3(k -1)!!!
= k
= (-1)!!! = 1
3 k 3 3k
-2 -214-2+3(k -1) -2-2+3(k -1)!!!
= k
= (-2)!!! = 1
3 k 3 3k
Formula 2.3.5
When n is two or more natural number and a is an integer such that |a|< n ,
a +n(k -1)!n (b )k xk
a
2F 1 , b, c ; x = 1 +Σ (a >0) (9)
n k =1 (c)k (nk)!n
a +n(k -1)!n (b )k xk
= 1 + aΣ (a <0) (9')
k =1 (c)k (nk)!n
When n is two or more natural number and a,b,c are integers such that |a |,|b |,|c|< n ,
a b c
2F 1 , , ;x
n n n
a +n(k -1)!n b +n(k -1)!n xk
= 1 +Σ (a,b,c >0) (10)
k =1 c +n(k -1)!n (nk)!n
a +n(k -1)!n b +n(k -1)!n xk
= 1 + aΣ (only a <0) (10')
k =1 c +n(k -1)!n (nk)!n
1 a +n(k -1)!nb +n(k -1)!n xk
=1+ Σ (only c <0) (10")
c k =1 c +n(k -1)!n (nk)!n
Calculation
From (8),(8') and n k k! = (nk)!n
a
(b)k k
n k
a x
2F 1 , b, c ; x = Σ
n k =0 (c)k k!
a +(k -1)n!n (b )k xk
= 1 +Σ (a >0) (9)
k =1 (c)k (nk)!n
a +(k -1)n!n (b )k xk
= 1 + aΣ (a <0) (9')
k =1 (c)k (nk)!n
- 17 -
a b
n n xk
Σ
a b c k k
2F 1 , , ;x =
n
n n n c k =0 k!
k
a +(k -1)n!n b +(k -1)n!n
nk nk xk
= 1 +Σ (a,b,c > 0)
k =1 c +(k -1)n!n k!
nk
a +n(k -1)!n b +n(k -1)!n xk
= 1 +Σ (a,b,c > 0) (10)
k =1 c +n(k -1)!n (nk)!n
Next, because a symbol whose sign is negative among a,b,c serves as a coefficient of Σ,
(10') and (10") are obvious.
Example 1
1
2F 1 3
, c, c ; x
Example 2
-1
2F 1 3
, c, c ; x
Example 3
1 1 3
2F 1 , , ; x2
2 2 2
1+2(k -1)!21+2(k -1)!2 x 2k
= 1 +Σ
k =1 3+2(k -1)!2 (2k)!2
2k
(2k -1)!!(2k -1)!! x
= 1 +Σ
k =1 (2k +1)!! (2k)!!
2 4
1!! x 3!! x 5!! x 6 sin -1 x
=1+ + + + =
2!! 3 4!! 5 6!! 7 x
- 18 -
Example 4
1 n -1 n +1
2F 1 , , ; xn
n n n
1+n(k -1)!n n -1+n(k -1)!n x nk
= 1 +Σ
k =1 n +1+n(k -1)!n (nk)!n
n(k -1)+1!n (nk -1)!n x nk
= 1 +Σ
k =1 (nk +1)!n (nk)!n
(n -1)!n x n (2n -1)!n x 2n (3n -1)!n x 3n
=1+ + + +
n !n n +1 (2n)!n 2n +1 (3n)!n 3n +1
2004.05.08
K. Kono
Alien's Mathematics
- 19 -
3 Generalized Multinomial Theorem
Theorem 3.1.1
If x1 , x2 are real numbers and n is a positive integer, then
n
x1+ x2 = ΣnC r x1 x2
n n-r r
(1.1)
r=0
Binomial Coefficients
Binomial Coefficient in (1.1) is a positive number and is described as nC r . Here, n and r are both
non-negative integer. nC r is the number of ways of picking r unordered outcomes from n possibilities
and is calculated as follows.
n!
n Cr =
(n -r)! r!
Pascal's Triangle
About nC r , what arranged n in the row and r in the column is called Pascal's Triangle .
C0
0 1
1C0 C1
1 1 1
C0
2 C1
2 C2
2 1 2 1
C0
3 C1
3 3 C2 3C3 1 3 3 1
4 C0 C1
4 C2
4 C3
4 C4
4 1 4 6 4 1
i nC r = nC n-r
ii nC r = n -1C r-1 + n -1C r
n -1
iii C r = Σ kC r-1
n
k = r-1
n
iv Σ
r=0
nC r = 2
n
n
v Σ
r=0
(-1)rnC r = 0
-1-
3.2 Generalized Binomial Theorem
Theorem 3.2.1
The following formulas hold for a real number .
r
Σ x r
|x| 1 ( |x| = 1 is allowed at >0 ) ( 1.1)
r=0
(1+ x) =
Σ -r x
-r
|x|>1 ( 1.2)
r=0
Proof
When n is a natural number, the following expression holds from the binomial theorem.
n
(1+ x)n = ΣnC r x r
r=0
ar+1
= |x|
(p -r)x (p /r -1)x -x
lim = lim = lim =
r ar r r +1 r 1+1/r 1
According to d'Alembert's ratio test, (1.1) converges absolutely if |x|< 1 .
then x < 1 . Therefore, from (1.1) ,
-1
Next, if |x|> 1
r
-1 r
1+ x = Σ x
-1
r=0
-2-
3.2.2 General Binomial Coefficient
The coefficient
r in Theorem 3.2.1 is called General Binomial Coefficient and is as follows.
Although properties similar to binomial coefficient also about general binomial coefficient are known, especially
an important thing is sum of the general binomial coefficient. We prepare some Lemma, in order to obtain this.
Lemma 3.2.2
-1
r
When is not positive integer, binomial series Σ ar converges or diverges simultaneously
r=0
ar
with Dirichlet series Σ (-1)r .
r=1 r
Lemma 3.2.3
r
Σ converges absolutely for non-integer >0 .
r=0
Proof
-1
r =
-r r
Then
-1
r r
Σ
r=0
=Σ
r=0 -r
Let ar = . Then
-r
-1
r r
Σ
r=0
ar =Σ
r=0
(s1)
ar
Σ (-1)r
= Σ(-1)r
(s2)
r=1 r r=1 r ( -r)
Here, let
(-1)r br
r ( -r)
Then
-3-
(-1)r+1
br+1 (r +1) ( -r -1) r ( -r)
= =-
br r (r +1)( -r -1)
(-1)
r ( -r)
From this
r
+1 +1
-1 +1 -1
r r 1 r
= lim = lim =1
r
(r +1)+1
(r +1) r 1
-1 1+ -1
(r +1)+1 r r +1
Since the judgment is impossible, we try Raabe's test for convergence.
(r +1)( -r -1)
br
lim r -1 = lim r -1
r br+1 r
r ( -r)
r r ( -r) - r
r(r +1) r(r +1)
= lim -
r
- r
1 1 r
= lim r 1+ - 1+
r r r -r
= lim 1+ r +
1- /r - r
1 1
r r
Since 1- /r > 0 for sufficiently large r ,
br
1 1 1
lim r -1 = lim 1+ r + 1+ -r
r br+1 r r r 1- /r
= lim r 1+
1 1 1
-1 + lim 1+
r r r r 1- /r
Here,
1 ( -1) 1 ( -1)( -2) 1
s r
1 1
1+ =Σ s
=1+ 1
+ 2
+ 3
+
r s=0 1! r 2! r 3! r
Then,
s r s r s r
1 1 1 1
r 1+ -1 = rΣ s
=Σ s-1
= +Σ s-1
r s=1 s=1 s=2
Therfore,
s r
1 1
lim r 1+ - 1 = + lim Σ s-1
=
r r r s=2
Moreover,
1 1
lim 1+ =1
r r 1- /r
After all,
-4-
br
lim r -1 = +1
r br+1
Thus, if >0 , (s2) converges absolutely. Then, (s1) also converges absolutely according to Lemma 3.2.2 .
Theorem 3.2.4
The following expressions hold for arbitrary real number >0 .
r = 2
Σ
r=0
(2.1)
Σ(-1) r = 0
r
(2.2)
r=0
Proof
r
According to Lemma 3.2.3 , Σ converges absolutely for non-integer >0 .
r=0
Therefore, from Theorem 3.2.1 (1.1) ,
r
Σ 1r = (1+1) = 2 (2.1)
r=0
Σ r (-1) = (1-1)
r
=0 (2.2)
r=0
Note
In fact, it is known that (2.1) holds if > -1 . ( Where, it is conditional convergence. )
For example, when = -0.9 , the right side is 2-0.9 = 0.53588673 and the left side seems to
converge to this value. However, the confirmation is difficult as the convergence is very slow. So, we apply
Knopp Transformation to this and accelerate the convergence. It is as follows.
Theorem 3.2.5
When is a real number, the following expression holds for x1 , x2 s.t. x 1 x 2 .
r
x1+ x2 = Σ x1 -rx2r ( x 1 = x 2 is allowed at >0 ) (3.1)
r=0
Proof
x2
x2
If x 1 x 2 then 1> = . Therefore, using Theorem 3.2.1 (1.1) ,
x1 x1
-5-
x = x
x2
x2
x1+ x2 = 1 1+
x1 1 1+
x1
x
1 x 2 x 3 x +
2 3
x 2 x 2 2
=x 1 +
1 + +
1 1 1
=
0 1
x +
1x x +
2 x x +
1 3 x x +
-1 1
2
-2 2
1 2
-3 3
1 2
r x x
-r r
=Σ 1 2
r=0
Note
As is clear from the process of the proof, if x 1 x 2 then (3.1) converges absolutely.
-6-
3.3 Multinomial Theorem
Theorem 3.3.0
For real numbers x1 , x2 , , xm and non negative integers n , r1 , r2 , , rm , the followings hold.
n!
x11 x22 xmm
n r r r
x1+ x2++ xm = Σ (0.1)
r1! r2! rm!
where denotes the sum of all combinations of r1 , r2 , , rm s.t. r1+ r2++ rm= n .
n! n-r -r r
x1 1 m-1 x21 xmm-1
n r
x1+ x2++ xm = Σ (0.2)
n -r1 -rm-1! r1! rm-1!
where denotes the sum of all possible combinations of n , r1 , r2 , , rm-1 .
Since (0.1) is well known, the proof is omitted. In addition, it is also clear (0.2) and (0.1) are synonymous.
These are near a definition rather than a theorem.
Theorem 3.3.1
For real numbers x1 , x2 , , xm and a natural number n , the following expressions hold.
r r r
n r1 rm -2 n r1 rm-2 rm-2-rm-1 rm-1
x1+ x2++ xm = Σ Σ Σ x2 xm-1
n n -r1 r1-r2
x1 xm
r1=0 r2=0 rm -1=0 1 2 m-1
(1.1)
r1+ r2+ + rm-1
r + r + + r
n n n n rm-2 + rm-1
= Σ Σ Σ
r1=0 r2=0 rm -1=0 1 2 m-1 r2+ + rm-1 rm-1
n -r1- - rm-1 r
x1 x21 x32 xmm-1
r r
(1.2)
Proof
According to Theorem 3.1.1 , the following expressions hold.
n n n-r1 r1
x1+ x2+ x3+ x4++ xm = Σ nC r1 x1 x2+ x3+ x4++ xm (1)
r1=0
r r1 r
r
x2+ x3+ x4++ xm
1
= Σ r1C r2 x21-r2 x3+x4++xm 2 (2)
r2=0
r2 r2 r
r
x3+ x4++ xm = Σ r2C r3 x32-r3 x4++xm 3 (3)
r3=0
rm -3
r r -rm-2 rm-2
xm-2 + xm-1 + xm
m-3
= Σ r Cr
r =0
m-3 m-2
m-3
xm-2 xm-1 + xm (m -2)
m -2
-7-
rm -2
r r -rm-1 r
xm-1 + xm
m-2
= Σ r Cr
r =0
m-2 m-1
m-2
xm-1 xmm-1 (m -1)
m -1
Substituting (2), (3), ,(m -2) ,(m -1) for (1) one by one, we obtain (1.1) .
Next, according to Theorem 3.4.1 ( later ) , when x1 x2+ x3++ xm ,
r1+ r2+ + rm-1
r + r + + r
rm-2 + rm-1
x1+ x2++ xm = Σ Σ Σ r2+ + rm-1
rm-1
r1=0 r2=0 rm -1=0 1 2 m-1
-r1-- rm -1 r1 r2 r
x1 x2 x3 xmm-1
Replacing the real number with non-negative integer n ,
r1+ r2+ + rm-1
r + r + + r
rm-2 + rm-1
x1+ x2++ xm = Σ Σ Σ
n
r1=0 r2=0 rm -1=0 1 2 m-1 r2+ + rm-1 rm-1
n -r1-- rm -1 r
x1 x21 x32 xmm-1
r r
r = 0 for r > n, r =1, 2, 3, , this is a definite multiple series. Therefore, the condition
n
Since
x1 x 2+ x3++ xm is unnecessary. Although this is not bad as it is, replacing on the with
n , we obtain (1.2) .
cf.
(1.2) results in (0.2) . Because,
r1+ r2+ + rm-1
r + r + + r
n rm-2 + rm-1 n!
=
1 2 m-1 r2+ + rm-1 rm-1 n -r1 - - rm-1! r1!+ + rm-1!
4
Example 1: The expasion of x1+ x2+ x3
Using (1.1) ,
r s
4 r 4 r r-s
x1+ x2+ x3 = ΣΣ x14-rx2 x3s
4
r=0 s=0
0 Σ s x
4 0 4
0
0-s s
= x 1 2 x3
s=0
1 Σ s x 2 s x
4 1 1 4 2 2
+ x 3
1
1-s s
2 x3 + x12Σ 2-s s
2 x3
s=0 s=0
3 s x 4 s x
4 3 3 4 4 4
Σ Σ
1 3-s s 0 4-s s
+ x 1 2 x3 + x 1 2 x3
s=0 s=0
= x14
+ 4x13 x2 + x3
+ 6x12 x22 + 2x2 x3 + x32
+ 4x1x23 + 3x22 x3 + 3x2 x32 + x33
+ x24 + 4x23 x3 + 6x22 x32 + 4x2 x33 + x34
-8-
Using (1.2) ,
r + s s
4 4 4 r+s
x1+ x2+ x3 = ΣΣ x14-r-s x2r x3s
4
r=0 s=0
0+ s s x
44 0+ s
= Σ 4-s 0
1 x2 x3s
s=0
1+ s s x 2+ s s x
44 1+ s 4 4 2+ s
+Σ 3-s 1 s
1 x2 x3 +Σ 2-s 2 s
1 x2 x3
s=0 s=0
3+ s s x 4+ s s x
44 3+ s 4 4 4+ s
+Σ 1-s 3 s
1 x2 x3 +Σ 0-s 4 s
1 x2 x3
s=0 s=0
3
Example 2: The expasion of x1+ x2+ x3+ x4
Now, the formulas of the theorem are expanded using mathematical software. (1.1) and (1.2) are expanded
and are verified respectively.
-9-
Sum of multinomial coefficients
n r1 r m -2
Σ Σ Σ nC r1 r1C r2 rm-2C rm-1 = m n
r1=0 r 2=0 rm -1=0
(1.1")
Proof
n n
n-r r
Σ
r=0
nC r =ΣnC r1
r=0
1 = (1+1)n = 2n
n r n r n
s=0 = Σ
n-r r
Σ ΣnC r rC s =Σ
r=0 s=0 r=0
nC r Σ rC s
r=0
nC r1 2 = (1+2)n = 3n
n r s n r s n
Σ Σ Σ
r=0 s=0 t=0
nC r rC s sC t =ΣnC r ΣΣrC s sC t
r=0 s=0 t=0 r=0
=ΣnC r1n-r3r = (1+3)n = 4n
Hereafter, by induction we obtain the desired expression.
4
Example: Sum of multinomial coefficients of x1+x2+x3
Let's calculate sum of multinomial coefficients in Example 1 . Then it is as follows.
1+(4+4)+(6+12+6)+(4+12+12+4)+(1+4+6+4+1) = 81 = 34
- 10 -
3.4 Generalized Multinomial Theorem
Although I do not know whether the theorem like generalized multinomial theorem exists or not , since this
is essential for Higher Calculus of Function Product, I present this here.
Theorem 3.4.1
The following expressions hold for real numbers and x1 , x2 , , xm s.t.x 1 x2+ x3++ x m .
r r r
r1 r m -2 r1 rm-2 -r1 r1-r2 r -rm -1 rm -1
x1+ x2++xm = Σ Σ Σ xm-1
m -2
. x1 x2 xm
r1=0 r 2=0 rm -1=0 1 2 m-1
(1.1)
r1+ r2+ + rm-1
r + r + + r
rm-2 + rm-1
= Σ Σ Σ
r 1=0 r 2=0 rm -1=0 1 2 m-1 r2+ + rm-1 rm-1
-r1- - rm-1 r1 r2 r
x1 x2 x3 xmm-1
(1.2)
Where, x 1 x 2+ x 3++ x m is allowed at >0 .
Proof
From Theorem 3.2.5 , when x 1 x 2+ x 3++ x m , the following expression holds.
-r1 r1
x1+ x2+ x3+ x4++ xm = Σ x1 x2+ x3+ x4++xm
r1=0 r1
r1
r1 r2 r m -2 r1 r2 rm-2 r -rm -1 rm -1
x1+ x2++ xm = Σ Σ Σ xm-1
r -r2 r -r3 m -2
x21 x32 xm
r2=0 r 3=0 r m -1=0 r2 r3 rm-1
Substituting the latter for the former ,
r r r x
r1 rm -2 r1 rm-2 -r1 r1-r2 r -rm -1 rm -1
x1+ x2++ xm = Σ Σ Σ xm-1
m -2
1 x2 xm
r 1=0 r 2=0 r m -1=0 1 2 m-1
(1.1)
Naturally, the right side also converges absolutely.
Next, let us describe a multiple series and its iterated series as follows respectively.
Σ ar ,r ,,r
r , r , ,r =0
1 2 m
, Σ Σ Σ ar ,r ,, r
r =0 r =0 r =0 1 2 m -1 ,rm
1 2 m 1 2 m
In order to convert the iterated series to its diagonal series, we should just perform the following operations.
( See " 02 Multiple Series & Exponential Function " ).
Replace rm-1 with rm-1- rm , and replace the 1st with rm-1 from the right.
Replace rm-2 with rm-2- rm-1 , and replace the 2nd with rm-2 from the right.
Replace r1 with r1- r2 , and replace the (m-1)th with r1 from the right.
If so, in order to return the diagonal series to the original iterated series, we should just perform this opposite
operation. That is,
Replace r1 with r1+ r2 , and replace r1 on the 2nd with from the left.
Replace r2 with r2+ r3 , and replace r2 on the 3rd with from the left.
- 11 -
Replace rm-1 with rm-1+ rm , and replace rm-1 on the (m -1)t h with from the left.
For example,
r r x x x
r1 r2 r1 r2 -r1 r1-r2 r2-r3 r3
x1+ x2+ x3+ x4 = Σ Σ Σ r x1 2 3 4
r1=0 r 2=0 r 3=0 1 2 3
r +r r r x x x
r1 r +r r r2 1 2 2 -r1-r2 r1 r2-r3 r3
r r + r ,ΣΣ; = Σ Σ Σ
1 1 2 1 2 3 x4
r1=0 r2=0 r 3=0 1 2 2 3
Σ Σ Σ r +r +r r +r r x
r2 r +r +r r +r
1 2 3 2 3 -r1-r2-r3 r1 r2 r3
2 2
r r + r ,Σ Σ;
3 = 1 x2 x3 x4
r1=0 r 2=0 r3=0 1 2 3 2 3 3
Thus, performing this operation to (1.1) , we obtain the following.
r1+ r2+ + rm-1
r + r + + r
rm-2 + rm-1
x1+ x2++ xm = Σ Σ Σ r2+ + rm-1
rm-1
r 1=0 r 2=0 rm -1=0 1 2 m-1
- r + + rm-1 r1 r2
x1 1
r
x2 x3 xmm-1 (1.2)
Since (1.1) converges absolutely, this rearrangement is allowed.
3.9
Example 1: The expasion of x1+ x2+ x3
Using (1.1) ,
r s x x x
r 3.9 r 3.9-r r-s s
x1+ x2+ x3 = ΣΣ
3.9
1 2 3
r=0 s=0
0 Σs 1 Σs x x
3.9 0 3.9
3.9
0 1 0-s s 2.9
1
1-s s
= x x 1x + x 2 3 1 2 3
s=0 s=0
2 s 3 s x x
3.9 2 3.9
2 3 3
+ x Σ x x + 1.9
1 x Σ 2-s s
2 3
0.9
1
3-s s
2 3
s=0 s=0
4 x Σs 5 x Σs x
3.9 1 4 4.1 41 5 5
+ x x0.1
+ 4-s s
2 3 1.1
5-s s
2 x3 +
s=0 s=0
1 1
= x13.9
+ 3.9 x12.9 x2 + x3
+ 5.655 x11.9 x22 + 2x2 x3 + x32
+ 3.5815 x10.9 x23 + 3x22 x3 + 3x2 x32 + x33
0.805838 4
x2 + 4x2 x3 + 6x2 x3 + 4x2 x3 + x3
3 2 2 3 4
+ 0.1
x1
0.0161168 5
x2 +5x2 x3 +10x2 x3 +10x2 x3 +5x2 x3 + x3
4 3 2 3 2 4 5
- 1.1
x1
Using (1.2) ,
r+s s x
3.9 r+s
x1+ x2+ x3
3.9
= ΣΣ 3.9-r-s r
1 x2 x3s
r=0 s=0
- 12 -
s=0 0+ s 1+ s s x
3.9 3.9
s
0+ s 1+ s
= Σ x13.9-s x20 x3s +Σ 2.9-s 1 s
1 x2 x3
s=0
2+ s s x 3+ s s x
3.9
2+ s 3.9 3+ s
+Σ 1.9-s 2 s
1 x2 x3 +Σ 0.9-s 3 s
1 x2 x3
s=0 s=0
4+ s s x
3.9
4+ s
+Σ -0.1-s 4 s
1 x2 x3 +
s=0
4
0.805838 x3
= x13.9 +
2.9
3.9 x1 x3 +
1.9 2
5.655 x1 x3 +
0.9
3.5815 x1 x 33 + +
x10.1
3
2.9 1.9 0.9 2 3.22335 x 2x3 0.0805838
+ 3.9 x1 x 2 + 11.31 x1 x2 x3 + 10.7445 x1 x2 x3 + -
x10.1 x11.1
2 2 2 3
4.83503 x2 x3 0.161168 x2 x3
+
1.9 2 0.9 2
+ 5.655 x1 x2 + 10.7445 x1 x2 x3 + -
x10.1 x11.1
3 3 2 3 3
3.22335 x 2 x3 0.161168 x2 x3 0.0590948 x2 x3
+
0.9 3
+ 3.5815 x1 x2 + - +
x10.1 x11.1 x12.1
4 4 4 2 4 3
0.805838 x2 0.0805838 x2 x 3 0.0443211 x2 x3 0.0310247 x2 x3
+ - + - +
x10.1 x11.1 x12.1 x13.1
we can see that what totaled this along the diagonal line is equal to the above.
Although a = b +c+d in this numerical example, (1.1) and (1.2) are consistent.
- 13 -
As expected, the following expression does not hold.
r r r = m
r1 rm -2 r1 rm-2
ΣΣ Σ
r 1=0 r2=0 rm -1=0 1
2 m-1
2007.07.06
2016.02.13 updated
2016.02.22 renewed
Kano. Kono
Alien's Mathematics
- 14 -
4 Higher Integral
Definition 4.1.1
(x) denotes the primitive function of f <n -1>(x) for n =1, 2, 3, ,
<n>
When f
<n>
we call f (x) Higher Order Primitive Function of f(x) ,
or for short , Higher Primitive of f(x) .
Where, f
<n>
(x) might mean indefinite integral f <n -1>
(x)dx + cn ,
f
x
<n -1>
or might mean integral function (x)dx .
an
The primitive function of f(x) is denoted by a capital letter F(x) in many cases. Then with what symbols
should the primitive function of F(x ) and its further primitive functions be denoted ?
For example, G(x), H(x), etc. can be considered. However, by these notation, it is difficult to denote
the primitive function such as the 27 th order. So, applying the notation of the differentiation, I had denoted
F(x), G(x), H(x), (x), f (-2)(x), f (-3)(x), . Soon, the minus sign became obst-
(-1)
by f
ructive. Then I thought to use [n ] instead of (-n) , but this notation was as confusing as floor function.
So I came to adopt the notation f (x), f <2>(x), f <3>(x),
<1>
using <> . The power of this notation
will become clear gradually .
Example
(sin x)<1> = -cosx + c1 c1 is an arbitrary constant.
Definition 4.1.2
We call it Higher Integral to integrate a function f with respect to an independent variable x repeatedly.
And it is described as follows.
a a f(x)dx dx dx dx
x x x x x x
n
f(x)dx =
n 1 an a3 a2 a1
And
when ak = a for k =1, , n , we call it higher integral with a fixed lower limit ,
when ak a for some k , we call it higher integral with variable lower limits .
-1-
is what integrated f(x) with respect to x repeatedly. Then it is also called repeated integration.
By familiar expression, a triple integral is what integrates f(x,y,z) with respect to length, width and height.
On the other hand, the 3 rd order integral is what integrates f(x) 3 times with respect to only length.
As a notation of the higher integral, it is denoted in many cases as follows using a notation of multiple integral.
ft dt dt dt
x tn t2
1 1 2 n
a a a
However, this notation is complicated and inconvenient. Hence in this paper, for these problem solutions and
saving of variables, we replaced t1 , t2 , , tn with x and replaced dt1 dt2 dtn with dx n , as seen in
the definition.
Example
e -1 dx = e - x -1
x x x x x
e xdx 2 = e xdx dx = x x
0 0 0 0 0
e dx = e dx dx = e dx = e
x x x x x
x 2 x x x
- - - - -
a
Where, F(a) is called Constant-of-integration .
Theorem 4.1.3
r =0, 1, , n
<r> r+1>
Let f be continuous functions defined on a closed interval I and f< be
ar , x I .
<r>
the arbitrary primitive function of f . Then the following expression holds for
f(x)dx
x x
<n> n -1
<n -r> (x -a)r
n
=f (x) - Σf (a) (1.2)
a a r=0 r!
-2-
Proof
First, the following equations hold according to Fundamental Theorem of Calculus.
x
f <0>dx = f <1> - f <a11> (01)
a1
f
x
<1>
dx = f <2> - f <a22> (12)
a2
f
x
<2>
dx = f <3> - f <a33> (23)
a3
f
x
<3>
dx = f <4> - f <a44> (34)
a4
Integrating both sides of (01) with respect to x from a 2 to x ,
dx
x x x x
f <0>dx 2 = f <1>dx - f <a11>
a2 a1 a2 a2
Substituting (12) for this,
f dx
x x x
<0> 2
dx = f <2> - f <a22> - f <a11>
a2 a1 a2
Integrating both sides of this with respect to x from a3 to x ,
dx
x x x x x x x
f(x)dx 3 = f <2>dx - f <a22> dx - f <a11> 2
a3 a2 a1 a3 a3 a3 a2
Substituting (23) for this,
dx
x x x x x x
f(x)dx 3 = f <3> - f <a33 > - f <a22> dx - f <a11 > 2
a3 a2 a1 a3 a3 a2
Integrating both sides of this with respect to x from a4 to x ,
f(x)dx = f dx - f dx dx
x x x x x x x x x
4 <3> <3> <2> 2 <1> 3
dx - f a3 a2 -f a1
a4 a1 a4 a4 a4 a3 a4 a3 a2
Substituting (34) for this,
dx
x x x x x x x x
f(x)dx 4 = f <4> - f <a44> - f <a33> dx - f <a22> dx 2 - f <a11 > 3
a4 a1 a4 a4 a3 a4 a3 a2
Hereafter by induction, we obtain the following expression.
x x n -1 x x
n <n>
f(x)dx = f
<n> <n -r>
-fa -Σfa dx r
an a1 n r=1 n-r an a n- r+ 1
<n>
And pushing f an into Σ, we obtain (1.1) .
x x (x -a)r
dx = r
r =1, 2, ,n -1
a a r!
Then, using this, we obtain
f(x)dx
x x
<n> n -1
<n -r> (x -a)r
n
=f (x) - Σf (a) (1.2)
a a r=0 r!
-3-
Constant-of-integration Polynomial
(x -a)r
n -1 x x n -1
<n-r>
Σf a n-r Σ
<n-r>
We call dx r and f (a) in the theorem
r=0 an an-r +1 r=0 r!
Constant-of-integration Polynomial.
In the case of (1.1) , it is very difficult to express the higher integral of 1 by a polynomial.
For reference, if this is calculated by force and the 4th order integral of f(x) is expressed by a polynomial,
it is as follows. The 5th order or more are unmanageable...
1
x -a 4 x -a 4x -2a 3+ a 4
x x
f(x)dx = f 4 <4>
-fa <4>
-fa <3>
- f <a2>
a4 a1 4 3 1! 2 2!
x -a 4x + a 4 x -3a2 x + 6a 2 a3 -3a 2 a4 -3a 3 + a 4
2 2 2
<1>
-fa
1 3!
Since the remainder term of (1.2) is already expressed by the polynomial, we can find out easily that it is
the (n-1)-th order. We can also express it explicitly as follows, though there is no merit not much.
n
f < -s>
x x n -1 n -1
s-r a
f(x)dx = f (x) -
n <n> r
cr+1 x , cr+1 = Σ (-1) sC s-r a
s-r
Σ
a a r =0 s =r s!
Example1
1
a a f(x)dx a dx x -a2
x x x
2 <2> <2> <1> <2> <2> <1>
=f -fa -fa =f -fa -fa
2 1
2 1
2
2 1 1!
<1> <2>
Let f = e x , e x = e x + c1 , e x = e x + c1 x + c2
Substituting these for the above experession, we obtain as follows.
1
a a a 1 x -a 2
x x
x 2 x a2
Left: e dx = e - e - e
2 1
1!
1
x <2> <2> <1> x -a 2
Right: e - e xa2 - e xa1
1!
1
x -a2
= e + c1 x + c2 - e - c1 a2- c2 - e + c1
x a2 a1
1!
1
a a1 x -a2
= ex - e 2 - e
1!
Example2
a a a a a a dx
x x x x x x
f(x)dx 3 = f <3> - f <a33> - f <a22> dx - f <a11> 2
3 2 1 3 3 2
1
x -a 3 x -a 3x -2a 2+ a 3
= f <3> - f <a3> - f <a2> - f <a1>
3 2 1! 1 2!
<1> <2> <3>
Let f = sin x , (sin x) = -cosx , (sin x) = -sin x , (sin x) = cosx
Substituting these for the above experession, we obtain as follows.
-4-
a a a sin x dx
x x x
3
Left:
3 2 1
x -a3cosa1
2 2
= cosx -cosa3 + + x -a3sin a2 - a2x -a3cosa1
2!
1
x -a 3 x -a 3x -2a2+ a3
Right: cosx - cosa3 + sin a2 + cosa1
1! 2!
x -a32cosa1
2
= cosx -cosa3 + x -a3sin a2 + - a2x -a3cosa1
2!
Definition 4.1.4
x x n -1 x x
<n>
f(x)dx = f (x) - Σ f
n <n -r>
an-r dx r (1.1)
an a1 r=0 an a n- r+ 1
In this expression,
when Constant-of-integration Polynomial is 0,
x x
we call f(x)dx n Lineal Higher Integral and
an a1
we call the function equal to this Lineal Higher Primitive Function .
x x
we call f(x)dx n Collateral Higher Integral and
an a1
we call the function equal to this Collateral Higher Primitive Function .
(or for short , Collateral Higher Primitive. )
These are the same also in (1.2).
In short , Lineal Higher Primitive Function is what integrated f(x) with respect to x repeatedly
without considering the constant of the integration.
Example
Left: Collateral Higher Integral Right: Collateral Higher Primitive
sin xdx
x x x 1
3 (x -3) (x -3)(x -22+3)
= cosx - cos 3+ sin 2 + cos 1
3 2 1 1! 2!
e dx
x x
x 2
= e x-1- x
0 0
x x x
sin xdx 3 = cosx
3 2 1
2 2 2
e dx
x x
x 2
= ex
- -
-5-
4.1.5 The Conditions for the Higher Integral being Lineal
In Theorem 4.1.3 (1.1), since the higher integral of 1 can be arbitrary value, in order for Constant-of-integration
Theorem 4.1.5
f(x)dx
x x
<n> n -1
<n -r> (x -a)r
a a
= f (x) -
r=0
n
f Σ (a)
r!
(1.2)
In these expressions, the necessary conditions for higher integrals being lineal are as follows respectively
r = 1, 2, , n
<r>
f ar = 0
r = 1, 2, , n
<r>
f (a) = 0
That is, The necessary condition for the higher integral being lineal is that a r or a is a zero of the higher
<r>
primitive function f for each r from 1 to n.
Note
These are necessary conditions and are not sufficient conditions. For example, let
f(x)dx
x x
r = 1, 2, , n
<r> r
f (x) =
ar a1
ar = 0 r =1, 2, ,n
<r>
Then, whether or not this higher integral is lineal, certainly f
In the case of f(x)=e x , let
e dx
x x
<r>
f (x) = x r
r = 1, 2
0 0
Then, f
<1>
(x)= e x -1 , f <2>(x)= e x -1- x and the lower limit 0 is a zero of these primitive functions.
Nevertheless, substituting these for (1.2), we obtain
x x x0 x1
e dx = e -1- x - e -1-0
x 2 x
- e -1
0
= e x -1- x
0
0 0 0! 1!
This is not lineal higher integral.
-6-
4.2 Reimann-Liouville Integral and Higher Integral
f(x)dx
x x
When ak = a , k =1, 2, , n the higher ingegral n
can result in the single integral
an a1
called Reimann-Liouville Integral .
(x -t)
x x 1 x
f(x)dx n =
n-1
f(t)dt (2.1)
a a (n) a
Proof
First, Reimann-Liouville Integral can be expanded as follows.
(x -t)
1 x
n-1 <n> n -1 (x -a)r <n -r>
f(t)dt = f (x) - Σ f (a) (2.2)
(n) a r=0 r!
Because,
f(t)dt = f
1 x x x
0
(x -t) f(t)dt =
<1>
(t)a
(1) a a
<1>
= f (x) - f <1>(a)
f
x x
1 1 1 <1> x <1>
(x -t) f (t)a +
1
(x -t) f(t)dt = (t)dt
(2) a 1! a
1 <1> <2> <2>
= -(x -a) f (a)+ f (x)- f (a)
x x
1 1 2 <1> x 1 <1>
(x -t) f (t)a + 2 (x -t) f (t)dt
2
(x -t) f(t)dt =
(3) a 2! a
1 x x
2 <1> 1 <2> <>
=- (x -a) f (a )+ (x -t) f (t)a + f 2 (t)dt
2! a
2 1
(x -a ) <1> (x -a ) <2> <> < >
=- f (a)- f (a)+ f 3 (x )- f 3 (a)
2! 1!
(x -t)
1 x
n-1 <n> n -1 (x -a)r <n -r>
f(t)dt = f (x) - Σ f (a) (2.2)
(n) a r=0 r!
On the other hand, Theorem 4.1.3 (1.2) was as follows.
f(x)dx
x x
<n> n -1
<n -r> (x -a)r
n
=f (x) - Σf (a) (1.2)
a a r=0 r!
From these, we obtain the desired expressin.
Note
This theorem shows that the higher integral with fixed lower limit is equivalent to Riemann-Liouville integral.
Therefore, we can examine the higher integral in left side by Riemann-Liouville integral in right side. Moreover,
this theorem shows that we can not apply Riemann-Liouville Integral to the higher integral with variable lower
limits. So, we can not apply Riemann-Liouville integral to the higher integral of sin x etc.. Strictly speaking,
if we apply it by force, then we obtain the collateral higher integral inevitably.
-7-
Reimann-Liouville Integral and Taylor's Theorem
The right side of (2.2) is rewritten as follows.
(x -t)
<n> n -1
<n -r> (x -a)r 1 x
n-1
f (x) = Σf (a) + f(t)dt (2.2)
r=0 r! (n) a
<n>
Then we notice that this is Taylor expansion of f (x) around a and Reimann-Liouville Integral is
the remainder term.
Actually, from this, we can derive usual Taylor's theorem. Let us shift by -n the index of the integration
operator of f(x) . Then
(x -t)
n -1
(r) (x -a)r 1 x
n-1 (n)
f <0>
(x) = Σ f (a) + f (t)dt (2.20)
r=0 r! (n) a
(n)
Just, this is the Taylor expansion of f(x) around a . And Reimann-Liouville Integral of f is the remainder
term called Bernoulli form .
And according to the first mean value theorem for integration, when g(t) is a continuous real valued function
on the [a,b] and (t) is an integrable function that does not change the sign on the a,b , there exists
a (a,b ) such that
(t)dt
b b
g(t) (t)dt = g( )
a a
Applying this to (2.20) , we obtain the following.
f ( )
(n)
(x -t)
1 x x
n-1 (n) n-1
(x -t) f (t)dt = dt
(n) a (n -1)! a
f
(n)
( ) n x (x -a)n (n)
= -(x -t) a = f ( )
n! n!
This is the remainder term called Lagrange form .
-8-
4.3 Higher Integrals of Power Function etc.
x 1
(ax +b) dx n = (ax +b)+n ( 0)
b b a (1+ +n)
- -
a a
(- - n)
x n
x 1
(ax +b) dx = - n
(ax +b)+n ( <-n)
a (- )
Proof
When 0 , the zero of the higher primitive functions of (ax +b) is -b /a . Hence
x 1 1
(ax +b) dx = (ax +b)+1
b a +1
-
a
x x 2
1 1
(ax +b) dx 2
= (ax +b)+2
b b a ( +1)( +2)
- -
a a
x n
x 1 1
(ax +b) dx = n
(ax +b)+n
b
-
b a ( +1)( +2)( +n)
-
a a
1 (1+)
Substituting = for this, we obtain (2).
( +1) ( +2) ( + n ) ( 1+ + n )
Furthermore, substituting a =1, b =0 for this, we obtain (1).
x 1 1
(ax +b)-dx = (ax +b)-+1
a - +1
(ax +b)
2
x x 1 1
-
dx =2
(ax +b)-+2
a (- +1)(- +2)
x x 1 1
-
dx = n
(ax +b)-+n
a (- +1)(- +2)(- + n)
-9-
Since
1 (-1)n
=
(- +1)(- +2)(- + n) ( -1)( -2)( - n)
[1+ -(n +1)] ( -n)
= (-1)n = (-1)n
1+( -1) ()
we can substitute this for the above as follows.
( - n)
n
x x 1 -
(ax +b) dx = - n
(ax +b)-+n
a ( )
Then replacing with - we botain (2), furthermore, substituting a =1, b =0 for this, we obtain (1).
Note
When -n < 0 , we do not dare define the higher integral of the power function. It is because the lower
limit of the integral changes irregularly in this case. For example as follows.
5
x x
x x x x x x 8
2 3 2 3
dx = -x(log x -1) , dx = x
0 1 0 3
Examples
e
x x
x
dx n = (1)ne x
n
a <0 : +
x a >0 : -
x 1
e
ax+b
dx n
= e
ax+b
a
Proof
When a >0 the zero of higher primitive functions of e ax+b is - , and when a <0 the zero of same one
is + . Hence we obtain (2). Furthermore, substituting a = 1 , b =0 for this, we obtain (1).
- 10 -
Formula 4.3.3 : Higher Integral of Logarithmic Function
log x dx
x xn
x n 1
n
= log x -Σ
0 0 n! k =1 k
x n
x 1 b n 1
log(ax +b)dx n = x+ log(ax +b) -Σ
b b n! a k =1 k
- -
a a
Proof
m+1 m+1
x x x
m
x log x dx = log x -
0 m +1 (m +1)2
Therfore,
x1 x1 x1
x 1 1
log x dx = x(log x - 1) = log x - = log x -
0 1! 1! 1 1! 1
2 2 2 2
1 x x x x x2 x2
= log x - 2 - = log x - -
1! 2 2 2 2! 2!2 2!
x2 x2 x2
1 1
= log x - 1+ = log x - 1+
2! 2! 2 2! 2
x x x 1 x 1 1 x
log x dx 3 = x 2 log x dx - 1+ x 2dx
0 0 0 2! 0 2! 2 0
x3 x3
1 1 1 3
= log x - 2 - 1+ x
2! 3 3 3! 2
x3 x3
1 1 3
= log x - - 1+ x
3! 3!3 3! 2
x3 x3 x3
1 1 1 1
= log x - 1+ + = log x - 1+ +
3! 3! 2 3 3! 2 3
Hereafter, we obtain (1) by induction. And replacing x with ax +b , we obtain (2) by similar way.
- 11 -
Formula 4.3.4 : Higher Integrals of sin x , cos x
x x 1
sin(ax +b)dx n
= sin ax +b -
n b 1 b a 2
- -
2a a 2a a
n
n
x x 1
cos(ax +b)dx = n
cos ax +b -
(n -1) b 0 b a 2
- -
2a a 2a a
Proof
Lineal higher primitive function of f(x)= sin x and its zeros are as follows.
1 1
f<1>(x) = -cosx
= sin x -
2 , x=
2
2 2
f<2>(x) = -sin x
= sin x -
2 , x=
2
3 3
f<3>(x) = cosx
= sin x -
2 , x=
2
4 4
f<4>(x) = sin x
= sin x -
2 , x=
2
Thus we obtain (1). And replacing x with ax +b cos x
, we obtain (2) by similar way. For , it is obtained by
deducting /2 from the lower limit of the higher integral in the left side and replacing sin with cos in the
right hand.
sinxdx
x x (-1)k 2k+1+ n
n
=Σ x
0 0 k =0 (2k +1+ n)!
n k
n-k
+Σ
n x
= sin x - sin
2 k =1 (n -k)! 2
x x
k
(-1)
cosxdx n = Σ x 2k+
n
0 0 k =0 (2k + n)!
- 12 -
n k
n-k
-Σ
n x
= cos x - cos
2 k=1 (n -k)! 2
sinxdx x1
cosxdx
x x x x
2 2
= -sinx + , = -cosx + 1
0 0 1! 0 0
sinxdx x2
cosxdx x1
x x x x x x
3 3
= cosx + -1 , = -sinx +
0 0 0 2! 0 0 0 1!
n i
e x + (-1)ne -x
x x x
n n
cosh xdx = i cosh x - =
(n -1) i 1 i 0 i 2 2
2 2 2
(2) Linear form
n i
n
x
x i
sinh(ax +b)dx = n
sinh ax +b -
n i b 1 i b a 2
- -
2a a 2a a
n
e
1 1
= ax+b
- (-1)ne -(ax+b)
2 a
n i
n
x
x i
cosh(ax +b)dx = n
cosh ax +b -
(n -1) i b 0 i b a 2
- -
2a a 2a a
n
e
1 1
= ax+b
+ (-1)ne -(ax+b)
2 a
Proof
Lineal higher primitive function of f(x)= sinh x and its zeros are as follows.
1 i 1 i
f<1>(x) = cosh x
= i sinh x -
2
, x=
2
2 i 2 i
f<2>(x) = sinh x
= i 2sinh x -
2 , x=
2
- 13 -
3 i 3 i
f<3>(x) = cosh x = i 3sinh x - 2 , x=
2
4 i 4 i
f<4>(x) = sinh x = i 4sinh x - 2 , x=
2
Thus we obtain the hyperbolic form of (1). And replacing x with ax +b , we obtain the hyperbolic form of (2)
by similar way. For cosh x , it is obtained by deducting i/2 from the lower limit of the higher integral in
the left side and replacing sinh with cosh in the right side.
Next,
n i n i
n i
sinh ax +b -
2 =
2
e
1 ax+b- 2
-e
- ax+b -
2
n i i -n
=
e
-
2
e
ax+b
-e -(ax+b -n i)
=
e 2 e ax+b - e i e -(ax+b)
n
2 2
-n
i
e - (-1)ne -(ax+b)
ax+b
=
2
n i n i
n i
cosh ax +b -
2
=
2
e
1 ax+b- 2
+e
-ax+b -
2
-n
i
e + (-1)ne -(ax+b)
ax+b
=
2
Therefore, using these we botain exponential forms of (1), (2).
sinhxdx
x x 1 2k+1+ n
n
=Σ x
0 0 (2k +1+ n)! k =0
n i k i
n-k
n x
n
= i sinh x - -Σ i -k sinh
2 k =1 (n -k)! 2
x x 1
coshxdx n = Σ x 2k+
n
0 0 k =0 (2k n
+ )!
n i k i
n-k
n x
n
= i cosh x - -Σ i -k cosh
2 k =1 (n -k)! 2
- 14 -
4.4 Higher Integrals of Inverse Trigonometric Functions
-1 -1
Formula 4.4.1 : Higher Integrals of tan x , cot x
When , , (x) denote floor function, ceiling function and digamma function respectively, the following
expressions hold for a natural number n2.
x x tan -1x n /2
-1
tan x dx = Σ
n
(-1)k nC n-2k x n-2k
0 0 n ! k =0
log1+ x 2 n /2
+ Σ (-1)k nC n+1-2k x n+1-2k
2n ! k =1
1 n /2
- Σ (-1)r nC n+1-2r (1+ n)- (2r)x n+1-2r
n! r=1
x x xn tan -1x n /2
(-1)k nC n-2k x n-2k
n! Σ
-1 -1n
cot x dx = cot x -
0 0 n! k =1
log1+ x 2 n /2
- Σ (-1)k nC n+1-2k x n+1-2k
2n ! k =1
1 n /2
+ Σ (-1)r nC n+1-2r (1+ n)- (2r)x n+1-2r
n r=1
!
Proof
x 1
tan -1x dx = xtan -1x - log1+ x 2
0 2
1 1
1C 1 xtan 1C 0 x log 1+ x
-1 0 2
= x-
1! 21!
tan
x x x2 1 x x
-1
x dx = 2
- tan -1x - log1+ x 2 +
0 0 2 2 2 2
1 1 1 1 x
2C 2 x -2C 0 tan x - 2C 1 x log 1+ x +
2 -1 1 2
=
2! 22! 1! 2 1!
tan
x x x x3 x x2 1 5x 2
-1
x dx =3
- tan -1x - - log1+ x +
2
0 0 0 6 2 4 12 12
1
3C 3 x -3C 1 x tan x
3 -1
=
3!
x2
1 1 1 1
3C 2 x -3C 0 x log 1+ x +
2 0 2
- +
23! 1! 2 3 2!
x4 x2 1 3
x x 3
x x 13x x
-1
tan x dx = 4
- + - log 1+ x 2 +
tan -1x - -
0 0 24 4 24
12 12 72 24
1 1
4C 4 x -4C 2 x +4C 0 tan x - 4C 3 x -4C 1 x log1+ x
4 2 -1 3 1 2
=
4! 24!
x3
1!
1 1 1 1 1 1 x
+ + + -
1! 2 3 4 3! 3! 4
- 15 -
x
5 3 4 2
x x x x x x 1 77x 4 3x 2
tan -1x dx 5 = - + -1
tan x - - + 2
log 1+ x + -
0 0 120 12 24 48 24 240 1440 80
1 1
5C 5 x -5C 3 x +5C 1 x tan x - 5C 4 x -5C 2 x +5C 0 x log1+ x
5 3 -1 4 2 0 2
=
5! 25!
x4 x2
1 1 1 1 1 1 1 1
+ + + + - +
1! 2 3 4 5 4! 3! 4 5 2!
Hereafter, by induction, we obtain the desired expression.
The desired expression for cot -1x is also obtained in a similar way.
-1 -1
Formula 4.4.2 : Higher Integrals of sin x , cos x
When , denote floor function and ceiling function respectively, the following expressions hold for
a natural number n .
n-2r
sin
x x n/2 x
-1
x dx = Σ
n
2 sin -1x
an a1 r=0 (2r)!! (n -2r)!
n-2r+1
n /2 n -2r+1 (s -1)!!2 x
+Σ Σ (-1) n -2r+1C s
s
2 (n -2r +1)!
1-x 2
r=1 s=0 (s +2r -1)!!
Where a1 = i 1.5088 , a 2 = 0 , a3 = -i 0.4758 , a4 = 0 ,
n-2r
x x xn n/2 x
cos x dx = -1
cos -1x - Σ
n
2 sin -1x
an a1 n! r=1 (2r)!! (n -2r)!
n-2r+1
n /2 n -2r+1 (s -1)!!2 x
- Σ Σ (-1) n -2r+1C ss
2 (n -2r +1)!
1-x 2
r=1 s=0 (s +2r -1)!!
Where a1 = 1 , a2 = 0 , a3 = a complex number , a4 = 0 ,
- 16 -
Proof
sin
x
-1
x dx = xsin -1x + 1-x 2
a1
x1 -1 1(-1)!!2 x 0
= sin x + 1-x 2
0!! 1!
2
1!!2 0!
sin
x x
-1 2 x2 1 3x
x dx = + sin -1x + 1-x 2
a2 a1 2 4 4
x2 x0 1(-1)!!2 10!!2 x1
= + sin -1x + - 1-x 2
0!! 2! 2!! 0!
2 2
1!!2 2!!2 1!
sin
x x x
-1 3 x3 x 11x 2 1
x dx = + sin -1x + + 1-x 2
a3 a2 a1 6 4 36 9
x3 x1
= + sin -1x
0!! 3! 2!! 1!
2 2
sin
x x x4 x2 1 25x
3
55 x
-1
x dx = 4
+ + sin -1x + + 1-x
2
a4 a1 24 8 64 288 576
x4 x2 x0
= + + sin -1x
0!! 4! 2!! 2! 4!! 0!
2 2 2
x3 x1
1(-1)!!2 30!!2 31!!2 12!!2 1(-1)!!2 10!!2
+ - + - + -
1!!2 2!!2 3!!2 4!!2 3! 3!!2 4!!2 1!
2
1-x
Hereafter, by induction, we obtain the desired expression.
The desired expression for cos -1x is also obtained in a similar way.
Note
Integration lower limits a1 , a2 , , an are the solutions of the transcendental equations which made
-1
the formal right-hand sides equal to 0. For example, in the case of sin x , a1 , a2 are the solutions
of the following transcendental equation respectively.
2 4
-1 2 x2 1 -1 3x
xsin x + 1-x = 0 , + sin x + 1-x 2 = 0
4
On the other hand, if the integration lower limit is replaced by a fixed lower limit 0 in this formula,
the following collateral high integrals are obtained. The difference among both is only the existence
of a constant-of-integration polynomial .
-1 -1
Formula 4.4.2' : Collateral Higher Integrals of sin x , cos x
When , denote floor function and ceiling function respectively, the following expressions hold for
a natural number n .
- 17 -
n-2r
x x n/2 x
sin x dx = Σ
-1 n
2 sin -1x
0 0 r=0 (2r)!! (n -2r)!
n-2r+1
n /2 n -2r+1 (s -1)!!2 x
+ Σ Σ (-1) n -2r+1C s
s
2 (n -2r +1)!
1-x 2
r=1 s=0 (s +2r -1)!!
n-2r+1
n /2 x
+Σ
r=1 (2r -1)!!2(n -2r +1)!
n-2r
x x xn n/2 x
-1
cos x dx = n
cos -1x - Σ 2 sin -1x
0 0 n! r=1 (2r)!! (n -2r)!
n-2r+1
n /2 n -2r+1 (s -1)!!2 x
-Σ Σ s
(-1) n -2r+1C s 2 (n -2r +1)!
1-x 2
r=1 s=0 (s +2r -1)!!
n-2r+1
n /2 x
+Σ
r=1 (2r -1)!!2(n -2r +1)!
-1
Example: Collaterral the 3rd order integral of cos x
-1 -1
Formula 4.4.3 : Higher Integrals of sec x , csc x
When , denote floor function and ceiling function respectively, the following expressions hold for
a natural number n2.
x x
-1 xn n
sec xdx = sec -1x
1 1 n!
n-2r-1
(n -1)/2 (2r -1)!! x
- Σ logx + x 2-1
r=0 (2r)!!(2r +1)! (n -2r -1)!
- 18 -
n-2r
s n -2rC s
(s -1)!!2 x
n /2 n -2r
+Σ Σ (-1) 2 x 2-1
r=1 s=0 2r +s (s +2r -1)!! (n -2r)!
x x xn
csc x dx = -1
csc -1x n
an a1 n !
n-2r-1
(n -1)/2 (2r -1)!! x
+ Σ logx + x 2-1
r=0 (2r)!!(2r +1)! (n -2r -1)!
n-2r
s n -2rC s
(s -1)!!2 x
n /2 n -2r
-Σ Σ (-1) 2 x 2-1
r=1 s=0 2r +s (s +2r -1)!! (n -2r)!
Where a1 , a2 , , an are all complex numbers.
Proof
sec
x
-1
xdx = x sec -1x - logx + x 2-1
1
x1 (-1)!! x 0
= -1
sec x - logx + x 2-1
1! 0!!1! 0!
x x x2 1
sec xdx =-1
sec -1x - x log x + x 2-1 +
2
x 2-1
1 1 2 2
x2 (-1)!! x 1 x 0 (-1)!!2
= -1
sec x - logx + x 2
-1 0!
+ x 2-1
2! 0!!1! 1! 21!!2
x x x x3 x2 1 5x
-1
sec xdx = 3
sec -1x - + logx + x 2-1 + x 2-1
1 1 1 6 2 12 12
x-
0!!1! 2! 2!!3! 0!
x3 (-1)!! x 1!! x 2 0
= sec -1 + logx + x 2-1
3!
x1 (-1)!!2 0!!2
+ - x 2-1
1! 21!!2 32!!2
x 4 3 2
x x x x 13x 1
sec -1x dx 4 = sec -1x - + logx + x 2-1 + + x 2-1
1 1 24 6 12 72 36
x4 (-1)!! x 3 1!! x 1
= sec -1x - + logx + x 2-1
4! 0!!1! 3! 2!!3! 1!
x2 (-1)!!2 20!!2 1!!2 x 0 (-1)!!2
+ - + + x 2-1
2! 21!!2 32!!2 43!!2 0! 43!!2
Hereafter, by induction, we obtain the desired expression.
The desired expression for csc -1x is also obtained in a similar way.
-1
Example: The 5th order integral of sec x
- 19 -
Since both sides overlapped exactly, the blue (Riemann-Liouville Integral) can not be seen.
On the other hand, if the integration lower limit is replaced by a fixed lower limit 1 in this formula, the following
collateral high integral is obtained. The difference among both is only the existence of a constant-of-integration
polynomial.
-1
Formula 4.4.3' : Collateral Higher Integral of csc x
When , denote floor function and ceiling function respectively, the following expressions hold for
a natural number n2.
x x
-1 n xn
csc x dx = csc -1x
1 1 n!
n-2r-1
(n -1)/2 (2r -1)!! x
+ Σ logx + x 2-1
r=0 (2r)!!(2r +1)! (n -2r -1)!
n-2r
s n -2rC s
(s -1)!!2 x
n /2 n -2r
-Σ Σ (-1) 2 (n -2r)! x
2
-1
r=1 s=0 2r +s (s +2r -1)!!
n -1
- Σ
2n! r=0
nC r(x -1)
r
- 20 -
4.5 Higher Integrals of Inverse Hyperbolic Functions
-1 -1
Formula 4.5.1 : Higher Integrals of tanh x , coth x
When , , (x) denote floor function, ceiling function and digamma function respectively, the following
expressions hold for a natural number n2.
x x tanh -1x n /2 n-2k
n! Σ
nC n-2k x
-1 n
tanh x dx =
0 0 k =0
1 n /2
nC n+1-2r (1+ n )- (2r)x
n+1-2r
n! Σ
-
r=1
x x xn tanh -1x n /2 n-2k
n
Σ nC n-2k x
-1 -1
coth x dx = coth x +
0 0 n! n ! k =1
log1-x 2 n /2 n+1-2k
+
2n ! k =1Σ nC n+1-2k x
1 n /2
nC n+1-2r (1+ n )- (2r)x
n+1-2r
- Σ
n! r=1
|x|<1
Proof
The integral up to the 4th order of tanh -1x are as follows.
x 1
tanh -1x dx = xtanh -1x + log1- x 2
0 2
2 2
x x 2
x 1 x x
-1 2 -1 2
tanh x dx = + tanh x + log 1- x -
0 0 2 2
6 2 4 12
x x x 3 2 2
x x x 1 5x
-1 3 -1 2
tanh x dx = + tanh x + + log 1- x -
0 0 0 12
x x 3
x x 1 x x 13x x
-1 4
+ log 1- x - -1 2
-
0 0 12 72 24
The absolute values of these coefficients are all consistent with the absolute values in Formula 4.4.1 .
Therefore, these coefficients can be easily obtained by changing the signs of the coefficients in Formula 4.4.1.
This is the same also for coth -1x .
-1
Example: The 3rd order integral of coth x
- 21 -
-1 -1
Formula 4.5.2 : Higher Integrals of sinh x , cosh x
When , denote floor function and ceiling function respectively, the following expressions hold for
a natural number n .
(-1)r x n-2r
sinh
x x n/2
-1
x dx = Σ n
2 sinh -1x
an a1 r=0 (2r)!! (n -2r)!
n-2r+1
n /2 n -2r+1
r+s (s -1)!!2 x
+Σ Σ (-1) n -2r+1C s 2 (n -2r +1)!
1+ x 2
r=1 s=0 (s +2r -1)!!
Where a1 = 1.5088 , a2 = 0 , a 3 = -0.4758 , a4 = 0 ,
n-2r
cosh
x x n/2 x
-1
x dx = Σ n
2 cosh -1x
1 1 r=0 (2r)!! (n -2r)!
n-2r+1
n /2 n -2r+1 (s -1)!!2 x
-Σ Σ s
(-1) n -2r+1C s 2 (n -2r +1)! x
2
-1
r=1 s=0 (s +2r -1)!!
Proof
The integral up to the 4th order of sinh -1x are as follows.
sinh
x
-1
x dx = xsinh -1x - 1+ x 2
a1
sinh
x x
-1 2 x2 1 3x
x dx = - sinh -1x - 1+ x 2
a2 a1 2 4 4
sinh 1+ x
x x x
-1 3 x3 x 11x 2 1
x dx = - sinh -1x - - 2
a3 a2 a1 6 4 36 9
-1
Example: The 4th order integral of cosh x
- 22 -
On the other hand, if the integration lower limit is replaced by a fixed lower limit 0 in this formula, the following
collateral high integral is obtained. The difference among both is only the existence of a constant-of-integration
polynomial.
-1
Formula 4.5.2' : Collateral Higher Integral of sinh x
When , denote floor function and ceiling function respectively, the following expressions hold for
a natural number n .
(-1)r x n-2r
sinh
x x n/2
-1
x dx = Σ
n
2 sinh -1x
0 0 r=0 (2r)!! (n -2r)!
n-2r+1
n /2 n -2r+1
r+s (s -1)!!2 x
+Σ Σ (-1) n -2r+1C s 2 (n -2r +1)!
1+ x 2
r=1 s=0 (s +2r -1)!!
n /2 (-1)r-1 x n-2r+1
+Σ
r=1 (2r -1)!!2(n -2r +1)!
-1 -1
Formula 4.5.3 : Higher Integrals of sech x , csch x
When , denote floor function and ceiling function respectively, the following expressions hold for
a natural number n2.
n-2r-1
x x xn (n -1)/2 (2r -1)!! x
sech x dx = sech x + Σ
-1-1 n
sin -1x
an a1 n! r=0 (2r)!!(2r +1)! (n -2r -1)!
n-2r
s n -2rC s
(s -1)!!2 x
n /2 n -2r
+ Σ Σ (-1) 2 (n -2r)!
1- x 2
r=1 s=0 2 r +s (s +2r -1)!!
Where a1 = a3 = a5 = = 0 , a2 , a4 , a6 , are complex numbers
n-2r-1
x x xn (n -1)/2 (-1)r(2r -1)!! x
csch x dx = csch x + Σ
-1
-1 n
sinh -1x
an a1 n ! r=0 (2r)!!(2r +1)! (n -2r -1)!
n-2r
r+s n -2rC s(s -1)!!2 x
n /2 n -2r
+Σ Σ (-1) 2 x 2+1
r=1 s=0 2r +s (s +2r -1)!! (n -2r)!
Where a1 = 0 , a 2 = 0.6079 , a 3 = 0 , a4 = 1.5539 ,
- 23 -
Proof
The integral up to the 4th order of sech -1x are as follows.
sech
x
-1
x dx = x sech -1x + sin -1x
a1
sech
x x
-1 2 x2 1 1
x dx = sech -1x + xsin -1x + 1- x 2 -
a2 a1 2 2 2
x x x x3
-1 3 x2 1 5x
sech x dx = sech -1x + + sin -1x + 1- x 2
a3 a2 a1 6 2 12 12
sech
x3 x
2
x x
-1 x4 13 x 1
x dx 4 = sech -1x + + sin -1x + + 1- x
2
a4 a1 24 6 12 72 36
The absolute values of these coefficients are all consistent with the absolute values in Formula 4.4.3.
Therefore, these coefficients can be easily obtained by changing the signs of the coefficients in Formula 4.4.3.
This is the same also for csch -1x .
On the other hand, if the integration lower limit is replaced by fixed lower limit 0 in this formula,
the following collateral high integrals are obtained. The difference among both is only the existence
of a constant-of-integration polynomial.
-1 -1
Formula 4.5.3' : Collateral Higher Integrals of sech x , csch x
When , denote floor function and ceiling function respectively, the following expressions hold for
a natural number n2.
n-2r-1
x x xn (n -1)/2 (2r -1)!! x
-1
sech x dx = sech x + Σ
n-1
sin -1x
0 0 n! r=0 (2r)!!(2r +1)! (n -2r -1)!
n-2r
s n -2rC s
(s -1)!!2 x
n /2 n -2r
+Σ Σ (-1) 2
1- x 2
r=1 s=0 2r +s (s +2r -1)!! (n -2r)!
n-2r
n /2 1 x
-Σ
r=1 2r(2r -1)!!2 (n -2r)!
n-2r-1
x x xn (n -1)/2 (-1)r(2r -1)!! x
-1
csch x dx = csch x + Σ
n-1
sinh -1x
0 0 n! r=0 (2r)!!(2r +1)! (n -2r -1)!
n-2r
r+s n -2rC s (s -1)!!2 x
n /2 n -2r
+Σ Σ (-1) 2 x 2+1
r=1 s=0 2r +s (s +2r -1)!! (n -2r)!
n-2r
n /2 (-1)r x
-Σ
r=1 2r(2r -1)!!2 (n -2r)!
-1
Example: Collateral the 4th order integral of sech x
- 24 -
Since both sides overlapped exactly, blue (Riemann-Liouville Integral) can not be seen.
This result also shows that Formula 4.5.3 is right.
- 25 -
4.6 Termwise Higher Integral and Taylor series of higher primitive function
4.6.1 Theorem of the Termwise Higher Integral with a fixed lower limit
Theorem 4.6.1
r =0, 1, , n
<r> <r +1>
Let f be continuous functions defined on [a , b] and f be the arbi-trary
<r>
primitive function of f . At this time, iff(x) can be expanded to the Taylor series around a ,
the following expressions hold for x [a , b] .
<n> n -1 (x -a)r
<n -r> (r) (x -a)n+ r
f (x) = Σf (a) + Σ f (a) (1.1)
r=0 r! r=0 (n + r)!
(x -a)r
= Σf <n -r>(a) (1.2)
r=0 r!
Proof
The following expression holds according to Theorem 4.1.3 .
f(x)dx
n -1 (x -a)r x x
f
<n>
(x) = Σf <n -r>
(a) + n
r=0 r! a a
On the other hand, if f(x ) can be expanded to the Taylor series around a ,
(r) (x -a)r
f(x) = Σf (a)
r=0 r!
This can be integrated termwise as follows.
f(x)dx
x x (r) (x -a)n+ r
n
=Σf (a)
a a r=0 (n + r)!
Substituting this for the above,
<n> n -1
<n -r> (x -a)r (r) (x -a)n+ r
f (x) = Σf (a) + Σ f (a) (1.1)
r=0 r! r=0 (n + r)!
Here,
(r) (x -a)n+ r
(r-n ) (x -a) r
<n -r> (x -a) r
Σ
r=0
f (a)
(n + r)!
=Σf
r=n
(a)
r!
=Σf
r=n
(a)
r!
Using this for (1.1),
<n> n -1 (x -a)r
<n -r>
<n -r> (x -a) r
f (x) = Σf (a) +Σf (a)
r=0 r! r=n r!
<n -r> (x -a)r
= Σf (a) (1.2)
r=0 r!
Remark
(1.1) shows that
<n>
f (x) consists of the termwise integral of f(x) and the constant-of-integration-polynomial.
<n>
(1.2) shows that these constitute the Taylor series of f (x ) .
Conclusion
(1) A termwise higher integral with a fixed lower limit is collateral generally.
- 26 -
(2) It is the following case that a termwise higher integral with a fixed lower limit is lineal.
for a -
<n>
(x)= e x , f <n -r>(a)= e a , f
(n)
f (a)= e a
Substituting these for the theorem,
n+ r
x <n> (x -a)r
n -1
a (x -a )
a (x -a )
r
e = Σe +Σe
a
= Σe
r=0 r! r=0 (n + r)! r=0 r!
The 1st term (constant-of-integration-polynomial) is not 0 for a - .
On the other hand, f(x)= e x is expanded to the Taylor series around a as follows.
(x -a)r
e = Σe
x a
r=0 r!
Integrating both sides of this with respect to x from 0 to x repeatedly, we obtain
x x (x -a)n+ r
e dx = Σ e
x n a
a a r=0 (n + r)!
x <n>
This termwise higher integral is a part of the Taylor series of e . Therefore, this is a collateral haigher
integral.
-1
Example 2: Termwise higher integral of f(x)= tan x
<n>
Let f (x) be the lineal higher primitive function of tan -1x . Then, a =0 is a zero of these functions
From this
n /2
(n)
f (0) = (-1)n(n -1)!Σ (-1)k nC n+1-2k 0n+1-2k
k =1
= (-1)r(2r)!
-1
Thus, the Taylor series around 0 of f(x)= tan x becoms as follows.
- 27 -
(2r)! 2r+1
tan -1x = Σ(-1)r x
r=0 (2r +1)!
And we obtain the following termwise higher integral.
tan
x x (2r)! n+2r+ 1
-1 n
x dx = Σ(-1)
r
x (w2)
0 0 r=0 (n +2r +1)!
Substituting (w1) and (w2) for the thorem,
<n> (2r)! n+2r+ 1
tan
-1
x = Σ(-1)r x
r=0 (n +2r +1)!
<n>
The termwise higher integral (w2) is all of Taylor series of tan x
-1
. Therefore, this is the lineal higher
integral.
Formula 4.6.2
k 1
When a harmonic number is denoted by Hk = Σ = (1+ k)+ , the following expressions hold.
s=1 s
(-1)r-1 2n 1 n -1
Σ
r=1 r(r +1)(r + n)
=
n!
log 2 -H n +
n! Σ
r=0
nC r Hn-r (2.3)
Proof
From Formula 4.3.3 , the lineal higher primitive function of f(x )=log x as follows.
r r 1
x
(log x)
<r>
= log x -Σ r =1, 2, ,n
r! s =1 s
From Formula 9.2.3 (later 9.2 ), the higher derivative function of this as follows.
(r -1)!
(log x) r = (-1)r-1 r =1, 2, 3,
()
xr
Substituting these for Thorem 4.6.1 (1.1) ,
xn
n 1
log x -Σ
n! s=1 s
n -1 n -r 1
=Σ log a -Σ +Σ(-1)
r=0 (n -r)! s=1 s r! r=0 ar (n + r) !
k 1
Substitute a =1 , H k = Σ for the right side, then
s=1 s
xn (-1)r-1(x -1)n+ r
=Σ
n 1 1 n -1
log x -Σ - Σ nC r H n-r(x -1)
r
(2.1)
n! s=1 s r=1 r(r +1)(r + n) n ! r=0
- 28 -
Substitute x =0 , x =2 for this respectively, then
0n (-1)r-1(-1)n+ r
n 1 1 n -1
log 0 -Σ =Σ - Σ nC r H n-r(-1)
r
n! s=1 s r=1 r(r +1)(r + n) n r=0
!
2n (-1)r-11n+ r
=Σ
n 1 1 n -1
log 2 -Σ - Σ nC r H n-r1
r
n! s=1 s r=1 r(r +1)(r + n) n ! r=0
From these, we obtain (2.2), (2.3) .
Example: n =3
1 1 1 1
+ + + +
1234 2345 3456 4567
(-1)2
2
3 1 1 3 1 3 1
= 1+ + - 1+ + =
3! 0 2 3 1 2 18
1 1 1 1
- + - +-
1234 2345 3456 4567
0
23 3 3 3
1 1 1 1 1 1
= log 2- 1+ + + 1+ + + 1+ +
3! 2 3 3! 2 3 1 2 2
4 8
= log 2 -
3 9
By-product
The following equations are known about (2.2) and (2.3).
1 1
Σ
r=1 r(r +1)(r + n )
=
n n !
(2.2')
r-1
(-1) 1 1 (1-t)n
Σ
r=1 r(r +1)(r + n)
=
n ! 0 1+ t
dt (2.3')
Therefore , the following expressions are derived from (2.2),(2.2') and (2.3),(2.3')
n-1
n -1 (-1)
Σ
r=0
(-1)rnC r Hn-r =
n Hr =(1+ r)+ (2.4)
1 (1- x)n n
dx = 2nlog 2- Hn+ΣnC r Hr Hr =(1+ r)+ (2.5)
0 1+ x r=1
(2.5) can be generalized further, and the following expression holds for p >0 .
p
1 (1- x) p
dx = 2plog 2- (1+p)-+Σ (1+r)+ (2.5')
0 1+ x r=1 r
Example1
- 29 -
Example2
2012.05.01 Renewal
K. Kono
Alien's Mathematics
- 30 -
05 Termwise Higher Integral (Trigonometric, Hyperbolic)
In this chapter, for the function which second or more order integral cannot be expressed with the elementary
functions among trigonometric functions and hyperbolic functions, we integrate the series expansion of these
function termwise. Therefore, sin x, cos x, sinhx, cosh x mentioned in " 4.3 Higher Integral of Elementary
Functions " are not treated here.
tan x dx
x
= -log cos x
0
tan x dx
x x
2
= non-elementary function
0 0
Formula 5.1.1
When B0=1, B2=1/6, B4=-1/30, B6=1/42, are Bernoulli Numbers, the following expressions
hold for |x|< /2 .
22k22k -1B2k 2k
tan x dx
x
=Σ x
0 k =1 2k (2k)!
22k22k -1B2k 2k+1
tan x dx
x x
2
=Σ x
0 0 k =1 2k (2k +1)!
22k22k -1B2k 2k+2
tan x dx
x x x
3
=Σ x
0 0 0 k=1 2k (2k +2)!
22k22k -1B2k 2k+ n-1
tan x dx
x x
n
=Σ x
0 0 k =1 2k (2k + n -1)!
Proof
tan x can be expanded to Taylor series as follows.
2
2 x
2
=Σ (1.2')
0 0 k=1 2k (2k +1)!
-1-
22k22k -1B2k
tan x dx
2 x x 2k+2
2
3
=Σ (1.3')
0 0 0 k=1 2k (2k +2)!
22k22k -1B2k
tan x dx
2 2k+3
2
x x
4
=Σ (1.4')
0 0 0 k =1 2k (2k +3)!
22k22k -1B2k
tan x dx
2 2k+ n-1
2
x x
n
=Σ (1.n')
0 0 0 k =1 2k (2k + n -1)!
2 2 -1B2k
2k 2k
x
tan x dx = Σ x 2k = -log cos x |x|<
0 k=1 2k (2k)! 2
From this, the following expression follows immediately.
2 2 -1B 2k
2k 2k
Σ
2k
x = -log cos x |x|< (1.t)
k =1 2k (2k)! 2
Substituting x =1 , 1/2 , /4 for this one by one, we obtain the following special values.
22k22k -1B2k
Σ
k =1 2k (2k)!
= -log cos 1
22k -1B2k 1
Σ
k =1 2k (2k)!
= -log cos
2
2 -1B2k
2k
2k
2
Σ
k =1 2k (2k)!
= -log cos
4
This is used in 5.3.1 later.
Formula 5.1.2
(-1)k-1
Let (x) =Σ be a Dirichlet Eta Function and be a ceiling function, then the following
k =1 kx
expressions hold for |x|< /2 .
1 (1) x 0
1 (-1)k-1
+
x
tan x dx = 0Σ sin 2k x -
0 2 k=1 k 1 2 20 0!
2 (1)
1 (-1)k-1 x1
x x
tan x dx 2
= 1Σ sin 2k x - + 0
0 0 2 k=1 k 2 2 2 1!
3 (1) x 2 (3) x 0
1 (-1)k-1
x x x
tan x dx = 2 Σ
3
3 sin 2k x - + 0 - 2
0 0 0 2 k=1 k 2 2 2! 2 0!
4 (1) x 3 (3) x 1
1 (-1)k-1
+
x x
tan x dx = 3 Σ
4
4 sin 2k x - -
0 0 2 k=1 k 2 20 3! 22 1!
-2-
n
tan x dx (-1)k-1
x x 1
n
= Σ sin 2k x -
0 0 2n-1 k=1 k n 2
n/2
k-1 (2k -1) x
n+1-2k
+ Σ (-1)
k =1 22k-2 (n +1-2k)!
Proof
tan x can be expanded to Fourier series in a broad meaning.as follows.
1 e ix - e -ix 1- e 2ix
tan x = =i = i1- e 2ix1 - e 2ix + e 4ix - e 6ix +-
i ix -ix
e +e 2ix
1+ e
= i - i2e 2ix - 2e 4ix + 2e 6ix - 2e 8ix +-
= 2(sin 2x - sin 4x + sin 6x - sin 8x +-)
-2i(cos 2x - cos 4x + cos 6x - cos 8x +-) + i (2.0)
y
40
30
20
10
-1.4 -1.2 -1.0 -0.8 -0.6 -0.4 -0.2 0.2 0.4 0.6 0.8 1.0 1.2 1.4
-10 x
-20
-30
-40
tan(x)
2*sin(2*x) - 2*sin(4*x) + 2*sin(6*x) - 2*sin(8*x)
tan x is the locus of the median (central line) of this real number part. (See the figure) However calculus of the
right side of (2.0) may be carried out, the real number does not turn into an imaginary number and the contrary
does not exist either. Therefore, limiting the range of tan x and the higher order integral to the real number, we
calculate only the real number part of the right side of (2.0) . Although the Riemann zeta (2n) is obtained
from calculation of the imaginary number part, since it is long, it is omitted in this section.
Now, integrate the real number part of both sides of (2.0) with respect to x from 0 to x , then
cos 2x cos 4x cos 6x cos 8x
=- - + - +-
1 2 3 4 0
cos 2x cos 4x cos 6x cos 8x 1 1 1 1
=- - + - +- + - + - +-
1 2 3 4 1 2 3 4
i.e.
k-1
x 1 k-1 cos 2k x (-1)
tan x dx = - 0 Σ(-1) +Σ
0 2 k=1 k1 k=1 k1
-3-
(-1)k-1
Here, let -cos x = sin x - , (x)=Σ , then we obtain
2 k=1 kx
(1) x 0
1 (-1)k-1
+
x
tan x dx = 0 Σ 1 sin 2k x -
0 2 k=1 k 2 20 0!
This is consistent with the real number part of the Fourier series of -log cos x .
Next, integrate both sides of this with respect to x from 0 to x , then
(1) x 0
tan x dx = dx
1 (-1)k-1
-
x x x
0Σ
2
1 sin 2kx -
0 0 0 2 k =1 k 2 20 0!
x
2 (1) x
1!
1 (-1)k-1 1
sin 2kx -
Σ 2
= -
21 k=1 k1 2 0
0
k-1
2 (1) x 1
1 (-1)
= Σ
21 k =1
sin 2kx - +
k1 2 20 1!
Next, integrate both sides of this with respect to x from 0 to x , then
tan x dx = 2 Σ
x
3 (1) x 2
x x x (-1)k-1
+
3 1
2 1
sin 2k x -
0 0 0 k=1 k 2 20 2!
0
k-1
3 (1) x 2
1 (-1) 1 cos 2k0
= Σ
22 k=1 k
1
sin 2k x -
2
+
20 2!
- 2Σ
2 k=1
(-1)k-1
k
3
2
2 x
2
= (2.2')
0 0 201!
(1) (3) (3)
2 x x 2 0
2 2
3
tan x dx = - - (2.3')
0 0 0 202! 220! 22
(1) (3)
tan x dx
2 3 1
2 2
x x
4
= - (2.4')
0 0 0 203! 221!
(1) (3) (5) (5)
2 4 2
2 2
x x
tan x dx = 5
- + + (2.5')
0 0 0 204! 222! 240! 24
(-1)k-1 (2k -1) (n) n
2 n+1-2k
2
x x n /2
tan x dx = Σ 6
+ sin (2.n')
0 0 0 k =1 22k-2(n +1-2k)! 2n-1 2
-4-
(2) Fourier Series of the1st order Integral of tan x
The1st order Integral of tan x is as follows.
x 1 k-1 cos 2k x log 2 0
tan x dx = - 0 (-1) Σ + x = -log cos x
0 2 k =1 k1 0!
From this, the following expression follows immediately.
k-1 cos 2k x
Σ (-1) = log(2cos x) |x|< (2.f)
k =1 k1 2
Substituting x =1 , 1/2 for this one by one, we obtain the following special values.
cos 2k
k-1
Σ (-1) = log(2cos 1)
k =1 k1
cosk
k-1
Σ (-1) = log2cos(1/2)
k =1 k1
Formula 5.1.3
When B2k , (x), (x) denote Bernoulli Numbers, Dirichlet Eta Function and Riemann Zeta Function
respectively, the following expressions hold.
2 -1B 2k
2k
2
(3) = -Σ 2k+2
+ (1) - (3)
k=1 2k (2k +2)! 2!
22k -1B2k 2k+3
3
1
(3) = - (1)
Σ
-
k=1 2k (2k +3)! 3!
22k -1B2k 2k+4 4
2
(5) = Σ - (1) + (3) - (5)
k=1 2k (2k +4)! 4! 2!
22k -1B2k 2k+ 5
5 3
1
(5) = (1)+ (3)
Σ
-
k=1 2k (2k +5)! 5! 3!
2 -1B2k 2k+ 6 6
2k
4 2
(7) = -Σ + (1)- (3)+ (5)- (7)
k=1 2k (2k +6)! 6! 4! 2!
22k -1B2k 2k+7 7
5 3
1
(7) = - (1)+ (3)- (5)
Σ
-
k=1 2k (2k +7)! 7! 5! 3!
2 -1B 2k 2k+2n
2k
(2n +1) = (-1) Σ
n
k=1 2k (2k +2n )!
n -1 2n-2k
- (-1) n
Σ (-1) k
(2k +1) - (2n +1)
k =0 (2n -2k)!
(-1)n 22k -1B2k 2k+2n +1
(2n +1) = Σ
k=1 2k (2k +2n +1)!
-5-
(-1)n n -1 2n+1-2k
- Σ (-1) k
(2k +1)
k=0 (2n +1-2k)!
Proof
(1) (3) (3)
tan x dx
2 x 2 0
2 2
x
3
= - - (2.3')
0 0 0 202! 220! 22
22k22k -1B2k
tan x dx
2 x 2k+2
2
x
3
=Σ (1.3')
0 0 0 k=1 2k (2k +2)!
from these
(1) 2
(3) 0
(3) 22k22k -1B2k 2k+2
2 - 2 0! 2 - 2
=Σ
202! 2
22 k =1 2k (2k +2)!
2 2
x x
4
= - (2.4')
0 0 0 203! 221!
22k22k -1B2k
2 2k+3
2
x x
tan x dx = Σ 4
(1.4')
0 0 0 k =1 2k (2k +3)!
from these
(1) 3
(3) 1 22k22k -1B2k 2k+3
2 2
203!
-
221!
=Σ
k=1 2k (2k +3)! 2
(1) (3) 22k -1B2k 2k+3
- 3
=Σ 1
3! 1! k =1 2k (2k +3)!
-6-
22k -1B2k 2k+3
22 1
3
(3) = - 2 Σ - log 2
2 -1 k=1 2k (2k +3)! 3!
24 22k -1B2k 2k+4 4 2 22-1
(5) = 5 Σ - log 2 + (3)
2 -1 k =1 2k (2k +4)! 4! 2! 22
2 -1B2k 2k+ 5 5
2k
24 1 3 22-1
(5) = 4 Σ - log 2 + (3)
2 -1 k=1 2k (2k +5)! 5! 3! 22
22k -1B2k 2k+ 6 6
26
(7) = - 7 Σ + log 2
2 -1 k =1 2k (2k +6)! 6!
4 22-1 2 24-1
26
+ 7 - (3)+ (5)
2 -1 4! 22 2! 24
22k -1B2k 2k+7
log 2
7
26 1
(7) = - 6 Σ -
2 -1 k=1 2k (2k +7)! 7!
5 22-1 3 24-1
26 1
+ 6 - (3) + (5)
2 -1 5! 22 3! 24
2 -1B 2k 2k+2n
2k
22n 2n
(2n +1) = (-1) 2n+1
n
Σ - log 2
2 -1 k=1 2k (2k +2n)! (2n)!
22n n -1 2n-2k 22k-1-1
- (-1) 2n+1 Σ(-1)
n k
(2k +1)
2 -1 k=1 (2n -2k)! 22k-1
2n+1
Σ
2 -1B2k
2k
(-1)n 22n
(2n +1) = 2k+2n+1 - log 2
22n-1 k=1 2k (2k +2n +1) ! (2n +1) !
(-1)n 22n n -1 2n+1-2k
22k -1
-
Σ(-1) (2n +1-2k) ! 22k (2k +1)
22n-1 k=1
k
-7-
5.2 Termwise Higher Integral of tanh x
Since both zeros of tanh x and the primitive function log cosh x are x =0 , the latter seems to be a lineal
primitive function with a fixed lower limit. Then, we assume zeros of the second or more order primitive function
are also x =0 .
tanh x dx
x
= log cosh x
0
tanh x dx
x x
2
= non-elementary function
0 0
Formula 5.2.1
When B0=1, B2=1/6, B4=-1/30, B6=1/42, are Bernoulli Numbers, the following expressions
hold for |x|< /2 .
22k22k -1B2k 2k
tanh x dx
x
=Σ x
0 k=1 2k (2k)!
22k22k -1B2k 2k+1
tanh x dx
x x
2
=Σ x
0 0 k=1 2k (2k +1)!
22k22k -1B2k 2k+2
tanh x dx
x x x
3
=Σ x
0 0 0 k=1 2k (2k +2)!
22k22k -1B2k 2k+ n-1
tanh x dx
x x
n
=Σ x
0 0 k=1 2k (2k + n -1)!
Proof
tanh x can be expanded to Taylor series as follows.
2 1
tanh x dx =Σ (1.1')
0 k=1 2k (2k)! 2
1
22k22k -1B2k
tanh x dx
2k+1
2 x 1
2
=Σ (1.2')
0 0 k=1 2k (2k +1)! 2
-8-
1
22k22k -1B2k
tanh x dx
2 x 2k+2
x 1
3
=Σ (1.3')
0 0 0 k =1 2k (2k +2)! 2
1
22k22k -1B2k
tanh x dx
2 2k+ n-1
x x 1
n
=Σ (1.n')
0 0 0 k=1 2k (2k + n -1)! 2
2 2 -1 B 2k
2k 2k
x
tanh xdx = Σ x 2k = log cosh x |x|<
0 k=1 2k (2k)! 2
From this, the following expression follows immediately.
2 2 -1 B 2k
2k 2k
Σ
2k
x = log cosh x |x|< (1.t)
k =1 2k (2k)! 2
Substituting x =1 , 1/2 , /4 for this one by one, we obtain the following special values.
22k22k -1B2k
Σ
k =1 2k (2k)!
= log cosh 1
22k -1B2k 1
Σ
k =1 2k (2k)!
= log cosh
2
2 -1B2k
2k
2k
2
Σ
k =1 2k (2k)!
= log cosh
4
This is used in 5.4.1 later.
Formula 5.2.2
Let (x) be a Dirichlet Eta Function, then the following expressions hold for x >0 .
-2k x
(1) x 0
tanh x dx
x 1 k-1 e x1
= Σ(-1)
20 k=1
+ -
0 k1 1! 20 0!
-2k x
(1) x 1 (2) x 0
x x 1 k-1 e x2
tanh x dx = - 1 Σ(-1)
2
+ - + 1
0 0 2 k=1 k2 2! 20 1! 2 0!
-2k x
x x x 1 k-1 e x3
tanh x dx = 2 Σ(-1) 3
+
0 0 0 2 k=1 k3 3!
(1) x 2 (2) x 1 (3) x 0
- + 1 -
20 2! 2 1! 220! 0!
(-1)n-1 -2k x
x x
k-1 e xn
tanh x dx n = Σ(-1) +
0 0 2n-1 k=1 kn n!
-9-
n -1 (k +1) x
n-1- k
- Σ(-1) k
k =0 2k (n -1- k)!
Proof
tanh x can be expanded to a Fourier series as follows. (See the following figure).
e x - e -x 1- e -2x
tanh x = x -x
= -2x
= 1- e -2x1 - e -2x + e -4x - e -6x +-
e +e 1+ e
= 1 - 2e -2x - 2e -4x + 2e -6x - 2e -8x +- (2.0)
y
1.0
0.8
0.6
0.4
0.2
0.0
0 1 2 3 4 5
x
tanh(x)
1 - 2*(exp(-2*x) - exp(-20002*x))/(exp(-2*x) + 1)
1 - 2e
x x
tanh xdx = -2x
- e -4x + e -6x - e -8x +- dx
0 0
x
e -2x e -4x e -6x e -8x
= x+ - + - +-
1 2 3 4 0
=x +
-2x -4x -6x -8x
e e e e 1 1 1 1
- + - +- - - + - +-
1 2 3 4 1 2 3 4
i.e.
x
(-1)k-1
x
k-1 e -2k
tanh x dx = Σ(-1) +x -Σ
0 k =1 k1 k=1 k1
(-1)k-1
Here, let (x)=Σ , then we obtain
k=1 kx
x
tanh x dx = Σ(-1)
x e -2k
k-1
+ x - (1)
0 k =1 k1
Of course, this is also the Fourier series expansion of log cosh x .
Next, integrate both sides of this with respect to x from 0 to x , then
- 10 -
-2k x
tanh x dx
x x x 1 k-1 e
2
= Σ(-1) + x - (1) dx
0 0 0 20 k =1 k1
x
-2k x
1 k-1 e x2 x1
= - 1 Σ(-1) + - (1)
2 k=1 k2 2! 1! 0
1 k-1 e
-2k x
x 2
(1) x 1 (2) x 0
=- Σ
21 k =1
(-1) + -
20 1!
+ 1
k2 2! 2 0!
Hereafter, in a similar way we obtein the desired expressions.
2 1
k-1
tanh x dx = Σ(-1) 1
+ - 0 (2.1')
0 k =1 k 1! 2 2 1! 2
1
-k
1 2
2 x 1 k-1 e 1
tanh x dx = - 1 Σ(-1)
2
+
0 0 2 k=1 k 2 2! 2
(1) 1 1 (2) 0
1
- 0 + 1 (2.2')
2 1! 2 2 0! 2
1
2 x -k 3
1
x 1 1
k-1 e
tanh x dx = 2 Σ(-1)
3
3
+
0 0 0 2 k=1 k 3! 2
(1) 1 2 (2) 1 1
(3) 0
- 2 0!
1
- 0 + 1 2 (2.3')
2 2! 2 2 1! 2 2
1
(-1)n-1 n
2 e -k
x x 1 1
tanh x dx = n
Σ(-1)
+k-1
0 0 0 2n-1 k=1 kn n! 2
n -1 (-1)k (k +1) 1 n-1- k
-Σ
k =0 2k (n -1- k)! 2 (2.n')
x
k-1 e
tanh x dx = (-1) Σ + x - log 2 = log cosh x x >0
0 k =1 k1
From this, the following expression follows immediately.
-2k x
k-1 e
Σ (-1) 1
= log(2cosh x) - x x >0 (2.f)
k =1 k
Substituting x =1 , 1/2 , /2 for this one by one, we obtain the following special values.
k-1 e -2k
Σ (-1) = log(2cosh 1) - 1
k =1 k1
- 11 -
e -k
1 1
k-1
Σ (-1) = log 2cosh -
k =1 k1 2 2
e -k
k-1
Σ(-1) = log 2cosh -
k =1 k1 2 2
Formula 5.2.3
When B2k , (x) denote Bernoulli Numbers and Dirichlet Eta Function, the following expressions hold.
k-1 -k 22k -1B2k 1 1
0 = Σ(-1) e +Σ -
k=1 k=1 2k (2k -1)! 2 0!
2 -1 B 2k
2k
e -k 1 1
(1) = Σ(-1) k-1
1
-Σ +
k=1 k k=1 2k (2k)! 2 1!
2 -1 B 2k
2k
e -k 1 1 1
(2) = Σ(-1) k-1
2
+Σ - + (1)
k=1 k k =1 2k (2k +1)! 2 2! 1!
2 -1 B 2k
2k
e -k 1 1 1 1
(3) = Σ(-1) k-1
-Σ + - (1)+ (2)
k=1 k3 k =1 2k (2k +2)! 2 3! 2! 1!
2 -1 B 2k
2k
e -k 1 1
(4) = Σ(-1)k-1 4
+Σ -
k=1 k k =1 2k (2k +3)! 2 4!
1 1 1
+ (1) - (2) + (3)
3! 2! 1!
e -k 22k
-1B2k (-1)n 1
(n) = Σ(-1) k-1
+ (-1) Σ
n
-
k=1 kn k=1 2k (2k + n -1)! 2 n!
n -1 (-1)k
-Σ (n -k)
k=1 k!
Proof
1
tanh = 1 - 2e -1 - e -2 + e -3 - e -4 +- (2.0')
2
22k22k -1B2k 2k-1
1 1
tanh =Σ (1.0')
2 k=1 2k (2k -1)! 2
22k -1B2k
1
0 = Σ(-1) k-1 -k
e +Σ -
k =1 k =1 2k (2k -1)! 2
1
2 e -k 1
tanh x dx = Σ(-1) k-1
+ - (1) (2.1')
0 k =1 k1 2
- 12 -
1
22k22k -1B2k
2k
2 1
tanh x dx =Σ (1.1')
0 k =1 2k (2k)! 2
2 -1 B 2k
2k
e -k 1
(1) = Σ(-1) k-1
-Σ +
k=1 k1 k=1 2k (2k)! 2
1
-k
1 2
2 x 1 k-1 e 1
tanh x dx = - 1 Σ(-1)
2
+
0 0 2 k=1 k 2 2! 2
(1) 1 1 (2) 0
1
- 0 + 1 (2.2')
2 1! 2 2 0! 2
1
22k22k -1B2k
tanh x dx
2k+1
2 x 1
2
=Σ (1.2')
0 0 k=1 2k (2k +1)! 2
2 -1 B 2k
2k
e -k 1 1 1
(2) = Σ(-1) k-1
- + (1)+Σ
k=1 k2 2 2! 1! k =1 2k (2k +1)!
Formula 5.2.4
Σ(-1)
2 -1 B 2k
2k
21 e -k 1 log 2
(2) = 1 k-1
+ Σ - +
2 -1 k=1 k2 k=1 2k (2k +1)! 22! 1!
2 -1 B 2k
2k
22 e -k 1 log 2
(3) = 2 Σ (-1) k-1
- Σ + -
2 -1 k=1 k3 k=1 2k (2k +2)! 23! 2!
2 -1 (2)
1
+
21 1!
2 -1 B 2k
2k
23 e -k 1 log 2
(4) = 3 Σ (-1) k-1
4
+Σ - +
2 -1 k=1 k k=1 2k (2k +3)! 24! 3!
22-1 (3) 21-1 (2)
+
22 1!
-
21 2!
-1 (n -j)
Σ(-1)
2n-1 e -k n -2
j-1 2
n-1- j
(n) = k-1
+ Σ (-1)
2n-1-1 k=1 kn j=1 2n-1- j j!
2 -1
(-2) n-1
22k -1B2k 1 log 2
- Σn-1 k =1
-
2k (2k + n -1)! 2n !
+
(n -1)!
- 13 -
5.3 Termwise Higher Integral of cot x
Since both zeros of cot x and the primitive function log sin x are x = /2 , the latter seems to be a lineal
primitive function with a fixed lower limit. Then, we assume zeros of the second or more order primitive function
are also x = /2 .
cot x dx
x
= log sin x
2
cotx dx
x x
2
= non-elementary function
2 2
Formula 5.3.1
(x) on 0< x <
<n>
When B2k denote Bernoulli Numbers and functions f are as follows.
xn 22kB2k 2k+n
-Σ
n 1
n =1, 2, 3,
<n>
f (x)= log x -Σ x
n! k =1 k k=1 2k (2k +n)!
the following expressions hold .
22kB2k 2k
cot x dx
x
= f<0>(x) = log x -Σ x
k=1 2k (2k)!
2
cot x dx
0
2
x x 1
2 <1>
=f (x) - x- f<1>
0! 2
2 2
x x x 0 1
2 2
1 1
cot x dx 3 = f <2>(x) - x- f
<2>
- x- f
<1>
0! 2 1! 2
2 2 2
x x 0 1
2 f 2
1 1
4
cot x dx = f <3>(x) - 0! x - 2 f
<3>
- x-
<2>
1! 2
2 2
2
2
1 <1>
- x- f
2! 2
cot x dx
x x k
2 n =2, 3, 4,
<n -1>
n -2 1 n -1-k >
n
=f (x) -Σ x- f<
k=0 k! 2
2 2
where f = 0 , f = - log 2
<0> <1>
2 2 2
Proof
22k B2k 2k
2k
2 B 2k
x cot x = 1+Σ(-1) x = 1+Σ k
x 2k
k =1 (2k)! k=1 (2k)!
- 14 -
From this
2k
1 2 B2k 2k-1
cot x = -Σ x
x k =1 2k (2k -1)!
Integrate the both sides with respect to x from /2 to x, then
22kB2k 2k
2k
cot x dx = log x -Σ
2 B 2k
2k
2
x
2 Σ
x - log -
k=1 2k (2k)! k=1 2k (2k)!
2
Here
22kB2k 2k
2
log -Σ =0
2 k =1 2k (2k)!
Because, the following equation is known ( 岩波数学公式Ⅱ p151)
2k
2 B 2k
log x -Σ 2k
x = log sin x 0< x <
k =1 2k (2k)!
Then, substituting x = /2 for this, we obtain the above.
Thus,
22kB2k 2k
cot x dx = log x -Σ
x
x = f<0>(x)
k=1 2k (2k)!
2
Next, integrate the both sides with respect to x from /2 to x, then
f 2
x x x
cot x dx 2 = <0>
(x)dx = f<1>(x) - f<1>
2 2 2
Furthermore, integrate the both sides with respect to x from /2 to x , then
f
x
2 dx
x x x
cot x dx 3 = <1>
(x) - f<1>
2 2 2 2
1
2 f 2
<2> <2> 1 <1>
=f (x) - f - x-
1! 2
Hereafter, in a similar way we obtein the desired expressions.
Although Formula 5.3.1 is a general expression anyway, if it is expand actually, it is very complicated.
Then, this was devised and was made the same easy expression as tan x. It is as follows.
Formula 5.3.1'
When B2k denote Bernoulli Numbers, the following expressions hold for 0< x < .
2 2 -1B 2k
2k 2k
2k+ n-1
x
x
cotx dx = -Σn
x-
k =1 2k (2k + n -1)! 2
2 2
Proof
cot x = -tan x - 2
- 15 -
22k22k -1B2k 2k-1
tan x = Σ x |x|<
k =1 2k (2k -1)! 2
From these
22k22k -1B2k 2k-1
cotx = -Σ x- 0< x <
k =1 2k (2k -1)! 2
Integratingboth sides of this with respect to x from /2 to x repeatedly, we obtain the desired expressions.
0
2 2
0 x 1
2
cot x dx = - f<1> (1.2')
0!
2 2
0 x 0 1
2 2 2 2
x 1 1
3 <2>
cot x dx = - f + f<1> (1.3')
0! 1!
2 2 2
0
1
2 2 2 2
0 x x 1 1
<3>
cot x dx = -
4
f + f<2>
0! 1!
2 2 2 2
2 f 2
1 <1>
- (1.4')
2!
cot x dx (-1)k+1
0 x k
2 f 2
x n -2
<n -1-k >
n
=Σ (1.n')
k=0 k!
2 2 2
cot x dx
x 2 B 2k
= log x -Σ x 2k = log sin x 0< x <
k =1 2k (2k)!
2
From this, the following expression follows immediately.
2k
2 B 2k
Σ x 2k
= -log sin x + log x 0< x < (1.t)
k =1 2k (2k)!
Substituting x =1 , 1/2 , /4 for this one by one, we obtain the following special values.
2k
2 B2k
Σ
k =1 2k (2k)!
= -log sin 1
B2k 1
Σ
k =1 2k (2k)!
= -log sin
2
- log 2
B2k 2k
2
Σ
k =1 2k (2k)!
= -log sin
4
+ log
4
- 16 -
Moreover, from the last expressions of this and 5.1.1 (2) , the next expression follows
2 -2B 2k
2k
2k
Σ
k =1 2k (2k)! 2
= -log This is used in 5.5.1 later.
4
Formula 5.3.2
Let (x) be Dirichlet Eta Function and be the ceiling function, then the following expressions hold
- 2 0! x - 2
x 1 1
= 0 Σ 1 sin 2kx - 0
2 k=1 k 2
2
2 (1)
cot x dx
1
- 2 1! x - 2
x x 1 1
2
= 1 Σ 2 sin 2kx - 0
2 k=1 k 2
2 2
- 202! x - 2 + 220! x - 2
1 1
Σ
3
= sin 2kx -
22 k =1 k 3 2
2 2 2
4 (1) (3)
x x 3 1
1 1
Σ
4
cot x dx = sin 2kx - - x- + x-
23 k =1 k
4 2 203! 2 221! 2
2 2
n
cot x dx
x x 1 1
n-1 Σ
n
= n sin
2kx -
2 k =1 k 2
2 2
n/2
+Σ 2k-2 (n +1-2k)!
x-
k =1 2 2
Proof
cot x can be expanded to Fourier series in a broad meaning.as follows.
e ix + e -ix 1+ e 2ix
cotx = i ix -ix
= -i 2ix
= -i1 + e 2ix1 + e 2ix + e 4ix + e 6ix +
e -e 1- e
= -i - i2e 2ix + 2e 4ix + 2e 6ix + 2e 8ix +
= 2(sin 2x + sin 4x + sin 6x + sin 8x +)
-2i(cos 2x + cos 4x + cos 6x + cos 8x +) - i (2.0)
cot x is the locus of the median (central line) of this real number part. (See the figure of the following page.)
However calculus of the right side of (2.0) may be carried out, the real number does not turn into an imaginary
number and the contrary does not exist either. Therefore, limiting the range of cot x and the higher order
integral to the real number, we calculate only the real number part of the right side of (2.0).
Although the Riemann zeta (2n) is obtained from the calculation of the imaginary number part, since it is
long, it is omitted in this section.
- 17 -
y
40
30
20
10
0
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0
-10 x
-20
-30
-40
cot(x)
2*sin(2*x) + 2*sin(4*x) + 2*sin(6*x) + 2*sin(8*x)
Now, integrate the real number part of both sides of (2.0) with respect to x from /2 to x , then
2 2
x
cos 2x cos 4x cos 6x cos 8x
=- + + + +
1 2 3 4
2
cos 2x cos 4x cos 6x cos 8x 1 1 1 1
=- + + + + - - + - +-
1 2 3 4 1 2 3 4
i.e.
k-1
x 1 cos 2k x (-1)
cot x dx = log sin x = - 0 Σ -Σ
2 k=1 k 1 k=1 k1
2
(-1)k-1
Here, let -cos x = sin x - , (x)=Σ , then we obtain
2 k=1 kx
1 (1)
0
- 2 0! x - 2
x 1 1
cot x dx = 0 Σ 1 sin 2kx - 0
2 k =1 k 2
2
This is consistent with the real number part of the Fourier series of log sin x .
Next, integrate both sides of this with respect to x from /2 to x , then
1 (1)
cot x dx = - 2 0! x - 2 dx
0
x x x 1 1
0 Σ 1 sin
2
2kx - 0
2 k=1 k 2
2 2 2
x
2 (1)
- 2 1! x - 2
1
1 1
1 Σ 2 sin
= 2kx - 0
2 k=1 k 2
2
2 (1) 1
- 2 1! x - 2
1 1
= 1 Σ 2 sin 2kx - 0
2 k =1 k 2
- 18 -
Next, integrate both sides of this with respectto x from /2 to x , then
-
x x
x - 2
1 1
Σ
3
1 2
sin 2kx -
k =1 k 2 201!
2 2 2 2
x
3 (1)
2 Σ
2
-
1
x - 2
1
= 2 3
sin 2kx -
k =1 k 2 202!
2
3 (1) 2 cos 2k
1 1
1
=
22 Σ
k=1 k
3
sin 2kx -
2
-
2 2!0
x-
2
-
22 Σ
k=1 k
3
3 (1) 2
(3) 0
1
1
=
22
Σ
k=1 k
3
sin 2kx -
2
-
202!
x-
2
+
222!
x-
2
Hereafter, in a similar way we obtain the desired expressions.
(1)
cot x dx
1
2
0 x
2
= (2.2')
201!
2 2
2 2
x x
3
=- 0
+ 2
+ (2.3')
2 2! 2 0! 22
2 2 2
(1) (3)
cot x dx
0 x 3 1
2 2
x
4
= - (2.4')
203! 221!
2 2 2
2 2
x
5
=- 0
+ 2
- 4
- (2.5')
2 4! 2 2! 2 0! 24
2 2 2
(-1)k-1 (2k -1) (n) n
cot x dx
n+1-2k
(n +1-2k) ! 2
0 x x n /2
n
= (-1) n
Σ
k=1 22k-2
-
2 n-1
sin
2
2 2 2
(2.n')
cotx dx = - 2
x 1 cos 2k x
0 Σ - log 2 = log sin x 0< x <
0 k =1 k1
From this, the following expression follows immediately.
cos 2k x
Σ = -log(2sin x) 0< x < (2.f)
k =1 k1
Substituting x =1 , 1/2 for this one by one, we obtain the following special values.
cos 2k
Σ = -log(2sin 1)
k =1 k1
- 19 -
cosk
Σ = -log2sin(1/2)
k =1 k1
Formula 5.3.3
When B2k , (x), (x) denote Bernoulli Numbers, Dirichlet Eta Function and Riemann Zeta Function
respectively, the following expressions hold.
B2k
log = 1 +Σ 2k
2k (2k +1)!
k=1
2 B2k
2 1
(3) = - log -Σ +Σ 2k+2 - (3)
2! k =1 k k=1 2k (2k +2)!
1
B2k
3
3 1
(3) = - log -Σ -Σ 2k+3
3! k=1 k k =1 2k (2k +3 )!
4
B2k 2
4 1
(5) = log -Σ -Σ 2k+4
+ (3) - (5)
4! k =1 k k=1 2k (2k +4)! 2!
1 5 3
B2k
5 1
(5) = log -Σ -Σ 2k+5
+ (3)
5! k=1 k k =1 2k (2k +5)! 3!
6 B2k 4 2
log -Σ +Σ 2k (2k +6)!
6 1
(7)= - 2k+6
- (3)+ (5) - (7)
6! k=1 k k=1 4! 2!
7 5 3
7! log -Σ -Σ 2k (2k +7)!
1 7 1 B2k
(7)= - 2k+7
+ (3)- (5)
k =1 k k =1 5! 3!
2n
(2n)!
B2k
2n 1
(2n +1) = (-1) n
log -Σ -Σ 2k+2n
k =1 k k =1 2k (2k +2n ) !
n -1 2n-2k
+ (-1)nΣ(-1)k-1 (2k +1) - (2n +1)
k=1 (2n -2k) !
2n+1
(2n +1) !
B2k
1 2n +1 1
(2n +1) = (-1) n
log - Σ -Σ 2k+2n+1
k=1 k k=1 2k (2k +2n +1) !
1 n -1 2n-2k+1
+ (-1) n
Σ (-1) k-1
(2k +1)
k=1 (2n -2k +1) !
Proof
From Fourier series (2.2') and Taylor series (1.2')
log 2 1 0 <1>
2
1
0
= - f (2.2")
2 1! 2 0! 2
1
22kB2k 2k+1
2 log 2 -1 -Σ 2
1
=-
1! k =1 2k (2k +1)!
- 20 -
Hence we obtain
B2k
log = 1 + Σ 2k
k =1 2k (2k +1)!
Next, substitute (2.2") for Taylor series (1.3') , then
cot x dx
0 1
2 2 2
x x 1
3 <2>
= -f + f<1>
1!
2 2 2
2
2 2
<2> log 2
= -f -
1!
From this and Fourier series (2.3') , we obtain
2
1 (3) 2
2 2
1 log 2 log 2
3 -
2 ( )
+ 2 = -f<2> -
2 2! 2 0! 2 1!
1 (3) <2> 2
2
1 log 2
i.e. (3) + = -f - (a)
22 22 0! 2 2!
From (a), it is as follows.
2
(3) = -2 f 2 - log 2 - (3)
2 <2>
2!
2 log 2 -Σ -Σ
2
22kB2k 2k+2
2
1 2 1
2
= -2
2! k=1 k k=1 2k (2k +2)!
2
- log 2 - (3)
2!
2 B2k
2 1
=- log -Σ +Σ 2k+2 - (3)
2! k =1 k k=1 2k k
(2 +2)!
Next, substitute (2.2") and (a) for Taylor series (1.4') one by one, then
2 + 1! 2
f 2 f 2
4 1 1 <1>
= -f -
2 2!
2 2 2
1
1 (3)
2 2
1
= -f <3>
- (3) +
22 22 0!
From this and Fourier series (2.4') , we obtain
3
1 (3) 1
1
1 (3)
2 - 2 2
log 2
2 2
1
- 2 = -f
<3>
2
(3)+
3! 2 1! 22 0!
1
3
2 2 2
1 log 2
i.e. (3) = -f<3> - (b)
22 3!
Substitute f<3> 2 for (b), then
1
3
2 (3) = -f 2 2
1 <3> log 2
-
22 3!
3
22kB2k 2k+3
log 2 3
2 2 2
1 3 1
=- log -Σ +Σ -
3! 2 k =1 k k =1 2k (2k +3) ! 3!
- 21 -
3 22kB2k 2k+3
2 log -Σ +Σ 2
1 3 1
=-
3! k =1 k k=1 2k (2k +3)!
From this, we obtain
2
-1
3 22kB2k 2k+3
2 2
1 1
3
(3) = -2 2
log -Σ +Σ
3! k=1 k k=1 2k (2k +3)!
3
3!
B2k
1 3 1
=- log -Σ -Σ 2k+3
k =1 k k=1 2k k(2 +3)!
Hereafter, in a similar way we obtein the desired expressions.
Formula 5.3.4
2
2! 2k (2k +2)!
22 2 1 B2k
( )
3 = - log -Σ -Σ 2k+2
23-1 k =1 k k=1
4!
B
(3)
24 4 2
k
1 4
(5) =
2k 2k+4
25-1
log - Σ - Σ 2k (2k +4)!
k =1
+
2! k =1
6!
6
26 B
k
1 6
(7)= - 7
2k 2k+6
2 -1
log -Σ -Σ 2k (2k +6)!
k=1 k =1
4! (5)
4 2
26
- 7 (3) -
2 -1 2!
2n
-1 (2n) !
22n B2k
2n 1
(2n +1) = (-1) n
2n+1
log -Σ -Σ 2k+2n
2 k=1 k k=1 2k (2k +2n ) !
22n n -1 2n-2k
+ (-1)n Σ(-1)k-1 (2k +1) n =2, 3, 4,
22n+1 -1 k=1 (2n -2k) !
- 22 -
5.4 Termwise Higher Integral of coth x
Since the zero of the first order integral coth x is x = sinh -11 = 0.8813 ,
log sinh x of
we assume zeros of the second or more order integral of coth x are also x =0.8813 .
Where, since the zero of coth x is not 0.8813 , these seem to be a collateral higher integral.
coth x dx
x
= log sinh x
2
coth x dx
x x
2
= non-elementary function
2 2
Formula 5.4.1
Let = 2sinh -11 (= 1.762747174) and B0=1, B2=1/6, B4=-1/30, B6=1/42,
<n>
are Bernoulli Numbers and functions (x) on x >0 be as follows.
f
xn 22k B2k
n 1
n =1, 2, 3,
<n>
f (x)= log x -Σ +Σ x 2k+n
n! k =1 k k=1 2k (2 k +n ) !
Then the following expressions hold for x >0 .
22k B2k 2k
coth x dx
x
=f <0>
(x) = log x +Σ x
k=1 2k (2k)!
2
coth x dx
0
2
x x 1
2 <1>
=f (x) - x- f<1>
0! 2
2 2
x x x 0 1
2 2
3 1 1
coth x dx = f <2>(x) - x- f
<2>
- x- f
<1>
0! 2 1! 2
2 2 2
coth x dx
x x 0 1
f 2 f 2
4 <3> 1 <3> 1 <2>
=f (x) - x- - x-
0! 2 1! 2
2 2
2
2
1 <1>
- x- f
2! 2
coth x dx
x x k
f 2
<n -1>
n -2 1 <n -1-k >
n
=f (x) -Σ x-
k=0 k! 2
2 2
Proof
22k B2k 2k
x coth x = 1+Σ x
k =1 (2k)!
- 23 -
From this
1 22k B2k
coth x = +Σ x 2k-1
x k=1 2k (2k -1)!
Integrate the both sides with respect to x from /2 to x , then
22k B2k 2k 22k B2k
coth x dx = log x +Σ
2k
2
x
2 Σ
x - log +
k=1 2k (2k)! k=1 2k (2k)!
2
Here
22k B2k 2k
2
log +Σ =0
2 k =1 2k (2k)!
Because, the following equation is known ( 岩波数学公式Ⅱ p150)
2k-1
2 B2k 2k
log x +Σ x = log sinh x x >0
k=1 k (2k)!
Then, substituting x = /2 for this, we obtain the above.
Thus,
22k B2k 2k
coth x dx = log x +Σ
x
x = f<0>(x)
k=1 2k (2k)!
2
Next, integrate the both sides with respect to x from /2 to x , then
2
x x x
coth x dx 2 = f<0>(x)dx = f<1>(x) - f<1>
2 2 2
Furthermore, integrate the both sides with respect to x from /2 to x , then
f
x
2 dx
x x x
coth x dx 3 = <1>
(x) - f<1>
2 2 2 2
1
2 f 2
<2> <2> 1 <1>
=f (x) - f - x-
1! 2
Hereafter, in a similar way we obtein the desired expressions.
0
2 2
0 x 1
2
coth x dx =- f<1> (1.2')
0!
2 2
0 0 1
2 2 2 2
x x 1 1
3 <2>
coth x dx = - f + f<1> (1.3')
0! 1!
2 2 2
- 24 -
0 1
2 f 2 2 f 2
0 x x 1 1
coth x dx = - 4 <3> <2>
+
0! 1!
2 2 2
2
2 2
1
- f<1> (1.4')
2!
(-1)k
0 x k
2 2
x n -2
coth x dx = -Σ
n n -1-k >
f< (1.n')
k=0 k!
2 2 2
22k B2k 2k
Σ x = log sinh x - log x 0< x < (1.t)
k =1 2k (2k)!
Substituting x =1 , 1/2 , /4 for this one by one, we obtain the following special values.
22k B2k
Σ
k =1 2k (2k)!
= log sinh 1
B2k 1
Σ
k =1 2k (2k)!
= log sinh
2
+ log 2
B2k 2k
2
Σ
k =1 2k (2k)!
= log sinh
4
- log
4
Moreover, from the last expressions of this and 5.2.1 (2) , the next expression follows
2 -2 B 2k
2k
2k
Σ
k =1 2k (2k)! 2 = log coth
4
+ log
4
Formula 5.4.2
Let = 2sinh -11 = 1.762747174, then the following expressions hold for x >0 .
1 e -2k x
+
x
coth x dx = log sinh x = - 0 Σ 1 + x - - log 2
2 k=1 k 2 2
2
1 e -2k x /2-log 2
coth x dx
2
x x 1
2
= 1Σ 2 + x- + x-
2 k=1 k 2! 2 1! 2
2 2
1 e -k
- 1Σ 2
2 k=1 k
- 25 -
/2-log 2
x 1 e -2k x 1 3 2
+ x - 2
x x
coth x dx = - 2 Σ 3 +
3
x-
2 k =1 k 3! 2 2!
2 2 2
1 e -k 1 e -k
- 1 Σ 2 x-
2 k=1 k 2 + 2Σ 3
2 k=1 k
/2-log 2
x 1 e -2k x 1 4 3
x
coth x dx = 3 Σ 4 + 4
x- + x-
2 k=1 k 4! 2 3! 2
2 2
1 e -k 1 2
1 e -k 1 e -k
- Σ
21 k=1 k 2 2!
x -
2 + Σ
22 k=1 k 3
x -
2 - Σ
23 k=1 k 4
/2-log 2
coth x dx
x n-1
x (-1)n e -2k n
+ x - 2
x 1
n
= Σ + x -
2n-1 k =1 kn n! 2 (n -1)!
2 2
(-1)n-j 1 j
e -k
Σ k
n -2
-Σ x -
j=0 2n-1- j j ! 2 k=1
n-j
Proof
coth x can be expanded to a Fourier series as follows. (See the following figure).
e x + e -x 1+ e -2x
coth x = x -x
= -2x
= 1+ e -2x1 + e -2x + e -4x + e -6x +
e -e 1- e
= 1 + 2e -2x + 2e -4x + 2e -6x + 2e -8x +- (2.0)
y
60
50
40
30
20
10
0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7
x
coth(x)
2*exp(-2*x) + 2*exp(-4*x) + 2*exp(-6*x) + 2*exp(-8
Now, let /2 =sinh -11 (= 0.881373587) . Integrate the both sides of (2.0) with respect to x
from /2 to x , then
- 26 -
2e
x x
coth x dx = -2x
+ e -4x+ e -6x+ e -8x+ + 1dx
2 2
x
e -2x e -4x e -6x e -8x
=-
1
+
2
+
3
+
4
+ - x -
2
2
-2x -4x -6x -8x
e e e e
=- + + + + - x -
1 2 3 4 2
e- e -2 e -3 e -4
+ + + + +
1 2 3 4
i.e.
1 e -2k x
e -k
+Σ
x
coth x dx = - 0 Σ 1 + x -
2 k=1 k 2 k=1 k1
2
Here, the following equation holds,
-k
e
Σ = - log 2 (w)
k =1 k1 2
Because,
xk
log(1-x) = -Σ -1 x < 1
k =1 k
From this,
e -2kx
Σ = -log1-e -2x x >0
k =1 k
On the other hand,
e x- e -x e x1- e -2x
1 = sech x = =
2 2
From this,
1 ex 1 ex
= log = log -log1-e -2x = x - log 2
1- e -2x 2 1- e -2x 2
Then,
e -2kx
Σ = -log1-e -2x = x - log 2
k =1 k
Substituting x = sech -11 = for this, we obtain (w).
2
Thus, using (w), we obtain
1 e -2k x
x
coth x dx = log sinh x = - 0 Σ 1 + x - + - log 2
2 k=1 k 2 2
2
This is consistent with the real number part of the Fourier series of log sinh x .
- 27 -
-Σ dx
x
coth x dx
e -2k
2
x x x
2
= 1
+ x- + - log 2
k=1 k 2
2 2 2
x
/2-log 2
x - 2
1 e -2k x 2
1
= 1 Σ
2 k =1 k 2
+
2!
x-
2
+
1!
2
1 e -2k x 2
/2-log 2 1 e -k
+ x - 2
1
= 1Σ 2 + x- - 1Σ 2
2 k =1 k 2! 2 1! 2 k=1 k
Hereafter, in a similar way we obtein the desired expressions.
(2) 1 e -s
coth x dx
2
2 2
0 x 1 log 2
2
= - + - 1 Σ 2
21 2! 1! 2 0! s=1 s
2 2
(2.2')
(3) 3 log 2 2
0
x x 2
3
coth x dx = - 2 + -
2 3! 2 2! 2
e -s 1 e -s
2 2 2
1
+ 1 Σ
2 1! 2 s=1 s 2
+ 2 Σ 3
2 0! s=1 s
(2.3')
(4) log 2 3
0 x 4
x 3
coth x dx 4
= - +
23 4! 2 3! 2
2 2 2
2
e -s e -s 1 e -s
2 Σ 2 Σ
1 1
- 1 2
- 2 - Σ (2.4')
2 2! s=1 s 2 1! s=1 s 3 230! s=1 s 4
(n)
coth x dx 2
0 x n n-1
2 2
x n -1 log 2
n n
= (-1) n-1
- +
n! (n -1)!
2 2 2
r
e -s
2 Σ s
n -2 1
-(-1)n Σ (2.n')
j=0 2n-1- r r! s=1
n-r
- 28 -
Substituting x =1 , 1/2 , /2 for this one by one, we obtain the following special values.
e -2k
Σ = 1 - log(2 sinh 1)
k =1 k1
e -k
1 1
Σ = - log 2 sinh
k =1 k1 2 2
e -k
Σ = - log 2 sinh
k =1 k1 2 2
Formula 5.4.3
When = 2sinh -11 = 1.762747174 and B2k are Bernoulli Numbers, the following
expressions hold.
B2k
1! log -Σ +Σ
2k+1
e -k 1 1 1 2
(2) = Σ - -
k =1 k2 k =1 k k=1 2k (2k +1)! 2 2!
B2k
2k+2
e -k 2 1 3
+Σ
2 1
(3) = Σ + log -Σ -
k =1 k3 2! k=1 k k=1 2k (2k +2)! 2 3!
+ (2)
1!
B2k
3! log -Σ +Σ
2k+3
e -k 3 3 1 1 4
(4) = Σ - -
k =1 k4 k =1 k k =1 2k (2k +3)! 2 4!
2
+ (3) - (2)
1! 2!
j-1
j
e -k n -2
(n) = Σ + Σ(-1) (n - j)
k =1 kn j=1 j!
B2k
(n -1)! log -Σ +Σ
2k+ n-1
n-1 n-1 n -1 1 1 n
+ (-1) -
k=1 k k=1 2k (2k + n -1)! 2 n!
Proof
From Fourier series (2.2') and Taylor series (1.2')
2
1 e -k
2 2 2
1 1 log 2
f <1>
=- (2) + - + Σ (a)
21 2! 1! 21 k =1 k 2
From this
2
log 2 e -k
2 2 -
2
(2) = -2f <1>
+ +Σ
2! 1! k=1 k2
- 29 -
2 log 2 -Σ +Σ
1
22k B2k 2k+1
2
1 1 1
= -2
1! k=1 k k=1 2k (2k +1)!
2
log 2 e -k
2
2
+ - +Σ
2! 1! k =1 k2
B2k
1! log -Σ 1 2 e -k
2k+1
1 1
+Σ
2 2! Σ
=- + + 2
k =1 k k=1 2k (2k +1)! k=1 k
<2>
coth x dx
0 x x
2
3
= -f (1.3')
2 2 2
3
2
e -k
2 2 2 Σ
1 1 log 2 1
- 1 (2)+ - + 1
2 2 2! 1! 2 k=1 k2
From this and Fourier series (2.3') , we obtain
3
2
2 2 2
1 1 log 2
f<2>
=- 1 (2) + -
2 2 3! 2!
1 1 e -k
+ 2 (3)- 2 Σ 3 (b)
2 2 k=1 k
From this,
22 3
2log 2 e -k
2 2 +
(3) = 2 f 2 <2>
+2 (2) - +Σ
2 3! 2! k=1 k3
2 log 2
2
22k B2k 2k+2
2
1 2 1
=2 2
-Σ +Σ
2! k =1 k k=1 2k (2k +2)!
22 3
2log 2 e -k
2 +
+ (2)- +Σ
3! 2! k =1 k3
2 B2k 1 3
2k+2 -k
log -Σ
2 1 e
+Σ + (2)
2 3! Σ
= - + 3
2! k =1 k k=1 2k (2k +2)! k=1 k
coth x dx = - f 2
0 x 1
2 2
x 1
4 <3>
+ f<2>
1!
2 2 2
2
2 2
1
- f<1> (1.4')'
2!
e -k 2 e -k
2
1 1
2 Σ 22! 2 Σ
<3>
= -f - 2 3
- 2
2 k=1 k k=1 k
2 1 4
1 1
- (2) + 2 (3) -
22! 2 2 2 23! 2
From this and Fourier series (2.4') , we obtain
- 30 -
2
2 2 2 (3)
1 1
f<3>
=- (2) +
22! 22
4
3
1 e -k
2 2
1 1 log 2
- 3 (4) + - + 3Σ 4 (c)
2 4! 3! 2 k=1 k
From this,
2 log 2 -Σ +Σ
3
22k B2k 2k+3
2
1 3 1
(4) = -2 3
3! k=1 k k=1 2k (2k +3)!
3
23 4
2 2
log 2
- +
3! 4!
23 e -k 23 2
23
+ 3Σ 4 -
2 k =1 k 22! 2 (2) +
22 2 (3)
B2k
3! log -Σ
2k+3
3 4
+Σ
3 1
=- -
k =1 k k =1 2k (2k +3)! 24!
e -k 2
+Σ 4
- (2) + (3)
k =1 k 2!
Hereafter, by induction we obtain (n) for natural number n >2 .
- 31 -
5.5 Termwise Higher Integral of csc x
log tan (x /2) of cscx is x = /2 , we assume zeros of
Since the zero of the first order integral
the second or more order integral are also x = /2 . Where, since the zero of cscx is i , these seem
to be a collateral higher integral.
x x 1 1- cosx
csc x dx = log tan = log
2 2 1+ cosx
2
csc x dx
x x
2
= non-elementary function
2 2
Formula 5.5.1
(x) on 0< x <
<n>
When B2k denote Bernoulli Numbers and functions f are as follows
0
2
x 1
<0>
csc x dx =f (x) - x- f<0>
0! 2
2
22k -2B2k 2k
= log x +Σ x - log 2
k =1 2k (2k)!
0 <1>
x x 1
2 2
1 1
csc x dx 2 = f <1>(x) - x- f
0!
-
2 1!
x-
2
f
<0>
2 2
x x x 0 1
f 2 f 2
3 1 1
csc x dx = f <2>(x) - x-
<2>
- x-
<1>
0! 2 1! 2
2 2 2
2
2
1 <0>
- x- f
2! 2
x x k
2
n -1 1
csc x dx = f n <n -1>
(x) -Σ x- f<
n -1-k >
k=0 k! 2
2 2
Where f<0> 2 = log 2
Proof
x
22k -2B2k 2k
csc x dx = f
2
x
<0>
(x) - f<0> = log x +Σ x
k=1 2k (2k)!
2 2
- 32 -
22k -2B2k 2k
2 -2B2k
2k
2k
= log x +Σ
k=1 2k (2k)!
x - log 2 +Σ
k =1 2k (2k ) ! 2
22k -2B2k 2k
= log x +Σ x - log - log
k=1 2k (2k)! 2 4
2 -2B2k
2k
2k
Σ
k=1 2k (2k ) ! 2 = -log
4
5.3.1 (2)
22k -2B2k 2k
= log x +Σ x - log 2
k=1 2k (2k)!
2 dx
x x x
2
csc x dx = f<0>(x) - f<0>
2 2 2
x
2
1
<1> 1 <0>
= f (x) - x- f
1! 2
2
1
2 2
1
= f<1>(x) - f<1> - x- f<0>
1! 2
Hereafter, in a similar way we obtain the desired expressions..
Although Formula 5.5.1 is a general expression anyway, if it is expand actually, it is very complicated.
Then, this was devised and was made the same easy expression as sec x. It is as follows.
Formula 5.5.1'
When E0=1, E2=-1, E4=5, E6=-61, E8=1385, are Euler Numbers,the following expressions
hold for 0< x < .
cscx dx
2k+ n
x E2k
x - 2
x
n
=Σ
k =0 (2k + n)!
2 2
Proof
1 1
csc x = = = sec x -
sin x 2
cos x -
2
E2k
secx =Σ x 2k |x|<
k =0 (2k)! 2
From these
E2k 2k
x - 2
cscx =Σ 0< x <
k =0 (2k)!
Integrating both sides of this with respect to x from /2 to x repeatedly, we obtain the desired expressions.
- 33 -
csc x dx
0 1
2 f 2 2 f 2
0 x 1 1
2 <1> <0>
=- + (1.2')
0! 1!
2 2
0 x 0 1
2 2 2 2
x 1 1
3 <2>
csc x dx = - f + f<1>
0! 1!
2 2 2
2
2 2
1
- f<0> (1.3')
2!
csc x dx
0 x (-1)k k
2 2
x n -1
n -1-k >
n
= -Σ f< (1.n')
k=0 k!
2 2 2
Where f<0> 2 = log 2
(2) Taylor Series of the 1st order Integral of csc x
The 1st order Integral of csc x is as follows.
2 -2B 2k
2k
x x
csc x dx = log x +Σ x 2k - log 2 = log tan 0< x <
k=1 2k (2k)! 2
2
From this, the following expression follows immediately.
2 -2B 2k
2k
x x
Σ
2k
x = log tan - log 0< x < (1.t)
k =1 2k (2k)! 2 2
Substituting x =1 for this , we obtain the following special value.
22k -2B2k 1
Σ
k =1 2k (2k)!
= log tan + log 2
2
Formula 5.5.2
(-1)k
Let (n ) =Σ be Dirichlet Beta Function and be floor function. Then the following
k =0 (2k +1)
n
x x 1
2
= 2Σ 2 +1)x -
2 sin ( k
+ x-
k =0 (2k +1) 2 0! 2
2 2
3 2(2)
csc x dx
1
x
x x 1
3
= 2Σ sin (2k +1)x - + x-
k =0 (2k +1)3 2 1! 2
2 2 2
- 34 -
4
csc x dx
x x 1
4
= 2Σ 2 +1)x -
4 sin ( k
k =0 (2k +1) 2
2 2
2(2) 2 2(4) 0
+
2!
x-
2
-
0!
x-
2
5
x x 1
csc x dx 5 = 2Σ 2 +1)x -
5 sin ( k
k =0 (2k +1) 2
2 2
2(2) 3
2(4) 1
+
3!
x-
2 -
1!
x-
2
n
csc x dx
x x 1
n
= 2Σ sin (2k +1)x -
k=0 (2k +1)n 2
2 2
2 (2k) n-2k
n /2
k-1
+ Σ (-1) x-
k=1 (n -2k)! 2
Proof
csc x can be expanded to Fourier series in a broad meaning.as follows.
2i 2ie ix
cscx = ix -ix = - 2ix
= -2ie ix1 + e 2ix + e 4ix + e 6ix +
e -e 1- e
= -2 ie ix + e 3ix + e 5ix + e 7ix +
= 2(sin x + sin 3x + sin 5x ++) -2i(cosx + cos 3x + cos 5x + ) (2.0)
csc x is the locus of the median (central line) of this real number part. (See the figure)
However, calculus of the right side of (2.0) may be carried out, the real number does not turn into an imagi-
nary number and the contrary does not exist, either. Therefore, limiting the range of csc x and the higher order
integral to the real number, we calculate only the real number part of the right side of (2.0). Although Dirichlet
Odd Beta is obtained from calculation of the imaginary number part, since it is long, it is omitted in this section
Now, integrate the real number part of both sides of (2.0) with respect to x from /2 to x . Then
- 35 -
2(sin x + sin 3x + sin 5x + sin 7x +)dx
x x
csc x dx =
2 2
x
cos 1x cos 3x cos 5x cos 7x
= -2 + + + +
1 3 5 7
2
cos 1x cos 3x cos 5x cos 8x
= -2 + + + +
1 3 5 7
i.e.
x cos(2k +1)x
csc x dx = -2Σ
k=0 2k +1
2
Here, let -cos x = sin x - 2 . Then using this we obtain
1
x 1
csc x dx = 2Σ sin (2k +1)x -
k =0 2k +1 2
2
This is consistent with the real number part of the Fourier series of log tan(x /2) .
Next, integrate both sides of this with respect to x from /2 to x , then
1
2Σ 2k +1 sin (2k +1)x - 2 dx
x x x 1
csc x dx 2 =
k=0
2 2 2
2 x
1
= 2Σ sin (2k +1)x -
k=0 (2k +1)2 2
2
2 (-1)k
1
= 2Σ sin (2k +1)x - + 2Σ
k =0 (2k +1)2 2 k=0 (2k +1)2
(-1)k
Here, let (n) =Σ , then we obtain
k =0 (2k +1)n
2
+ 2(2)
x x 1
csc x dx 2 = 2Σ sin (2k +1)x -
k =0 (2k +1)2 2
2 2
Next, integrate both sides of this with respect to x from /2 to x , then
2
2Σ (2k +1) sin (2k +1)x - 2 + 2(2) dx
x x x x 1
csc x dx 3 = 2
k =0
2 2 2 2
3 x
+ 2(2)x - 2
1
= 2Σ 2 +1)x -
3 sin ( k
k =0 (2k +1) 2
3
2
+ 2(2)x - 2
1
= 2Σ sin (2k +1)x -
k =0 (2k +1)3 2
Hereafter, in a similar way we obtain the desired expression..
- 36 -
(1) Termwise Higher Definite Integral of Fourier Series of csc x
Substituting x =0 for Formula 5.5.2 , we obtain the following expressions.
2(2)
0
2
0 x
2
csc x dx = (2.2')
0!
2 2
2 (2)
csc x dx
1
0 x 23-1
2
x
3
=- + (3) (2.3')
1! 22
2 2 2
2 (2) 2(4)
csc x dx
0 x 2 0
2 2
x
4
= - (2.4')
2! 0!
2 2 2
2 (2) 2(4)
3 1
0 x x 25-1
csc x dx = - 2 2 (5)
5
+ - (2.5')
3! 1! 24
2 2 2
2(2k)
csc x dx
0 x n-2k
2
x n /2
k-1
n
= (-1) n
Σ
k =1
(-1)
(n -2k)!
2 2 2
n 2n-1
- sin (n) (2.n')
2 2n-1
Proof
Since
n
sin (2k +1)0 -
2 =0 for n =2, 4, 6,
n
sin (2k +1)0 -
2 = 1 for n =3, 5, 7, ,
3 2 (2)
csc x dx
1
0
x x 1
3
= 2Σ sin (2k +1)0- + 0-
k =0 (2k +1)3 2 1! 2
2 2 2
2(2) 1
23-1 2 (2) 1
2 2
1
= 2Σ 3
- = 2
(3) -
k =0 (2k +1) 1! 2 1!
Hereafter, in a similar way we obtain the desired expressions..
x cos(2k +1)x x
csc x dx = -2Σ = log tan 0< x <
k =0 2k +1 2
2
From this, the following expression follows immediately.
- 37 -
cos(2k +1)x 1 x
Σ = - log tan 0< x < (2.f)
k =0 2k +1 2 2
Substituting x =1 for this, we obtain the following special values.
cos(2k +1) 1 1
Σk =0 2k +1
= - log tan
2 2
Formula 5.5.3
When B2k , (x), (x) denote Bernoulli Numbers, Dirichlet Beta Function, Riemann Zeta Function
(x) on 0< x < are as follows
<n>
respectively, and functions f
2 -2B 2k
2k
xn
n 1
n =1, 2, 3, ,
<n>
f (x)= log x -Σ +Σ x 2k+n
n ! k =1 k k=1 2 2
k k n
( + )!
the following expressions hold.
1 1 (-1)k k
log 2
2 2 2 2
1
(2) = - Σ f <1-k > = - f<1>
2 k=0 k ! 2
1 2 (-1)k k-1
2 -1 (3) 3
(2) =
2Σ 2 2 2
<2-k >
f + 2
k=0 k!
1 3 (-1)k k
(2) 2
(4) =
2Σ 2 f 2 + 2! 2
<3-k >
k=0 k!
1 4 (-1)k k-1
(2) 2
25-1 (5)
(4) = - Σ 2 2 3! 2 +
<4-k >
f +
2 k=0 k ! 24
1 5 (-1)k k
(4) 2
(2) 4
(6) = - Σ 2 f 2 + 2! 2 - 2
<5-k >
2 k =0 k ! 4!
1 6 (-1)k
k-1
(4) 2
(2) 4
(6) =
2Σ 2 2 3! 2 5! 2
<6-k >
f + -
k=0 k!
27-1 (7)
+
26
(-1)n 2n -1 (-1)k k
(2n -2k) 2k
2 f 2 -Σ(-1) 2
n -1
(2n) =
2 Σ
<2n-1-k > k
k =0 k! k=1 (2k) !
(-1)n 2n (-1)k k-1
(2n -2k) 2k
2 2 -Σ(-1)
n -1
(2n) = -
2 Σ
<2n-k > k
f
k =0 k! k =1 (2k +1) ! 2
22n+1 -1 (2n +1)
+
22n
Proof
0 1
2 2 2 2
0 x 1 1
2 <1>
csc x dx = - f + f<0> (1.2')
0! 1!
2 2
- 38 -
(-1)k k
2 f 2
1
= -Σ <1-k >
k =0 k!
2(2)
0 x
csc x dx 2 = (2.2')
0!
2 2
1 1 (-1)k k
log 2
2 2 2 2
1
(2) = - Σ f <1-k > = - f<1>
2 k=0 k ! 2
csc x dx (-1)k
0 k
2 f 2
x x 3-1
3
= -Σ <3-1-k >
(1.3')
k =0 k!
2 2 2
2 (2)
0 23-1 1
2
x x
csc x dx =3
2
(3) - (2.3')
2 1!
2 2 2
Formula 5.5.4
When B2k , (x), (x) denote Bernoulli Numbers, Dirichlet Beta Function, Riemann Zeta Function
(x) on 0< x < are as follows
<n>
respectively, and functions f
2 -2B 2k
2k
xn
n 1
n =1, 2, 3, ,
<n>
f (x)= log x -Σ +Σ x 2k+n
n! k =1 k k=1 2k (2k +n )!
the following expressions hold.
(2)3
1
22 2 (-1)k k
2 2 f 2
2
(3) = 3 - 3 Σ <2-k >
2 -1 1! 2 -1 k=0 k !
(4) (2)
2
25 1 3
(5) = 5
2 -1 1! 2 -
3!
24 4 (-1)k k
+ 5 Σ
2 -1 k =0 k ! 2 f <4-k > 2
(6) (4) (2)
2
27 1 3 5
(7) = 7
2 -1 1! 2 -
3! 2 +
5!
26 6 (-1)k k
- 7 Σ
2 -1 k=0 k ! 2 f <6-k > 2
- 39 -
22n+1 (2n -2k) 2k+1
2
n -1
(2n +1) = Σ (-1) k
22n+1 -1 k =0 (2k +1)!
22n (-1)k k
2 2
2n
+ (-1) n
Σ k!
f <2n -k >
22n+1 -1 k=0
- 40 -
5.6 Termwise Higher Integral of csch x
Since both zeros of csch x and the primitive function log tanhx /2
x = , the latter seems to be
are
a lineal primitive function with a fixed lower limit. On the other hand, the Fourier series of csch x is known for
x >0 . Then, we assume zeros of the second or more order primitive function are also x = .
x x
csch x dx = log tanh
2
csch x dx
x x
2
= non-elementary function
2 -2 B 2k
2k
1
csch x = -Σ x 2k-1 0< x <
x k =1 2k (2k -1)!
However, this can not be integrated with the lower limit x = .
Formula 5.6.2
The following expressions hold for x >0 .
x
csch x dx = -2Σ
x e -(2k +1)
k =0 2k +1
csch x dx
x
x x e -(2k +1)
n
= (-1) 2Σ n
k=0 (2k +1)n
Proof
csch x can be expanded to a Fourier series as follows. (See the figure of the following page.)
2 2e -x
csch x = x -x
= -2x
= 2e -x1 + e -2x + e -4x + e -6x +
e -e 1- e
= 2e -1x + e -3x + e -5x + e -7x + (2.0)
- 41 -
Integrate the both sides of (2.0) with respect to x from to x , then
2e
x x
csch x dx = -1x
+ e -3x + e -5x + e -7x +dx
x
e -x e -3x e -5x e -7x
= -2 + + + +
1 3 5 7
i.e.
x
x e -(2k +1)
csch x dx = -2Σ
k=1 2k +1
x
This is consistent with the real number part of the Fourier series of log tanh .
2
Next, integrate both sides of this with respect to x from to x , then
x
csch x dx
x x e -(2k +1)
2
= 2Σ
k=0 (2k +1)2
Hereafter, in a similar way we obtein the desired expression.
Formula 5.6.3
1
When (n) =Σ , the following expressions hold.
k=0 (2k +1)n
csch x dx = 2(2)
0 x
2
csch x dx = -2(3)
0 x x
3
- 42 -
csch x dx
0 x x
n
= (-1)n2 (n)
Formula 5.6.4
csch x dx
2n-1 0 x x
(n) = (-1) n n n
2 -1
- 43 -
5.7 Termwise Higher Integral of sec x
1 1+ sin x
Since the zero of the first order integral log of secx is x = 0 , we assume the zeros of
2 1- sin x
the second or more order integral are also x = 0 . Where, since the zero of secx is i , these seem
to be a collateral higher integral.
x 1 1+ sin x
sec x dx = log
0 2 1- sin x
secx dx
x x
2
= non-elementary function
0 0
Formula 5.7.1
When E0=1, E2=-1, E4=5, E6=-61, E8=1385, are Euler Numbers, the following expressions
x E2k
sec x dx =Σ x 2k+1
0 k =0(2k +1)!
sec x dx
x x E2k
2
=Σ x 2k+2
0 0 k =0 (2k +2)!
sec x dx
x x x E2k
3
=Σ x 2k+3
0 0 0 k =0 (2k +3)!
x x E2k
secx dx n =Σ x 2k+
n
0 0 k =0 (2k + n)!
Proof
secx can be expanded to Taylor series as follows.
E2k
secx =Σ x 2k |x| <
k=0 (2k)! 2
Integrating both sides of this with respect to x from 0 to x repeatedly, we obtain the desired expression.
2
2 x
2
=Σ (1.2')
0 0 k=0 (2k +2)!
sec x dx E2k
2 x x 2k+3
2
3
=Σ (1.3')
0 0 0 k =0 (2k +3)!
- 44 -
secx dx
2k+ n
2
E2k
2
x x
n
=Σ (1.n')
0 0 0 k =0 (2k + n)!
n 2n
E2n =Σ(-1)k T +1 n =1, 2, 3, (1)
k =1 2k -1 2k-1
On the other hand, between Tangent numbers and Bernoulli Numbers, the following equations hold.
n-1 2 2 -1
2n 2n
T2n-1 ( )= -1 B2n n =1, 2, 3, (2)
2n
Therefore, the following equations have to hold between Euler Numbers and Bernoulli Numbers.
2k 2k
2 2 -1
n 2n
E2n = 1 -Σ B2k n =1, 2, 3, (3)
k =1 2k 2k -1
In fact, for example,
2222-1
2424-1
2626-1
23 23 23
E6 = 1 - B2 + B4 + B6
2 1 4 3 6 5
2 6 1 4 30 3 6 42 5 = -61
43 1 6 1615 1 6 6463 1 6
=1- - +
On the contrary, the following equations between Bernoulli Numbers and Euler Numbers are shown by
Mr. Nishimura .
- 45 -
2n n -1 2n -1
B2n = Σ E2k n =1, 2, 3, (4)
22n22n-1 k =0 2k
The following example shows the formula is right.
2 E
23 23-1 23-1 23-1
B6 = E0 + E2 + 4
223223-1 0 4
4 42
23 5 5 5 1
= 1- 1+ 5 =
6463 0 2
In addition, Mr. Nishimura expressed Even Zeta with the Euler numbers using this equation.
Formula 5.7.2
When E2k , (n) , denote Euler Numbers, Dirichlet Beta Function, floor function respectively,
the following expressions hold for |x |< /2 .
1
(-1)k
x
sec x dx = 2Σ cos (2k +1)x -
0 k=0 2k +1 2
2 2 (2) 0
(-1)k
x x
secx dx 2
= 2Σ 2 cos (2k +1)x - + x
0 0 k =0 (2k +1) 2 0!
3 2 (2) 1
k
x x x (-1)
sec x dx 3 = 2Σ 2 +1)x -
3 cos ( k
+ x
0 0 0 k=0 (2k +1) 2 1!
4
(-1)k
x x
sec x dx 4 = 2Σ 2 +1)x -
4 cos ( k
0 0 k =0 (2k +1) 2
2 (2) 2 2 (4) 0
+ x - x
2! 0!
5
sec x dx (-1)k
x x
5
= 2Σ 2 +1)x -
5 cos ( k
0 0 k=0 (2k +1) 2
2 (2) 3 2 (4) 1
+ x - x
3! 1!
n
sec x dx
(-1)k
x x
n
= 2Σ cos (2k +1)x -
0 0 k =0 (2k +1)n 2
n /2 2 (2k) n-2k
+ Σ (-1)k-1 x
k=1 (n -2k)!
Proof
sec x can be expanded to Fourier series in a broad meaning.as follows.
2 2e ix
secx = ix -ix = 2ix
= 2e ix - e 3ix + e 5ix - e 7ix +-
e +e 1+ e
= 2(cosx - cos 3x + cos 5x - cos 7x +-)
+ 2i(sin x - sin 3x + sin 5x - sin 7x +-) (2.0)
- 46 -
sec x is the locus of the median (central line) of this real number part. (See the figure)
y
40
30
20
10
-1.4 -1.2 -1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
x
1/cos(x)
2*cos(x) - 2*cos(3*x) + 2*cos(5*x) - 2*cos(7*x) +
However calculus of the right side of (2.0) may be carried out, the real number does not turn into an imaginary
number and the contrary does not exist, either. Therefore, limiting the range of sec x and the higher order
integral to the real number, we calculate only the real number part of the right side of (2.0) .
Although the Dirichlet Odd Beta is obtained from calculation of the imaginary number part, since it is long, it is
omitted in this section.
Now, integrate the real number part of both sides of (2.0) with respect to x from 0 to x, then
0 0
x
sin 1x sin 3x sin 5x sin 8x
=2 - + - +-
1 2 3 4 0
sin 1x sin 3x sin 5x sin 8x
=2 - + - +-
1 2 3 4
i.e.
sec x dx = 2Σ
x (-1)k
sin (2k +1)x
0 k =0 2k +1
Here, since sin x = cos x - 2 , we obtain
1
sec x dx = 2Σ
(-1)k
x
cos (2k +1)x -
0 k =0 2k +1 2
1 1+ sin x
This is consistent with the real number part of the Fourier series of log .
2 1- sin x
Next, integrate both sides of this with respect to x from 0 to x , then
1
dx
(-1)k
x x x
secx dx = 2
2Σ cos (2k +1)x -
0 0 0 k =0 2k +1 2
x
2
(-1)k
= 2Σ 2 cos (2k +1)x -
k =0 (2k +1) 2
0
- 47 -
(-1)k 2 (-1)k
- 2Σ
= 2Σ cos (2k +1)x - cos(-)
k =0 (2k +1)2 2 k =0 (2k +1)2
(-1)k
Here, let (n) =Σ , then we obtain
k =0 (2k +1)n
2 2 (2) 0
secx dx
(-1)k
x x
2
= 2Σ 2 +1)x -
2 cos ( k
+ x
0 0 k =0 (2k +1) 2 0!
Next, integrate both sides of this with respect to x from 0 to x , then
2 2 (2) 0
(-1)k
x x x x
sec x dx = 3
2Σ 2 cos (2k +1)x - + x dx
0 0 0 0 k=0 (2k +1) 2 0!
x
3 2 (2) 1
(-1)k
= 2Σ 3 cos ( k
2 +1)x - + x
k =0 (2k +1) 2 1!
0
k
3 2 (2) 1
(-1)
= 2Σ 3 cos ( k
2 +1)x - + x
k =0 (2k +1) 2 1!
Hereafter, in a similar way , we obtain the desired expressions.
2
2 x
2
= (2.2')
0 0 0!
2 (2)
1
2 x x 23-1
2 (3)
3
sec x dx = - (2.3')
0 0 0 1! 22
2 (2) 2 (4)
2 x 2 0
2 2
x
4
sec x dx = - (2.4')
0 0 0 2! 0!
2 (2) 2 (4)
3 1
2 x 25-1
2 2
x
sec x dx = 5
- + (5) (2.5')
0 0 0 3! 1! 24
2 (2k) n 2n-1
2 x n-2k
2
x n /2
sec x dx = Σ (-1)
n k-1
+ sin (n)
0 0 0 k =1 (n -2k)! 2 2n-1
(2.n')
Proof
Since
n
cos (2k +1)
2 2
- =0 for n =2, 4, 6,
n n =1, 3, 5,
cos (2k +1) -
2 2 = (-1) k+1
for
k =0, 1, 2,
,
- 48 -
giving x = /2 to Formula 5.7.2 and substituting these for them,
2
secx dx
(-1)k
+ 2(2) = 2(2)
2 x
2
= 2Σ cos (2k +1) -
0 0 k=0 (2k +1)2 2 2
3 2 (2)
sec x dx
(-1)k 1
2 x x
2
3
= 2Σ 2 +1) -
3 cos ( k
+
0 0 0 k=0 (2k +1) 2 2 1!
= 2Σ +
k=0 (2k +1)3 1! 2
23-1 2 (2) 1
= - 2 (3) +
2 1! 2
Hereafter, in a similar way we obtein the desired expressions.
Formula 5.7.3
When E2k , (x), (x) denote Euler Numbers, Dirichlet Beta Function, Riemann Zeta Function
respectively, the following expressions hold.
1 E2k 2k+2
(2) =
2Σk =0 (2k +2)! 2
1 E2k 2k+2
23-1 (3)
(2) =
2Σk =0 (2k +3)! 2 +
22
1 E2k 2k+4
(2) 2
(4) = - Σ
2 k=0 (2k +4)! 2 +
2! 2
1 E2k 2k+4
(2) 2
25-1 (5)
(4) = - Σ
2 k=0 (2k +5)! 2 +
3! 2 +
24
1 E2k 2k+6
(4) 2
(2) 4
(6) =
2Σk =0 (2k +6)! 2 +
2! 2 -
4! 2
- 49 -
1 E2k 2k+6
(4) 2
(2) 4
(6) =
2Σk =0 (2k +7)! 2 +
3! 2 -
5! 2
2 -1 (7)
7
+
26
(-1)n-1 E2k 2k+2n
(2n -2k) 2k
2 2
n
(2n) = Σ +Σ(-1)k-1
2 k=0 (2k +2n ) ! k=1 (2k) !
(-1)n-1 E2k (2n -2k )
2k+2n+1 2k
2 2
n
(2n) = Σ +Σ(-1)k-1
2 k =0 (2 k +2 n +1) ! k =1 (2k +1)!
22n+1-1 (2n +1)
+
22n
Proof
sec x dx
E2k
2k+2
2
2 x
2
=Σ (1.2')
0 0 k=0 (2k +2)!
2 (2)
secx dx
2 x
2
= (2.2')
0 0 0!
1 E2k 2k+2
(2) = Σ
2 k =0 (2k +2)! 2
sec x dx E2k
2 x x 2k+3
2
3
=Σ (1.3')
0 0 0 k =0 (2k +3)!
2 (2) 1
2 x x 23-1
sec x dx = - 2 (3) +
3
(2.3')
0 0 0 2 1! 2
1 E2k 2k+2 23-1 (3)
(2) = Σ
2 k=0 (2k +3)! 2
+
22
Hereafter, in a similar way we obtein the desired expressions.
c.f.
Mr.Sugioka expressed Dirichlet Even Beta by definite Reamann Odd Zeta and infinite Reiamnn Even Zeta
using his " Taylor System ". (4), (6) which he calculated are as follows.
22-1 1
2
1
(4) = (3)
22 1!
3
2 3!
log 2 22-1 1! 24-1 3! 26-1 5!
- - (2) - (4) - (6) -
22 1
22 22 5! 24 24 7! 26 26 9!
24-1 1
22-1 3
2 2
1 1
(6) = (5) - (3)
24 1! 22 3!
5
2 5!
log 2 22-1 1! 24-1 3! 26-1 5!
+ 4 1
- 2
(2) 2
- 4
(4) 4
- 6
(6) 6
-
2 2 2 7! 2 2 9! 2 2 11!
- 50 -
5.7.4 Riemann Odd Zeta
Riemann Odd Zeta can be conversely obtained from the half of Formula 5.7.3 . The following formula shows
the relation between Riemann Odd Zeta and Dirichlet Even Beta by the easy expression.
Formula 5.7.4
When E2k , (x), (x) denote Euler Numbers, Dirichlet Beta Function, Riemann Zeta Function
respectively, the following expressions hold.
23 (2) 1
22 E2k 2k+3
(3) = 3
2 -1 1! 2 - 3 Σ
2 -1 k=0 (2k +3)! 2
(4) (2)
2
25 1 3
(5) = 5
2 -1 1! 2 -
3!
24 E2k 2k+5
+ 5 Σ
2 -1 k =0 (2k +5)! 2
(6) (4) (2)
2
27 1 3 5
(7) = 7
2 -1 1! 2 -
3! 2 +
5!
26 E2k 2k+7
- 7 Σ
2 -1 k =0 (2k +7)! 2
22n+1 (2n -2k) 2k+1
2
n -1
(2n +1) = Σ (-1) k
22n+1 -1 k=0 (2k +1)!
22n E2k 2k+2n+1
- 51 -
5.8 Termwise Higher Integral of sech x
2tan e x-
-1
Since both zeros of sech x and the primitive function are x = , the latter seems to be
a lineal primitive function with a fixed lower limit. Then, we assume zeros of the second or more order primitive
function are also x = .
sech x dx = 2tan e -
x
-1 x
Formula 5.8.2
The following expressions hold for x >0 .
x
sech x dx
x e -(2k +1)
= -2Σ(-1)k
k =0 2k +1
x x x -(2k +1)x
e
3 k
sech
x dx = -2Σ(-1)
(2k +1) k=0
3
x
x x e -(2k +1)
sech x dx = (-1) 2Σ(-1)
n n k
k=0 (2k +1)n
Proof
sech x can be expanded to a Fourier series as follows. (See the figure of the following page.)
2 2e -x
sech x = x -x
= -2x
= 2e -x 1 - e -2x + e -4x - e -6x +-
e +e 1 +e
= 2e -1x - e -3x + e -5x - e -7x +- (2.0)
- 52 -
y
1.0
0.8
0.6
0.4
0.2
0.0
0 1 2 3 4 5
x
1/cosh(x)
2*(exp(-x) - exp(-201*x))/(exp(-2*x) + 1)
2e
x x
sech x dx = -x
- e -3x + e -5x - e -7x +-dx
x
e -x e -3x e -5x e -7x
= -2 - + - +-
1 3 5 7 0
i.e.
x
x e -(2k +1)
sech x dx = -2Σ(-1) k
k =0 2k +1
2tan e x-
-1
This is consistent with the real number part of the Fourier series of .
Next, integrate both sides of this with respect to x from to x , then
x
x x
-(2k +1)
x x
k e
e -(2k +1)
sech x dx = 2Σ(-1)
2
= 2Σ(-1) k
k=0 (2k +1)2
k=0 (2k +1)2
Hereafter, in a similar way we obtein the desired expressions.
Formula 5.8.3
(-1)k
When (n) =Σ is the Dirichlet Beta Function, the following expressions hold.
k=0 (2k +1)n
sech x dx = -2(1)
0
sech x dx = 2(2)
0 x
2
sech x dx = -2(3)
0 x x
3
- 53 -
sech x dx
0 x x
n
= (-1)n2 (n)
Note
1 1 1 1
(1) = 1
- 1 + 1 - 1 +- = : Madhava series
1 3 5 7 4
1 1 1 1
(2) = 2 - 2 + 2 - 2 +- = 0.9159655941 : Catalan's constant
1 3 5 7
2006.10.05
2012.05.29 Updated
K. Kono
Alien's Mathematics
- 54 -
06 Termwise Higher Integral (Inv-Trigonometric, Inv-Hyperbolic)
The 2nd or more order integrals of Inverse Trigonometric Functions and Inverse Hyperbolic Functions were
all shown in 4.4 and 4.5 . However, as for these, the integration lower limits and the function forms were
complicated. On the contrary, integrating term by term the Taylor series of these functions, we can obtain the
simple general forms. Naturally, many of these are the collateral higher integrals. However, it cannot be said
that the lineal higher integral is useful and the collateral one is useless. Simple is the best ! A simple infinite
series is superior to the complicated definite terms function.
Formula 6.1.1
The following expression holds for |x|< 1 .
x x (2k)!
tan -1x dx n = Σ(-1)k x 2k+
n+1
0 0 k=0 (2k + n +1)!
Proof
1 (2k)! 2k+1
tan -1x = Σ(-1)k x 2k+1 = Σ(-1)k x
k =0 2k +1 k=0 (2k +1)!
Integrating both sides of this with respect to x from 0 to x repeatedly, we obtain the desired expression.
The lineal higher integral of this was as follows as shown in Formula 4.4.1 ( 4.4 ).
tan
x x tan -1x n /2 n-2k
n
Σ (-1) nC n-2k x
k
-1
x dx =
0 0 n! k=0
log1+ x 2 n /2
+ Σ (-1)k nC n+1-2k x n+1-2k
2n ! k=1
1 n /2
- Σ (-1)r nC n+1-2r (1+ n)- (2r)x n+1-2r
!
n r=1
Therefore, the termwise higher integral above is a lineal higher integral. And matching both, we obtain the
following formula.
Formula 6.1.1'
The following expression holds for |x| 1 .
(2k)! tan -1x n /2
Σ(-1)
k
x 2k+n
= Σ (-1)k nC n-2k x n-2k
k =0 (2k +n +1)! n ! k =0
log1+ x 2 n /2
+ Σ (-1)k nC n+1-2k x n+1-2k
2n ! k=1
1 n /2
- Σ (-1)r nC n+1-2r (1+ n)- (2r)x n+1-2r
n! r=1
-1-
(2k)! 1 1 5 log 2
Σ
k =0
(-1)k
(2k +4)!
= -
1234 3456
+- = -
12 12
-
6
(2k)! 1 1 5
Σ
k =0
(-1)k
(2k +5)!
= -
12345 34567
+- =
36
-
24
(2k) ! n /2 log 2 n /2
Σ
k =0
(-1)k
(2k +n +1) !
=
4 n! Σ
k=0
(-1)k nC n-2k + Σ
2n! k=1
(-1)k nC n+1-2k
1 n /2
- Σ (-1)r nC n+1-2r (1+ n)- (2r) (1.n')
!
n r=1
Note
The following formula is known about these series.
(1- t)m-1
(-1)k 1 1
Σ
k =0 (2k +1)(2k +2)(2k +m)
=
(m -1)! 0 1+ t 2
dt
This is easily drawn from Formula 6.1.1 using Riemann-Liouville Integral.
And the following expression is also drawn easily..
n
1 (1- x) n /2 log 2 n /2
2 dx
= Σ (-1)k nC n-2k + Σ (-1)k nC n+1-2k
0 1+ x 4 k =0 2 k =1
n /2
- Σ (-1)r nC n+1-2r (1+ n )- (2r)
r=1
n /2 1 n /2
(-1)r nC n+1-2r (1+ n)- (2r)
4 n! Σ (-1)knC n-2k -
k =0 n! Σ
r=1
(2k)!
=Σ(-1)k
k =0 (2k + n +1)!
Replacing n with 4n, we obtain the following equation.
-2-
2n 1 2n
Σ(-1)k 4nC 4n-2k - (-1)r4nC 4n+1-2r (1+4 n)- (2r)
4(4n)! k =0 (4n )! Σ
r=1
(2k)!
=Σ(-1)k
k =0 (2k +4n +1)!
This is the one having to call Higher order Gregory Series . When n=2,3, it is as followes.
109 1 1
- = - +-
10080 325800 123456789 34567891011
87217 1 1
- = - +-
829870272000 29937600 12 1213 34 1415
-3-
6.2 Termwise Higher Integral of arccot x
Formula 6.2.1
The following expression holds for 0< x 1 .
xn
x x (2k)!
-1
cot x dx =n
- Σ(-1)k x 2k+
n+1
0 0 2 n! k =0 (2k + n +1)!
Proof
1 (2k)! 2k+1
cot -1x = -Σ(-1)k x 2k+1 = -Σ(-1)k x
2 k=0 2k +1 2 k=0 (2k +1)!
Integrating both sides of this with respect to x from 0 to x repeatedly, we obtain the desired expression.
The lineal higher integral of this was as follows as shown in Formula 4.4.1 ( 4.4 ).
cot
x x xn tan -1x n /2 n-2k
n
Σ k
(-1) nC n-2k x
-1 -1
x dx = cot x -
0 0 n! n! k=1
log1+ x 2 n /2
- Σ (-1)k nC n+1-2k x n+1-2k
2n ! k =1
1 n /2
(-1)r nC n+1-2r (1+ n)- (2r)x n+1-2r
n! Σ
+
r=1
Therefore, the termwise higher integral above is a lineal higher integral.
-4-
6.3 Termwise Higher Integral of arcsin x
Formula 6.3.1
|x|< 1 .
The following expression holds for
x x 2
(2k -1)!!
sin x dx = Σ
-1 n
x 2k+
n+1
Collateral
0 0 k=0 (2k + n +1)!
Proof
(2k -1)!! (2k -1)!!(2k)!
sin -1x = Σ x 2k+1 = Σ x 2k+1
k =0 (2k )!! (2k +1 ) k=0 (2k)!! (2k +1)!
2
(2k -1)!!
=Σ x 2k+1 (2k)! = (2k)!!(2k -1)!!
k=0 (2k +1)!
Integrating both sides of this with respect to x from 0 to x repeatedly, we obtain the desired expression.
The collateral higher integral of this was as follows as shown in Formula 4.4.2' ( 4.4 ).
n-2r
x x n/2 x
sin x dx =-1 n
Σ 2 sin -1x
0 0 r=0 (2r)!! (n -2r)!
n-2r+1
n /2 n -2r+1 (s -1)!!2 x
+Σ Σ (-1)sn -2r+1C s 2 (n -2r +1)!
1-x 2
r=1 s=0 (s +2r -1)!!
n-2r+1
n /2 x
+Σ
r=1 (2r -1)!!2(n -2r +1)!
Therefore, the termwise higher integral above is a collateral higher integral.
And matching both, we obtain the following formula.
Formula 6.3.1'
The following expression holds for |x| 1 .
2 n-2r
(2k -1)!! n/2 x
2k+ n+1
Σ
k=0 (2k + n +1)!
x =Σ
r=0
2
(2r)!! (n -2r)!
sin -1x
n-2r+1
n /2 n -2r+1 (s -1)!!2 x
+Σ Σ s
(-1) n -2r+1C s 2 (n -2r +1)!
1-x 2
r=1 s=0 (s +2r -1)!!
n-2r+1
n /2 x
+Σ
r=1 (2r -1)!!2(n -2r +1)!
(2k -1)!!
2
Σ
k=0 (2k +2)!
=
2
-1
(2k -1)!!
2
3
Σ
k=0 (2k +3)!
=
8
-1
(2k -1)!!
2
5 11
Σ
k=0 (2k +4)!
=
24
-
18
-5-
(2k -1)!!
2
C n-1 n /2 1
-Σ
2n -1
Σ
k=0 (2k + n +1)!
=
(2n)!! k=1 (2k -1)!!2(n -2k +1)!
(1.n')
2
(2j -1)!! n /2 1 (2n)!!
= Σ +Σ
j=0 (2j + n +1)!
2
k=1 (2k -1)!! (n -2k +1)! C n-1
2n -1
Calculating formula
Whencn , c =0.851.00 denotes the number of significant figures,
1 n /2 (2k -1)!!2 1 (2n)!!
(cn)= +Σ +
(n +1)! k=1 (2k + n +1)! (2k -1)!!2(n -2k +1)! 2n -1C n-1
In this formula, the number of terms required for the number of significant figures cn is about n /2 . And
we can assume c=1 when large n is taken.
In the following example, supposing n =1,700 and c=0.9 , we obtained the significant figure of 1,530
digits. And the number of terms required for this was 851 .
Example (1530)
n:=1700: DIGITS:=floor(0.9*n): MAXDEPTH:=1000:
f := sum((2*k-1)!!^2/(2*k+n+1)!+1/((2*k-1)!!^2*(n-2*k+1)!)
, k=1..ceil(n/2)):
pi := (1/(n+1)!+f)*(2*n)!!/binomial(2*n-1,n-1):
float(pi)
-6-
6.4 Termwise Higher Integral of arccos x
Formula 6.4.1
The following expression holds for |x|< 1 .
xn
cos
x x 2
(2k -1)!! n+1
-1 n
x dx = -Σ x 2k+ Collateral
0 0 2 n! k =0 (2k + n +1)!
Proof
(2k -1)!! (2k -1)!!(2k)!
cos -1x = -Σ x 2k+1 = -Σ x 2k+1
2 k=0 (2k)!! (2k +1) 2 k=0 (2k)!! (2k +1)!
(2k -1)!!
2
= -Σ x 2k+1 (2k)! = (2k)!!(2k -1)!!
2 k =0 (2k +1)!
Integrating both sides of this with respect to x from 0 to x repeatedly, we obtain the desired expression.
The collateral higher integral of this was as follows as shown in Formula 4.4.2' ( 4.4 ).
n-2r
x x xn n/2 x
cos x dx = -1 n -1
cos x - Σ 2 sin -1x
0 0 n ! r=1 (2r)!! (n -2r)!
n-2r+1
n /2 n -2r+1 (s -1)!!2 x
-Σ Σ (-1) n -2r+1C s
s
2 (n -2r +1)!
1-x 2
r=1 s=0 (s +2r -1)!!
n-2r+1
n /2 x
+Σ
r=1 (2r -1)!!2(n -2r +1)!
Therefore, the termwise higher integral above is a collateral higher integral.
-7-
6.5 Termwise Higher Integral of arctanh x
Formula 6.5.1
|x|< 1 .
The following expression holds for
x x (2k)!
tanh -1x dx n = Σ x 2k+
n+1
0 0 k =0 (2k + n +1)!
Proof
1 (2k)! 2k+1
tanh -1x = Σ x 2k+1 = Σ x
k=0 2k +1 k=0 (2k +1)!
Integrating both sides of this with respect to x from 0 to x repeatedly, we obtain the desired expression.
The lineal higher integral of this was as follows as shown in Formula 4.5.1 ( 4.5 ).
tanh
x x tanh -1x n /2 n-2k
n
Σ nC n-2k x
-1
x dx =
0 0 n! k=0
1 n /2
nC n+1-2r (1+ n )- (2r)x
n+1-2r
- Σ
n! r=1
Therefore, the termwise higher integral above is a lineal higher integral.
And matching both, we obtain the following formula.
Formula 6.5.1'
The following expression holds for |x| 1 .
(2k)! 2k+ n+1 tanh -1x n /2 n-2k
Σ
k=0 (2k + n +1)!
x = Σ
n ! k =0
nC n-2k x
1 n /2
nC n+1-2r (1+ n )- (2r)x
n+1-2r
-
! Σ
n r=1
-8-
Note
The following formula is known about these series.
(1- t)m-1
1 1 1
Σ
k=0 (2k +1)(2k +2)(2k +m)
=
(m -1)! 0 1- t 2
dt
This is easily drawn from Formula 6.5.1 using Riemann-Liouville Integral.
And the following expression is also drawn easily..
1 (1- x)n n /2
nC n+1-2r (1+ n )- (2r)
n-1
0 1- x 2
dx = 2 log 2 - Σ
r=1
-9-
6.6 Termwise Higher Integral of arcsinh x
Formula 6.6.1
The following expression holds for |x|< 1 .
sinh
2
x x
k (2k -1)!! n+1
-1
x dx = Σ(-1)
n
x 2k+ Collateral
0 0 k=0 (2k + n +1)!
Proof
(2k -1)!! (2k -1)!!(2k)! 2k+1
sinh -1x =Σ(-1)k x 2k+1 =Σ(-1)k x
k=0 (2k)!! (2k +1) k=0 (2k)!! (2k +1)!
2
k (2k -1)!!
= Σ(-1) x 2k+1 (2k)! = (2k)!!(2k -1)!!
k =0 (2k +1)!
Integrating both sides of this with respect to x from 0 to x repeatedly, we obtain the desired expression.
The collateral higher integral of this was as follows as shown in Formula 4.5.2' ( 4.5 ).
(-1)r x n-2r
x x n/2
sinh x dx = -1 n
Σ 2 sinh -1x
0 0 r=0 (2r)!! (n -2r)!
n-2r+1
n /2 n -2r+1
r+s (s -1)!!2 x
+Σ Σ (-1) n -2r+1C s 2 (n -2r +1)!
1+ x 2
r=1 s=0 (s +2r -1)!!
n /2 (-1)r-1 x n-2r+1
+Σ
r=1 (2r -1)!!2(n -2r +1)!
Therefore, the termwise higher integral above is a collateral higher integral.
And matching both, we obtain the following formula.
Formula 6.6.1'
The following expression holds for |x| 1 .
k (2k -1)!!
2
2k+ n+1
n/2 (-1)r x n-2r
Σ
k=0
(-1)
(2k + n +1)!
x =Σ
r=0
2
(2r)!! (n -2r)!
sinh -1x
n-2r+1
n /2 n -2r+1
r+s (s -1)!!2 x
+Σ Σ (-1) n -2r+1C s 2 (n -2r +1)!
1+ x 2
r=1 s=0 (s +2r -1)!!
n /2 (-1)r-1 x n-2r+1
+Σ
r=1 (2r -1)!!2(n -2r +1)!
- 10 -
k (2k -1)!!
2 n/2 (-1)r
Σ
k=0
(-1)
(2k + n +1)!
=Σ
r=0
2
(2r)!! (n -2r)!
sinh -11
n /2 n -2r+1
r+s (s -1)!!2 2
+Σ Σ (-1) n -2r+1C s 2 (n -2r +1)!
r=1 s=0 (s +2r -1)!!
n /2 (-1)r-1
+Σ
r=1 (2r -1)!!2(n -2r +1)!
- 11 -
6.7 Termwise Higher Integral of arcsech x
Formula 6.7.1
The following expression holds for 0< x <1 .
xn 2
-Σ 2k (2k +n)! x
x x 2 n 1 (2k -1)!!
-1
sech x dx = n
log +Σ 2k+n
Collateral
0 0 n! x j=1 j k=1
Proof
2 (2k -1)!! 2 (2k -1)!!(2k)! 2k
sech -1x = log -Σ x 2k = log -Σ x
x k=1 (2k)!! 2k x k=1 (2k)!! 2k (2k)!
2
2 (2k -1)!!
= log -Σ x 2k (2k)! = (2k)!!(2k -1)!!
x k=1 2k (2k)!
Integrating both sides of this with respect to x from 0 to x repeatedly, we obtain the desired expression.
The collateral higher integral of this was as follows as shown in Formula 4.5.3' ( 4.5 ).
sech
n-2r-1
x x xn (n -1)/2 (2r -1)!! x
n
Σ
-1 -1
x dx = sech x + sin -1x
0 0 n! r=0 (2r)!!(2r +1)! (n -2r -1)!
n-2r
s n -2rC s
(s -1)!!2 x
n /2 n -2r
+Σ Σ (-1) 2 (n -2r)!
1- x 2
r=1 s=0 2r +s (s +2r -1)!!
n-2r
n /2 1 x
-Σ
r=1 2r(2r -1)!!2 (n -2r)!
Therefore, the termwise higher integral above is a collateral higher integral.
And matching both, we obtain the following formula.
Formula 6.7.1'
The following expression holds for 0< x 1 .
n-2r
2
xn
+ Σ 2r(2r -1)!!
(2k -1)!! 2 n 1 n /2 1 x
Σ log +Σ
2k+n
x =
k =1 2k (2k +n)! n! x j=1 j r=1
2 (n -2r)!
n-2r-1
xn (n -1)/2 (2r -1)!! x
- sech x - Σ
-1
sin -1x
n! r=0 (2r)!!(2r +1)! (n -2r -1)!
n-2r
s n -2rC s
(s -1)!!2 x
n /2 n -2r
-Σ Σ (-1) 2 (n -2r)!
1- x 2
r=1 s=0 2r +s (s +2r -1)!!
- 12 -
2
(2k -1)!! 1 n 1 n /2 1 1
Σ 2k (2k +n)!
=
n!
log 2+Σ
j=1 j
+Σ
(n -2r)!
k=1 r=12r(2r -1)!!2
(n -1)/2 (2r -1)!! 1
- Σ
2 r=0 (2r)!!(2r +1)! (n -2r -1)!
- 13 -
6.8 Termwise Higher Integral of arccsch x
Formula 6.8.1
The following expression holds for 0< x <1 .
xn 2
-Σ(-1)
x x 2 n 1
k (2k -1)!!
-1 n
csch x dx = log +Σ x 2k+n Collateral
0 0 n! x j=1 j k =1 2k (2k +n)!
Proof
2 (2k -1)!! 2k
csch -1x = log +Σ(-1)k-1 x
x k=1 (2k)!! 2k
2 (2k -1)!!(2k)! 2k
= log -Σ(-1)k x
x k=1 (2k)!! 2k (2k)!
2
2
k (2k -1)!!
= log -Σ(-1) x 2k (2k)! =(2k)!!(2k -1)!!
x k=1 2k (2k)!
Integrating both sides of this with respect to x from 0 to x repeatedly, we obtain the desired expression.
The collateral higher integral of this was as follows as shown in Formula 4.5.3' ( 4.5 ).
csch
n-2r-1
x x xn (n -1)/2 (-1)r(2r -1)!! x
n
Σ
-1 -1
x dx = csch x + sinh -1x
0 0 n! r=0 (2r)!!(2r +1)! (n -2r -1)!
n-2r
n /2 n -2r
r+s n -2r Cs
(s -1)!!2 x
+Σ Σ (-1) x 2+1
r=1 s=0 2r +s (s +2r -1)!!2 (n -2r)!
n-2r
n /2 (-1)r x
-Σ
r=1 2r(2r -1)!!2 (n -2r)!
Therefore, the termwise higher integral above is a collateral higher integral.
2011.02.15
K. Kono
Alien's Mathematics
- 14 -
7 Super Integral (Non-integer order Integral)
Definition 7.1.1
<p>
f (x) obtained by continuing analytically the index of the integration operator of a higher primitive function
<n>
f (x) to a complex plane [0 , p] from a natural number interval [1 , n ] is called Super Primitive Function
f(x) . f p (x)
<>
of may mean the Super Indefinite Integral, and may mean a Super Integration Function.
Example
p
(sin x) p = sin x - + c (x)
< >
p cp(x) is an arbitrary function.
2
Definition 7.1.2
We call it Super Integral to integrate a function f with respect to an independent variable x from a(0)
to a(p) continuously. And it is described as follows.
x x x x
f(x)dx p = f(x)dx dx
a(p) a(0) a(p) a(0)
And
when a(k)= a for all k [0 , p] , we call it super integral with a fixed lower limit ,
when a(k) a for some k [0 , p] , we call it super integral with variable lower limits .
Example
sin xdx
x x (-1)r p
p
=Σ x 2r+1+
0 0 r=0 (2r +2+ p)
-1-
p
x x
sin xdx p = sin x -
p 0 2
2 2
Theorem 7.1.3
r [0 , p ] be an continuous function on the closed interval I
<r>
Let f and be arbitrary the r-th order
primitive function of f . And let a(r) be a continuous function on the closed interval [0 , p] .
Then the following expression holds for a(r), x I .
x x p -1 x x
<p>
f(x)dx p = f (x) - Σ f <p -r>
a(p -r) dx r (1.1)
a(p) a(0) r=0 a(p) a(p -r+1)
Especially, when a(r)= a for all k [0 , p ] ,
f(x)dx
x x
<p> p -1
<p -r> (x -a)r
p
=f (x) - Σf (a) (1.2)
a a r=0 (1+ r)
Constant-of-integration Function
p -1
We call Σ
r=0
etc. Constant-of-integration Function of f(x) . Since p is a real number, in order to obtain
p -1
Σ
r=0
etc. generally, the calculus about the calculus operator <r>,(s) is required. This is very difficult.
However, it becomes easy exceptionally at the time of f(x)= e x , and we can obtain the following expression
from (1.2) .
p -1 (x -a)r x x (x -a)r (x -a)r+ p
Σ e a
= e x
- e xdx p = e aΣ -
r=0 (1+ r) a a r=0 (1+ r) (1+ r + p)
(x -a)r x x (x -a)r+ p
e =e Σ
a x
, e dx = e aΣ
x p
r=0 (1+ r) a a r=0 (1+ r + p)
Definition 7.1.4
x x p -1 x x
<p> <p -r>
f(x)dx = f p
(x) - Σ f a(p -r) dx r (1.1)
a(p) a(0) r=0 a(p) a(p -r+1)
In this expression,
when Constant-of-integration Functin is 0,
x x
we call f(x)dx p Lineal Super Integral and
a(p) a(0)
we call the function equal to this Lineal Super Primitive Function.
-2-
x x
we call f(x)dx p Collateral Super Integral and
a(p) a(0)
we call the function equal to this Collateral Super Primitive Function.
In short , Lineal Super Primitive Function is what integrated f(x) with respect to x continuously without
considering the constant-of-integration function.
Example
Left : collateral integral Right : collateral primitive function
p p
x x p -1 x x
sin xdx = sin x - p
- Σ sin (p -r)- dx r
p 0 2 r=0 2 p p-r+1
e dx =Σ
r+ p r+ p
x x xr x x
x p
= e -Σ
x
-
0 0 r=0 (1+ r) 1+ r + p r=0 1+ r + p
Left : lineal integral Right : lineal primitive function
p
x x
sin xdx p = sin x -
p 0 2
2 2
e dx
x x
x p
= ex
- -
7.1.5 The Necessary Conditions for the Super Integral being Lineal
In Theorem 7.1.3 (1.1), since the higher integral of 1 can be arbitrary value, in order for Constant-of-integration
= 0 for all r [0 , p ] .
<r>
Function to be 0, we find out that it must be f a(r)
Since this is important, it is stated here as a theorem.
Theorem 7.1.5
x x p -1 x x
<p>
f(x)dx p = f (x) - Σ f <p -r>
a(p -r) dx r (1.1)
a(p) a(0) r=0 a(p) a(p -r+1)
f(x)dx
x x
<p> p -1
<p -r> (x -a)r
p
=f (x) - Σf (a) (1.2)
a a r=0 (1+ r)
The necessary condition for the super integral being lineal is as follows respectively.
for all r [0 , p]
<r>
f a(r) =0
for all r [0 , p]
<r>
f (a) =0
That is, the necessary condition for the super integral being lineal is that a(r) or a are zeros of the super
<r>
primitive function f for all continuous r from 0 to p.
In addition, this is not sufficient condition as well as the case of Higher Integral.
Note
Although I was taught to Mr. Sugimoto and knew in 2005, seemingly such integration is the field which is
called "Fractinal Integral" and studied in Europe in recent years. I thought I follow this term in this text.
However, although the number of times of integration is extensible even to a complex plane, wrapping cloth of
"Fractional" is too small. So, in this text, I decided to use "Super Integral" that I was using from before.
-3-
7.2 Fractional Integral
1 x
D -p
f(x) = (x -t)p-1 f(t)dt (a is a zero of the left side ) (2.0)
a x
(p) a
-p
The left side a Dx f(x) is Non-integer order Primitive Function , and this is the same as Super Primitive
<p>
Function f (x) defined in the previous section. Notation aDx-p is called Riemann-Liouville operator.
On the other hand, the integration of the right side is called Riemann-Liouville Integral , and this is equivalent
to Super Integral defined in the previous section.
Although many of super primitive function (non-integer order primitive function) cannot be expressed with an
elementary function, the some can be expressed with elementary functions. In traditional Fractional Integral ,
the super primitive function expressed with these elementary functions are drawn from Riemann-Liouville
integral. However, the method is very difficult. Two examples are shown below.
(x -t)p-1
f(t) = t , g(x -t) =
(p)
then
f(t)g(x -t)dt
<p> 1 x x
x t (x -t)
p-1
= dt =
(p) 0 0
(f g)(x) . Then we take the Laplace transform of (f g)(x)
We find out that this is a convolution ;
(f g)(x) F(s)G(s)
( +1)
f(x) = x
= F(s)
s +1
x
p-1
1 (p) 1
g(x) = = = G(s)
(p) (p) s p sp
( +1) 1 ( +1) ( +p +1)
F(s)G(s) = =
s +1 s
p ( +p +1) s +p+1
( +1) a+p
Finally, taking the inverse Laplace transform, we obtain x .
( +p +1)
x
(2) super primitive function of f(x)= e
<p> 1 x
e x = (x -t)p-1 e tdt
(p) -
p-1
-
The indefinite integral of the right-hand side is expressed as follows using (p, x)= e d .
x
-4-
(x -t)
x <p> 1 x ex
(p, x -t)-
p-1 x
e = t
e dt =
(p) - (p)
x
e ex
= (p, x - x) - (p, x +) = (p)
(p) (p)
= ex
Theorem 7.2.3
When f(x) denotes a continuously differentiable function and (z) denotes Gamma Function,
the following expression holds for any p >0 .
x x 1 x
f(x)dx p = (x -t)p-1 f(t)dt (2.1)
a a (p) a
Proof
Analytically continuing the index of the integration operator in Theorem 4.2.3 to [0 , p ] from [1 , n ],
we obtain the desired expression.
Remark
This theorem means that Super Integral with a fixed lower limit is equivalent to Fractional Integral.
However, in Super Integral, instead of Riemann-Liouville integral, the Higher Integral is used to draw the super
primitive function expressed with elementary functions. That is, we ask for the function form of the higher order
primitive function by the higher integral first, and extend the order to the real number from an integer. In Super
Integral, by this way, we can easily obtain the super primitive function which is expressed with elementary
-5-
functions, without the difficult calculation such as previous examples. And in order to verify the result
numerically, Riemann-Liouville integral is used.
e dx (p)
x x 1 x
x p
= (1)pe x = (x -t) p-1 t
e dt
log x - (1+ p) - p
log xdx (p)
x x 1 x
p p-1
= x = (x -t) log t dt
0 0 (1+ p) 0
Middle is the Super Integral obtained by analytically continuing the index of the operator of the Higher Integral
to real number from natural number, and the right side is Riemann-Liouville Integral. These formulas are inputted
into mathematical software, arbitrary one point is chosen suitably, and the function values are calculatied. The
value of bothe sides are in agreement and it is shown numerically that Super Integral and Riemann-Liouville
Integral are equal.
-6-
7.3 Super Integral of Power Function
Formula 7.3.1
(1) Basic form
(1+ ) +p
(x -t)
x x 1 x
x dx p =
p-1
x = t dt ( 0)
0 0 1+ + p (p) 0
(- - p) +p
(x -t)
x x 1 x
x dx p = (-1)p x = p-1
t dt ( <-p)
(- ) (p)
(2) Linear form
(1+ )
x p
x 1
(ax +b) dx p = (ax +b)+p ( 0)
b b a (1+ +p)
- -
a a
1 x
= (x -t)p-1 (at +b) dt
(p) b
-
a
(- - p)
x p
x 1
(ax +b) dx = - p
(ax +b)+p ( <-p )
a (- )
(x -t)
1 x
= p-1
(at +b) dt
(p)
Note
When -p < 0 , we do not define the super-integral of a power function. It is because a lower limit of
the super integral changes irregularly at this time.
Example 1 : When 0
10
1
(1+1)
1 11 11
1+ 100
x
1 10 10 10
= x = x = 0.955579 x
1 1
1+1+ 11
10 10
10
9
(1+1)
9 19 19
1+ 100
x
1 10 10 10
= x = x = 0.547239 x
9 9
1+1+ 171
10 10
2 <e> (1+2) 2+e 2
x = x = x 2+e = 0.026755 x 4.718281828
1+2+e 3+ e
1 <i> (1+1) 1+ i x 1+
i
x = x = = (1.200176 - 0.630568 i) x 1+ i
1+1+ i 1+ i !
<1/10> <9/10>
When x 1 , x 1 , x 1 , x 2/2 are drawn on a figure side by side, it is as follows.
-7-
y
50
40
30
20
10
0
0 1 2 3 4 5 6 7 8 9 10
x
x
100/11*x^(11/10)/gamma(1/10)
100/171*x^(19/10)/gamma(9/10)
1/2*x^2
1 1 - +
= - (5x +4) 2 2
(5x +4)5/2 5 5/2
4 5i 0.0672835
=- =- i
75 (5x +4) 5x +4
-2 <1- i > 2- 1- i -2+1- i
x = (-1)1- i x
2
1 i! -11.524427 + 3.585646 i
=- =
(-1) i x 1+ i x 1+
i
2
1
(1-1/2) -1+ 2
1 1 1
-
x
-1
= (-1) x 2
= i x 2
1
2
1
x
1 1 1
x
- x x -
2
= 2
dx 2
(-p = < 0)
0
= x
<2> x x
x -1 -1
dx 2 = x(log x -1) (-p < < 0)
0 1
Formula 7.3.2
Let n be a non-negative integer, -1!! 1, (2n -1)!! 135(2n -1) ,
0!! 1, 2n !! 2462n ,
then following expressions hold.
-8-
(1) Basic form
2
1 1
2(2n)!! n+
x
n 2
= x
(2n +1)!!
2
1
(2n +1)!!
x
1
n+
2 n+1
= x
2(2n)!!(n +1)
(2) Linear form
1
1 1
2
2 1 2(2n)!! n+
(ax +b) n 2
= (ax +b)
a (2n +1)!!
2
1
1
(2n +1)!!
(ax +b)
1 2
n+ 1
2
= (ax +b)n+1
a 2(2n)!!(n +1)
Proof
From Formula 7.3.1 (2) Linear form,
1
1
(1+n)
2 1
2 1 n+
(ax +b) n 2
= (ax +b)
a 1
1+n +
2
1
1+ n +
2
1 1
(ax +b) = 1a
1 2
n+
2 2 n+1
(ax +b)
1+n + +
1 1
2 2
n + =
3 (2n +1)!!
Where, since n+1
,
2 2
(1+n) (1+n) 2n+1 n ! 2(2n)!!
= = =
(2n +1)!!
(2n +1)!!
1 3
1+n + n+
2 2
n +
1 3
1+ n +
2 2
=
(n +2)
1+n + +
1 1
2 2
(2n +1)!! n+1 (2n +1)!! n+1
= x = x
2n+1 (n +1)! 2(2n)!!(n +1)
Substituting these for above expressions, we obtain the linear form. And giving a =1 , b =0 ,
we obtain the basic form.
Example 1
2
1 1 1
20!! 2
x
0 2 2
= x = x
1!!
-9-
2
1 3 3
22!! 4
x
1 2 2
= x = x
3!! 3
2
1 5 5
24!! 16
x
2 2 2
= x = x
5!! 15
2
1 7 7
26!! 32
x
3 2 2
= x = x
7!! 35
Example 2
2
1
1!! 1
x
1
2
= x = x1
20!! 2
2
1
3!! 2 3 2
x
3
2
= x = x
22!!2 8
2
1
5!! 3 15 3
x
5
2
= x = x
24!!3 48
2
1
7!! 4 35 4
x
7
2
= x = x
26!!4 128
Formula 7.3.3
When n denots a natural number, the following expression holds.
1 1
2 n (-1)k n
k x
2 n+
x
n
Σ
2
=
k=0 2k +1
Proof
Let n be a natural number, = n , p =1/2 . Then since >-p , Applying the Formula 7.3.1 to these,
we obtain the following expression.
1 1
(x -t)
2 1 x - 1 x tn
x
n 2 n
= t dt = dt
(1/2) 0 0 x -t
Where, the following equation is known. ( 岩波数学公式Ⅰ p96 ).
(x -t)n-r
tn
2 x -t n n
dt = Σ (-x)r
x -t (-1)n+1 r=0 r 2n -2r +1
Then
- 10 -
x
(x -t)n-r
tn
x 2 x -t n n
dt = Σ (-x)r
0 x -t (-1)n+1 r=0 r 2n -2r +1
0
1
xn
r
2 n n
Σ(-1)r
2
= x
(-1)n r=0 2n -2r +1
1
(-1)r-n
r x
n n n+
= 2Σ 2
r=0 2n -2r +1
2
1 1
(-1)r-n
r x
2 n n n+
x n
= Σ
2
r=0 2n -2r +1
Where, we devise further.
r-n
(-1)n-r (-1)k
r k
n (-1) n n n n n
Σ
r=0 2n -2r +1
=Σ
r=0 2(n -r)+1 n -r
=Σ
k=0 2k +1
Using this, we obtain
1 1
(-1)k
k x
2 2 n n n+
x
n
Σ
2
=
k=0 2k +1
By-product
According to Formula 7.3.2,
1 1
2 2(2n)!! n+
x
n 2
= x
(2n +1)!!
Since this must be consistent with Formula 7.3.3, then
k
k
2 n n
(-1) 2(2n)!!
Σ =
k=0 2k +1 (2n +1)!!
From this,
(-1)k
k
n n (2n)!!
Σ
k =0 2k +1
=
(2n +1)!!
(3.2)
If expanded, it is as follows.
0
1 0 0!!
=
1 1!!
1 0 3 1
1 1 1 1 2!!
- =
3!!
k
n (-1) n (1n)! 1
Σ
k =0 1k +1
=
(1n +1)!
=
n +1
(3.1)
- 11 -
(-1)k
k
n n (3n)!!!
Σ
k =0 3k +1
=
(3n +1)!!!
!!! means triple factorial. (3.3)
k
Σ mk +1 k
(-1) n
n (mn)!m
= !m means multi factorial. (3.n)
k =0 (mn +1)!m
Seemingly, it is not known m= 3 or more.
Lemma
When m,n are natural numbers, the following expression holds.
1
m-1
(x -t)n-r
m
r
- m(x -t) n n
(x -t) m t n dt = n+1 r=0
(- x) r
Σ m(n -r)+1
(-1)
Proof
1
m
(x -t)n-r
r
m(x -t) n n
F(t) = Σ (-x) r
(-1)n+1 r=0 m(n -r)+1
Differentiate this with respect to t. Then
1
m
(x -t)n-r
r
d d m(x -t) n n
dt
F(t) =
dt (-1)n+1 Σ
r=0
(-x) r
m(n -r)+1
1
m
(x -t)n-r
m(x -t) d n n
+ Σ (-x)r
(-1)n+1 dt r=0 r m(n -r)+1
1
-1
m
(x -t)n-r
r
(x -t) n n
=- Σ (-x) r
(-1)n+1 r=0 m(n -r)+1
1
m
(x -t)n-r-1
r
m(x -t) n n
- Σ(n -r)(-x)r
(-1)n+1 r=0 m(n -r)+1
1
-1
m
(x -t)n-r
r
(x -t) n n
=- Σ (-x) r
(-1)n+1 r=0 m(n -r)+1
1
-1
m
(x -t)n-r
r
(x -t) n n
- Σ m(n -r)(-x) r
(-1)n+1 r=0 m(n -r)+1
1
-1
m
(x -t)n-r
r
(x -t) n n
=
(-1)n
Σ
r=0
1+m(n -r)(-x)
r
m(n -r)+1
- 12 -
1 1
-1 -1
m m
r
(x -t) n n (x -t)
= Σ (-x)r(x -t)n-r = (-t)n
(-1)n r=0 (-1)n
m-1
-
m n
= (x -t) t Q.E.D
Using this lemma, we obtain the following formula.
Formula 7.3.4
When m,n are natural numbers, the following expressions hold
1 1
n (-1)k
x
m m n n+
(1/m) Σ
x
n m
= (4.1)
k=0 mk +1 k
(-1)k (1+ n , 1/m)
k =
n n
Σ ( ) denots beta function. (4.2)
k =0 mk +1 m
Proof
(1+ ) +p
(x -t)
<p> 1 x
x
p-1
= x = t dt ( 0)
1+ +p (p) 0
Giving = n , p =1/m ,
m
1 m-1
(x -t)
1 x -
x n = m n
t dt
(1/m) 0
Using above Lemma, we calculate as follows.
x
1
m-1 n-r
m
r
x - m(x -t) n n (x -t)
(x -t) m n
t dt = Σ(-x)r
0 (-1)n+1 r=0 m(n -r)+1
0
1 n
r
m n n x
=
(-1)n
x m
Σ(-1)r
r=0 m(n -r)+1
1
(-1)r-n
r x
n n n+
= mΣ m
r=0 m(n -r)+1
Furthermore, since r,n are integer,
r-n
(-1)n-r (-1)k
r k
n (-1) n n n n n
Σ
r=0 m(n -r)+1
=Σ
r=0 m(n -r)+1 n -r
=Σ
k=0 mk +1
Using this,
m-1 1
(x -t)
(-1)k
k x
x - n n n+
m n
t dt = mΣ m
0 k =0 mk +1
Thus, we obtain
1 1
n (-1)k
k x
m m n n+
(1/m) Σ
x
n m
= (4.1)
k=0 mk +1
Next,
- 13 -
m
1
(1+ n) (1+ n) 1/m
1 1
n+ n+
x
n m
= x = x m
1+ n +1/m 1/m 1+ n +1/m
Since this must be equal to (4.1), we obtain
k
(1+ n) 1/m (1+ n , 1/m)
k
n (-1)n
Σ
k =0 mk +1
=
m 1+ n +1/m
=
m
(4.2)
Remark
(4.2) shows that (4.1) can be expressed with a beta function and n can be the real number. Actuality,
(1+ ) 1 (1+ )(p) 1+ , p
= =
1+ +p (p) 1+ +p (p)
Then,
<p> (1+ ) 1+ , p +p
x = x +p = x ( 0) (4.3)
1+ +p (p)
Therefore, also,
(x -t)
x
p-1
t dt = 1+ , p x +p ( 0) (4.4)
0
Example 1
3
1
(1+2)
1 7 7
2+ 2
x 2
= x 3
= x3 = 0.7199 x 3
1+2+1/3 10/3
3
1 7
2 (-1)k
k x
3 2
x
(1/3) Σ
2 3
=
k =0 3k +1
7 7
4
1 13
3 (-1)k
k x
4 3
x
(1/4) Σ
3 4
=
k =0 4k +1
13 13
3 x
4 1 3 1 3 1 3 1 3 4 4
= - + - = 0.7241 x
(1/4) 1 0 5 1 9 2 13
Example 2
0 1 2
1 2 1 2 1 2 B(1+2 , 1/3) 9
- + = =
1 4 7 3 14
1 9 1 9 1 9 1 9 B(1+9, 1/3) 1, 594, 323
- + -+ - = =
1 0 4 1 7 2 28 9 3 3, 803, 800
3 1 3 1 3 1 3 B(1+ 3, 1/4) 128
- + - = =
0 5 1 9 2 13 3 4 195
- 14 -
Formula 7.3.5
When n is a natural number, the following expressions hold for p >0 .
n (-1)k
n <p> 1 n
x = Σ x n+ p (5.1)
(p) k=0 p + k k
n -1 (-1)k
n -1
(n , p) = Σ ( )denots beta function. (5.2)
k =0 p + k k
Example 1
<e> (1+2) 2
x 2
e e
= x 2+ = x 2+ = 0.026755 x 2+ e
1+2+ e 3+ e
1 2 (-1)k 2
2 <e>
x = Σ
(e) k=0 e + k k
x 2+ e
2
1 1 2 1 2 1 2 e e
= - + x 2+ = 0.026755 x 2+
(e) e 0 e +1 1 e +2
Example 2
1 1 1 11 1
(2 , e) = -- = 0.098938
=
0 e 1 e e +1
e +1
1 2 1 2 1 2
(3 , ) = - + = 0.029896
0 +1 1 +2 2
Formula 7.3.6
The following expressions hold for positive numbers p ,q .
q+p r
q <p> x (-1) q
(p) Σ
x = (6.1)
r=0 p + r r
r
(-1) q -1
(q , p) = Σ ( )is Beta Function (6.2)
r=0 p + r r
2 2
1 1 5
16
x 2-2 x + = (x -) (x -)
2 2 2
=
15
- 15 -
If this is calculated termwise as follows, the result is different from the former.
1
1 1 1
2 2 2 2
x -2 x + - 2x + x
2 2
2
= x 2 1 0
8 2 4
1
2
x +
2 2
= x x -
15 3
Needless to say, this cause is the difference between
1 1
x x x x
x -2 x + dx and
2 2 2
x -2 x + dx 2 .
2 2
0 0
That is, it is because the latter regarded it as 0 although the former regarded the zero of the super primitive
function as pi.
The latter is right when there is a special reason why the zero of the super primitive function should be 0.
However, such a case is rare, and in almost all cases the former is right.
- 16 -
7.4 Super Integral of Exponential Function
Formula 7.4.1
(1) Basic form
(x -t)
x x 1 x
x
e dx p = (1)pe x = p-1
e dt
t
(p)
(2) Linear form
(x -t)
x p
x 1 1 x
e
ax+b
dx p
= e
ax+b
= p-1
e
at+b
dt
a (p)
a >0 : - , a <0 : +
(3) General form
(x -t)
x p
x 1 1 x
ax+bdx p = ax+b = p-1
at+bdt
a log (p)
a >0 : - , a <0 : +
Example
2
1 1
e -x
= (-1) e 2 -x
= i e -x
2
3x-4 2
1
e = e 3x-4 = 0.211469 e 3x-4
3
2
1
1/2
1
3
x
= 3x = 0.954064 3x
log 3
1
1
2
2 1
(-3)x = (-3)x
log(-3)
= (0.447018 - 0.317241 i) (-3)x
1 i x
2 x <i>
=
log 2
2 = (0.933582 + 0.358362 i) 2x
<1/2> <1>
When 3x , 3x , 3x are drawn on a figure side by side, it is as follows.
- 17 -
y
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
x
3^x/ln(3)
3^x*(1/ln(3))^(1/2)
3^x
Formula 7.4.2
(1) Basic form
p i p i
x x
x
e dx p = i p cosh x - sinh x -
2 2
= (cosh x)<p> (sinh x)<p>
(2) Linear form
p i p i
e
p p
x
x i
ax+b
dx p
= cosh ax +b - sinh ax +b -
a 2 2
= cosh(ax +b)<p> cosh(ax +b)<p>
Proof
From the Formula 7.7.1 mentioned later, the following expressions hold.
p i
1 x
(cosh x) p = i p cosh x - e + (-1)pe -x
< >
=
2 2
p i
1
(sinh x) p
<>
= i p sinh x - = e x - (-1)pe -x
2 2
Adding and subtracting these two formulas, we obtain the following expressions.
p i p i
i p cosh x -
2 + sinh x -
2 =e x
p i p i
ip cosh x -
2 - sinh x -
2 = (-1) e p -x
Substituting these for the Formula 7.4.1, we obtain (1) basic form.
In a similar way, (2) linea form is obtained
Note
Therefore, in the concept, the following expression holds.
- 18 -
x x x x x x
x
e dx p = cosh xdx p sinh xdx p
(p -1) i -1 i p i 0 i
2a 2a 2 2
Actually, when p is a natural number n, both sides recover the function of operation immediately and the result
in the higher integral as follows..
e
x x x x x x
x
dx =n
cosh xdx
n n
sinh xdx
(n -1) i 0 i n i 1 i
2a 2a 2 2
- 19 -
7.5 Super Integral of Logarithmic Function
Analytically continuing the index of the integration operator in Formula 4.3.3 to [0 , p ] from[1 , n ],
we obtain the following formula. In addition, Rieamnn-Liouville integrals are also expressed together
Formula 7.5.1
(1) Basic form
log x - (1+p) - p
(x -t)
x x 1 x
log xdx p = x =
p-1
log t dt
0 0 (1+p) (p) 0
(2) Linear form
log(ax +b)- (1+ p) -
x p
x b
p
log(ax +b)dx = x+
b b (1+ p) a
- -
a a
1 x
= (x -t)p-1 log(at +b)dt
(p) b
-
a
Proof
Formula 4.3.3 in "04 Higher Integral" was as follows.
log x dx
x xn
x n 1
n
= log x -Σ
0 0 n! k=1 k
log(ax +b)dx
x n
x 1 b n 1
n
= x+ log(ax +b) -Σ
b b n! a k=1 k
- -
a a
On the other hand, in "01 Gamma Function & Digamma Function" there were the following formulas.
n 1
n ! = (1+ n), Σ = (1+n) +
k =1 k
Then, subsutituting these for the upper formulas and analytically continuing the index of the integration operator
to [0 , p ] from[1 , n ], we obtain the desired expressions.
Example
log x - (1+1) - 1
(log x)<1> = x = x(log x -1)
(1+1)
1 1
log(3x +4) - 1+ -
2
1
4 2
2
log(3x +4) = x+
1 3
1+
2
= 1.1283(log x - 0.6137) x +1.3333
1
log x - 1+ -
10
1 1
10
(log x) = x 10
= 1.0511 10 x (log x -0.1534)
1
1+
10
- 20 -
9
log x - 1+ -
=
9 9 9
10 10 10 10
(log x) x = 1.0397 x (log x -0.9333)
9
1+
10
When log x , (log x)<1/10> , (log x)<9/10> , (log x)<1> are drawn on a figure side by side,
<1/10>
it is as follows. We can see slightly that the zero of (log x) (red) is also x =0 .
y
6
0
1 2 3 4 5 6 7
x
-2
ln(x)
-10*x^(1/10)/gamma(1/10)*(EULER - ln(x) + psi(1/10
-10/9*x^(9/10)/gamma(9/10)*(EULER - ln(x) + psi(9/
x*(ln(x) - 1)
- 21 -
7.6 Super Integral of Trigonometric Function
Formula 7.6.1
(1) Basic form
p
x x
sin xdx p = sin x -
p 0 2
2 2
p
x x
cosxdx p = cos x -
(p -1) -1 2
2 2
(2) Linear form
p
p
x x 1
sin(ax +b)dx p = sin ax +b -
p b 0 b a 2
- -
2a a 2a a
p
p
x x 1
cos(ax +b)dx p = cos ax +b -
(p -1) b 1 b a 2
- - -
2a a 2a a
Example
1
1
(sin x ) 2
= sin x -
2 2 = sin x - 4
1 1 1
((sin x ) 2 ) 2 = (sin (x - )) 2
1
4
2
1
= sin x - 2 = -cos x
1
= sin x - -
1 4 2 2
<1> 1
1
= sin x
1
sin x + = sin x + -
2 1 2 2
1
1
(cos x) 2
= cos x -
2 2 = cos x - 4
2
2
(cos x)
= cos x -
2 = cos(x -1)
When cos x , (cos x)1/2 , sin x are drawn on a figure side by side, it is as follows. Red shows 1/2
order super integral. It is clear also in the figure that super integral which is the easiest to understand is super
integral of trigonometric functions.
- 22 -
7.6.2 Termwise Super Integrals of sin x , cos x
If common lower limit 0 is employed as the lower limits of the super integral, we obtain the following termwise
super integral. These are collateral super integrals as understood from the constant-of-integration function in
the right side
p
sinxdx
(-1)k
+ C(p, x)
x x
2k+1+ p
p
=Σ x = sin x -
0 0 k=0 (2k +2+ p) 2
k
p-k
p x
C(p, x) = Σ sin
k =1 (1+p -k) 2
p
cosxdx (-1)k
+ C(p, x)
x x
p
p
=Σ x 2k+ = cos x -
0 0 k =0 (2k +1+ p) 2
k
p-k
p x
C(p, x) = Σ cos
k =1 (1+p -k) 2
When the 1/2th order collateral super integral of cos x is drawn as cos x and sin x side by side, it is as
follows. Red shows the 1/2th order collateral super integral.
- 23 -
7.7 Super Integral of Hyperbolic Function
Formula 7.7.1
(1) Basic form
p i
e x - (-1)pe -x
x x
p p
sinh xdx = i sinh x - =
p i 0 i 2 2
2 2
p i
e x + (-1)pe -x
x x
cosh xdx p p
= i cosh x - =
(p -1) i -1 i 2 2
2 2
(2) Linear form
p i
p
x
x i
sinh(ax +b)dx p
= sinh ax +b -
p i b 0 i b a 2
- -
2a a 2a a
p
1 1
e - (-1)pe -(ax+b)
ax+b
=
2 a
p i
p
x
x i
cosh(ax +b)dx p = cosh ax +b -
(p -1) i b -1 i b a 2
- -
2a a 2a a
p
e
1 1
= ax+b
+ (-1)pe -(ax+b)
2 a
Example 1
1 i i
1 1 1
(sinh x ) 2 2
= i sinh x -
2 2 2
= i sinh x - 4
i 2
1 1 1 1
((sinh x ) 2 ) 2 = (i 2 sinh (x - ))
4
1
i 1 i
1 2
4
2 i
=i sinh x - -
1 2 2
i
= i sinh x -
2
= cosh x
i <1> 1
i 1 i
= i cosh x
i
cosh x + = cosh x + -
2 1 2 2
i i
(cosh x) 10 = 0.854635 cosh x +
10
=i cosh x +
20 20
- 25 -
9i 9i
9 9
(cosh x) 10 =i 10
cosh x + 20 = 0.243237 cosh x + 20
Super integral of hyperbolic function is the most incomprehensible in super integrals. The reason is that the
super primitive function turns into a complex function except order p is an integer or a purely imaginary number.
i 9i
Then, when
cosh x , (cosh x) 10 , (cosh x) 10 , sinh x which can be displayed on a real number
domain are drawn on a figure side by side, it is as follows.
i
10
All of four curves have overlapped in the positive area. It is natural that (cosh x) is near cosh x in
9i
10 is far apart from sinh x
a negative area. But (cosh x) in why.
Example 2
1
1 x
2 e - (-1) e 2 -x
e x - i e -x
(sinh x ) = =
2 2
2
1
2
1
1 1
e x - i e -x ex i
((sinh x ) 2 ) 2 = = - e -x
2 2 2
ex i e x + e -x
= - i e =
-x
= cosh x
2 2 2
i i
i
<1>
1-1 x+
- x+
2 1 2
cosh x + = e + (-1) e
2 2
i i
=
1
2
-
e 2 e x - e 2 e -x =
1
2
1
i e x - e -x
i
i
= e x + e -x = i cosh x
2
i i
i -x
e x - (-1) e -x
i
e x - e e e x - e -1-x
(sinh x) = = =
2 2 2
- 26 -
Note
Super integral of a hyperbolic function is the integral with variable lower limits, and super integral of an
exponential function is the integral with a fixed lower limit. So, for example, the p-th order super integral
of sinh x is as follows in the concept.
<p>
x x
p
x x e x - e -x p
(sinh x) = sinh xdx = dx
p i 0 i p i 0 i 2
2 2 2 2
But the example forgets such a thing and perform super integral of e x , e -x with lower limits - , .
The result is all right. Truly, this is convenience and interesting.
p i
sinhxdx
p
x 2k+1+
+ C(p, x)
x x
p
=Σ p
= i sinh x -
0 0 k=0 (2k +2+ p) 2
k i
p-k
p x
C(p, x) = Σ sinh
k =1 (1+p -k) 2
p i
coshxdx
p
x 2k+
+ C(p, x)
x x
p
=Σ = i p cohs x -
0 0 k=0 (2k +1+ p) 2
k i
p-k
p x
C(p, x) = Σ cosh
k =1 (1+p -k) 2
When the values of the lineal and the collateral the 1.2th order integral of cosh x on x=1, x=8 are calculated
respectively, it is as follows.
Though the difference of both is large where x is small, both are almost corresponding ing where x is large.
Since this is similar also in sinh x, it is thought that the termwise super integrals of sinh x and cosh x are
asymptotic expansions of the lineal super integrals.
- 27 -
7.8 Super Integral of Inv-Trigonometric Function etc.
Formula 7.8.1
When (x),(x) denote gamma function and digamma function respectively, the following expressions
hold for x 1 .
tan -1x
x x p
tan x dx =-1 p
Σ ( -1) k
x
p-2k
0 0 (1+ p) k =0 p -2k
log1+ x 2
p p+1-2k
+ Σ
2(1+ p) k =1
(-1)k
p +1-2k
x
1 p
(1+ p )- (2r)x
p+1-2r
- Σ
(1+ p) r=1
(-1)r
p +1-2r
xp tan -1x
x x p
cot x dx =
-1 p -1
cot x - Σ (-1)k x
p-2k
0 0 ( p)
1+ ( p) k =1
1+ p -2k
log1+ x
2
p p+1-2k
- Σ
2(1+ p) k =1
(-1)k
p +1-2k
x
1 p
(1+ p )- (2r)x
p+1-2r
+ Σ
(1+ p) r=1
(-1) r
p +1-2r
-1
Example: The 5/2 th order integral of cot x
- 28 -
Formula 7.8.2
When (x),(x) denote gamma function and digamma function respectively, the following expression
holds for x 1 .
tanh -1x
x x p p-2k
p
Σ
-1
tanh x dx = x
0 0 (1+ p) k =0 p -2k
log1- x 2
p p+1-2k
+ Σ
2(1+ p) k =1 p +1-2k
x
1 p
(1+ p )- (2r)x
p+1-2r
(1+ p) Σ
-
r=1 p +1-2 r
-1
Example: The 3/2 th order integral of tanh x
2010.07.07
K. Kono
Alien's Mathematics
- 29 -
Examples
Below, the examples of the termwise super integral in Formula 8.1.1 are shown. One arbitrary point is chosen
suitably. fl is the the function value on the point by the formula and fr is the function value on the same point
by Riemann-Liouville Integral. All digits are corresponding to both, and this shows the justification of the above
termwise super integrals numerically. In the figure, blue shows the function to be integrated, redshows the
termwise super integral, green shows the 1st order integral.
-2-
Formula 8.1.2
When B2k , E2k , (x) denote Bernoulli Numbers, Euler Numbers, Gamma Function respectively,
the following expressions hold for any p 0 and any /2< x < .
-3-
9 Higher Derivative
Definition 9.1.1
(x) denotes the derivative function of f (n -1)(x) for n =1, 2, 3, ,
(n)
When f
(n)
we call f (x) ,
or for short , .
Definition 9.1.2
We call it to differentiate a function f with respect to an independent variable x
repeatedly. And it is described as follows.
n
d d d d d d
f(x) = f(x) : n pieces
dx n dx dx dx dx dx
Theorem 9.1.3
r =0, 1, , n
(r )
When f are continuous functions on a closed interval I and are the r th derivative
functions of f, the following expression holds for x I.
dn
) = f n (x)
()
n f(x
(1.1)
dx
Proof
Theorem 4.1.3 in 4.1 can be rewritten as follows.
x x n -1 x x
<n>
f (x) = n
f(x)dx + <n -r>
f an-r dx r
an a1 r=0 an a n- r+ 1
Differentiating both sides with respect to x n times, we obtain
Here, since the constant-of-integration polynomial of the right side is degree n -1 , if this is differentiated n
times, it must become 0. Then we obtain
d n <n>
nf
(x) = f<0>(x)
dx
Shifting by -n the index in the integration operator <> and replacing <> by differentiation operator ( ) ,
we obtain the desired expression.
-1-
Remark
Since differentiation is an inverse operation of integration, it cannot be unrelated to the constant-of-integration.
In fact, the lineal and the collateral exist in Super Differentiation (non-integer times differentiation). (See later
12.1.2.) But this thorem guarantees that only the lineal exists in Higher Differentiation.
-2-
9.2 Higher Derivative of Elementary Functions
1
(ax +b) (n)
= (ax +b)-n 0
a (1+ -n)
1 -n (- + n)
= -
a (-)
(ax +b)-n ( < 0)
Proof
-1
1
(ax +b) (ax +b)-1
(1)
=
a
-2
1
(ax +b) (2)
= ( -1)(ax +b)-2
a
-n
(n) 1
(ax +b) = ( -1) -(n -1)x -n
a
(1+)
Substituting ( -1) -(n -1) = for this, we obtain (2).
(1+ -n)
Furthermore, substituting a =1, b =0 for this, we obtain (1).
However, when is a negative integer, a denominator and a numerator become the infinite form of infinity,
and as for this right side, the value can not be calculated.
Therfore, when <0 let =-. . Then, since (ax +b) = (ax +b)- ,
-1
1
-(ax +b)
(1)
(ax +b)- = -1
a
-2
- (2) 1
(ax +b) = +( +1)(ax +b) -2
a
-n
- (n) 1
(ax +b) = (-1)n( +1)( +n -1)(ax +b) -n
a
Here
( + n)
( +1)( +n -1) =
()
we can substitute this for the above as follows.
-3-
-n
( + n)
- (n) 1
(ax +b) = (-1)n (ax +b) -n
a ()
Then replacing with - we botain (2), furthermore, substituting a =1, b =0 for this, we obtain (1).
ax+b (n) 1
e
ax+b
= e
a
Proof
According to Formula 4.3.2 in 4.3 , it was as follows.
e
x x
x n x
n
dx = ( 1) e
n
e a <0 : +
x a >0 : -
x 1
ax+b n ax+b
dx = e
a
Since the differentiation is a reverse-operation of integration, replacing the index n of the integration operator
with -n , we obtain the desired expression.
(n) n-1 b
log (ax +b ) = (-1) (n -1)! x +
a
Proof
Differentiating log x with respect to x one by one, we obtain the desired expression. As for (2), it is similar.
n
= cos x +
2
(cosx) n
()
-4-
Proof
Replacing the index n of the integration operator with -n in Formula 4.3.4 (4.3 ) , we obtain the desired
expressions.
1 1
e - (-1)-ne -(ax+b)
ax+b
=
2 a
-n
n i
(n) i
cosh(ax +b ) = cosh ax +b +
a 2
-n
1 1
e + (-1)-ne -(ax+b)
ax+b
=
2 a
Proof
Replacing the index n of the integration operator with -n in Formula 4.3.5 (4.3 ) , we obtain the desired
expressions.
0T 0 1
1T 0 1T 1 1 1
2T 0 2T 1 2 2
3T 0 3T 1 3T 2 = 6 8 2
4T 0 4T 1 4T 2 24 40 16
5T 0 5T 1 5T 2 5T 3 120 240 136 16
6T 0 6T 1 6T 2 6T 3 720 1680 1232 272
-5-
And these coefficients are obtained by the following sequential computation
0! Calculating formula
1
1! 1 1=0! 1
2 0
2! 2 2=1! 2+1 0
3 1
3! 8 2 8=2! 3+2 1 , 2=2 1
4 2 0
4! 16 40 40=3! 4+8 2 , 16=8 2+2 0
5 1 3
5! 240 136 16 240=4! 5+40 3 , 136=40 3+16 1 , 16=16 1
6 4 2 0
6! 1680 1232 272 1680=5! 6+240 4 , 1232=240 4+136 2 ,
272=136 2+16 0
Note
When T1=1 , T3=2 , T5=16 , T7=272 , T9=7936 , , T2n-1 , are tangent numbers,
there is the following relation between these and the above coefficients.
-6-
0E 0 1
1E 0 1
2E 0 2E 1 2 1
3E 0 3E 1 = 6 5
4E 0 4E 1 4E 2 24 28 5
5E 0 5E 1 5E 2 120 180 61
6E 0 6E 1 6E 2 6E 3 720 1320 662 61
7E 0 7E 1 7E 2 7E 3 5040 10920 7266 1385
Note
When E0 =1 , E2 =-1 , E4 =5 , E6 =-61 , E8 =1385 , , E2n , are Euler numbers,
there is the following relation between these and the above coefficients.
-7-
9.3 Higher Derivative of Inverse Trigonometric Functions
Proof
Differentiating arctan x with respect to x one by one, it is as follows.
(2) 1! 2x 1!2>A1 x
tan x
-1
=- 2
=- 2
x +1 x +1
2 2
Hereafter by induction, we obtain the desired expression.
c.f.
According to < X ¶ ’ Ł p39,40, the following expressions hold for a natural number n.
(n )
tan x = (n -1)!cos n(tan -1 x )sin n tan -1 x +
-1
2
(n )
cot x
-1
= (-1)n(n -1)! sin n(cot -1 x )sinn cot -1 x
Therefore, combining with Formula 9.3.1, these formulas lead to the following equations.
n /2 n
(-1)k n>An+1-2k x n+1-2k = -x 2+1 sin n(cot -1 x )sinn cot -1 x
k =1
n
n /2 n
(-1) n>An+1-2k
k
= -2 sin 2
k =1 4
Example
5
-5>A4 x 4+ 5>A2 x 2- 5>A0 x 0 = -x 2+1 sin 5(cot -1 x )sin5 cot -1 x
5
5
-5>A4 + 5>A2 - 5>A0 = -2 sin 2
=4
4
-8-
Formula 9.3.2 : Higher Derivatives of arcsin x, arccos x
When denots floor function, the following expressions hold for a natural number n.
(2r -1)!!(2n -3-2r)!!x n-1-2r
(n ) n/2 n -1
sin x
-1
= 1
r=0 n -1-2r n-r-
2
1- x
2
Proof
Differentiating arcsin x with respect to x one by one, it is as follows.
(1) 1
sin x
-1
= 1
2
1- x
2
(2) x
sin x
-1
= 3
2
1- x
2
(3) 3x 2 1
sin x
-1
= 5
+ 3
2 2
1- x 1- x
2 2
(4) 15x 3 9x
sin x
-1
= 7
+ 5
2 2
1- x 1- x
2 2
4
(5) 105x 90x 2 9
sin x
-1
= 9
+ 7
+ 5
2 2 2
x +1 x +1 x +1
2 2 2
These coefficients are expressed as follows using binomial coefficients and double factorials.
1 0>A0(-1)!!(-1)!!
2 1>A1 (-1)!!1!!
Substituting these for the above,
-9-
(1) >A0(-1)!!(-1)!!x 0
0
sin x
-1
= 1
1-
2
x +1
2
(2) >A1(-1)!!1!!x 1
1
sin x
-1
= 1
2-
2
x +1
2
(3) >A2(-1)!!3!!x 2
2 >A01!!1!!x 0
2
sin x
-1
= 1
+ 1
3- 2-
2 2
x +1 x +1
2 2
(4) >A3(-1)!!5!!x 3
3 >A11!!3!!x 1
3
sin x
-1
= 1
+ 1
4- 3-
2 2
x +1 x +1
2 2
(5) >A4(-1)!!7!!x 4
4 >A21!!5!!x 2
4 >A03!!3!!x 0
4
sin x
-1
= 1
+ 1
+ 1
5- 4- 3-
2 2 2
x +1 x +1 x +1
2 2 2
(6) >A5(-1)!!9!!x
5
5
>A31!!7!!x
5
3
>A13!!5!!x 1
5
sin x
-1
= 1
+ 1
+ 1
6- 5- 4-
2 2 2
x +1 x +1 x +1
2 2 2
(2r -1)!!(2n -3-2r)!!x n-1-2r
(n ) n/2 n -1
sin x
-1
= 1
r=0 n -1-2r n-r-
2
1- x
2
(n)
cos x
-1
And reversing this sighn , we obtain .
-1
Example: the 9th order derivative of sin x
Re pe ate d diffe re ntiation
g := diff(arcsin(x),x,x,x,x,x,x,x,x,x)
2 4 6 8
2182950 x
396900 x 3783780 x
2027025 x
11025
11
13
15
17 9
2 2 2 2 2 2 2 2 2 2
1 x 1 x 1 x 1 x 1 x
Formula
f := n-> sum(binomial(n-1,n-1-2*r)
*(2*r-1)!!*(2*n-3-2*r)!!*x^(n-1-2*r)
/(1-x^2)^(n-r-1/2), r=0..floor(n/2))
f(9)
2 4 6 8
2182950 x
396900 x 3783780 x
2027025 x
11025
11
13
15
17 9
2 2 2 2 2 2 2 2 2 2
1 x 1 x 1 x 1 x 1 x
- 10 -
Formula 9.3.2 can also be expressed as follows, if partial fraction decomposition is applied to the right side.
Formula 9.3.2'
r
(n ) 1 n -1 n -1 (2r -1)!!(2n -2r -3)!!
sin x (-1)r
-1
=
2n-1 1-x 2 r=0 (1+x)r(1-x)n-1-r
r (1+x) (1-x)
(n ) 1 n -1 n -1 (2r -1)!!(2n -2r -3)!!
cos x
-1 r
=- (-1) r n-1-r
n-1
2 1-x 2 r=0
For example,
(3) 3x 2 1 1 2x 2+1
sin x
-1
= 5
+ 3
=
2 2 1-x 2 (1+ x)2(1- x)2
1- x 1- x
2 2
(3) 1 3 2 3
sin x
-1
= 2
- 1 1
+
22 1-x 2 (1-x) (1+x) (1-x) (1+x)2
1 (-1)!!3!! 2 1!!1!! 3!!(-1)!!
= 0 2
- +
22 1-x 2 (1+x) (1-x) (1+x)1(1-x)1 (1+x)2(1-x)0
r
1 3-1 2 (2r -1)!!(2 3-2r -3)!!
= (-1)r
22 1-x 2 r=0 (1+x)r(1-x)3-1-r
(n) n n Ar
sec x
-1
= (-1)n-1 1
r=1 r-
n-1+2r 2
1-x
-2
x
- 11 -
(n) n n Ar
csc x = (-1)n
-1
1
r=1 r-
n-1+2r 2
1-x
-2
x
where n Ar are coefficients as follows.
1A1 1
2A1 2A2 2 1
3A1 3A2 3A3 6 7 3
4A1 4A2 4A3 4A4 = 24 48 45 15
5A1 5A2 5A3 5A4 5A5 120 360 549 390 105
6A1 6A2 6A3 6A4 6A5 6A6 720 3000 6570 7425 4200 945
And these coefficients are obtained by the following sequential computation
2! 1!!
1, 5
3! 7 3!! 7=2! 1+1!! 5
1, 6 3, 8
4! 48 45 5!! 48=3! 1+7 6, 45=7 3+3!! 8
1, 7 3, 9 5, 11
5! 360 549 390 7!! 360=4! 1+48 7, 549=48 3+45 9,
1, 8 3, 10 5, 12 7, 14 390=45 5+5!! 11
6! 3000 6570 7425 4200 9!! 3000=5! 1+360 8, 6570=360 3+549 10,
7425=549 5+390 12, 4200=390 7+7!! 14
Proof
Differentiating arcsec x with respect to x one by one, it is as follows.
(1) 1
sec x
-1
= 1
2
x 21-x -2
(2) 2 1
sec x
-1
=- 1
- 3
2 2
x 1-x
3 -2
x 1-x
5 -2
(3) 6 7 3
sec x
-1
= 1
+ 3
+ 5
2 2 2
x 41-x -2 x 61-x -2 x 81-x -2
(4) 24 48 45 15
sec x
-1
=- 1
- 3
- 5
- 7
2 2 2 2
x 1-x x 1-x x 1-x x 1-x
5 -2 7 -2 9 -2 11 -2
- 12 -
These coefficients are calculable with the algorithm in the formula. Then, by induction, we obtain the desired
expressions.
Note
These coefficients and the direct calculation method is not known. However, these have the following
character.
1 = 0!
2 - 1 = 1!
6 - 7 + 3 = 2!
24 - 48 + 45 - 15 = 3!
120 - 360 + 549 - 390 + 105 = 4!
720 - 3000 + 6570-7425+4200-945 = 5!
- 13 -
9.4 Higher Derivative of Inverse Hyperbolic Functions
x +1
2
Proof
If we differentiate arctanh x with respect to x one by one, these coefficients become the same as the ones
(n)
tan x
-1
of ( Formula 9.3.1 ). Then we obtain the desired expression.
Proof
If we differentiate arcsinh x with respect to x one by one, these coefficients become the same as the ones
(n)
sin x
-1
of ( Formula 9.3.2 ). Then we obtain the desired expression.
Formula 9.4.2 can also be expressed as follows, if partial fraction decomposition is applied to the right side.
Formula 9.4.2'
n-1 n -1 1 1
n-1 n -1 1 1
-1
Example: the 5th order derivative of sinh x
Re pe ate d diffe re ntiation
g := diff(arcsinh(x),x,x,x,x,x):simplify(%)
4 2
24 x 72 x 9
9
2 2
x 1
Formula
asinh := n-> (I/2)^(n-1)*
sum((-1)^r*binomial(n-1,r)*(2*r-1)!!*(2*n-2*r-3)!!
*(1+I*x)^(-1/2-r)*(1-I*x)^(1/2+r-n), r=0..n-1)
- 14 -
asinh(5):simplify(%)
4 2
24 x 72 x 9
9 9
2 2
(1 i x) (i x 1)
(n) n-1
n (-1)r n Ar
sech x
-1
= (-1) 1
r=1 r-
2
x -2 -1
n-1+2r
x
(n) n (-1)r n Ar
csch
-1
x = (-1)n-1 1
r=1 r-
2
x -2 +1
n-1+2r
x
where n Ar are same as ones in Formula 9.3.3 .
- 15 -
9.5 Termwise Higher Derivative of a Logarithmic Function
Since the higher derivative of a logarithmic function has already shown in the previous section, there is no
necessity of differentiating this termwise. However, interesting results are obtained if the termwise higher
derivative is compared with the higher integral.
Formula 9.5.1
k +(n -1)! k (n -1)!
(-1)k x = -1< x <1 (1.1)
k=0 k! (1+ x)n
k
-n
x k = (1+ x)-n -1< x <1 (1.1')
k=0
Proof
x -1 (x -1)2 (x -1)3 (x -1)4
- + - +- = log x
1 2 3 4
Differentiating both sides of this with respect to x,
0!
1 - (x -1) + (x -1)2 - (x -1)3 +- = 0< x <2
x
Further differentiating both sides of this with respect to x,
1!
1 - 2(x -1) + 3(x -1)2 - 4(x -1)3 +- = 0< x <2
x2
Furthermore differentiating both sides of this with respect to x,
2!
1 2 - 2 3(x -1) + 3 4(x -1)2 - 4 5(x -1)3 +- = 0< x <2
x3
Thus generally,
k +(n -1)! (n -1)!
(-1)k (x -1)k = 0< x <2
k=0 k! xn
Replacing x with 1+ x , we obtain
k +(n -1)! k (n -1)!
(-1)k x = -1< x <1 (1.1)
k=0 k! (1+ x)n
Next, if we divide both sides by (n -1)! , the left side becomes as follows.
k
k +(n -1)! k n -1+ k -n
(-1)k x = (-1)k xk = xk
k=0 k !(n -1)! k=0 k k=0
Therefore we obtain
k x
-n k
= (1+ x)-n -1< x <1 (1.1')
k=0
c.f.
In (1.1'), n may be the real number already. Here, let = -n , then
k
x k = (1+ x) -1< x <1
k=0
This is Newton's generalized binomial theorem.
- 16 -
Special Values
In fact, Formula 9.5.1 holds also on x=1. Therfore, giving x=1 to (1.1) without considering convergence
conditions, we obtain the following special values.
0!
1 - 1 + 1 - 1 +- =
21
1!
1 - 2 + 3 - 4 +- = 2
2
2!
1 2 - 2 3 + 3 4 - 4 5 +- = 3
2
3!
1 2 3 - 2 3 4 + 3 4 5 - 4 5 6 +- = 4
2
k +(n -1)! (n -1)!
(-1)k =
k =0 k! 2n
2012.01.06 Renewal
K. Kono
Alien's Mathematics
- 17 -
10 Termwise Higher Derivative (Trigonometric, Hyperbolic)
In this chapter, for the function which the second or more order derivative is difficult to express with an easy
unification notation among trigonometric functions and hyperbolic function, we differentiate the series expansion
of these functions term by term. Therefore, sin x, cosx, sinhx, cosh x mentioned in " 9.2 Higher Derivative of
Elementary Functions " are not treated here.
0T 0 1
1T 0 1T 1 1 1
2T 0 2T 1 2 2
3T 0 3T 1 3T 2 = 6 8 2
4T 0 4T 1 4T 2 24 40 16
5T 0 5T 1 5T 2 5T 3 120 240 136 16
6T 0 6T 1 6T 2 6T 3 720 1680 1232 272
Formula 10.1.1
When B0=1, B2=1/6, B4=-1/30, B6=1/42, are Bernoulli Numbers and is ceilling function,
-1-
Proof
tan x is expanded to Tylor series as follows.
2 2 -1B 2k
2k 2k
tan x = Σ x 2k-1 |x| <
k =1 2k (2k -1)! 2
Differentiating both sides of this with respect to x repeatedly, we obtain (1.1) ~ (1.4).
Here, considering the relation between the derivative order n and the first term k0 of Σ, it is as follows.
n 0 1 2 3 4 5
k0 1 1 2 2 3 3
n +1
Such a relation can be expressed by k0 = using a ceiling function x = x. Then we obtain
2
the desired expression.
4
1/2
Σ
k =1 2k (2k -2)!
= Σ 1T r = 2
r=0
4
2/2
Σ
k =2 2k (2k -3)!
= Σ 2T r = 4
r=0
2 2 -1B2k
2k 2k
2k-4
4
3/2
Σ
k =2 2k (2k -4)!
= Σ 3Tr = 16
r=0
4
4/2
Σ
k =3 2k (2k -5)!
= Σ 4Tr = 80
r=0
Although this does not hold as an equation, differentiating the real part of the both sides repeatedly,
we obtain the following perfunctory expressions.
(1) 2
(tan x) = 2 (1cos 2x -2cos 4x +3cos 6x -4cos 8x +-)
(tan x)(2) = -2312sin 2x -22sin 4x +32sin 6x -42sin 8x +-
(tan x)(3) = -2413cos 2x -23cos 4x +33cos 6x -43cos 8x +-
-2-
(tan x)(4) = 2514sin 2x -24sin 4x +34sin 6x -44sin 8x +-
(tan x)(2n -1) = (-1)n-122nΣ(-1)k-1 k 2n-1cos 2kx (2.2n-1)
k =1
(tan x)(2n) = (-1)n22n+1Σ(-1)k-1 k 2nsin 2kx (2.2n)
k =1
Of course, these also do not hold as an equation. However, if the calculation is advanced assuming that
thiese hold, interesting results are obtained as follows.
Formula 10.1.3
(-1)k-1 (-1)k
When (x) =Σ , (n) =Σ , B2k are Bernoulli Numbers and nTr are the
k =1 kx k =0 (2k +1) n
coefficients mentioned in 10.1.0 , the following expressions hold.
(-1)n-1 2 2 -1B2k
2k 2k
2k-2n
(-1)n-1
4
n
(1-2n) = Σ = Σ 2n-1T r
4n-1 k =n
2 2k(2k -2n)! 24n-1 r=0
Proof
From (0.n) , (1.n) , (2.2n-1) and (2.2n) , we obtain the following perfunctory expressions.
(-1)n-1 n
Σ
2n-2r
= 2n 2n-1T r(tan x)
2 r=0
k-1 (-1)n 22k 22k -1B2k 2k-2n-1
Σ Σ
2n
(-1) k sin 2kx = x
k =1 22n+1 k =n +1 2k(2k -2n -1)!
(-1)n n
Σ
2n+1-2r
= 2n+1 2n T r(tan x)
2 r=0
Example1
1 2 -1B2k
2k 2k
1 1 1
(-1) = Σ 2k (2k -2)! = Σ 1T r =
23 k =1 23 r=0 4
-3-
1 2 -1B2k
2k 2k
1 2 1
(-3) = - 7 Σ = - 7 Σ 3T r = -
2 k =2 2k (2k -4)! 2 r=0 8
1 2 -1B2k
2k 2k
1 1 1
(-2) = - 3 Σ = - 3 Σ 2T r = -
2 k =2 2k(2k -3)! 2 r=0 2
1 2 -1B2k
2k 2k
1 n 5
(-4) = Σ 2k(2k -5)! = Σ 4T r =
25 k =3 25 r=0 2
Note
Using a relation (-2n ) = E2n / 2 between Dirichlet Even Beta and Euler number, .we obtain the
following expression.
-4-
10.2 Termwise Higher Derivative of tanh x
Where, is celling function and nTr are the same as the coefficients in 10.1.0 .
Formula 10.2.1
When B0=1, B2=1/6, B4=-1/30, B6=1/42, are Bernoulli Numbers and is ceilling function,
Proof
tanh x is expanded to Tylor series as follows.
2 2 -1 B 2k
2k 2k
tanh x = Σ 2k-1
x |x| <
k =1 2k (2k -1)! 2
Differentiating both sides of this with respect to x repeatedly, we obtain the desired expression. ( The number
of the first term of Σ is the same as it of tan x. )
Formula 10.2.2
The following expressions hold for x >0 .
(tanh x)(1) = 2211e -2x -21e -4x +31e -6x -41e -8x +-
(tanh x)(2) = -2312e -2x -22e -4x +32e -6x -42e -8x +-
(tanh x) n = (-1)n-12n+1Σ(-1)k-1k ne -2kx
( )
(2.n)
k =1
Proof
tanh x is epanded to Fourier series for x > 0 as follows.
e x - e -x 1- e -2x
tanh x = x -x
= -2x
= 1- e -2x1 - e -2x + e -4x - e -6x +-
e +e 1+ e
-5-
= 1 - 2e -2x - 2e -4x + 2e -6x - 2e -8x +- (2.0)
Formula 10.2.3
When B2n are Bernoulli Numbers and nTr
is ceilling function, are the coefficients mentioned in 10.1.0 ,
the following expressions hold. for 0 < x < .
k-1 kn n-1 22k -1B2k 2k- n-1
Σ (-1) = (-1) Σ x
k =1 e kx k=
n +1 2k(2k - n -1)!
2
n+1-2r
1 n /2x
= - n+1 Σ (-1) nTr tanh r
2 r=0 2
Example
2 -1B2k 2k-2
2k
11 21 31 41
1 2 x
e
1x -
e
2x
+
e
3x
-
e
4x
+- = Σ
k=1 2k (2k -2) !
x = - 2 tanh
2 2
-1
2 -1B2k 2k-3
2k
12 22 32 42
1 3 x x
1x - 2x
+ 3x
- 4x
+- = -Σ x = - 3 2 tanh -2 tanh
e e e e k=2 2k (2k -3) ! 2 2 2
Formula 10.2.3'
(-1)k-1
Let (x) =Σ , and let Bernoulli number B2n and tangent number Tr are as follows.
k =1 kx
B2=1/6, B4=-1/30, B6=1/42, B8=-1/30, B10=5/66,
T1=1, T3=2, T5=16, T7=272, T9=7936,
Then the following expressions hold.
2n
2 -1B2n (-1)n
(-2n +1) = = - 2n T2n-1 (3.2n-1')
2n 2
(-2n) =0 (3.2n')
Proof
Sbstitute x =0 for Formula 10.2.3 . Then, the Fourier series is
kn (-1)k-1
Σ (-1) k-1
k0
=Σ = (-n)
k =1 e k=1 k -n
-6-
The Taylor series and the polynomial are
when n =2m-1
22k -1B2k
Σ
2m-1-1
= (-1) 02k-(2m-1)-1
2m -1+1
2 2
k k -(2 -1
m ) -1 !
k=
2
1 n /2 0
- n+1 Σ (-1) nTr tanh r
2 r=0 2
1 m 1
Σ(-1)r 2m-1Tr0 (-1)m 2m-1Tm 00
2m-2r
=- =-
22m r=0 2 2m
(-1)m
=- T2m-1 2m-1T m = T2m-1
22m
when n =2m
22k -1B2k
Σ
2m-1
= (-1) 02k-2m-1 = 0
2m +1
2 2
k k m
( -2 -1 )!
k=
2
n+1-2r
1 n /2 0
- n+1 Σ (-1) nTr tanh r
2 r=0 2
1 m
2m+1 Σ
=- (-1)r 2mTr0 2m+1-2r
=0
2 r=0
-7-
10.3 Termwise Higher Derivative of cot x
Where, is celling function and nTr are the same as the coefficients in 10.1.0 .
Formula 10.3.1
When B0=1, B2=1/6, B4=-1/30, B6=1/42, are Bernoulli Numbers and is ceilling function,
the following expressions hold for 0 < x < .
1! 22kB2k 2k-2
(cot x) (1)
=- -Σ x
x2 k=1 2k(2k -2)!
2! 22kB2k 2k-3
(cot x)(2) = -Σ x
x3 k=2 2k(2k -3)!
(n) n! 22kB2k
(cot x) = (-1) n
n+1
-Σ x 2k-n-1 (1.n)
x n +1
2k(2k -n -1)!
=k
2
Proof
22kB2k 2k
x cot x = 1 +Σ x 0< x <
k =1 (2k)!
From this
0! 22kB2k 2k-1
cot x = -Σ x
x1 k=1 2k(2k -1)!
Differentiating both sides of this with respect to x repeatedly, we obtain the desired expression. ( The number
of the first term of Σ is the same as it of tan x. )
Formula 10.3.1'
When B2k denote Bernoulli Numbers and denotes ceilling function, the following expressions hold for
/2 < x < .
(n)
(cot x) =-Σ x-
n +1
2k(2k - n -1)! 2
= k
2
Proof
cot x = -tan x - 2
-8-
22k22k -1B2k 2k-1
tan x = Σ x |x|<
k=1 2k (2k -1)! 2
From these
cotx = -Σ x- 0< x <
k =1 2k (2k -1)! 2
Differentiating both sides of this with respect to x repeatedly, we obtain the desired expression. ( The number
of the first term of Σ is the same as it of tan x. )
22kB2k
n /2
n+1 n+1-2r
Σ n
Σ
2k
x = (-1) n ! - x n Tr(cotx) (1.t)
n +1 2k(2k -n -1)! r=0
k=
2
22kB2k 2k
2
4 4
Σ
k =1 2k(2k -2)!
= -1! + 2
22kB2k 2k
3
4 4
Σ
k =2 2k(2k -3)!
= 2! - 4
22kB2k 2k
4
4 4
Σ
k =2 2k(2k -4)!
= -3! + 16
22kB2k 2k
5
4 4
Σ
k =3 2k(2k -5)!
= 4! - 80
-9-
Of course, these also do not hold as an equation. However, if the calculation is advanced assuming that
thiese hold, interesting results are obtained as follows.
Formula 10.3.3
(-1)k-1 (-1)k
When (x) =Σ , (n ) =Σ , B2k are Bernoulli Numbers and nTr are the
k =1 kx k =0 (2k +1)
n
(2n -1)! + Σ
(-1)n-1 2n 22kB2k 2k
4
4
(1-2n) =
24n-1 k=n 2k(2k -2n)!
(-1)n-1 n
= Σ 2n-1T r
24n-1 r=0
(2n)! - Σ
(-1)n 2n 22kB2k 2k
4
4
(-2n) =
22n+1 k =n +1 2k(2k -2n -1)!
n
= (-1)nΣ 2nTr
r=0
Proof
From (0.n) , (1.n) , (2.2n-1) and (2.2n) , we obtain the following perfunctory expressions.
2k
B2k
n 2
(-1 )
Σ k 2n-1cos 2kx = 2n 2n (2n -1)! +
2 k( 2 k -2 n Σ
)!
x 2k
k =1 2 x k=n
(-1)n n
Σ
2n-2r
= 2n 2n-1T r(cotx)
2 r=0
(2n)! - Σ
(-1)n 22kB2k
Σ
2n
k sin 2kx = x 2k
k =1 22n+1x 2n+1 k=n +1 2k(2k -2n -1)!
n
= (-1)nΣ 2nTr(cotx)2n+1-2r
r=0
If x = /4 is substituted for these, the left sides are as follows respectively.
k
Σk cos
2n-1
= -22n-1Σ(-1)k-1 k 2n-1 = -22n-1(1-2n)
k =1 2 k=1
k
Σ k 2n
sin k = Σ (-1)k-1 (2k -1)2n = (-2n)
k =1 2 k =1
Therefore, we obtain the desired expressions.
Example
2 22kB2k 2k
4
1 4 1 1 1
(-1) = 1!+Σ = Σ 1T r =
23 k =1 2k(2k -2)! 23 r=0 4
(-3) = - 3!+Σ
2k(2k -4)! 4
2k
1 4 2 B 4
1 2k
2k 2 1
2 7
= -
2
Σ k =2
7
r=0
3T r =-
8
- 10 -
2
22kB2k 2k
4
1 4 1 1 1
(-2) = - 3 2!-Σ =- Σ T
2 r = -
2 k =2 2k(2k -3)! 23 r=0 2
2k(2k -5)! 4
2k
2 B 4
2k
2
1 4 1 n 5
(-4) =
2k
5
4!-Σ =
k =3
Σ 4T r =
25 r=0 2
Formula 10.3.4
When B2k denotes a Bernoulli Number, the following expressions hold.
22k -2B2k 2k
(2n -1)! = Σ
k=n 2k(2k -2n )! 2 (4.1)
2k
2 B2k 2k
(2n)! = Σ (4.2)
k= n +1 2k(2k -2n -1)! 2
Proof
Replacing n with 2n -1 in Formula 10.3.1' , we obtain
2 2 -1B 2k
2k 2k
2k-2n
= -Σ x - 2
(2n -1)
(cot x)
k =n 2k(2k -2n)!
Substituting x = /4 for this,
22k22k -1B2k 2k-2n
|x=/4 = -Σ 4
(2n -1)
(cotx)
k=n 2k(2k -2n)!
On the othe hand, substituting x = /4 for (2.2n-1) ,
2n 22kB2k 2k-2n
4
(2n -1) 4
(cotx) |x=/4 = -(2n -1) ! -Σ
k=n 2k(2k -2n) !
Since these have to be equal, we obtain (4.1).
Next, replacing n with 2n in Formula 10.3.1' ,
2 2 -1B2k
2k 2k
2k-2n-1
(cot x)(2n) = - Σ x-
k=n +1 2k(2k -2n -1)! 2
x = /4 for this, we obtain
Substituting
2 2k
22k
-1B2k 2k-2n-1
4
(cotx)(2n)|x=/4 = Σ
k=n +1 2k(2k -2n -1)!
On the othe hand, substituting x = /4 for (2.2n) ,
2n+1 22kB2k 2k-2n-1
4
(2n ) 4
(cotx) |x=/4 = (2n) ! -Σ
k=n +1 2k(2k -2n -1) !
Since these have to be equal, we obtain (4.2).
- 11 -
- 12 -
10.4 Termwise Higher Derivative of coth x
Where, is celling function and nTr are the same as the coefficients in 10.1.0 .
Formula 10.4.1
When B0=1, B2=1/6, B4=-1/30, B6=1/42, are Bernoulli Numbers and is ceilling function,
1! 22k B2k
(coth x) (1)
=- +Σ x 2k-2
x2 k=1 2k(2k -2)!
2! 22k B2k
(coth x)(2) = +Σ x 2k-3
x3 k=2 2k(2k -3)!
Proof
22k B2k 2k
x coth x = 1 + Σ x 0< x <
k =1 (2k)!
From this
0! 22k B2k
coth x = +Σ x 2k-1
x k=1 2 2
k( k -1 )!
Differentiating both sides of this with respect to x repeatedly, we obtain the desired expression. ( The number
of the first term of Σ is the same as it of tan x. )
Formula 10.4.2
The following expressions hold for x >0 .
(coth x)(1) = -2211e -2x +21e -4x +31e -6x +41e -8x +
(coth x)(2) = 2312e -2x +22e -4x +32e -6x +42e -8x +
(coth x) n = (-1)n2n+1Σk ne -2kx
()
(2.n)
k=1
- 13 -
Proof
coth x is expanded to Fourier series for x >0 as follows.
e x + e -x 1+ e -2x
coth x = x -x
= -2x
= 1+ e -2x1 + e -2x + e -4x + e -6x +
e -e 1- e
= 1 + 2e -2x + 2e -4x + 2e -6x + 2e -8x +- (2.0)
Formula 10.4.3
When is ceilling function, B2n are Bernoulli Numbers and nTr are the coefficients mentioned in 10.1.0 ,
the following expressions hold for 0 < x < .
kn 1 B2k x 2k
Σ = n ! +(-1) Σ n
k=1 e kx x
n+1
k=
n +1 2k(2k -n -1)!
2
n+1-2r
1 x
n /2
= n+1 Σ (-1) nTr coth r
2 r=0 2
Example
2
11 21 31 41 B2k x 2k
1 1 x
1x + 2x
+ 3x
+ 4x
+ = 2
1! -Σ = 2
coth 2 -1
e e e e x k=1 2k(2k -2) ! 2
x 2k(2k -3) !
2k
12 22 32 42 B x
2
1 1 2k x x
2! +Σ
3
1x + 2x
+ 3x
+ 4x
+ = 3
= 2 coth 3
-2 coth
e e e e k=2 2 2
Formula 10.4.3'
When is ceilling function, B2n are Bernoulli Numbers and nTr are the coefficients mentioned in 10.1.0 ,
the following expressions hold.
n n+1-2r
k B2k
1 n /2 1
Σ k
= n ! + (-1)n
Σ n +1 2k(2k -n -1) !
= n+1 Σ (-1)r nTr coth
2
k=1 e k=
2 r=0
2
Note
Mr. Sugimoto was foreseeing this formula. ( http://homepage3.nifty.com/y_sugi/sp/sp56.htm )
Example
11 21 31 41 B2k
1 2 1
1+ 2
+ 3
+ 4
+ = 1! -Σ = 2 coth -1
e e e e k =1 2k(2k -2) ! 2 2
- 14 -
12 22 32 42 B2k
1 3 1 1
1 + 2
+ 3
+ 4
+ = 2! +Σ = 3 2 coth -2 coth
e e e e k =2 2k(2k -3) ! 2 2 2
13 23 33 43 B2k
1 4 1 2 1
1+ 2
+ 3
+ 4
+ = 3! -Σ = 4 6 coth -8 coth +2
e e e e k =2 2k(2k -4) ! 2 2 2
Formula 10.4.3"
1p 2p 3p 4p
+ + + + (1+ p) p >0 (3.p)
e1 e2 e3 e4
Example (1+2.5)
- 15 -
10.5 Termwise Higher Derivative of csc x
0E 0 1
1E 0 1
2E 0 2E 1 2 1
3E 0 3E 1 = 6 5
4E 0 4E 1 4E 2 24 28 5
5E 0 5E 1 5E 2 120 180 61
6E 0 6E 1 6E 2 6E 3 720 1320 662 61
7E 0 7E 1 7E 2 7E 3 5040 10920 7266 1385
Formula 10.5.1
When B0=1, B2=1/6, B4=-1/30, B6=1/42, are Bernoulli Numbers and is ceilling function,
Proof
22k -2B2k 2k
xcscx = 1 +Σ x 0< x <
k=1 (2k)!
From this
2 -2B 2k
2k
0!
cscx = +Σ x 2k-1
x k=1 2k (2k -1)!
Differentiating both sides of this with respect to x repeatedly, we obtain the desired expression. ( The number
of the first term of Σ is the same as it of tan x. )
- 16 -
Formula 10.5.1'
When E0=1, E2=-1, E4=5, E6=-61, E8=1385, are Euler Numbers and is floor function,
E2k 2k-n
(n)
(csc x) =Σ x-
n +1 (2k -n )! 2
k=
2
proof
csc x = sec x - 2
E2k
secx =Σ x 2k |x|<
k=0 (2k)! 2
From these
E2k 2k
x - 2
cscx =Σ 0< x <
k=0 (2k)!
Differentiating both sides of this with respect to x repeatedly, we obtain the desired expression. ( The number
of the first term of Σ is the same as it of sec x. )
22k -2B2k 2k
n+1 n/2
x
n-1 n-2r
Σ n +1 2k (2k - n -1)!
x = (-1) n! -
sin x Σ
r=0
n E r(cotx) (1.t)
k=
2
2 -2B2k
2k
2k
2
Σ
k=1 2k (2k -2)!
= 1!
22k -2B2k 2k
3
2 2
Σ
k=2 2k (2k -3)!
= -2! + 1
22k -2B2k 2k
2
Σ
k=2 2k (2k -4)!
= 3!
22k -2B2k 2k
5
2 2
Σ
k=3 2k (2k -5)!
= -4! + 5
- 17 -
(csc x)(1) = 2(1cosx +3cos 3x +5cos 5x +7cos 7x +-)
(csc x)(2) = -212sin x +32sin 3x +52sin 5x +42sin 7x +-
(csc x)(3) = -213cosx +33cos 3x +53cos 5x +43cos 7x +-
(csc x)(4) = 214sin x +34sin 3x +54sin 5x +44sin 7x +-
(csc x)(2n -1) = (-1)n-12Σ(2k +1)2n-1cos(2k +1)x (2.2n-1)
k=0
(csc x)(2n) = (-1)n2Σ(2k +1)2nsin(2k +1)x (2.2n)
k=0
Of course, these also do not hold as an equation. However, if the calculation is advanced assuming that
thiese hold, interesting results are obtained as follows.
Formula 10.5.3
(-1)k
Let (n ) =Σ n
, and let Bernoulli number B2n and Euler number Er are as follows.
k=0 (2k +1)
B0=1, B2=1/6, B4=-1/30, B6=1/42, B8=-1/30,
E0=1, E2=-1, E4=5, E6=-61, E8=1385,
Then the following expressions hold.
22k -2B2k
(-1)n
2n+1 2k E2n
2
2
(-2n) = (2n)!+ Σ =
2 k =n+1 2k(2k -2n -1)! 2
Proof
From (0.n) , (1.n) and (2.2n) , we obtain the following perfunctory expression.
22k -2B2k 2k
(-1)n
Σ (2n)! + Σ
2n
(2k +1) sin(2k +1)x = x
k=0 2x 2n+1 k=n +1 2k (2k -2n -1)!
(-1)n n
= Σ
2sin x r=0 2n E r(cotx)2n-2r
If x = /2 is substituted for this, the left side is as follows.
(2k +1)
Σ(2k +1) sin
2n
= Σ(-1)k (2k +1)2n = (-2n)
k=0 2 k=0
Then,
22k -2B2k
(-1)n
(2n)! + Σ
2n+1 2k
2
2
(-2n) =
2 k =n+1 2k(2k -2n -1)!
(-1)n n (-1)n
Σ
2n-2r
= 2n Er 0 = 2n E n00
2 r=0 2
E2n
(-1) 2nE n = E2n
n
=
2
- 18 -
Example
22k -2B2k
3 2k E2
2
1 2 1
(-2) = - 2! + Σ = =-
2 k =2 2k(2k -3)! 2 2
2 -2B
2 Σ 2k(2k -5)! 2 2
5 2k 2k
1 2 E 5
(-4) =
2k 4
4! + = =
k =3 2
- 19 -
10.6 Termwise Higher Derivative of csch x
Where, is floor function and n Er are the same as the coefficients in 10.5.0.
Formula 10.6.1
When B0=1, B2=1/6, B4=-1/30, B6=1/42, are Bernoulli Numbers and is ceilling function,
(n) n! 22k -2B2k n-1
(csch x) = (-1) n
n+1
-Σ x 2k- (1.n)
x n +1 2k(2k - n -1)!
k=
2
Proof
2 -2B2k 2k
2k
xcsch x = 1 -Σ x 0< x <
k=1 (2k)!
From this
2 -2 B 2k
2k
0!
csch x = -Σ x 2k-1
x k=1 2k ( 2k -1)!
Differentiating both sides of this with respect to x repeatedly, we obtain the desired expression. ( The number
of the first term of Σ is the same as it of tan x. )
Formula 10.6.2
The following expressions hold for x >0 .
(csch x)(1) = -21e -1x +3e -3x +5e -5x +7e -7x +
(csch x)(2) = 212e -1x +32e -3x +52e -5x +42e -7x +
(csch x) n = (-1)n2Σ(2k +1)n e -(2k +1)x
( )
(2.n)
k=0
Proof
csch x is expanded to Fourier series for x >0 as follows.
- 20 -
2 2e -x
csch x = = = 2e -x1 + e -2x + e -4x + e -6x +
e x- e -x 1- e -2x
= 2(cosh ix +cosh 3ix +cosh 5ix +)- 2i(sinh ix +sinh 3ix +sinh 5ix +)
Differentiating both sides of (2.0) with respect to x repeatedly, we obtain the desired expression.
Formula 10.6.3
When , B2n
are Bernoulli Numbers and n Er are
are ceilling function and floor function respectivly,
the coefficients mentioned in 10.5.0 , the following expressions hold for 0 < x < .
22k -2B2k
(2k +1)
n
1
Σ (2k +1)x
= n+1 n!-(-1) n
n +1 2k(2k - n -1)!
x 2kΣ
k =0 e 2x k=
2
1 n/2
= Σ (-1)r n E r(csch x)n-2r
2sinh x r=0
Example
2 -2B2k 2k
2k
11 31 51 71 1 coth x
1
1x + 3x
+ 5x
+ 7x
+ = 2
1!+Σ x =
e e e e 2x k =1 2k (2k -2) ! 2sinh x
2 -2B
2k
12 32 52 72 1 2coth x - coth x
2k
2 0
2!-Σ
2k
1x + 3x
+ 5x
+ 7x
+ = 3
x =
e e e e 2x 2k (2k -3) !
k =1 2sinh x
Formula 10.6.3'
When , B2n
are Bernoulli Numbers and n Er are
are ceilling function and floor function respectivly,
the coefficients mentioned in 10.5.0 , the following expressions hold.
2 -2B2k
2k
2k +1n 1 1 n/2
Σ
k =0 e
(2k +1)
=
2
n!-(-1) n
Σ
n +1 2k(2k-n -1)!
= Σ
2sinh 1 r=0
(-1)rnE r(coth 1) n-2r
k=
2
Example
2 -2B2k
2k
11 31 1 51 71 coth 1
1
1 + 3 + 5 + 7 + = 2 1! +Σ 2k(2k-2)! =
2sinh 1
e e e e k=1
2
2 -2B
2k(2k-3)!
2 2 2 2 2k 2 0
1 3 5 7 1 2coth 1 - coth 1 2k
+ 1+ + + =
3
2! - Σ
5 7
=
e e e e 2sinh 1 k =2
Therefore also, this leads to the following expression conjointly with Formula 10.4.3'.
- 21 -
10.7 Termwise Higher Derivative of sec x
Where, is floor function and n Er are the same as the coefficients in 10.5.0.
Formula 10.7.1
When E0=1, E2=-1, E4=5, E6=-61, E8=1385, are Euler Numbers and is floor function,
E2k
(sec x)(1) =Σ x 2k-1 (1.1)
k =1 (2k -1)!
E2k
(sec x)(2) =Σ x 2k-2 (1.2)
k =1 (2k -2)!
E2k
(sec x)(3) =Σ x 2k-3 (1.3)
k =2 (2k -3)!
E2k
(sec x)(4) =Σ x 2k-4 (1.4)
k =2 (2k -4)!
E2k
(sec x) n =
()
Σ
n +1 (2k -n)!
x 2k-n (1.n)
k=
2
Proof
sec x is expanded to Tylor series as follows.
E2k
secx =Σ x 2k |x| <
k=0 (2k)! 2
Differentiating both sides of this with respect to x repeatedly, we obtain (1.1) ~ (1.4).
Here, considering the relation between the derivative order n and the first term n of Σ, it is as follows.
n 0 1 2 3 4
k0 0 1 1 2 2
n +1
Such a relation can be expressed by k0 = using a floor function x = x . Then we obtain
2
the desired expression.
- 22 -
And giving x = /4 to this, we obtain the following special values.
E2k 2k-1
0
Σ = 2 Σ n E r = 2 1
k=1 (2k -1)! 4 r=0
E2k 2k-1
1
Σ = 2 Σ n E r = 2 3
k=1 (2k -1)! 4 r=0
E2k 2k-3
1
Σ = 2 Σ n E r = 2 11
k=2 (2k -3)! 4 r=0
E2k 2k-4
2
Σ = 2 Σ n E r = 2 57
k=2 (2k -4)! 4 r=0
- 23 -
10.8 Termwise Higher Derivative of sech x
Where, is floor function and n Er are the same as the coefficients in 10.5.0.
Formula 10.8.1
When E0=1, E2=-1, E4=5, E6=-61, E8=1385, are Euler Numbers and is floor function,
E2k
(sech x)(1) =Σ x 2k-1
k=1 2 -1
( k ) !
E2k
(sech x)(2) =Σ x 2k-2
k=1 (2k -2)!
E2k
(sech x) n =
( )
Σ
n +1 (2k -n)!
x 2k-n (1.n)
k=
2
Proof
sech x is expanded to Tylor series as follows.
E2k 2k
sech x =Σ x |x| <
k=0 (2k)! 2
Differentiating both sides of this with respect to x repeatedly, we obtain the desired expression. ( The number
of the first term of Σ is the same as it of sec x. )
Formula 10.8.2
The following expressions hold for x >0 .
(sech x)(1) = -21e -1x -3e -3x +5e -5x -7e -7x +-
(sech x)(2) = 212e -1x -32e -3x +52e -5x -42e -7x +
(sech x) n = (-1)n2Σ(-1)k (2k +1)n e -(2k +1)x
( )
(2.n)
k=0
Proof
sech x is expanded to Fourier series for x >0 as follows.
-x
2 2e
sech x = = = 2e -x 1 - e -2x + e -4x - e -6x +-
e x + e -x 1 +e -2x
- 24 -
= 2e -1x - e -3x + e -5x - e -7x +- (2.0)
Formula 10.8.3
When is floor function, E2n are Euler Numbers and n Er are the coefficients mentioned in 10.5.0 ,
1 n/2
= Σ (-1)r n E r(tanh x)n-2r
2cosh x r=0
Example1
11 31 51 1 E2k71 tanh 1x
- + - +- = - Σ x 2k-1
=
e 1x e 3x e 5x e 7x 2 k =1 (2k -1)! 2cosh x
12 32 52 72 1 E2k 2 0
2tanh x - tanh x
2Σ
2k-2
- + - +- = x =
e 1x e 3x e 5x e 7x k=1 (2k -2)! 2cosh x
Formula 10.8.3'
(-1)k
When (n) =Σ and E0=1, E2=-1, E4=5, E6=-61, E8=1385, are
k =0 (2k +1) n
Euler Numbers, the following expressions hold.
(-2n +1) = 0 (3.2n-1')
- 25 -
E2n
(-2n) = (3.2n')
2
Proof
Sbstitute x =0 for Formula 10.8.3 . Then, the Fourier series is
(-1)2m-1 E2k
=
2 Σ
2m -1+1 2k -(2m -1)!
02k-(2m-1)
k=
2
1 E2k
2Σ
=- 02k-2m+1 = 0
k=m (2k -2m +1)!
when n =2m
(-1)n E2k (-1)2m E2k
Σ Σ
2k-n
0 = 02k-2m
2 n +1
(2k -n )! 2 2m +1
(2k -2m )!
k= k=
2 2
1 E2k 1 E2m 0
2Σ
2k-2m
= 0 = 0
k=m (2k -2m)! 2 0!
E2m
=
2
Replacing m with n , we obtain (3.2n-1') and (3.2n') .
Note
These results are obtained also from the polynomial in Formula 10.8.3 ..
2006.11.11
2012.07.04 Renewal
K. Kono
Alien's Mathematics
- 26 -
11 Termwise Higher Derivative (Inv-Trigonometric, Inv-Hyperbolic)
Formula 11.1.1
When is ceilling function, the following expressions hold for |x |< 1 .
(n) (2k)! n
tan x Σ (-1)k
-1
= x 2k+1- (1.t)
n
k=
-1
(2k +1- n)!
2
Proof
arctan x is expanded to Tylor series as follows.
(2k)! 2k+1
tan -1x = Σ(-1)k x |x| < 1
k =0 (2k +1)!
Differentiating both sides of this with respect to x repeatedly, we obtain the following.
(1) (2k)! 2k
tan
-1
x = Σ(-1)k x
k=0 (2k)!
(2) (2k)! 2k-1
tan
-1
x = Σ(-1)k x
k=1 (2k -1)!
(3) (2k)! 2k-2
tan
-1
x = Σ(-1)k x
k=1 (2k -2)!
(4) (2k)! 2k-3
tan
-1
x = Σ(-1)k x
k=2 (2k -3)!
Here, considering the relation between the derivative order n and the first term k0 of Σ, it is as follows.
n 0 1 2 3 4 5
k0 0 0 1 1 2 2
n -1
Such a relation can be expressed by k0 = using a ceiling function x = x. Then we obtain
2
(1.t). And,
x 0 : +
cot -1x =
2
- tan -1x
x <0 : -
Therefore, we obtain (1.c) immediately.
x +1
2 r=1
-1-
From (1.t) and this , the following equation follows for |x|< 1 .
(2k)! (n -1)! n /2
n
Σ (-1)k x 2k+1- = (-1)n n Σ (-1)rnC n+1-2r x n+1-2r
k=
n -1
(2k +1- n)! x +1 r=1
2
2
And giving x =1 to this without considering the convergence condition, we obtain the following special value.
(2k)! (n -1)! n /2
Σ
n -1
(-1)k
(2k +1- n)!
= (-1)n
2n Σr=1
(-1)rnC n+1-2r (1.t')
k=
2
Example
0! 1
1 - 1 + 1 - 1 +- 1 1C 0
= =
2 2
1! 1
2 - 4 + 6 - 8 +- = 2C 1 =
22 2
2! 1
12 - 34 + 56 - 78 +- = - 3 3C 2-3C 0 = -
2 2
3!
234 - 456 + 678 - 8910 +- = - 4 4C 3-4C 1 = 0
2
Formula 11.1.2
When is ceilling function, the following expressions hold for |x |< 1.
2
(n) (2k -1)!!
sin x Σ
-1
= x 2k+1-n (2.s)
k=
n -1
(2k +1- n)!
2
2
(n) (2k -1)!!
cos x Σ
-1
=- x 2k+1-n (2.c)
n -1
k=
(2k +1- n)!
2
Proof
(2k -1)!! (2k -1)!!(2k)!
sin -1x = Σ x 2k+1 = Σ x 2k+1
k =0 (2 k)!! (2k +1) k=0 (2k)!! (2k +1)!
2
(2k -1)!!
=Σ x 2k+1 (2k)! = (2k)!!(2k -1)!!
k =0 (2k +1)!
Differentiating both sides of this with respect to x repeatedly, we obtain (2.s). ( The number of the first term
of Σ is the same as it of arctan x. )
cos x = /2 - sin x
-1 -1
(2.c) is obtained immediately from .
Formula 11.1.2'
-2-
r
2n 2n
Σ
r=0
(-1)r (2r -1)!!(4n -2r -1)!! = 22n(2n -1)!!2 (2.se)
Proof
Formula 9.3.2' in "9 Higher Derivative" was as follows.
r
(n ) 1 n -1 n -1 (2r -1)!!(2n -2r -3)!!
sin x Σ(-1)r
-1
=
2n-1 1-x 2 r=0 (1+x)r(1-x)n-1-r
From this,
1 2n -1
2n -1
x=0 =
(2n )
sin x Σ r
-1
(-1) (2r -1)!!(4n -2r -3)!!
22n-1 r=0 r
1 2n 2n
x=0
(2n +1)
sin x = 2n Σ(-1)r
-1
(2r -1)!!(4n -2r -1)!!
2 r=0 r
From (2.s) ,
2
(2k -1)!!
x=0 = Σ (2 +1-2 )! 02k+1-2n = 0
(2n )
sin x
-1
k =n k n
2 2
(2k -1)!! (2n -1)!!
x=0 = Σ (2 +1-2 -1)! 0
(2n +1)
sin x
-1 2k-2n
= 00
k =n k n 0!
Thus, we obtain the desired expressions.
Example
3C 0(-1)!!5!!- 3C 11!! 3!!+ 3C 23!!1!!- 3C 35!!(-1)!! = 0
4C 0 (-1)!!7!!- 4C 1 1!!5!!+ 4C 2 3!!3!!- 4C 3 5!!1!!+ 4C 4 7!!(-1)!! = 144
Formula 11.1.3
The following expressions hold for |x |> 1 .
(n)
(2k -1)!! (2k +n)! -2k-n-1
csc x
-1
= (-1)nΣ x (3.c)
k=0 (2k)!! (2k +1)!
(n) (2k -1)!! (2k +n )!
sec x
-1
= (-1)n-1Σ x -2k-n-1 (3.s)
k=0 (2k)!! (2k +1)!
Proof
1 (2k -1)!!
csc -1x = sin -1 =Σ x -2k-1 |x|>1
x k=0 (2k)!!(2k +1)
Differentiating both sides of this with respect to x repeatedly, we obtain (3.c).
sec x = /2 - csc x
-1 -1
(3.s) is obtained immediately from .
-3-
11.2 Termwise Higher Derivative of Inverse Hyperbolic Functions
Formula 11.2.1t
When is ceilling function, the following expressions hold for |x |< 1 .
(n) (2k)! n
tanh x Σ
-1
= x 2k+1- (1.t)
n -1
k=
(2k +1- n)!
2
Proof
arctanh x is expanded to Tylor series as follows.
(2k)! 2k+1
tanh -1x = Σ(-1)k x |x| < 1
k=0 (2k +1)!
Differentiating both sides of this with respect to x repeatedly, we obtain the desired expression. ( The number
of the first term of Σ is the same as it of arctan x. )
Formula 11.2.1c
The following expression holds for |x |> 1 .
(n) (2k + n)! -2k-1- n
coth
-1
x = (-1)(n)Σ x (1.c)
k =0 (2k +1)!
Proof
arccoth x is expanded to Tylor series as follows.
(2k)! -2k-1
coth -1x = Σ x |x| > 1
k =0 (2k +1)!
Differentiating both sides of this with respect to x repeatedly, we obtain the desired expression.
Formula 11.2.2s
When is ceilling function, the following expression holds for |x |< 1 .
2
(n) k (2k -1)!!
sinh x Σ
-1
= (-1) x 2k+1-n (2.s)
n -1
k=
(2k +1- n)!
2
Proof
arcsinh x is expanded to Tylor series for |x |< 1 as follows.
(2k -1)!! (2k -1)!!(2k)! 2k+1
sinh -1x =Σ(-1)k x 2k+1 =Σ(-1)k x
k =0 (2k)!! (2k +1) k =0 (2k)!! (2k +1)!
2
k (2k -1)!!
=Σ(-1) x 2k+1 (2k)! = (2k)!!(2k -1)!!
k =0 (2k +1)!
Differentiating both sides of this with respect to x repeatedly, we obtain the desired expression. ( The number
of the first term of Σ is the same as it of arctan x. )
-4-
Sum of Taylor series of the higher derivative of arcsin x
Formula 9.4.2 in "9 Higher Derivative" was as follows.
From (2.s) and this , the following equation follows for |x|< 1 .
2
(2k -1)!!
Σ
n -1
(-1)k
(2k +1- n)!
x 2k+1-n
k=
2
And giving x =1 to this without considering the convergence condition, we obtain the following special value.
2 n/2
n -1-2r
k (2k -1) !!
n -1 (2r -1)!!(2n -3-2r)!!
Σ
n -1
(-1)
(2k +1-n ) !
= (-1)n-1 Σ (-1)r
r=0 n-r-
1
k= 2
2 2
Example
(-1)!!2 3!!2 5!!2 7!!2 2
- + - +- =
0! 2! 4! 6! 2
1!!2 3!!2 5!!2 7!!2 2
- + - +- =
1! 3! 5! 7! 4
1!!2 3!!2 5!!2 7!!2 2
- + - +- =-
0! 2! 4! 6! 8
3!!2 5!!2 7!!2 9!!2 3 2
- + - +- =
1! 3! 5! 7! 16
Formula 11.2.2c
The following expressions hold for |x |> 1 .
(n) (n -1)! n-1 (2k + n -1)! -2k-n
cosh
-1
x = (-1)n-1 + (-1) Σ 2 x (2.c)
xn k =1 (2k)!!
Proof
arccosh x is expanded to Tylor series for |x | 1 as follows.
(2k -1)!! -2k (2k -1)!!(2k -1)!
cosh -1x = log 2x -Σ x = log 2x -Σ x -2k
k=1 (2k) !!2k k =1 ( 2k)!!(2k) !
(2k -1)!
= log 2x -Σ 2x
-2k
(2k )! = (2k )!!(2k -1)!!
k =1 (2k)!!
Differentiating both sides of this with respect to x repeatedly, we obtain the desired expression.
-5-
11.2.3 Termwise Higher Derivative of arccsch x , arcsech x
Formula 11.2.3
When is ceilling function, the following expressions hold for 0<x <1 .
2
(n -1)!
(n) k (2k -1)!!
csch
-1
x = (-1) n
- Σ (-1)
n
x 2k-n (3.c)
x n 2k (2k - n)!
k=
2
2
(n) (n -1)! (2k -1)!!
sech
-1
x = (-1) n
-Σ
n
x 2k-n (3.s)
x n 2k (2k - n )!
k=
2
Proof
2 (2k -1)!! 2k
csch -1x = log +Σ(-1)k-1 x
x k=1 (2k)!! 2k
2 (2k -1)!!(2k)! 2k
= log -Σ(-1)k x
x k =1 (2k)!! 2k (2k)!
2
k (2k -1)!!
= log 2 - log x -Σ(-1) x 2k (2k )!=(2k )!!(2k -1)!!
k=1 2k k
(2 )!
Differentiating both sides of this with respect to x repeatedly, we obtain the following.
2
0!
-1 (1) k (2k -1)!!
csch x = - 1
- ( -1 ) Σ 2k (2k -1)!
x 2k-1
x k =1
2
1!
(2) k (2k -1)!!
csch
-1
x = 2
- Σ(-1) x 2k-2
x k=1 2k (2k -2)!
2
2!
(3) k (2k -1)!!
csch x - Σ(-1)
-1
=- 3 x 2k-3
x k =2 2k (2k -3)!
2
(4) 3! (2k -1)!!
csch
-1
x = 4
- Σ(-1)k x 2k-4
x k=2 2k (2k -4)!
Here, considering the relation between the derivative order n and the first term k0 of Σ, it is as follows.
n 0 1 2 3 4
k0 1 1 1 2 2
n
Such a relation can be expressed by k0 = using a ceiling function x = x. Then we obtain
2
(3.c) . Next,
2 (2k -1)!! 2 (2k -1)!!(2k)! 2k
sech -1x = log +Σ x 2k = log -Σ x
x k=1 (2k)!! 2k x k =1 (2k)!! 2k (2k)!
2
(2k -1)!!
= log 2 - log x -Σ x 2k (2k )!=(2k )!!(2k -1)!!
k =1 2k (2 k)!
Differentiating both sides of this with respect to x repeatedly, we obtain (3.s) . ( The number of the first term
of Σ is the same as it of arctan x. )
-6-
Sum of Taylor series of the higher derivative of arccsch x
Formula 9.4.2 in "9 Higher Derivative" was as follows.
(n) n-1
n (-1)r n Ar
csch x Σ
-1
= (-1) 1
r=1 r-
2
x +1
n-1+2r -2
x
where n Ar are coefficients as follows.
1A1 1
2A1 2A2 2 1
3A1 3A2 3A3 6 7 3
4A1 4A2 4A3 4A4 = 24 48 45 15
5A1 5A2 5A3 5A4 5A5 120 360 549 390 105
6A1 6A2 6A3 6A4 6A5 6A6 720 3000 6570 7425 4200 945
From (3.s) and this , the following equation follows for 0<x <1 .
k (2k -1)!!
2
(-1)n n (-1)r n Ar
Σn (-1) (n -1)! +Σ
2k-n
x =
k=
2k (2k - n)! xn r=1 r-
1
2
x 2r-1 x -2 +1
2
And giving x =1 to this without considering the convergence condition, we obtain the following special value.
k (2k -1)!!
2 n (-1)r n Ar
Σ (-1)
n 2k (2k - n)!
= (-1) (n -1)! + 2 Σ
n
r=1 2r
(3.c')
k=
2
Example
1!!2 3!!2 5!!2 7!!2 2
- + - +- = 0! -
21! 43! 65! 87! 2
1!!2 3!!2 5!!2 7!!2 3 2
- + - +- = -1! +
20! 42! 64! 86! 4
3!!2 5!!2 7!!2 9!!2 13 2
- - - +- = -2! +
41! 63! 85! 107! 8
3!!2 5!!2 7!!2 9!!2 75 2
- - - +- = 3! -
40! 62! 84! 106! 16
2006.11.20
2012.07.08 Renewal
K. Kono
Alien's Mathematics
-7-
12 Super Derivative (Non-integer times Derivative)
Defintion 12.1.1
(p)
f (x) obtained by continuing analytically the index of the differentiation operator of Higher Derivative of
a function f(x ) to a complex plane [0 , p ] from a natural number interval [1 , n ] is called Super Derivative
of f(x) .
Example
p
(sin x) p = sin x + + c (x)
()
p cp(x) is an arbitrary function.
2
Definition 12.1.2
We call it Super Differentiation to differentiate a function f with respect to an independent variable x
non-integer times continuously. And it is described as follows.
p
d d d d
f(x) = f(x) : p pieces
dx p dx dx dx
Example
dp p
dx p
cosx = cos x +
2
12.1.3 Fundamental Theorem of Super Differentiation
The following theorem holds from Theorem 7.1.3 in 7.1.
Theorem 12.1.3
r [0 , p] be an continuous function on the closed interval I and be arbitrary the r-th order
(r)
Let f
derivative function of f . And let a(r ) be a continuous function on the closed interval [0 , p ] .
Then the following expression holds for a(r), x I .
dp d p p -1 (r) x x
(p)
f(x) = f (x) + Σ f a(p -r) dx r (1.1)
dx p dx p r=0 a(p) a(p -r)
Especially, when a(r)= a for all k [0 , p] ,
dp (p) d p p -1 (r) (x -a)r
f(x) = f (x) + Σf (a) (1+ r) (1.2)
dx p dx p r=0
Proof
Theorem 7.1.3 in 7.1 can be rewritten as follows.
x x p -1 x x
<p>
f (x) = f(x)dx + Σ f
p <p -r>
a(p -r) dx r
a(p) a(0) r=0 a(p) a(p -r+1)
-1-
Especially, when a(r)= a for all k [0 , p] ,
f(x)dx
<p>
x x p -1
<p -r> (x -a)r
f (x) = n
+ Σf (a)
a a r=0 (1+ r)
Differentiating these both sides with respect to x p times,
d p <p> d p p -1 <p -r> x x
Σ dx r
<0>
pf
(x) = f (x) + p r=0 f a(p -r)
dx dx a(p) a(p -r)
Constant-of-differentiation Function
d p p -1
We call Σ etc. Constant-of-differentiation Function of f(x) . Since p is a real number,
dx p r=0
generally it is difficult to obtain this. However, it becomes easy exceptionally at the time of f(x)= e x .
That is, Constant-of-integration Function in 7.1.3 was as follows.
p -1
a (x -a )
r
(x -a)r (x -a)r+ p
Σ
r=0
e
(1+ r)
=e a
Σ
r=0
-
(1+ r) (1+ r + p)
Differentiating both sides with respect to x p times,
dp (x -a)r
p -1 dp (x -a)r (x -a)r+ p
Σe a
dx p r=0 (1+ r)
=e Σ p
a
r=0 dx
-
(1+ r) (1+ r + p)
From the 12.3.1 mentioned later, the following expressions hold.
d p p -1 a (x -a)r (x -a)r-p (x -a)r
Σ
p r=0 e (1+ r)
= e a
Σ (1+r -p)
-
(1+r)
dx r=0
Definition 12.1.4
dp d p p -1 (r) x x
(p)
f(x) = f (x) + Σ f a(p -r) dx r (1.1)
dx p dx p r=0 a(p) a(p -r)
In this expression,
when Constant-of-differentiation Functin is 0,
-2-
dp
we call f(x) Lineal Super Differentiation and
dx p
we call the function equal to this Lineal Super Derivaive Function.
dp
we call f(x) Collateral Super Differentiation and
dx p
we call the function equal to this Collateral Super Derivaive Function.
These are the same also in (1.2) .
In short , Lineal Super Derivaive Function is what differentiated f(x) with respect to x continuously
without considering the constant-of-differentiation function.
dp x d p p -1 a (x -a)r
pe
= e + x
Σ
p r=0 e (1+ r)
dx dx
Here, using the former expression i.e.
d p p -1 a (x -a)r (x -a)r-p (x -a)r
Σ e = e a
Σ -
dx p r=0 (1+ r) r=0 (1+r -p) (1+r)
we obtain
= e Σ
dp x (x -a)r-p (x -a)r (x -a)r-p
e = e +e Σ
x a
- a
dx p r=0 (1+r -p) (1+r) r=0 (1+r -p)
When a - , since the constant-of-differentiation functin can not be 0, this is a collateral differentiation.
When p =1/2 , a =0 , if the differential quotients on x =0.3 are compared with the calculation result
by Riemann-Liouville differintegral (later 12.2.1), it is as follows.
-3-
And if the lineal super derivative and the collateral super derivative are illustrated side by side, it is as follows.
Remark
It is thought that this collateral super derivative is an asymptotic expansion. And this collateral super derivative
is corresponding with the termwise super differentiation. In general, a termwise super differentiation seems to
become a collateral super differentiation.
c 0
(p) (p)
c f(x) =cf (x) : constant multiple rule
(p) (p) (p)
f(x)+ g(x) = f (x) + g (x) : sum rule
-4-
12.2 Fractional Derivative
(p) 1 dn x
f (x) = (x -t)n-p-1f(t)dt n = p (2.0)
(n -p) dx a n
This expression is called Riemann-Liouville differintegral. "differintegral" is a coined word which combined
"differential" and "integral".
Although the numerical integration and the numerical differentiation are possible for (2.0) with this, the
accuracy of numerical differentiation is bad and the desired result may not be obtained. In this case, the
following formula which replaced the calculation order of integration and differentiation is effective.
(p) 1 x
n-p-1 dn
f (x) = (x -t) f(t) dt n = p (2.0')
(n -p) a dt n
Although this formula has a possibility of cutting off a constant of integration as a result of differentiating
previously, in many cases, it is correctly calculable. This (2.0') is often used in the following chapters.
(1+) -p
dn
(x -t)
1 x
x
(p) n-p-1
= t dt = x ( 0)
(n -p) dx n 0 1+ -p
(- + p) -p
dn
dt
1 x
= (x -t)
n-p-1
t dt = (-1)-p x ( < 0)
(n -p)
n (-)
(x -t)
x (p) 1 dn x
t -p x
e e dt = (1) e
n-p-1
=
(n -p) dx n
Note
When n =p , -(n -p) < <0 , the following expression does not hold.
(p) 1 dn x
x = (x -t)n-p-1 t dt (2.1)
(n -p) dx n
n -p times integral of x ( <0) is carried out, and super primitive function x n-p
+
In this case,
+ n -p > 0 is obtained. According to this, the zero of the super primitive function changes from
to 0. The integral with a fixed lower limit is inapplicable to such an integral.
In this case, the following formula which replaced the calculation order of integration and differentiation
is effective.
-5-
(x -t)
(p) 1 x
n-p-1 dn
x = t dt n =[p] (2.1')
(n -p) dt n
x ( -n <0) is carried out, and super primitive function
-n
If this formula is used, n -p times integral of
-p
x ( -p < 0) is obtained. Since the zero of the super primitive function does not change, the integral
with a fixed lower limit is applicable. (See Example 5 in the following chapter).
Example 1
2
1
(1+1) (2) 2
1 1 1
1- 2 2
x
1 2
= x = x = x
1+1-1/2 3/2
2
1 1
(x -t)
1 d1 x 1- -1
x
1 2 1
= t dt
(1-1/2) dx 1 0
1
(x -t)
x
1 d x - 1 d 2
2 1
= t dt = - x -t (2x +t)
(1/2) dx 0 dx 3 0
3 1
1 d 4 2 2 2
= x = x
dx 3
Example 2
2
1 1
- 1 -x
e -x
= (-1) 2 -x
e = e
i
2
1 1
(x -t)
1 d1 x 1- -1
e
-x 2
= e -tdt
(1-1/2) dx 1 +
1
1 d x 1 d - x
-e erfi x -t
x
2 -t
= (x -t) e dt =
dx + dx
1 d 1 d
= e -x erfi x - = e -xerfi x -
dx i dx
1 d 1 d 1
= e -xerf- = -e -x = e -x
i dx i dx i
Example 3
-6-
log x - (- -2 log 2) -
1 1
-
2 log x + 2 log 2 -
2
= x = x
1 1
(x -t)
1 d1 x 1- -1
(log x) 2 = 2
log t dt
(1-1/2) dx 1 0
1
(x -t)
1 d x -
2
= log t dt
dx 0
x
-2
1 d x -t
= -4 x tanh -1 x -t (log t -2)
dx x 0
Here
x n +x 1
tanh -1 = log = log(n + x) - log(n - x) |x|< n
n n -x 2
Then
x -t 1
tanh -1 = log x + x -t - log x - x -t
x 2
From this
x -t
4 x tanh -1 =2 x log x + x -t - log x - x -t
x
= 4 x log x + x -t - 2 x log x + x -t + log x - x -t
= 4 x log x + x -t - 2 x logx -(x -t)
= 4 x log x + x -t - 2 x log t
Therefore
x -t
4 x tanh -1 + 2 x -t (log t -2)
x
= 4 x log x + x -t - 2 x log t + 2 x -t log t - 4 x -t
= 4 x log x + x -t - 2 x - x -t log t - 4 x -t
Using this
x
1
-4 -2
1 d x -t
(log x) 2 = x tanh -1
x -t (log t -2)
dx x 0
1 d x
= -4 x log x + x -t +2 x - x -t log t +4 x -t 0
dx
1 d
= -4 x log x + 2 x log x
dx
1 d
+ 4 x log x + x -2 x - x log 0-4 x
dx
1 d
= 4 x log2 x - 4 x 2 x log x = x log x
dx
-7-
Furthermore
2
1 1
1 d log x +2 log 2 - 2
(log x) = 4 x (log 2-1) +2 x log x = x
dx
Reference
1
(x -t)
-
2 2
t dt = - x -t (2x +t)
3
1
- 2 z 2
e dt = -e erf x -t
2 t x
(x -t) erf(z)= e -t dt
0
1
(x -t)
-
2
e -t dt = -e -x erf i x -t erf i(z)= erf(iz)/i
1
-8-
12.3 Super Derivative of Power Function
Formula 12.3.1
When (z) is gamma function and n = p is ceiling function, the following expressions hold
(- + p) -p
(x -t) dt
dn
1 x
= (-1) -p
x = n-p-1
t dt ( < 0)
(-) (n -p)
n
1 dn x
= (x -t)n-p-1(at +b) dt
(n -p) dx n b
-
a
-p
(- + p)
1
(ax +b) (p)
= - (ax +b)-p ( < 0)
a (- )
(x -t)
1 x dn
= n-p-1
( + ) dt
n at b
(n -p) dt
Caution !
-p
1 q
Do not describe to be a p in the upper formula. Because, since the law of exponents a p = a pq
a
does not hold for the arbitrary real numbers p, q at the time a < 0,
-p -p
1 1
= (a ) -1 -p
a (-1)(-p )
=a p
i .e . ap
a a
-p
1
If it is described as = ap, the rotation direction on the complex plane becomes reverse and the
a
mistaken result is caused.
For example, when a = -3 , p =1/2 ,
1 1
- - 1 1
-p 2 2
1 1 1 -
= = (-1) = (-1) 2
3 2
= -i 3
a -3 3
1 1 1 1
= (-1)3
p 2 2 2 2
a = (-3) = (-1) 3 = i 3
-9-
Example 1 > p
10
1
(1+2)
1 19 19
x2 2-
10 100
= x = x 10 =0.547239 x 10
2 21+2-1/10 1719/10
10
9
(1+2)
9 11 11
2
x 2-
10 100
= x = x 10 = 0.955579 x 10
2 21+2-9/10 111/10
<1/10> <9/10>
When x 2/2 , x 2/2 , x 2/2 , x1 are drawn on a figure side by side, it is as follows.
Example 2 = p
1 3
1+
x
1 1 1
2 2 -
2 2 2
= x = x0 =
1
1 1 2
1+ -
2 2
Example 3 0 < p
2
1
(1+0) (1) - 2
1 1 1
0- 1 -2
x
0 2
= x = x = x
1+0-1/2 (1/2)
1 (2) (1+1) 1-2 (2) -1 1
x = x = x = x -1 = 0
1+1-2 (0)
Example 4 < 0
2
1
-(-1)+1/2 -1- 2 -2
1 1 3
-
x
-1 2
= (-1) x = -i x
-(-1) 2
-1 (1) -(-1)+1 -1-1 (2) -2
x = (-1)-1 x =- x = -x -2
-(-1) (1)
1 (2)
-(-1/2)+2 - 2 -2 (5/2) - 2
x
1 5 5
-
2 -2 3 -
= (-1) x = x = x 2
-(-1/2) (1/2) 4
- 10 -
Example 5 -p - p < < 0
2
1
x
1 1
- - +
x
1 1 1 5
(5/6) - 6
1
- - 3 2 - -
= -i
3 2 3 2
= (-1) x
(1/3)
1
- -
3
5 5
1.1287 - 6 -
6
=- ix = -0.42135 i x
2.6789
2
1
x
1 1 1 4
(x -t) dt = - 3 (x -t)
1
-
3 1 x -
2 d1 - 3 1 x -
2
-
3
= t t dt
1-
1 dt 1
2
This last integral cannot be expressed with an elementary function. Then, if the value on x=1 is numerically
integrated by mathematical software, it is as follows. This result is consistent with the previous value
-0.42135i exactly.
Rie mann-Liouville diffe rinte gral
a := -1/3: p := 1/2:
df := diff(t^a,t)
1
4
3t
3
fl := x-> 1/gamma(ceil(p)-p)
*int((x-t)^(ceil(p)-p-1)*df, t=infinity..x)
x
1 p p 1
x (x t) df d t
( p p)
float(fl(1))
0.4213560763 i
Formula 12.3.2
Let n be a non-negative integer, -1!! 1, (2n -1)!! 135(2n -1) ,
0!! 1, 2n !! 2462n ,
then following expressions hold.
2
1 1
(2n)!! n-
x n
= x 2
(2n -1)!!
2
1
(n +1)(2n +1)!! n
x
1
n+
2
= x
2(n +1)!!
- 11 -
(2) Linear form
2
1 1
(2n)!! a n-
(ax +b) n 2
= (ax +b)
(2n -1)!!
2
1
(ax +b)
1
n+ (n +1)(2n +1)!!
2
= a (ax +b)n
2(n +1)!!
Proof
Upper rows 0 of the linear form of Formula 12.3.1 were as follows.
1
2
1 -
(1+n)
2 1
1 n-
(ax +b)n
2
= (ax +b)
a 1
1+n -
2
1
(ax +b)
1 1
2 - 1+n +
(ax +b)
1 2
n+
2 1 2 n
=
a (1+n)
Here, when n is a non-negative integer,
1 (2n -1)!!
n+ = n
, (2n)!! = 2n n !
2 2
Then
(1+n) (1+n) 2n n ! (2n)!!
= = =
(2n -1)!! (2n -1)!!
1 1
1+n - n+
2 2
(1+n)+
1 1
1+n +
2 2 1 2(1+n)-1 !!
= =
(1+n) (1+n) 2 n! 1+n
Substituting these for the previous formula, we obtain the linear form, andgiving a =1, b =0 to them,
we obtain the basic form.
Example 1
2
1 1 1 1
0!! - 1 - -
x
0
= x 2
= x 2
( a 0x 2
)
(-1)!!
2
1 1 1 1
2!! 2
x
1
= x 2
= x 2
( a 1x ) 2
1!!
2
1 3 3 3
4!! 8
x
2
= x 2
= x 2
( a 2x ) 2
3!! 3
2
1 5 5 5
6!! 16
x
3
= x 2
= x 2
( a 3x ) 2
5!! 5
- 12 -
Example 2
2
1
11!! 0
x
1
2 1 0
= x = x0 (= x )
2!! 2 a1
1
2
23!! 1 3 1
x
3
2 2 1
= x = x (= x )
4!! 4 a2
2
1
35!! 2 15 2
x
5
2 3 2
= x = x (= x )
6!! 16 a3
2
1
47!! 3 35 3
x
7
2 4 3
= x = x (= x )
8!! 32 a4
Formula 12.3.3
When n denots a natural number, the following expression holds.
1 1
2n +1 n (-1)k n
k x
2 n-
x
n
Σ
2
=
k =0 2k +1
Proof
Let n be a natural number, = n , p =1/2 . Then, since > p , from Formula 12.3.1 we obtain
the following expression.
1 1
(x -t)
2 1 d1 x - 1 d x tn
x
n 2 n
= t dt = dt
(1/2) dx 1 0 dx 0 x -t
Here, according to「岩波数学公式Ⅰ」p96, the following expressions hold
(x -t)n-r
tn
r
2 x -t n n
dt = n+1 r=0
(-x)r
Σ
2n -2r +1
x -t (-1)
Then
x
(x -t)n-r
tn
x 2 x -t n n
dt = Σ (-x)r
0 x -t (-1)n+1 r=0 r 2n -2r +1
0
1
xn
r
2 n n
Σ(-1)r
2
= x
(-1)n r=0 2n -2r +1
Differentiate both sides of this with respect to x as follows.
1
tn
d x d 2 n n 1 n+
dx
dt = n
dx (-1) r=0Σ (-1) r
r 2n r
-2 +1
x 2
0 x -t
1
n
2 1 n 1 n-
n n 2 Σ
r
= + (-1) x 2
(-1) r=0 r 2n -2r +1
- 13 -
1
(-1)r-n
r x
n n n-
= (2n +1)Σ 2
r=0 2n -2r +1
2
1 1
(-1)r-n
r x
2n +1 n n n-
x
n
= Σ
2
r=0 2n -2r +1
Here, we devise further,
r-n
(-1)n-r (-1)k
r k
n (-1) n n n n n
Σ
r=0 2n -2r +1
=Σ
r=0 2(n -r)+1 n -r
=Σ
k =0 2k +1
Using this, we obtain
1 1
(-1)k
k x
2 2n +1 n n n-
x
n
Σ
2
=
k =0 2k +1
By-product
Comparing Formula 12.3.2 and Formula 12.3.3 , we obtain the following formula.
k
k
n (-1)n (2n)!!
Σ
k =0 2k +1
=
(2n +1)!!
This is the same as the by-product in 7.3.3 .
Lemma
When m,n are natural numbers, the following expression holds.
m-1
1
(x -t)n-r
m(x -t) m n
r
- n
m n
(x -t) t dt = n+1
(- x) r
Σ m(n -r)+(m -1)
(4.0)
(-1)
r=0
Proof
Let m-1
m
(x -t)n-r
r
m(x -t) n n
F(t) = Σ (-x) r
(-1)n+1 r=0 m(n -r)+(m -1)
Differentiate this with respect to t. Then
m-1
m
(x -t)n-r
r
d d m(x -t) n n
dt
F(t) =
dt (-1)n+1 Σ
r=0
(-x) r
m(n -r)+(m -1)
m-1
m
(x -t)n-r
r
m(x -t) d n n
+ Σ (-x) r
(-1)n+1 dt r=0 m(n -r)+(m -1)
1
-
m
(x -t)n-r
r
(m -1)(x -t) n n
=- Σ (-x) r
(-1)n+1 r=0 m(n -r)+(m -1)
- 14 -
m-1
m
(x -t)n-r-1
r
m(x -t) n n
- Σ (n -r)(-x) r
(-1)n+1 r=0 m(n -r)+(m -1)
1
-
m
(x -t)n-r
r
(x -t) n n
=
(-1)n
Σ
r=0
(m -1)+m(n -r)(-x)
r
m(n -r)+(m -1)
1 1
- -
m m
r
(x -t) n n (x -t)
= Σ (-x)r(x -t)n-r = (-t)n
(-1)n r=0 (-1) n
1
-
m n
= (x -t) t
Formula 12.3.4
When n, m are natural numbers such that n 1, m 2 , the following expressions hold.
1 1
(-1)k
k x
m m(mn +m -1) n n n-
x
(-1/m) Σ
n m
=- (4.1)
k =0 mk +(m -1)
k 1
Proof
(1+ ) -p
(x -t)
(p) 1 dk x
k-p-1
x = x = t dt ( p)
1+ -p (k -p) dx k 0
(x -t)
1 d1 x -
x n
= m n
t dt
(1-1/m) dx 1 0
1
(x -t)
m d x -
m n
=- t dt
(-1/m) dx 0
Using the above Lemma, we calculate as follows.
x
m-1
1
(x -t)n-r
(x -t)
m
r
x - m(x -t) n n
m n
t dt = Σ (-x) r
0 (-1)n+1 r=0 m(n -r)+(m -1)
0
m-1
m n-r
r
mx n n x
=
(-1)n
Σ
r=0
(-x) r
m(n -r)+(m -1)
- 15 -
1
(-1)r-n
r x
n n n- +1
= mΣ m
r=0 m(n -r)+(m -1)
Furthermore, since r,n are integers,
r-n
(-1)n-r
r
n (-1) n n n
Σ
r=0 m(n -r)+(m -1)
=Σ
r=0 m(n -r)+(m -1) n -r
(-1)k
n n
=Σ
k =0 mk +(m -1) k
Using this,
1 1
(x -t)
(-1)k
k x
x - n n n- +1
m n
t dt = mΣ m
0 k =0 mk +(m -1)
Differentiating this with respect to x,
1 1
(x -t)
(-1)k
k x
d x - n n n-
m n
t dt = (mn +m -1)Σ m
dx 0 k =0 mk +(m -1)
Thus
m
1 1
(-1)k
k x
m(mn +m -1) n n n-
x
(-1/m) Σ
n m
=- (4.1)
k =0 mk +(m -1)
Moreover, (4.1') follows immediately from this.
Next,
1
(1+ n) (1+ n) (-1/m)
1 1
m n- n-
x
n
= m
x = x m
(1+ n -1/m) (-1/m) (1+ n -1/m)
Since this have to be equal to (4.1) ,
Remark
(4.2) suggests that (4.1) can be expressed with a beta function and n,m can be real numbers. Actuality,
(1+ ) (1+ ) -p (1+ , -p)
= =
(1+ - p) (-p) (1+ -p) (-p)
Then,
(p) (1+ ) (1+ , -p) -p
x = x -p = x ( 0)
(1+ -p) (-p)
Formula 12.3.5
When n is a natural number, the following expressions hold for 0< p n .
- 16 -
n +1- p n (-1)k
k
n (p)
n
x x n-p
(1-p) Σ
= (5.1)
k =0 k +1- p
k
k +1- p k
n -p (-1) n -1
n -1
(n , -p) = - Σ ( )denots beta function. (5.2)
p k =0
Example 1
2 3 (1+2) 2
x x 2- 3
= = x 2- 3 = 2.2151 x 2- 3
1+2- 3 3- 3
2 3 3- 3 1 2 1 2 1 2
x = - + x 2- 3
1- 3 1- 3 0 2- 3 1 3- 3 2
= 2.2151 x 2- 3
Example 2
2- 3 1 1 1 1
2, - 3 = - - = 0.788675
3 1- 3 0 2- 3 1
3- e 1 2 1 2 1 2
(3, -e) = - - + = -0.596137
e 1- e 0 2- e 1 3- e 2
Formula 12.3.6
The following expressions hold for p ,q such that 0< p q .
r
r x
(p) q +1- p (-1) q q-p
x q
(1-p) Σ
= (6.1)
r=0 r +1- p
r
p Σ r +1- p r
q -p (-1) q -1
(q , -p) = - ( )denots beta function. (6.2)
r=0
For example, in the case of the 1/2 times derivative of f(x)= x 2-2 x + 2, the following calculation is
right in many case.
2 2
1 1 3
8
x -2 x + = (x -) (x -)
2 2 2 2
=
3
If this is calculated termwise as follows, the result is different from the former.
- 17 -
2
1
2 2 2
1 1 1
3 1 1
1 8 2 2 -
= x - 4 x 2 + x 2
3
Needless to say, this cause is the difference between
1 1 1
dx
x x - x x - x x -
x -2 x + dx and
2 2 2
x -2 xdx 2
2
+ 2 2
0 0
That is, it is because the latter regarded it as 0 and although the former regarded the zero of the super
derivative as pi.
The latter is right when there is a special reason why the zero of the super derivative should be 0. However,
such a case is rare, and in almost all cases the former is right.
- 18 -
12.4 Super Derivative of Exponential Function
Analytically continuing the index of the differentiation operator in Formula 9.2.2 ( 9.2 ) to [0 , p ] from [1,n ]
we obtain the following formula. In addition, Rieamnn-Liouville diffeintegrals are also expressed together
Formula 12.4.1
When n = p is ceiling function, the following expressions hold
(x -t)
x (p) 1 dn x
e = (1)-pe x = n-p-1
e dt
t
(n -p) dx n
(2) Linear form
(x -t)
-p
dn
1 1 x
ax+b (p)
e
ax+b n-p-1 at+b
= e = e dt
a (n -p) dx n
a >0 : - , a <0 : +
(3) General form
(x -t)
-p
dn
1 1 x
ax+b (p)
= ax+b = n-p-1
a t+bdt
a log (n -p) dx n
a >0 : - , a <0 : +
Example
2
1 1
- 1 -x
e -x
= (-1) 2 -x
e = e = -ie -x
i
- 2
3x-4 2
1
e = e 3x-4 = 4.728804 e 3x-4
3
-i
x (i) 1
2 = 2x = (0.933582 - 0.358362 i) 2x
log 2
1
1 -
2
2 1
(-3)x = (-3)x
log(-3)
= (1.487742 + 1.05582 i) (-3)x
2
1
-1/2
1
3
x
= 3x = 1.048147 3x
log 3
x (1) (1/2)
When 3 , 3x , 3x are drawn on a figure side by side, it is as follows.
- 19 -
y
10
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
x
3^x*ln(3)
3^x/(1/ln(3))^(1/2)
3^x
- 20 -
12.5 Super Derivative of Logarithmic Function
Reversing the sign of the index of the integration operator in Formula 7.5.1 ( 7.5 ), we obtain the following
formula. In addition, Rieamnn-Liouville differintegrals are also expressed together
Formula 12.5.1
When (z), (z), n = p denote zeta function, psi function, ceiling function respectively, the following
expressions hold.
(p) b
log(ax +b ) = x+
(1-p) a
1 dn x
= (x -t)n-p-1log(at +b)dt
(n -p) dx n b
-
a
(1- p)
Where, = (-1)p(p -1)! for p =1, 2, 3,
(1- p)
Proof
Formula 7.5.1 ( 7.5 ) was as follows.
- (1+p)- p
x x log x
p
log xdx = x
0 0 (1+ p)
log(ax +b)- (1+ p) -
x p
x b
p
log(ax +b)dx = x+
b b (1+ p) a
- -
a a
Since differentiation is the reverse operation of integration, replacing the index p of the integration operator
with -p , we obtain the desired expressions. And Formula 1.3.1 ( 1.3 ) was
(-n)
= (-1)n+1n! , n =0, 1, 2, 3,
(-n)
Then, replacing n with p -1 , we obtain
(1-p)
= (-1)p(p -1)! , p =1, 2, 3,
(1-p)
Example
log x - 1-1- -1
(log x)(1) = x = -(-1)10! x -1 = x -1
1-1
1 1
log(3x +4) - 1- - -
2
1 2
2 4
log(3x +4) = x+
1 3
1-
2
= 0.5641(log x + 1.3862) / x +1.3333
- 21 -
1
log x - 1- -
=
1 1 1
10 10 -
10
-
10
(log x) x = 0.9357x (log x +0.1777)
1
1-
10
9
log x - 1- -
=
9 9 9
10 10 -
10
-
10
(log x) x = 0.1051 x (log x +9.8465)
9
1-
10
y
2
0
1 2 3 4 5 6 7
x
-2
-4
-6
ln(x)
-1/x^(1/10)/gamma(9/10)*(EULER - ln(x) + psi(9/10)
-1/x^(9/10)/gamma(1/10)*(EULER - ln(x) + psi(1/10)
1/x
- 22 -
12.6 Super Derivative of Trigonometric Function
Formula 12.6.1
(1) Basic form
p
(sin x) p
()
= sin x +
2
p
(cosx) p
()
= cos x +
2
(2) Linear form
-p
p
(p) 1
sin(ax +b ) = sin ax +b +
a 2
-p
p
(p) 1
cos(ax +b ) = cos ax +b +
a 2
Example
1
1
(sin x ) 2
= sin x +
2 2 = sin x + 4
1 1 1
((sin x ) 2 ) 2 = (sin (x + )) 2
4
1
-
2
1
= cos x
1
= sin x + + = sin x +
1 4 2 2 2
(1) -1
1
= sin x
1
sin x - = sin x - +
2 1 2 2
1
1
(cos x) 2
= cos x +
2 2 = cos x + 4
2
2
(cos x)
= cos x +
2 = cos(x +1)
When cos x , (cos x)(1/2) , -sin x are drawn on a figure side by side, it is as follows. Red shows
1/2 order super derivative. It is clear also in the figure that super derivative which is the easiest to understand
is super derivative of trigonometric functions.
- 23 -
12.6.2 Termwise Super Derivative of sin x , cos x
Reversing the sign of the index of the operator of the collateral super integrals of sin x , cos x in 7.6.2 ( 7.6 ) ,
we obtain the following termwise super derivatives. These are collateral super derivatives as understood from
the constant-of-differentiation function in the right side.
(-1)k p
+ C(p, x)
(p)
2k+1- p
(sin x) =Σ x = sin x +
k =0 (2k +2- p) 2
p x -p-k k
C(p, x) =Σ sin
k =1 (1- p -k) 2
k
p
- C(p, x)
(-1)
(cosx) p =Σ p
()
x 2k- = cos x +
k =0 (2k +1- p) 2
p x -p-k k
C(p, x) =Σ cos
k =1 (1- p -k) 2
When the 1/2th order collateral super derivative of cos x is drawn as cos x and -sin x side by side, it is as
follows. Red shows the 1/2th order collateral super derivative.
- 24 -
Compared with the upper figure, the collateral super derivative is curving unnaturally near the coordinate origin
in this figure. Since this is similar also in sin x, it is thought that the termwie super derivatives of sin x and cos x
are asymptotic expansions of the lineal super derivatives.
- 25 -
12.7 Super Derivative of Hyperbolic Function
Formula 12.7.1
(1) Basic form
p i e x - (-1)-pe -x
(p) -p
(sinh x) =i sinh x + =
2 2
p i e x + (-1)-pe -x
(p) -p
(cosh x) =i cosh x + =
2 2
(2) Linear form
-p
p i
(p) i
sinh(ax +b ) = sinh ax +b +
a 2
-n
1 1
e - (-1)-pe -(ax+b)
ax+b
=
2 a
-p
p i
(p) i
cosh(ax +b ) = cosh ax +b +
a 2
-n
1 1
e + (-1)-pe -(ax+b)
ax+b
=
2 a
Example 1
1 i i
1 1 1
- -
(sinh x ) 2 =i 2
sinh x + =i 2
sinh x +
2 2 4
i 2
1 1 1 1
-
((sinh x ) 2 ) 2 = (i 2 sinh (x + ))
4
i 1 i
1 1
- -
2 2
=i i sinh x + +
4 2 2
i
= i -1 sinh x + 2 = cosh x
i (1) -1
i 1 i
i
cosh x + = cosh x + +
2 1 2 2
= -i cosh(x + i) = i cosh x
i i
-
(cosh x) 10 =i 10
cosh x - = 1.170088 cosh x -
20 20
9i 9i
9 9
-
(cosh x) 10 =i 10
cosh x - = 4.111207 cosh x -
20 20
- 26 -
Super derivative of hyperbolic function is the most incomprehensible in super derivatives. The reason is that
the super derivative turns into a complex function except the order p is an integer or a purely imaginary number.
(i/10) (9 i/10)
Then, when cosh x , (cosh x) , (cosh x) , sinh x which can be displayed on a real number
domain are drawn on a figure side by side, it is as follows.
400
300
200
100
-4 -3 -2 -1 1 2 3 4 5 6
x
cosh(x)
1/(I)^(1/10*I)*cosh(x - 1/20*PI)
1/(I)^(9/10*I)*cosh(x - 9/20*PI)
sinh(x)
(i/10)
All of four curves have overlapped in the positive area. It is natural that (cosh x) is near cosh x in
(9 i/10)
a negative area. But (cosh x) is far apart from sinh x in why.
Example 2
1
-
1 x
2 e - (-1) 2 -x
e e x + i e -x
(sinh x ) = =
2 2
2
1
2
1
) =
1 1
x x
e + i e -x e i
((sinh x ) 2 2
= + e -x
2 2 2
ex i e x + e -x
= + -i e -x = = cosh x
2 2 2
i i
e
i
(1) -1
cosh x + 2
1 1 x+
2 -1
- x+
2
= + (-1) e
2 1
i i
=
1
2
-
e 2 e x - e 2 e -x =
1
2
1
i e x - e -x
i
i
= e x + e -x = i cosh x
2
i i
-
-
-x i -x
i
e x - (-1) e e x - e e e x - e 1-x
(sinh x) = = =
2 2 2
- 27 -
12.7.2 Termwise Super Derivative of sinh x , cosh x
Reversing the sign of the index of the operator of the collateral super integrals of sinh x , cosh x in 7.7.2
we obtain the following termwise super derivatives. These are collateral super derivatives as understood from
the constant-of-integration function in the right side.
p i
p
x 2k+1-
- C(p, x)
(p)
(sinh x) =Σ =i -p
sinh x +
k=0 (2k +2- p) 2
p x -p-k k i
C(p, x) Σ i -k sinh
k=1 (1- p -k) 2
p i
p
x 2k-
- C(p, x)
(p)
(cosh x) =Σ = i -p cohs x +
k=0 (2k +1- p) 2
p x -p-k k i
C(p, x) Σ i -k cosh
k=1 (1- p -k) 2
When the values of the lineal and the collatera l .7th order derivatives of cosh x on x=1, x=6 are calculated
respectively, it is as follows.
Though the difference of both is large where x is small, both are almost corresponding ing where x is large.
Since this is similar also in sinh x, it is thought that the termwise super derivatives of sinh x and cosh x are
asymptotic expansions of the lineal super derivatives.
- 28 -
12.8 Super Derivative of Inverse Trigonometric Function
-1 -1
12.8.1 Super Derivatives of tan x , cot x
Reversing the sign of the index of the integration operator in Formula 7.8.1 ( 7.8 ), we obtain the following
formula.
Formula 12.8.1
When (x),(x) denote gamma function and digamma function respectively, the following expressions
hold for x 1 .
tan -1x
(p) -p
tan x Σ (-1)k
-1
= x -p-2k
(1-p) k =0 -p -2k
log1+ x 2
-p
+
2(1-p) k=1 Σ (-1)k
-p +1-2k
x -p+1-2k
1 -p
Σ (-1)r (1- p)- (2r)x
-p+1-2r
-
(1-p) r=1 -p +1-2r
x -p tan -1x
(p) -p
cot x Σ (-1)k
-1 -1
= cot x - x -p-2k
(1- p) (1-p) k =1 -p -2k
log1+ x 2
-p
-
2(1-p) k=1 Σ (-1)k
-p +1-2k
x -p+1-2k
1 -p
(1- p)- (2r)x
(1-p) Σ
r -p+1-2r
+ (-1)
r=1 - p +1-2 r
-1
Example: 1/2th order derivative of cot x
- 29 -
x
1 1 p1
x (x t) arccot(t) d t
(1 p)
0
2010.07.07
K. Kono
Alien's Mathematics
- 30 -
13 Termwise Super Derivative
In this chapter, for the function whose super derivatives are difficult to be expressed with easy formulas, we
differentiate the series expansion of these functions non integer times termwise. Therefore, e x, log x , sin x ,
cos x , sinh x , cosh x mentioned in " 12 Super Derivative " are not treated here.
Formula 13.1.1
Let (x) be gamma function, , are ceilling function and floor function and let Bernoulli number B2n
and Euler number E2k are as follows.
E2k
(sec x) p
()
Σ
2k-p
= x : collateral
p +1 (2k -p +1)
k=
2
E2k
(sech x) p =
( )
Σ p +1 (2k p
- +1)
x 2k-p : collateral
k=
2
Proof
There were the following formulas in " 10 Termwise Higher Derivative "
E2k
(sec x) n
()
Formula 10.7.1 = Σ
n +1 (2k -n)!
x 2k-n
k=
2
E2k
(sech x) n =
()
Formula 10.8.1 Σ n +1 (2k n
- )!
x 2k-n
k=
2
In these formulas, replacing m! with gamma function (1+ m) and analytically continuing the index of
the differentiation operator to [0 , p ] from [1 , n ], we obtain the desired expressions.
Since the termwise super integrals of arcsin x and arccos x were collateral, the super derivatives have to be
collateral too. (This is the same in the following chapters.)
-1-
Example 1: 3/4 th order derivative of tan x
We calculated the the super differential coefficients on arbitrary one point x =0.4 according to the formula
and Riemann-Liouville differintegral. As the result, two values were almost corresponding. Moreover, in the figure,
blue shows tan x, red shows the 3/4 th order derivative and green shows the 1 st order derivative.
Te rmwise supe r de rivative of tan x
tanp := (p,x)-> sum(2^(2*k)*(2^(2*k)-1)*abs(bernoulli(2*k))/((2*k
*gamma(2*k-p))*x^(2*k-1-p),k=ceil((p+1)/2)..200
2k 2k
200
2 2 1 bernoulli(2 k) 2 k 1 p
( p, x) x
p1
(2 k) (2 k p)
k
2
-2-
Example 3: 1/2 th order derivative of sec x
We calculated the the super differential coefficients on arbitrary one point x =0.3 according to the formula
and Riemann-Liouville differintegral. As the result, two values were almost corresponding. Moreover, in the figure,
blue shows sec x, red shows the 1/2 th order derivative and green shows the 1 st order derivative.
Formula 13.1.2
Let (x) be gamma function, , are ceilling function and floor function and let Bernoulli number B2n
and Euler number E2k are as follows.
x - 2
(p)
(cot x) =- Σ
p +1 2k (2k - p)
k=
2
E2k 2k-p
p +1 (2k -p +1)
x-
(p)
(csc x) = Σ 2
: collateral
k=
2
-3-
Proof
There were the following formulas in " 10 Termwise Higher Derivative "
x - 2
(n)
Formula 10.3.1' (cot x) = - Σ
n +1 2k(2k - n -1)!
k=
2
E2k 2k-n
x - 2
(n)
Formula 10.5.1' (csc x) = Σ
n +1 (2k -n)!
k=
2
In these formulas, replacing m! with gamma function (1+ m) and analytically continuing the index o
the differentiation operator to [0 , p ] from [1 , n ], we obtain the desired expressions.
-4-
Example 2: 14/15 th order derivative of csc x
Only the figure is shown. Blue shows csc x, red shows the 14/15 th order derivative and green shows the
1 st order derivative.
The following lineal termwise super derivatives exist for csch x and sech x.
Formula 13.1.3
The following expressions hold for p 0 , x >0 .
(2k +1)p
(csch x) (p)
= (-1) 2Σ
-p
x
e(2k +1) k =0
p
k (2k +1)
(sech x) = (-1) 2Σ(-1) (2k +1)x
(p) -p
k =0 e
Proof
Formula 8.1.3 ( 8.1 ) was as follows.
csch x dx
x
x x e -(2k +1)
p
= (-1) 2Σ p
k =0 (2k +1)p
x
x x e -(2k +1)
sech x dx = (-1) 2Σ(-1)
p p k
k =0 (2k +1)p
Since differentiation is the reverse operation of integration, replacing the index p of the integration operator
with -p , we obtain the desired expressions.
-5-
-6-
13.2 Termwise Super Derivative of Inverse Trigonometric Functions
Formula 13.2.1
When (x) is gamma function and is ceilling function, the following expressons hold for p0 and
0 < x <1 .
(p) (2k)! 2k+1- p
tan x Σ (-1)k
-1
= x
k=
p -1
(2k +2- p )
2
(p) x
-p (2k)! 2k+1- p
cot x Σ (-1)k
-1
= - x
2 (1-p)
k=
p -1
(2k +2- p)
2
2
(p) (2k -1)!!
sin x Σ
-1
= x 2k+1-p : collateral
p -1 (2k +2- p )
k=
2
(p) x
-p (2k -1)!!
2
cos x Σ
-1
= - x 2k+1-p : collateral
2 (1-p) p -1 (2k +2- p )
k=
2
Proof
There were the following formulas in " 11 Termwise Higher Derivative " .
(n) (2k)! 2k+1- n
tan x Σ (-1)k
-1
Formula 11.1.1 = x
n -1
k=
(2k +1- n)!
2
2
(n) (2k -1)!!
sin x Σ
-1
Formula 11.1.2 = x 2k+1-n
n -1
k=
(2k +1- n)!
2
In these formulas, replacing m! with gamma function (1+ m) and analytically continuing the index of
-1 (p ) (p )
the differentiation operator to [0 , p ] from [1 , n ], we obtain tan x , sin -1x . Next,
cot -1x = x 0 - tan -1x
2
From this,
(p) 0 (p) (p)
cot x x - tan -1x
-1
=
2
Substituting
-p
x (2k)!
0 (p) (p) 2k+1- p
x tan x Σ (-1)k
-1
= , = x
(1-p)
k=
p -1
(2k +2- p)
2
-1 (p ) -1 (p )
for this, we obtain cot x . cos x is also obtained in a similar way.
Note
When p =1, 2, 3, , (1-p) =(0),(-1),(-2), i.e. (1-p) = . Then,
-p
x
= 0 for p =1, 2, 3,
(1-p)
(p ) (p )
Therefore, If we replace p with n , cot -1x , cos -1x results in the following formulas in 11.1 .
-7-
(n) (2k)! 2k+1- n
cot x Σ (-1)k
-1
Formula 11.1.1 =- x
k=
n -1
(2k +1- n)!
2
2
(n) (2k -1)!!
cos x Σ
-1
Formula 11.1.4 = - x 2k+1-n
k=
n -1
(2k +1- n)!
2
-8-
Example 3: 4/5 th order derivative of arcsin x
Only the figure is shown. Blue shows arcsin x, red shows the 4/5 th order derivative and green shows the 1 st
order derivative.
-9-
13.3 Termwise Super Derivative of Inverse Hyperbolic Functions
Formula 13.3.1
When (x), (x), , are gamma function, digamma function, ceilling function and Euler-Mascheroni
constant (= 0.57721566...) , the following expressons hold for p0 and 0 < x <1 .
(p) (2k)! 2k+1- p
tanh x Σ
-1
= x
p -1 (2k +2- p )
k=
2
2
(p) k (2k -1)!! p
sinh x Σ
-1
= (-1) x 2k+1- : collateral
k =
p -1
(2k +2- p)
2
x -p 2
- Σ 2k (2k -p+1) x
(p) 2 (2k -1)!!
sech
-1
x = log + (1- p)+ 2k-p
(1- p) x p
k=
2
: collateral
-p 2
x (2k -1) !!
(p) 2
csch x log + (1- p) +
-1
= - Σ (-1)k x 2k-p
(1- p) x p 2k (2k -p+1)
k=
2
: collateral
Proof
There were the following formulas in " 11 Termwise Higher Derivative "
(n) (2k)! n
tanh x Σ
-1
Formula 11.2.1t = x 2k+1-
n -1
k=
(2k +1- n )!
2
2
(n) k (2k -1)!!
sinh x Σ
-1
Formula 11.2.2s = (-1) x 2k+1-n
n -1
k=
(2k +1- n)!
2
In these formulas, replacing m! with gamma function (1+ m) and analytically continuing the index of
-1 (p ) (p )
the differentiation operator to [0 , p ] from [1 , n ], we obtain tanh x , sinh -1x .
Next, arcsech x is expanded to Tylor series for 0< x <1 as follows.
2
(2k -1)!!
sech x = log 2x - log x -Σ
-1 0
x 2k
k =1 2k (2k)!
Differentiating both sides of this with respect to x n times,
2
(n) (n) (2k -1)!!
sech
-1
x = log 2x 0 - (log x)(n) - Σ x 2k-n
n
k=
2k (2k - n)!
2
Substituting
x -n 2
-Σ
(n) 2 (2k -1)!!
sech
-1
x = log + (1- n)+ x 2k-n
(1- n) x n
k=
2k (2k - n)!
2
-1 (p ) -1 (p )
Replacing n with p , we obtain sech x . csch x is also obtained in a similar way.
- 10 -
Note
According to Formula 1.3.1 ( 1.3 ) ,
(1-n)
= (-1)n(n -1)! n =1, 2, 3,
(1-n)
-1 (n )
If we substitute this for sech x in the proof, it results in Formula 11.2.3 ( 11.2 ) as follows.
2
(n) (n -1)! (2k -1)!!
sech x n
-Σ
-1
= (-1) n x 2k-n
x n 2k (2k - n )!
k=
2
(p) 1 x
n-p-1 dn
f (x) = (x -t) f(t) dt n = p
(n -p) a dt n
We calculated the the super differential coefficients on arbitrary one point x =0.3 according to the formula
and this expression. As the result, two values were corresponding.
- 11 -
Example 3: 6/7 th order derivative of arcsech x
Only the figure is shown. Blue shows arcsech x, red shows the 6/7 th order derivative and green shows the
1 st order derivative.
-5
2006.11.22
2012.07.22 Renewal
K. Kono
Alien's Mathematics
- 12 -
14 Higher and Super Calculus of Logarithmic Integral etc.
x t e
Ei(x) = dt (1.0)
- t
Integrating both sides of (1.0) with respect to x repeatedly by ONLINE INTEGRATOR (Wolfram Mathematica)
and arranging the results, we obtain the following higher indefinite integrals.
1 n -1
Ei(x)dx n =
n-1- r
x nEi(x)-e xΣr! x
n! r=0
Although these right sides are the lineal primitive functions of Ei(x), since both zeros of Ei(x) and ex are - ,
zeros of the right sides are all - . Therfore, the lineal higher primitive function of Ei(x) can be expressed by
the higher integral with a fixed lower limit - .
Formula 14.1.1
x et
When Exponential Integral is Ei(x)= dt , the following expressions hold.
- t
x x 1 n -1
n-1- r
Ei(x)dx n = x nEi(x)-e xΣr! x (1.n)
- - n! r=0
-1-
-2-
14.2 Higher Integral of Cosine Integral
Cosine Integral is defined as follows.
x cos t
Ci(x)= dt
t
Integrating both sides of this with respect to x repeatedly by ONLINE INTEGRATOR and arranging
the results, we obtain the following higher indefinite integrals.
Ci(x)dx
1
x Ci(x) - 0!sin x
1
=
1!
Ci(x)dx
1
x Ci(x) - 0!xsin x + 1!cos x
2 2
=
2!
Ci(x)dx
1
x Ci(x) - 0!x -2!sin x + 1!x cos x
3 3 2
=
3!
Ci(x)dx
1
x Ci(x) - 0!x -2!xsin x + 1!x -3!cos x
4 4 3 2
=
4!
1 (n -1)/2
Ci(x)dx n = Ci(x)x n - sin x Σ (-1)r(2r)! x n-1-2r
n! r=0
(n -2)/2
+ cos x Σ (-1)r(2r +1)! x n-2-2r
r=0
Although these right sides are the lineal primitive functions of Ci(x), these zeros are all . (See the above
figure.) Therfore, the lineal higher primitive function of Ci(x) can be expressed by the higher integral with a
fixed lower limit .
Formula 14.2.1
x cost
When is floor function and Ci(x)= dt ,is Cosine Integral, the following expressions hold.
t
Ci(x)dx
x 1
x Ci(x) - 0!sin x
1
=
1!
Ci(x)dx
x x 1
(n-1)/2
n-1-2r
n
= Ci(x x
) n
- sin x Σ (-1)r(2r)! x
n ! r=0
-3-
(n-2)/2
+ cosx Σ (-1)r(2r +1)! x n-2-2r n 2
r=0
-4-
14.3 Collateral Higher Integral of Sine Integral
Sine Integral is defined as follows.
x sin t
Si(x)= dt
0 t
Integrating both sides of this with respect to x repeatedly by ONLINE INTEGRATOR and arranging
the results, we obtain the following higher indefinite integrals.
Si(x)dx
1
x Si(x) + 0!cos x
1
=
1!
Si(x)dx
1
x Si(x) + 0!x cos x + 1!sin x
2 2
=
2!
Si(x)dx
1
x Si(x) + 0!x -2!cos x +1! x sin x
3 3 2
=
3!
Si(x)dx
1
x Si(x) + 0!x -2!xcos x + 1!x -3!sin x
4 4 3 2
=
4!
1 (n -1)/2
Si(x)dx n = Si(x)x n + cos x Σ (-1)r(2r)! x n-1-2r
n! r=0
(n -2)/2
n-2-2r
+ sin x Σ
r=0
(-1)r(2r +1)! x
Although these right sides are the lineal primitive functions of Si(x), those zeros are all 0 at the time of even
order and are not 0 at the time of odd order. That is, the lineal higher primitive function of Si(x) can not be
expressed by the higher integral with a fixed lower limit.(See the above figure.) Therfore, the higher integra
of Si(x) with a fixed lower limit 0 is not lineal but collateral. However, the idea which makes 0 a common lower
limit is natural. It is because the Si(x) itself is defined by the integral with a lower limit 0.
Formula 14.3.1
x sin t
When is floor function and Si(x)= dt ,is Sine Integral, the following expressions hold.
0 t
-5-
x 0
1 x
x Si(x) + 0!cos x -
1
Si(x)dx =
0 1! 10!
x x 1
Si(x )dx = x Si(x) + 0!x -2!xcos x + 1!x -3!sin x
4 4 3 2
0 0 4! 3 1
x x
- +
13! 31!
x x 1
(n-1)/2
Si(x)dx n = Si(x)x n + cosx Σ (-1)r(2r)! x
n-1-2r
0 0 n ! r=0
(n-2)/2
+ sin x Σ (-1)r(2r +1)! x n-2-2r
r=0
n-1-2r
(n-1)/2 x
- Σ
r=0
(-1) r
(2r +1)(n -1-2r)!
-6-
14.4 Higher Integral of Logarithmic Integral
Logarithmic Integral is defined as follows.
x 1
li(x)= dt (1.0)
0 log t
6
2 4 6 8 10
-2
-4
-6
Lemma 14.4.1
x et
When Exponential Integral is Ei(x)= dt , the following expressions hold.
- t
Proof
t t
2 x 1
Let 2log x = t . Then x = e , dx = dt = e 2 dt . Hence
2 2
t
Ei(t)e
1 2
Ei(2log x)dx = dt
2
Calculating the integral of the right side by ONLINE INTEGRATOR, we obtain
t t
Ei(t)e 2 t
2 2 3
dt = 2Ei(t)e - 2Ei
Using this,
t
= xEi(2log x) - Ei(3log x)
2 3
Ei(2log x)dx = Ei(t)e - Ei t
2
Next, let 3log x = t . Then we obtain the following expresssion by the same calculation.
t t t
-7-
= xEi(3log x) - Ei(4log x)
Hereafter, by induction, we obtain the desired expresson.
Note
Since log x - at the time x +0 , x =0 is clearly a zero of these functions. Then, (1.n) can be
written as follows.
Lemma 14.4.2
et x
When Exponential Integral is Ei(x)= dt , the following expressions hold.
- t
1 n+1 1
x nEi(log x)dx = x Ei(log x)- Ei(n +2)log x (2.n)
n +1 n +1
Calculation
Calculating by ONLINE INTEGRATOR , we obtain (2.n) immediatly.
Note
Since log x - at the time x +0 , x =0 is crealy a zero of these functions. Then, (2.n) can be
written as follows.
x 1 n+1 1
n
x Ei(log x)dx = x Ei(log x)- Ei(n +2)log x (2.n')
0 n +1 n +1
Formula 14.4.3
When
x 1 x et
li(x)= dt , Ei(x)= dt
0 log t - t
the following expression holds for x 0 .
x x 1 n
li(x)dx n = Σ (-1)r nC r x n- r Ei(r +1)log x (3.n)
0 0 n ! r=0
Proof
Let t = log x . Then [0 , x][- , t] , dx = e tdt . Hence
x 1 t et
dx = dt = [Ei (t )]t- = Ei (log x ) = li(x)
0 log x - t
Next, let
li(x)dx
x x 1 x
dx 2 =
0 0 log x 0
Calculating the integral of the right side by ONLINE INTEGRATOR, we obtain
0
Next, integrating both sides of this with respect to x and applying Lemma 14.4.1 , 14.4.2
to the result, we obtain the following
1 3 1 1 1
= x Ei(log x)- Ei(4log x)- x 2Ei(2log x)+ Ei(4log x)
3! 3! 2 2
1 1
+ xEi(3log x) - Ei(4log x)
2 2
1
x Ei(log x)-3x Ei(2log x)+3xEi(3log x)-Ei(4log x)
3 2
=
3!
Hereafter, by induction, we obtain the desired expresson .
-9-
14.5 Higher Integral of Double Logarithmic Function
Double Logarithmic Function is defined as follows.
f(x) = log|log x| (1.0)
Integrating both sides of this with respect to x repeatedly and arranging the results, we obtain the following
higher indefinite integrals. Where, li(x)=Ei(log x) is Logarithmic Integral mentioned in the previous.
1
x log|logx| -3x li(x)+3xEi(2log x)- Ei(3log x)
3 3 2
log |log x|dx =
3!
1 n
log|log x|dx n =
n-r
x nlog|logx| + Σ(-1)rnC r x Ei(rlog x)
n! r=1
Although these right sides are the lineal primitive functions of log|logx | , since both zeros of x nlog|logx |
and Ei(nlog x ) are 0 , zeros of the right sides are all 0 . Therfore, the lineal higher primitive function of
log|logx | can be expressed by the higher integral with a fixed lower limit 0 .
Formula 14.5.1
x et
When Ei(x)= dt , the following expressions hold for x 0 .
- t
x x 1 n
n-r
log|log x|dx n = x n log|logx| + Σ(-1)rnC r x Ei(rlog x) (1.n)
0 0 n! r=1
- 10 -
14.6 Super Calculus of Logarithmic Integral
Among the higher integrals mentioned in previous sections, Higher Integral of Logarithmic Integral is
extensible even to Super Calculus. It is because this higher integral is expressed with binomial coefficients.
Formula 14.6.1
When
x 1 x et
li(x)= dt , Ei(x)= dt
0 log t - t
p 0 and x 0 .
the following expression holds for
x x 1 p
li(x)dx p = Σ (-1)r p- r
x Ei(r +1)log x
0 0 (1+ p) r=0 r
Proof
r
n
First, replace n!, nC r with (1+n) , respectively in Formula 14.4.3 . Next, analytically
continuing the index of the integration operator to [0 , p ] from[1 , n ], we obtain the desired formula.
Formula 14.6.2
When
- 11 -
x 1 x et
li(x)= dt , Ei(x)= dt
0 log t - t
p >0 , p 1, 2, 3, and x 0 .
the following expression holds for
1 -p r
Σ
(p) r
li(x) = (-1) x -p- Ei(r +1)log x
(1- p) r=0 r
Proof
In fact, Formula 14.6.1 holds for p -1, -2, -3, , x 0 . Therefore, in Formula 14.6.1 , replacing the
integration operator <p > with the differentiation operator (p ) = <-p > , we obtain the desired expression.
- 12 -
14.7 Super Calculus of Double Logarithmic Function
Among the higher integrals mentioned in previous sections, Higher Integral of Double Logarithmic Function is
extensible even to Super Calculus. It is because this higher integral is expressed with binomial coefficients.
Formula 14.7.1
x et
When Ei(x)= dt , the following expressions hold for p >0 , x 0.
- t
x x 1 p
p-r
log|log x|dx p = x plog|logx| + Σ(-1)r x Ei(rlog x)
0 0 (1+ p) r=1 r
Proof
r
n
First, replace n!, nC r with (1+n) , respectively in Formula 14.5.1 . Next, analytically
continuing the index of the integration operator to [0 , p ] from[1 , n ], we obtain the desired formula.
Formula 14.7.2
x et
When Ei(x)= dt , the following expression holds for p >0 , p 1, 2, 3, , x 0 ,
- t
- 13 -
the following expression holds.
1 -p Ei(rlog x)
log|log x|
(p)
= log|logx| + Σ (-1)r
x p(1-p) r=1 r xr
Proof
In fact, Formula 14.7.1 holds for p -1, -2, -3, , x 0 . Therefore, in Formula 14.7.1 , replacing the
integration operator <p> with the differentiation operator (p) = <-p > , we obtain the desired expressions.
2007.10.05
K. Kono
Alien's Mathematics
- 14 -
15 Higher and Super Calculus of Elliptic Integral
Formula 15.1.1
The following expressions hold for |k| 1 , |x| 1 .
x dx
F(x, k) =
1- x 1- k x
0 2 2 2
(-1)r -1/2
r -1/2 2s 2r+1
= ΣΣ k x (1.1)
r=0 s=0 2r +1 r-s s
r 1 (2r -2s -1)!! (2s -1)!! 2s 2r+1
= ΣΣ k x (1.1')
r=0 s=0 2r +1 (2r -2s)!! (2s)!!
Proof
Since |k| 1 , |x| 1 , from Generalized Binomial Theorem (See 3.2 ) .
1
x
-
2 -1/2
1- x = Σ(-1)r
2 2r
r=0 r
1
-
2 -1/2
1- k x = Σ(-1)r
2 2
k 2rx 2r
r=0 r
Multiplying each other, it is as follows.
1
x
- -1/2 -1/2 -1/2 -1/2
2 2
1- x = x0 - x2 + x4 - 6
+-
0 1 2 3
1
k x +-
- -1/2 -1/2 -1/2 -1/2
2
1- k x
2 2
= k 0x 0- k 2x 2+ k 4x 4- 6 6
0 1 2 3
0 0 k x
-1/2 -1/2 0 0
=
0 1 1 0 k x
-1/2 -1/2 -1/2 -1/2 0 2 2
- k +
0 2 1 1 2 0 k x
-1/2 -1/2 -1/2 -1/2 -1/2 -1/2 0 2 4 4
+ k + k +
r-s
r -1/2 -1/2
= ΣΣ(-1)r k 2sx 2r
r=0 s=0 s
That is
r-s
1 r -1/2 -1/2
= ΣΣ(-1)r k 2sx 2r
1- x 1- k x
2 2 2 r=0 s=0 s
Then, integrating both sides of this with respect to x from 0 to x, we obtain (1.1).
Next, from the definition of General Binomial Coefficient (See 3.2 ) ,
-1-
(-1/2+1) (1/2)
-1/2
= =
s (-1/2-s +1)(s +1) (1/2-s)(s +1)
On the other hand, from the properties of the Gamma Function (See 1.1.6 ) ,
2n
=
1 1
n+ (2n -1)!! , -n = (-1) n
2 2n 2 (2n -1)!!
Then, for non-negative integer s,
2s
=
1 1
, -s = (-1) s
2 2 (2s -1)!!
Using these,
(2s -1)!!
-1/2 s (2s -1)!!
= (-1)s = (-1)
s 2ss! 2ss!
i.e.
r -s
-1/2 (2s -1)!! -1/2 (2r -2s -1)!!
= (-1)s , = (-1)r-s
s (2s)!! (2r -2s)!!
Hence, substituting these for (1.1) , we obtain (1.1') .
1
Example 1 : Double series expansion of F x ,
2
When arbitrary point x= 0.7 is given to this elliptic integral and its double series, both are compared and the
right side is illustrated, it is as follows.
-2-
15.1.2 Double series expansion of Elliptic Integral of the 2nd kind
Formula 15.1.2
The following expressions hold for |k| 1 , |x| 1 .
x 1- k 2x 2
E(x, k) = dx
0 1- x 2
(-1)r -1/2
r 1/2 2s 2r+1
= ΣΣ k x (1.2)
r=0 s=0 2r +1 r-s s
r 1 (2r -2s -1)!! (2s -1)!! 2s 2r+1
= ΣΣ k x (1.2')
r=0 s=0 (2r +1)(1-2s) (2r -2s)!! (2s)!!
Proof
Since |k| 1 , |x| 1 , from Generalized Binomial Theorem .
1
r x
-
2 -1/2
1- x = Σ(-1)r
2 2r
r=0
1
r k x
2 1/2
1- k x = Σ(-1) r
2 2 2r 2r
r=0
Multiplying each other and integrating both sides of the result with respect to x from 0 to x, we obtain (1.2).
This is only what reversed the sign of 1/2 in the 2nd binomial coefficient in (1.1).
Next, from the definition of General Binomial Coefficient ,
(3/2) (3/2)
s
1/2
= =
(3/2-s)(s +1) (1/2-s)(1/2-s)(s +1)
On the other hand, from the properties of the Gamma Function (See 1.1.6 ) , the following expressions hold
for non-negative integer s,
2s
2 = 2 -s = (-1)
3 1
, s
2 (2s -1)!!
Using these ,
(-1)s (2s -1)!!
s
1/2 (2s -1)!!
s
= (-1) =
(1-2s)s! 2s (1-2s) (2s)!!
r -s
-1/2 (2r -2s -1)!!
Substituting this and previous =(-1)r-s for (1.2) , we obtain (1.2') .
(2r -2s)!!
2
Example 2 : Double series expansion of E x ,
3
When arbitrary point x= 0.8 is given to this elliptic integral and its double series and both are compared,
it is as follows.
-3-
15.1.3 Triple series expansion of Elliptic Integral of the 3rd kind
Formula 15.1.3
The following expressions hold for |c| 1 , |k| 1 , |x| 1 .
x dx
(x,c,k) =
1+ cx 1- x 1- k x
0 2 2 2 2
Proof
Since |c| 1 , |k| 1 , |x| 1 , from Generalized Binomial Theorem .
1
1+ cx 2
= Σr=0
(-1)rc rx 2r
1
x
-
2 -1/2
1- x = Σ(-1)r
2 2r
r=0 r
1
k
-
2 -1/2
1- k x = Σ(-1)r
2 2 2r 2r
x
r=0 r
Multiplying each other and integrating both sides of the result with respect to x from 0 to x, we obtain (1.3) .
And replacing the binomial coefficients with the double factorial in (1.3) , we obtain (1.3') .
1 1
Example 3 : Triple series expansion of x, ,
3 2
When arbitrary point x= 0.9 is given to this elliptic integral and its triple series and both are compared,
it is as follows.
-4-
15.2 Arc length of an ellipse
The ellipse which has foci on x-axis is shown by the following formula.
2 2
x y
2
+ 2 =1 (0< b < a) (1)
a b
Comparing this with cos + sin = 1 ,
2 2
x2 y2
2
= cos 2 , 2
= sin 2
a b
i.e.
x = a cos , y = b sin , 0 2
This is illustrated as shown in Fig.A. By this expression, since the arc length is calculated by the counter-
rclockwise rotation from x-axis, the calculation is complicated.
Therefore we think about only the 1st quadrant, and let us replace with /2- .
i.e.
x = a sin , y = b cos , 0 /2 (1')
Then, we can calculate the arc length clockwise from y-axis as shown in Fig.B. Because the calculation is
easy, we adopt expression (1').
Fig.A Fig.B
When the parametric equation of a curve on a plane is x = f() , y = g (), the length l is given by the
following expression.
2 2
+
dx dy
l=
d d
From (1') ,
dx dy
= a cos , = -b sin
d d
Then
2 2
dx dy
+ = a 2cos 2 + b 2sin 2 = a 2 - a 2-b 2sin 2
d d
= a 21- k 2sin 2 (k a -b /a
2 2 2
: Eccentricity )
-5-
Therefore, the length l of bQ is as follows.
l =a 0
1- k 2sin 2 d (2.0)
1- k 2t 2
a x a 2-b 2
l =a dt = aE ,k k= (2.1)
0 1- t 2 a a2
Since the right side is a Elliptic Integral of the 2nd kind, using (1.2) in previous section , we obtain
(-1)r 2r+1
r-s s
r -1/2 1/2 x a 2-b 2
l = a ΣΣ k 2s
k= (2.1')
r=0 s=0 2r +1 a a2
From a =3 , b =2 , xq =2 ,
32-22 5 xq 2
k= = , =
32 3 a 3
Then the length l of bQ is as follows from (2.1') .
2s
r-s s
r (-1)r -1/2 1/2 5 2 2r+1
l = 3ΣΣ
r=0 s=0 2r +1 3 3
When this is calculated, it is as follows.
-6-
15.3 Arc length of a lemniscate
2 -2x 2+ a 2 a 8x 2+ a 2
y =
2
Since a>0 , 8x 2+ a 2 > 0 , Employing +, we obtain
-2x 2+ a 2 a 8x 2+ a 2
y= (3.0h)
2
These zeros are obtained from x 2 - a 2x 2 =0 which is drawn from (3.0h) = 0. They are as follows.
x0 = a , =0
That is, the x-coordinate of A points is a in Fig.1. The expression of + of (3.h) is an upper half (blue) and
the expression of - is a lower half (red) in Fig.1.
-7-
Next, substituting the following expressions for (3.xy),
t t2
dt = 2 cos d , sin = , cos = 1-
2 2
-8-
From these
d 1 dt 1 dt
= =
1-2 sin 2 2 cos
t2
2 1 2
1- t 1-
2
1-2 t
2 2
Then, the length l of AP in Fig.2 is obtained as follows.
a u dt a 1
l= = F u, (3.1)
2 0 2 2 2
1
1- t 1-
2
t2
2
3 a 2+8x 2
u= 2 sin = - (3.1u)
2 2a
Where, (3.1u) is obtained from t = 2 sin and (3.0') .
Since the right side of (3.1) is the elliptic Integral of the 1st kind, using (1.1) in previous section, we obtain
a r (-1)r 2s
r-s
-1/2 -1/2 1
l= ΣΣ
2 r=0 s=0 2r +1 s 2
u 2r+1 (3.1')
Note
When K(k) =F(1, k) is a complete elliptic integral of the 1st kind, the length L/4 of AO in Fig.2
is given by the following expression.
L a 1 a
= K = 1.854074677301372 (3.1q)
4 2 2 2
Furthermore, the length l of OP in Fig.2 is drawn from (3.1) and (3.1q) as follows.
a 1 1
l = K - F t, (0 t 1) (3.1)
2 2 2
-9-
1 cos 2 + sin 2 1+ tan 2 1+ t 2
= = =
1-2 sin 2 cos 2 - sin 2 1- tan 2 1- t 2
Then, the length l of AP in Fig.2 is
u 1+ t 2 dt u dt
l =a =a .
0 1- t 2 1+ t 2 0 1- t 2
1+ t 2
That is,
u dt
l =a = aF(t, i) (3.2)
0 2 2 2
1- t 1- i t
Although the right side of (3.2) is slightly irregular ( k = -1 ), since it is the elliptic Integral of the 1st kind,
using (1.1) in previous section, we obtain the following expression.
r (-1)r
r-s i
-1/2 -1/2 2s 2r+1
l = a ΣΣ u (3.2')
r=0 s=0 2r +1 s
x
1 -1/2
= Σ(-1)r 4r
1- x 4 r=0 r
From this,
(-1)r
u
u dt -1/2
l =a = aΣ 4r+1
(3.3)
0 1- t 4 r=0 4r +1 r
or
u dt (2r -1)!! u 4r+1
l =a = aΣ (3.3')
0 1- t 4 r=0 (2r)!! 4r +1
The upper limit of the integration is the same as (3.2u), as follows.
(-1)!! 0.3891 0!! 0.3895 3!! 0.3899 5!! 0.38913 7!! 0.38917
s= + + + +
0!! 1 1!! 5 4!! 9 6!! 13 8!! 17
= 0.390
Note
After all, there are the following relations between the three integrals.
d
1 t dt u du
= =
1-2sin 2
0 2 0 1 2 0 1-u 4
1- t 1- t
2
2
2u
t= 2 sin = 2 sintan -1u =
1+ u 2
=
t t
u = tan = tan sin -1
2 2- t 2
- 11 -
15.4 Termwise Higher Calculus of Elliptic Integral
Formula 15.4.1
When |x| 1 , |k| 1 , |c| 1
x dx x 1- k 2x 2
F(x, k) = , E(x, k) = dx
0 1- x 1- k x
2 2 2 0 1- x 2
x dx
(x,c,k) =
1+ cx 1- x 1- k x
0 2 2 2 2
r (-1)r(2r)!
(2r + n +1)! r - s s
x x -1/2 -1/2
F(x, k)dx =ΣΣ
n
k x 2s 2r+n+1
(4.1)
0 0 r=0 s=0
r
(2r + n +1)! r - s s
x x (-1) (2r)! -1/2 1/2 r
E( x, k)dx =Σ Σ n
k x 2s 2r+n+1
(4.2)
0 0 r=0 s=0
Proof
From Formula 15.1.1 ,
(-1)r
r-s s k x
r -1/2 -1/2
F(x, k) = ΣΣ 2s 2r+1
r=0 s=0 2r +1
r
(2r +1)! r - s s
(-1) (2r)! -1/2
r -1/2
= ΣΣ k 2s 2r+1
x
r=0 s=0
Integrating both sides of this with respect to x from 0 to x n times, we obtain (4.1) . (4.2) and (4.3) are also
obtained from the Formula 15.1.2 and 15.1.3 in a similar way.
Note
The following expression using the double factorial is also possible. However, it is complicated and does not
have a merit.
x x r (2r)! (2r -2s -1) !! (2s -1) !! 2s 2r+n+1
F(x , k) dx n =ΣΣ k x
0 0 r=0 s=0 (2r +n +1)! (2r -2s) !! (2s) !!
x ,
x x 2
Example : dx 3
0 0 3
In the left side, the integrand of elliptic integral of the 2nd kind is integrated 3+1 times by Riemann-Liouville
integral. In the right side, this higher integral is calculated by the double series (4.2). Moreover, one arbitrary
point x= 0.8 is given to the both sides, and the values of the both sides are compared. Both are corresponding.
- 12 -
15.4.2 Termwise Higher Derivative of Elliptic Integral
Formula 15.4.2
When |x| 1 , |k| 1 , |c| 1 , x = x
x dx x 1- k 2x 2
F(x, k) = , E(x, k) = dx
0 1- x 1- k x
2 2 2 0 1- x 2
x dx
(x,c,k) =
1+ cx 1- x 1- k x
0 2 2 2 2
r-s k
d (-1) (2r)!
r -1/2 -1/2 2s 2r- n+1
F(x, k) = Σ Σ x (4.4)
dx n r=
n -1 s=0 (2r - n +1)!
s
2
dn r (-1)r(2r)!
r-s s k
-1/2 1/2
Σ Σ
2s 2r-n+1
E(x, k) = x (4.5)
dx n r=
n -1 s=0 (2r - n +1)!
2
dn s (-1)r(2r)!
-1/2
k
r -1/2
r-s
(x,c,k) = Σ Σ Σ c 2t
x 2r-n+1
dx n r=
n -1 s=0 t=0 (2r - n +1)!
s-t t
2
(4.6)
Proof
From Formula 15.1.1 ,
(-1)r(2r)!
r-s k
r -1/2 -1/2
F(x, k) = ΣΣ 2s 2r+1
x
r=0 s=0 (2r +1)! s
Differentiating both sides of this with respect to x n times, we obtain (4.4). The 1st term r0 of outside Σ is
decided as the power 2r -n +1 of x is nonnegative. And there is a following relation between the 1st term
r0 and the order n of derivative.
n 0 1 2 3 4 5
r0 0 0 1 1 2 2
- 13 -
d3
1 1
Example : x , ,
dx 3 3 2
In the left side, the integrand of the elliptic integral of the 3rd kind is differentiated 3- 1 times directly. In the
right side, this higher derivative is calculated by the double series (4.6). Moreover, one arbitrary point x= -0.9
is given to the both sides, and the values of the both sides are compared. Both are corresponding.
- 14 -
15.5 Termwise Super Calculus of Elliptic Integral
Formula 15.5.1
When |x| 1 , |k| 1 , |c| 1
x dx x 1- k 2x 2
F(x, k) = , E(x, k) = dx
0 1- x 1- k x
2 2 2 0 1- x 2
x dx
(x,c,k) =
1+ cx 1- x 1- k x
0 2 2 2 2
p0.
the following expressions hold for
(-1)r(2r)!
r-s s k x
x x r -1/2 -1/2
F(x, k)dx =ΣΣ
p 2s 2r+p+1
(5.1)
0 0 r=0 s=0 (2r + p +2)
r
ΣΣ (2r +p +2) r - s s k x
x x (-1) (2r)! -1/2 1/2 r
E( x, k)dx = p 2s 2r+p+1
(5.2)
0 0 r=0 s=0
r
s-t t k
x x (-1) (2r)! -1/2 -1/2 r s
(x,c,k )dx =Σ Σ Σ c p r-s 2t
x 2r+p+1
0 0 (2r + p +2) r=0 s=0 t=0
(5.3)
Proof
Analytically continuing the index of the integration operator in Formula 15.4.1 to [0 , p ] from[1 , n ],
we obtain the desired expressions.
x
x 1
Example : F x, dx 2
0 0 2
In the left side, the integrand of elliptic integral of the 1st kind is integrated 3/2+1 times by Riemann-Liouville
integral. In the right side, this higher integral is calculated by the double series (5.1). Moreover, one arbitrary
point x= 0.7 is given to the both sides, and the values of the both sides are compared. Both are corresponding,
Formula 15.5.2
When |x| 1 , |k| 1 , |c| 1 , x = x
- 15 -
x dx x 1- k 2x 2
F(x, k) = , E(x, k) = dx
0 1- x 1- k x
2 2 2 0 1- x 2
x dx
(x,c,k) =
1+ cx 1- x 1- k x
0 2 2 2 2
p0.
the following expressions hold for
dp (-1)r(2r)!
r-s
r -1/2 -1/2
F(x, k) = Σ Σ k 2s x 2r-p+1 (5.4)
dx p p -1 s=0 (2r - p +2) s
r=
2
p
(-1)r(2r)!
r-s s
d r -1/2 1/2
E(x, k) = Σ Σ k 2s x 2r-p+1 (5.5)
dx p p -1 s=0 (2r - p +2)
r=
2
p s (-1)r(2r)!
-1/2
d r -1/2
r-s
(x,c,k) = Σ ΣΣ c k 2t x 2r-p+1
dx p p -1 s=0 t=0 (2r - n +2) s-t t
r=
2
(5.6)
Proof
Analytically continuing the index of the differentiation operator in Formula 15.4.2 to [0 , p ] from [1 , n],
we obtain the desired expressions.
d 3/2 2
Example : E x,
dx 3/2 3
In the left side, the 2nd order derivative of elliptic integral of the 2nd kind is integrated 1/2 times by Riemann-
Liouville integral. In the right side, this super derivative is calculated by the double series (5.5). Moreover, one
arbitrary point x= 0.6 is given to the both sides, and the values of the both sides are compared. Both sides are
corresponding,
2007.11.07
K. Kono
Alien's Mathematics
- 16 -
16 Higher Integral of the Product of Two Functions
Formula 16.1.1
<r> <r>
Let f be the arbitrary r th order primitive function of f(x) and g be the r th order derivative function
x x n -1 r x x
<0> (0) n
f
<n> (0)
g dx = f g - ΣΣrC s f a<n -r+s> g(as) dx r
an a1 r=0 s=0 n-r n-r an a n- r+ 1
f
n x x
- ΣnC r <r> (r)
g dx n (1.1)
r=1 an a1
f
x x
<n> (0) n -1 r
<n -r+s> (x -a)r
(s)
<0> (0)
g dx = f n
g - ΣΣrC s f a ga
a a r=0 s=0 r!
f
n x x
- ΣnC r <r> (r)
g dx n (1.2)
r=1 a a
f
x x n x x
f <0>g(0)dx n = f g(0) - ΣnC r
<n> <r> (r)
g dx n (1.3)
a a r=1 a a
Proof
The following equations hold according to the formula of integration by parts.
x x
f <0>g(0)dx = f <1>g(0) - f <a11 >g(a01) - f <1>g(1)dx (001)
a1 a1
f - f
x x
<1> (0)
g dx = f <2>g(0) - f <a22>g(a02) <2> (1)
g dx (102)
a2 a2
f - f
x x
<1> (1)
g dx = f <2>g(1) - f <a21 >g(a11) <2> (2)
g dx (111)
a1 a1
f - f
x x
<2> (0)
g dx = f <3>g(0) - f <a33>g(a03) <3> (1)
g dx (203)
a3 a3
f
x x
f <2>g(1)dx = f <3>g(1) - f <a32>g(a12) - <3> (2)
g dx (212)
a2 a2
f f
x x
<2> (2)
g dx = f <3>g(2) - f <a31 >g(a21) - <3> (3)
g dx (221)
a1 a1
Integrating both sides of (001) with respect to x from a2 to x ,
-1-
f
x x x x x x
f <0>g(0)dx 2 = f <1>g(0)dx - f <a11 >g(a01) dx - <1> (1)
g dx 2
a2 a1 a2 a2 a2 a1
Substituting (102) for this,
f
x x x
<0> (0) 2
f g dx = f <2>g(0) - f <a22>g(a02) - <2> (1)
g dx
a2 a1 a2
dx - f
x x x
<1> (0) <1> (1)
-f a1 ga1 g dx 2
a2 a2 a1
f f dx - f
x x x x x x
<1> (1) 2 <2> (1) <2> (1) <2> (2)
g dx = g dx - f a1 ga1 g dx 2
a2 a1 a2 a2 a2 a1
From this,
f
x x x x x x
f <2>g(1)dx = f <a21 >g(a11) dx + f <1>g(1)dx 2 + <2> (2)
g dx 2
a2 a2 a2 a1 a2 a1
Substituting this for the above,
dx
x x x
<0> (0) 2
f g dx = f <2>g(0) - f <a22>g(a02) - f <a11 >g(a01 ) + f <a21>g(a11)
a2 a1 a2
f f
x x x x
<1> (1)
-2 g dx 2 - <2> (2)
g dx 2 (2)
a2 a1 a2 a1
f f
x x x x
<0> (0) 3 <2> (0)
g dx = g dx
a3 a2 a1 a3
dx
x x x
- f <a22>g(a02) dx - f <a11>g(a01 ) + f <a21>g(a11 )
a3 a3 a2
- 2 f f
x x x x x x
<1> (1)
g dx 3 - <2> (2)
g dx 3
a3 a2 a1 a3 a2 a1
Substituting (203) for this,
f
x x x x
<0> (0)
f g dx 3 = f <3>g(0) - f <a33>g(a03) - <3> (1)
g dx
a3 a2 a1 a3
dx
x x x
- f <a22>g(a02 ) dx - f <a11>g(a01 ) + f <a21>g(a11 )
a3 a3 a2
- 2 f f
x x x x x x
<1> (1)
g dx 3 - <2> (2)
g dx 3
a3 a2 a1 a3 a2 a1
f
x x x x x x
f <3>g(1)dx = f <a32>g(a12) dx + f <2>g(1)dx 2 + <3> (2)
g dx 2
a3 a3 a3 a2 a3 a2
Substituting this for the above,
-2-
dx
x x x x
f <0>g(0)dx 3 = f <3>g(0) - f <a33>g(a03) - f <a22>g(a02) + f <a32>g(a12)
a3 a2 a1 a3
dx
x x
- f <a11 >g(a01 ) + f <a21 >g(a11)
a3 a2
f g - f g
x x x x
<2> (1) 2 <3> (2)
- dx dx 2
a3 a2 a3 a2
- 2 f f
x x x x x x
<1> (1)
g dx 3 - <2> (2)
g dx 3
a3 a2 a1 a3 a2 a1
Here, integrating both sides of (111) , (221) with respect to x from a 2 to x respectivly,
f
x x x x x x
f <1>g(1)dx 2 = f <2>g(1)dx - f <a21 >g(a11) dx - <2> (2)
g dx 2
a2 a1 a2 a2 a2 a1
f dx - f
x x x x x x
<2> (2)
g dx 2 = f <3>g(2)dx - f <a31 >g(a21) <3> (3)
g dx 2
a2 a1 a2 a2 a2 a1
f
x x x x x x x x x x
f <1>g(1)dx 3 = f <2>g(1)dx 2- f <a21 >g(a11) dx 2- <2> (2)
g dx 3
a3 a2 a1 a3 a2 a3 a2 a3 a2 a1
f = f dx - f
x x x x x x x x x x
<2> (2)
g dx 3 <3> (2)
g dx 2- f <a31 >g(a21) 2 <3> (3)
g dx 3
a3 a2 a1 a3 a2 a3 a2 a3 a2 a1
i.e.
f
x x x x x x x x x x
f <2>g(1)dx 2 = f <1>g(1)dx 3+ f <a21 >g(a11) dx 2+ <2> (2)
g dx 3
a3 a2 a3 a2 a1 a3 a2 a3 a2 a1
f = f dx + f
x x x x x x x x x x
<3> (2)
g dx 2 <2> (2)
g dx 3+ f <a31 >g(a21) 2 <3> (3)
g dx 3
a3 a2 a3 a2 a1 a3 a2 a3 a2 a1
Substituting this for the above,
f
x x x
<0> (0) 3
g dx = f <3>g(0) - f <a33>g(a03)
a3 a2 a1
dx
x
- f <a22>g(a02) + f <a32>g(a12)
a3
dx
x x
- f <a11>g(a01 ) + 2f <a21 >ga(11) + f <a31 >g(a21) 2
a3 a2
= f <3>g(0) - f <a33>g(a03)
1 x x 2 x x
-Σ1C s f a<22+ s>g(as2) dx 1 -Σ2C s f a<11+ >ga1
s (s)
dx 2
s=0 a3 a 3-1 + 1 s=0 a3 a 3-2 + 1
f
3 x x
- Σ3C r <r> (r)
g dx 3
r=1 a3 a1
-3-
2 r x x
= f <3>g(0) - f <a33>g(a03) - ΣΣrC s f a<3-r+s>g(as) dx r
r=1 s=0 3 -r 3 -r a a 3- r+ 1
3
f
3 x x
- Σ3C r <r> (r)
g dx 3
r=1 a3 a1
Thus,
f
x x x 2 r x x
<0> (0)
g dx = f <3>g(0) - f <a33>g(a03) - ΣΣrC s f a<33-r+s
3
-r
ga3-r
> (s)
dx r
a3 a2 a1 r=1 s=0 a3 a 3- r+ 1
f
3 x x
- Σ3C r <r> (r)
g dx 3 (3)
r=1 a3 a1
Hereafter by induction, we obtain the following expression.
f
x x x x
<n> (0) <n> (0) n -1 r <n -r+s> (s)
<0> (0) n
g dx = f g - f a g a -ΣΣrC s f a ga dx r
an a1 n n r=1 s=0 n-r n-r an a n- r+ 1
f
n x x
- ΣnC r <r> (r)
g dx n
r=1 an a1
<n>
And pushing f a g(a0) into ∑∑, we obtain (1.1) .
n n
x x x x (x -a)r
dx =r r
dx = for r =0, 1, , n -1
an a n- r+ 1 a a r!
Then, substituting this for (1.1) , we obtain (1.2) .
Last, when f (a ) = 0 (r =1, 2, ,n ) or (a ) = 0 (s =0, 1, ,n -1) ,
<r> (s)
g (1.3) is clear.
Note
Different from the integration by parts of the 1st order, These formulas are seldom used directly.
Theorem 16.1.2
<r> <r>
Let f be the arbitrary r th order primitive function of f(x) and g be the r th order derivative function
a a r
x x m -1 -n
f <0>g(0)dx n = Σ f <n + r> (r)
g
r=0
n 1
s f a a
-n + r
n -1 m -1 x x
- ΣΣ <n -r+ s>
a n-r ga
(s)
dx r
r=0 s=0 n-r
n n- r+1
a a
n -1 r -1 r -1 x x
t s m +n-1 -r + tC m-1 f a
m+ n -r+s (m+ s)
+ (-1) m
Σ ΣΣ C
r =1 s =0 t=s n- r
ga
n- r
dx r
n n- r+1
a a f
(-1)m n -1 n -1C k x x
<m+ k > (m+ k )
(n , m) Σ
+ g dx n (2.1)
k=0 m + k 1
n
-4-
ar = a for r =1, 2, ,n ,
Especially, when
x x m -1 -n
f <0>g(0)dx n = Σ n r (r)
f < + >g
a a r=0 r
r
n -1 m -1 -n + r
<n -r+ s> (s) (x -a )
- ΣΣ fa ga
r=0 s=0 s r!
n -1 r -1 r -1 (x -a)r
t s m +n-1 -r + tC m-1 f a
m + n -r+s (m+ s)
+ (-1) m
Σ ΣΣ C
r =1 s =0 t=s
ga
r!
f
(-1)m n -1 n -1C k x x
<m+ k > (m+ k )
(n , m) Σ
+ g dx n (2.2)
k=0 m + k a a
Especially, when f (a) = 0 (r =1, 2, ,m+n -1) or g (s)(a ) = 0 (s =0, 1, ,m+n -2),
<r >
r f
x x m -1 -n
f <0>g(0)dx n = Σ <n + r> (r)
g
a a r=0
f
(-1)m n -1 n -1C k x x
<m+ k > (m+ k )
(n , m) Σ
+ g dx n (2.3)
k=0 m + k a a
Proof
k =0, 1, 2,
<k > (k )
According to Formula 16.1.1 , the 3rd order integral of f g are as follows
f
x x x
<0> (0)
g dx 3 = f <3>g(0)
a3 a2 a1
dx
x
- f <a3>g(a0) - f <a2>g(a0) + f <a3>g(a1)
3 3 2 2 2 2 a3
dx
x x
- f <a11>g(a01 ) + 2f <a21 >ga(11) + f <a31 >g(a21) 2
a3 a2
f
x x x
<1> (1)
g dx 3 = f <4>g(1)
a3 a2 a1
dx
x
- f <a4>g(a1) - f <a3>g(a1) + f <a4>g(a2)
3 3 2 2 2 2 a3
g + f g dx
x x
- f <a21>g(a11 ) + 2f <a31 > (2)
a1
<4> (3)
a1 a1
2
a3 a2
f dx -3 f g dx - f
x x x x x x x x x
<2> (2) 3 <3> (3) 3 <4> (4)
-3 g g dx 3
a3 a2 a1 a3 a2 a1 a3 a2 a1
f
x x x
<2> (2)
g dx 3 = f <5>g(2)
a3 a2 a1
dx
x
- f <a5>g(a2) - f <a4>g(a2) + f <a5>g(a3)
3 3 2 2 2 2 a3
-5-
dx
x x
- f <a31>g(a21 ) + 2f <a41 >ga(31) + f <a51 >g(a41) 2
a3 a2
f
x x x
<0> (0) 3 <3> (0) <4> (1)
g dx = f g -3 f g
a3 a2 a1
- f <a3>g(a0) - 3f <a4>g(a1)
3 3 3 3
dx
x
- f a ga - 2f a ga - 3f a ga
<2> (0) <3> (1) <4> (2)
2 2 2 2 2 2 a3
dx
x x
- f <a11>g(a01 ) - f <a21 >g(a11 ) - 5f <a31 >ga(21) - 3f <a41>g(a31 ) 2
a3 a2
dx
x
- f <a2>g(a0) - 2f <a3>g(a1) + 3f <a4>g(a2) + 6f <a5>g(a3)
2 2 2 2 2 2 2 2 a3
dx
x x
- f <a11>g(a01 ) - f <a21 >g(a11 ) + f <a31 >ga(21) + 9f <a41>ga(31) + 6f <a51>g(a41 ) 2
a3 a2
0 1 f g
x x x -3 -3
f <0>g(0)dx 3 = f <3>g(0) + <4> (1)
a3 a2 a1
0 f 1f g
-3 <3> (0)
-3 <4> (1)
- a3 ga3 + a3 a3
0 f 1 f g - 3f dx
-2 <2> (0)
-2 <3> (1) <4> (2)
x
- a2 ga2 + a2 a2 a2 ga2
a3
0 f 1 f g - 5f dx
-1 <1> (0)
-1 <2> (1) <3> (2)
x x
- a1 ga1 + a1 a1 a1 ga1 - 3f <a41 >g(a31 ) 2
a3 a2
f
2 C k 2(2+1)(2+2)
2 x x x
<2+ k > (2+ k )
+(-1)2Σ g dx 3
k =0 2! 2+ k a3 a2 a1
-6-
0 1 2
-3 -3 -3
= f <3>g(0) + f <4>g(1) + f <5>g(2)
0 f g 1 f g 2 f
-3 -3 -3<3> (0) <4> (1) <5> (2)
- + + a3 a3 a3 a3 a3 ga3
0 1 2 f dx
-2 -2 -2<2> (0) <3> (1) <4> (2)
x
- f g + f g + a2 a2 a2 a2 a2 ga2
a3
0 f g 1 f g 2 f dx
-1 -1 -1<1> (0) <2> (1) <3> (2)
x x
2
- + + a1 a1 a1 a1 a1 ga1
a3 a2
dx
x
- 6f <a5>g(a3)
2 2 a3
dx
x x
- 9f <4> (3)
a1 ga1 + 6f <5> (4)
a1 ga1
2
a3 a2
f
2 2 Ck 3(3+1)(3+2) x x x
<3+ k > (3+ k )
Σ
3
+(-1) g dx 3
k =0 2! 3+ k a3 a2 a1
Here, red and magenta coefficients are the elements of the following Pascal type triangles. Bule coefficients are
3m
given, magenta coefficients are equal to c00 , and a red coefficient is obtained as the sum of two coefficients
of the upper row.
32
c00 3 C1
3
32
c10 32
c11 = 2 3 = 2C1 C1
3
32
c20 32
c21 32
c22 1 5 3 1C1 5 C1
3
33
c00 6 C2
4
33
c10 33
c11 = 3 6 = 3C2 C2
4
33
c20 33
c21 33
c22 1 9 6 2C2 9 C2
4
If bule coefficients are given, the other coefficients can be calculated by the following expression. ( See
Lemma 16.8.1 )
r -1
crs3m = ΣC t s-1 cr-1-t 0
3m
r, s 1
t=s -1
When
32
m =2 , c00 = 3C 1 , c10
32
= 2C 1 , c20
32
= 1C 1 i.e.
32
ct0 = 3-tC 1 t =0, 1, 2
0 = 3- r -1-tC 1 = 4 -r +tC 1
32
From this, cr-1-t
r -1
crs32 = ΣC
t s-1 4 -r +tC 1 r, s 1
t=s -1
When
33
m =3 , c00 = 4C 2 , c10
33
= 3C 2 , c20
33
= 2C 2 i.e.
33
ct0 = 4-tC 2 t =0, 1, 2
0 = 4- r -1-tC 2 = 5-r +tC 2
33
From this, cr-1-t
-7-
r -1
crs33 = ΣC t s-1 5-r+ tC 2 r, s 1
t=s -1
r -1
Thus, crs3m = t s-1 m +2-r+ C
ΣC t m-1
t=s -1
If these are used, the above formula is expressed as follows.
r s f g
x x x 1 -3 <3+ r> (r)
1 -3 <3+ s> (s)
f <0>g(0)dx 3 = Σ f g -Σ a3 a3
a3 a2 a1 r=0 s=0
s f dx Σ 0 f dx
-2 1
<2+ s> (s)
x -1 1
<1+ s> (s)
x x
-Σ a2 ga2 - a1 ga1 2
s=0 a3 s=0 a3 a2
dx
x x x
32 <4> (2)
+ c11 f a ga 2 2
dx + c21
32 <3> (2) 32 <4> (3)
f a1 ga1 + c22 f a1 ga1 2
a3 a3 a2
f
2 C k 2(2+1)(2+2)
2 x x x
<2+ k > (2+ k )
+(-1)2Σ g dx 3
k =0 2! 2+ k a3 a2 a1
r s f
2 -3 <3+ r> (r)
2 -3 <3+ s> (s)
=Σ f g -Σ a3 ga3
r=0 s=0
s 0 dx
2 -2 s (s)
x 2 -1 s (s)
x x
-Σ f a<2+ >
ga2 dx - Σ f a<11+ >ga1 2
s=0 2 a3 s=0 a3 a2
dx
x x x
- c11 f a2 ga2
33 <5> (3)
dx - c21
33 <4> (3) 33 <5> (4)
f a1 ga1 + c22 f a1 ga1 2
a3 a3 a2
f
2 C k 3(3+1)(3+2)
2 x x x
<3+ k > (3+ k )
+(-1)3Σ g dx 3
k =0 2! 3+ k a3 a2 a1
That is,
f r f g
x x x 1 -3 <3+ r> (r)
<0> (0)
g dx = Σ
3
a3 a2 a1 r=0
s f a a
-3+ r
2 1 x x
- ΣΣ <3-r+ s> (s)
a 3- r ga dx r
r=0 s=0 3- r
3 3- r+1
a a
2 r-1 x x
Σ Σ dx r
2
+ (-1) c 32
r 1+s fa <5-r+s> (2+s)
ga
r=1 s=0 3- r 3- r
3 3- r+1
f
2 C k 2(2+1)(2+2)
2 x x x
<2+ k > (2+ k )
+(-1)2Σ g dx 3
k =0 2! 2+ k a3 a2 a1
r f
2 -3 <3+ r> (r)
=Σ g
r=0
a a
2 2 -3+ r x x
- ΣΣ f a<3-r+ >ga
s (s)
dx r
r=0 s=0 s 3- r 3- r
3 3- r+1
a a
2 r-1 x x
+ (-1)3ΣΣc 33 <6-r+s> (3+s)
r 1+s f a ga dx r
r=1 s=0 3- r 3- r
3 3- r+1
-8-
f
2 C k 3(3+1)(3+2)
2 x x x
<3+ k > (3+ k )
+(-1)3Σ g dx 3
k =0 2! 3+ k a3 a2 a1
r -1
Here, if cr3m t s m +2- r +tC m-1 is used, the expansion of 0m -1 terms is as follows.
s+1 =Σ C
t=s
r f
x x x m -1 -3 <3+ r> (r)
f <0>g(0)dx 3 = Σ g
a3 a2 a1 r=0
s f a a
-3+ r
2 m -1 x x
- ΣΣ <3-r+ s> (s)
a 3- r ga dx r
r=0 s=0 3- r
3 3- r+1
a a
2 r-1 r-1 x x
t sm +2-r +tC m-1 f a
<m+3-r+s> (m+s)
+ (-1)mΣΣΣ C ga dx r
r=1 s =0 t=s 3- r 3- r
3 3- r+1
f
2 2 C k m(m +1)(m +2) x x x
<m+ k > (m+ k )
+(-1)mΣ g dx 3 (3)
k =0 2! m +k a3 a2 a1
In a similar way, calculating for the 4th integral,
f
x x
<0> (0) 4 <4> (0) <5> (1)
g dx = f g - 4f g
a4 a1
- f <a4>g(a0) - 4f <a4>g(a1)
4 4 4 4
dx
x
- f <a3>g(a0) - 3f <a4>g(a1)
3 3 3 3 a4
- f a ga - 2f a ga dx
x x
<2> (0) <3> (1) 2
2 2 2 2 a4 a3
- f g - f g dx
x x x
<1> (0) <2> (1) 3
a1 a1 a1 a1
a4 a3 a2
+ 4f a ga dx
x
<5> (2)
3 3 a4
+ 7f a ga + 4f a ga dx
x x
<4> (2) <5> (3) 2
2 2 3 3 a4 a3
dx
x x x
+ 9f <a31 >ga(21) + 11f <a41>ga(31) + 4f <a51>g(a41 ) 3
a4 a3 a2
f g dx
x x x x
+ 10 f <2> g(2)dx 4 + 20 <3> (3) 4
a4 a1 a4 a1
+ 15 f + 4 f g dx
x x x x
<4> (4)
g dx 4 <5> (5) 4
a4 a1 a4 a1
dx
x
- f <a3>g(a0) - 3f <a4>g(a1) + 6f <a5>g(a2)
3 3 3 3 3 3 a4
-9-
dx
x x
- f <a2>g(a0) - 2f <a3>g(a1) + 3f <a4>g(a2) 2
2 2 2 2 2 2 a4 a3
dx
x x x
- f <1> (0)
a1 ga1 - f <2> (1)
a1 ga1 + f <3> (2)
a1 ga1
3
a4 a3 a2
dx
x
- 10f <a6>g(a3)
3 3 a4
dx
x x
- 16f <a5>g(a3) + 10f <a6>g(a4) 2
2 2 2 2 a4 a3
dx
x x x
- 19f <a41 >ga(31) + 26f <a51 >ga(41) + 10f <a61>g(a51) 3
a4 a3 a2
f f
x x x x
- 20 <3> (3)
g dx 4 - 45 <4> (4)
g dx 4
a4 a1 a4 a1
- 36 f - 10 f
x x x x
<5> (5)
g dx 4 <6> (6)
g dx 4
a4 a1 a4 a1
Here, red and magenta coefficients are the elements of the following Pascal type triangles. Bule coefficients are
4m
given, magenta coefficients are equal to c00 , and a red coefficient is obtained as the sum of two coefficients
of the upper row.
42
c00 4 C1
4
42 42
c10 c11 = 3 4 = C1
3 C1
4
42
c20
42
c21
42
c22 2 7 4 2 C1 7 C1
4
42
c30 c31
42
c32
42
c33
42
1 9 11 4 C1
1 9 11 C1
4
43
c00 10 C2
5
c10
43
c11
43
= 6 10 = C2
4 5 C2
43
c20
43
c21
43
c22 3 16 10 3 C2 16 C2
5
43 43 43 43
c30 c31 c32 c33 1 19 26 10 C2
2 19 26 C2
5
As well as before, if bule coefficients are given, the other coefficients can be calculated by the following
equation.
r -1
crs4m = Σt s-1 m +3-r+ C
C t m-1
t=s -1
If these are used, the above formula is expressed as follows.
f r f
x x 1 -4 <4+ r> (r)
<0> (0) 4
g dx = Σ g
a4 a1 r=0
s f a a
-4+ r
3 1 x x
- ΣΣ <4-r+ s> (s)
a 4- r ga4- dx r
r=0 s=0 r
4 4- r+1
a a
3 r-1 x x
+ ΣΣc 42 <6-r+s> (2+s)
r 1+ s f a ga dx r
r=1 s=0 4- r 4- r
4 4- r+1
- 10 -
f
3 C k 2(2+1)(2+2)(2+3) x x
3 <2+ k > (2+ k )
+ (-1)2Σ g dx 4
k =0 3! 2+ k a4 a1
r f g
2 -4 <4+ r> (r)
=Σ
r=0
s f a a dx
3 -4+ r
2 x x
- ΣΣ <4-r+ s> (s)
a 4- r ga4- r r
r=0 s=0
4 4- r+1
a a dx
3 r-1 x x
7-r+s (3+s)
- ΣΣc 43
r 1+ s f a ga r
r=1 s=0 4- r 4- r
4 4- r+1
f
3 C k 3(3+1)(3+2)(3+3) x x
3 <3+ k > (3+ k )
+ (-1)3Σ g dx 4
k =0 3! 3+ k a4 a1
r -1
Here, if t s m +3-r+ C
s+1 =Σ C
cr4m t m-1 is used, the expansion of 0m -1 terms is as follows.
t=s
r
x x m -1 -4
f <0>g(0)dx 4 = Σ f <4+ >g
r (r)
a4 a1 r=0
s f a a
-4+ r
3 m -1
<4-r+ s> (s)
x x
- ΣΣ a 4- r ga4- dx r
r=0 s=0 r
4 4- r+1
a a
3 r-1 r -1 x x
t s m +3-r +tC m-1 f a
m +4 -r+s (m+s)
+ (-1)mΣΣΣ C ga dx r
r=1 s=0 t=s 4- r 4- r
4 4- r+1
f
3 C k m(m +1)(m +2)(m +3) x x
3 <m+ k > (m+ k )
+ (-1)mΣ g dx 4
k =0 3! m +k a4 a1
Hereafter, in a similar way, the expansion of 0 m -1 terms of n th order integral is obtained as follows.
a a r
x x m -1 -n
f <0>g(0)dx n = Σ f<
n + r> (r)
g
r=0
n 1
s f a a
-n + r
n -1 m -1 x x
- ΣΣ <n -r+ s>
a n-r ga
(s)
dx r
r=0 s=0 n-r
n n- r+1
a a
n -1 r -1 r -1 x x
t s m + n -1- r + tC m-1 f a
m +n -r+s m +s
+ (-1)m ΣΣΣ C ga dx r
r =1 s =0 t=s n- r n- r
n n- r+1
a a r f
x x m -1 -n
f <0>g(0)dx n = Σ <n + r> (r)
g
r=0
n 1
- 11 -
a a
n -1 m -1 -n + r <n -r+ s> (s)
x x
- ΣΣ fa ga dx r
r=0 s=0 s n-r n-r
n n- r+1
a a
n -1 r -1 r -1 x x
t s m + n -1- r + tC m-1 f a
m +n -r+s m +s
+ (-1) m
Σ ΣΣ C
r =1 s =0 t=s n- r
ga
n- r
dx r
n n- r+1
a a f
(-1) n -1 n -1C k
m x x
<m+ k > (m+ k )
(n , m) Σ
+ g dx 4 (2.1)
k=0 m + k 1
n
ar = a for r =1, 2, ,n ,
Especially, when
x x m -1 -n
f <0>g(0)dx n = Σ n r (r)
f < + >g
a a r=0 r
dx
n -1 m -1 -n + r x x
n s (s)
- ΣΣ f a< -r+ > ga r
r=0 s=0 s a a
dx
n -1 r -1 r -1 x x
t s m + n -1- r + tC m-1 f a
m +n -r+s m +s
+ (-1)m ΣΣΣ C ga r
r =1 s =0 t=s a a
f
(-1) n -1 n -1C k
m x x
<m+ k > (m+ k )
(n , m) Σ
+ g dx 4
k=0 m + k a a
Here, since
x x (x -a)r
dx =r
r = 0, 1, 2, ,n -1
a a r!
substituting this for the above, we obtain (2.2) .
Note
As mentioned in 4.1.3, the polynomial expression for the higher integral of 1 is difficult.
e sin x dx
x x
x 2
Example 1
2 1
a a r f
x x 3-1 -2 <2+ r> (r)
f <0>g(0)dx 2 = Σ g
r=0
2 1
s f a a
-2+ r
2-1 3-1
<2-r+ s> (s)
x x
- ΣΣ a 2-r ga dx r
r=0 s=0 2-r
2 2- r+1
a a
2-1 r -1 r -1 x x
t s 3+2-1- r + tC 3-1 f a
3+2-r+s 3+s
Σ ΣΣ C dx r
3
+ (-1) ga
r =1 s =0 t=s 2- r 2- r
2 2- r+1
a a f
(-1)3 2-1 2-1C k x x
<3+ k > (3+ k )
(2, 3) Σ
+ g dx 2
k=0 3+ k 2 1
- 12 -
r
2 -2 <2+ r> (r)
=Σ f g
r=0
s f s f a dx
-2 2
<2+ s> (s) 2 -1 <1+ s> (s)
x
-Σ a2 ga -Σ a1 ga 1
s=0 2 s=0 1
2
a dx
x
- 3 f <a41> g(a31)
2
f
x x x x
-4 f <3>g(3)dx 2 - 3 <4> (4)
g dx 2
a2 a1 a2 a1
a a f e sin x dx
x x x x
<0> (0) 2 x 2
Left: g dx =
1 2 1
2
r r
r r
2 -2 2 -2
Right: = e xΣ sin x + - e 2Σ sin 2+
r=0 2 r=0 2
s x -2 3 x -2
s 2 1!
-1
2 1!
2
-e Σ 1
sin 1+ - 3 e sin 1+ 1
s=0
- 13 -
+ 2e xsin x -2e 2sin 2-2e 1x sin 1-2e 1x cos 1+4e 1sin 1+4e 1cos 1
3 3 3 3
+ e xcos x - e 2cos 2+ e 1x sin 1- e 1x cos 1-3e 1sin 1+3e 1cos 1
2 2 2 2
1 1 1 1
= - e xcos x + e 2cos 2 - e 1x sin 1 + e 1x cos 1 + e 1sin 1 - e 1cos 1
2 2 2 2
This is consistent with the left. And, since this result contains the constant-of-integration polynomial with
degree 1, we can see that this 2nd order integral is collateral.
x x
Example 2 e x sin x dx 2
2 1
4 4
When n =2, m =3 , from (2.1) ,
a a f r f
x x 2 -2 <2+ r> (r)
<0> (0)
g dx = Σ 2
g
r=0
2 1
s f s a dx
-2 2
<2+ s> (s) 2 -1 s> (s)
x
-Σ a2 ga - Σ f a<1+
1
ga1
s=0 2 s=0
2
a dx
x
- 3 f <a41> g(a31)
2
f f
x x x x
<3> (3) 2 <4> (4)
-4 g dx - 3 g dx 2
a2 a1 a2 a1
a a f
x x
<0> (0) 2
x x
x e xcos x 2
Left: g dx = e sin x dx = -
2 1 2 1 2
4 4
Next,
r
<r> <2+ r> (r)
f =f = ex , g = sin x +
2
2 1
2 s 1 s
f a2<2+ s>
=e 4
, f a1 <1+ s>
=e 4
, g(as2) = sin 4
+
2 , g(as1) = sin 4
+
2
3
a a
x x x x
f <3>g(3)dx 2 = e x sin x + dx 2
2 1 2 1 2
4 4
1 1
2
2 2
4 4
1 x 4 e x 2 e
=- e sin x - e sin + -
2 4 2 4
4
a a f = e sin x +
x
2
x x x x e cos x
<4> (4) 2 x 2
g dx dx = -
2 1 2 1 2
4 4
Substituting these for the right side,
2
s 2 s
s s
-2 -2
2 2
Right: e x
Σ
s=0
sin x +
2
-e 4
Σ
s=0
sin
4
+
2
- 14 -
1
1 s 2
s
-1
2
-e 4
Σ
s=0
sin
4
+
2
x-
4
1
1 3 2
- 3e 4
sin 4
+
2 x-
4 1 1
2
2 2
+
1 x 4 e 4
x 2 e 4
3e xcos x
-4 - e sin x - e sin - +
2 4 2 4 2
= -2e xcos x -2e xsin x
2 1
4 4
+ 2e + 2e x 2
1
- e 4
2
2 1 1
3e xcos x
x
+ 2e sin x - 2e 4
- 2e 4
x 2 +e 4
2 +
2
e xcos x
=-
2
This is consistent with the left. And, since this result does not contains the constant-of-integration polynomial,
we can see that this 2nd order integral is lineal.
e sin x dx
x x
x 2
Example 3
0 0
When n =2, m =3 , from (2.2) ,
f r f
x x 2 -2 <2+ r> (r)
<0> (0)
g dx = Σ 2
g
a a r=0
s f
-2 2
<2+ s> (s) (x -a)0 2 -1 <1+ s> (s) (x -a)1
-Σ a ga -Σ fa ga
s=0 0! s=0 s 1!
1
(x -a)
- 3 f <a4> g(a3)
1!
f f
x x x x
<3> (3) 2 <4> (4)
-4 g dx - 3 g dx 2
a a a a
x
Let f = e , g = sin x , a = 0 , then
f
x x
<0> (0) 2
x x
x e xcos x2 x +1
Left: g dx = e sin x dx = - +
a a 0 0 2 2
Next
r
<r> <2+ r> (r)
f =f = ex , g = sin x +
2
s> s
f a<1+ = f <a4> = e 0 , g(as) = sin
2
3
e xsin x
x x x x x
<3> (3) 2 x 2
f g dx = e sin x + dx = - +
a a 0 0 2 2 2
4
f = e sin x +
e xcos x
2
x x x x x +1
<4> (4) 2 x 2
g dx dx =- +
a a 0 0 2 2
- 15 -
Substituting these for the right side,
r
r
2 -2
Right: = e xΣ sin x +
r=0 2
s x 3 x r
s 2 r!
1 2 -2+ r 1
- e ΣΣ
0
sin - 3 e sin 0
r=0 s=0 2 1!
- 4 - + - 3 -
2
x x
e sin x x e cos x x +1
+
2 2 2
= -2e xcos x - 2e xsin x + 2 + x + 3x
x 3e xcos x 3x +3
+ 2e sin x - 2x + -
2 2
x
e cos x x 1
=- + +
2 2 2
This is consistent with the left. And, since this result contains the constant-of-integration polynomial with
degree 1, we can see that this 2nd order integral is collateral.
e sin x dx
x x
x 2
Example 4
- -
When n =2, m =3 , from (2.3) ,
f g
(-1)3 1 1C k
r
x x 2 -2 <n + r> (r)
x x
<3+ k > (3+ k )
dx =Σ
(2, 3) Σ
<0> (0) 2
f g f g + dx 2
a a r=0 k =0 3+ k a a
s f dx - 3 f
-22
<n + s> (s)
x x x x
=Σ g -4 f<3>g(3) 2 <3> (3)
g dx 2
s=0 a a a a
f
x x
<0> (0) 2
x x
x e xcos x
2
Left: g dx = e sin x dx = -
a a - - 2
Next,
r
<r> <2+ r> (r)
f =f = ex , g = sin x +
2
3
f
e xsin x
x x x x
<3> (3) 2 x 2
g dx = e sin x + dx = -
a a - - 2 2
4
f = e sin x +
e xcos x
2
x x x x
<4> (4) 2 x 2
g dx dx =-
a a - - 2
Substituting these for the right side,
r
e xsin x e xcos x
r
-2
2
Right: =e x
Σ
r=0
sin x +
2
-4 -
2
-3 -
2
x 3e xcos x x
= e (sin x - 2cos x - 3sin x) + 2e sin x +
2
e xcos x
=-
2
- 16 -
This is consistent with the left. And, since this result does not contains the constant-of-integration polynomial,
we can see that this 2nd order integral is lineal.
Remark 1
The following thing can be seen from Example 2 and Example 4 which are lineal together.
e sin x dx
<2> e xcos x x x x x
e sin x
x x x
2 2
=- = e sin x dx =
2 2 1 - -
4 4
That is, the higher integral which gives the higher primitive function of the product of two functions is not
necessarily unique.
Remark 2
As understood from the above proof process and Example 1,2 , (2.1) holds unconditionally and is perfect.
However, it is too complicated and the practical use is difficult. (2.2) is not simplified considering the
conditions.
(2.3) is the most practicable compared with these. The condition looks severe apparently. But, let
Theorem 16.1.3
<r> <r>
Let f be the arbitrary r th order primitive function of f(x) and g be the r th order derivative function
r
1 x m -1 -n
(x -t)n-1 f g dt = Σ f <n + r> (r)
g
(n) a r=0
(x -a)r
s f
-n + r
n -1 m -1
<n -r+ s> (s)
- ΣΣ a ga
r=0 s=0 r!
n -1 r -1 r -1 (x -a)r
t s m + n -1- r + tC m-1 f a
m +s + n -r m +s
+ (-1) m
Σ ΣΣ C
r =1 s =0 t=s
ga
r!
- 17 -
(x -t)
(-1)m n -1 n -1C k x
n-1 <m+ k > (m+ k )
+ Σ
(n , m)(n) k=0 m + k a
f g dt (3.1)
Especially, when f (a) = 0 (r =1, 2, ,m+n -1) or g (s)(a ) = 0 (s =0, 1, ,m+n -2),
<r >
r
1 x m -1 -n
(x -t)n-1 f g dt = Σ f <n + r> (r)
g
(n) a r=0
(x -t)
(-1)m n -1 n -1C k x
n-1 <m+ k > (m+ k )
+ Σ
(n , m)(n) k=0 m + k a
f g dt (3.2)
- 18 -
16.2 Higher Integral of x^a f (x) (general)
There are two features in the higher integral of x f(x) . That is
1 When is a positive integer, the remainder and a part of the constant-of-integration polynomial disappear .
2 When the commonn lower limit of Higher Integral is 0, the constant-of-integration polynomial disappear.
Formula 16.2.0
Let (z) be the gamma function, (n , m) be the beta function, f <r> be the arbitrary r th primitive
<r > <r>
function of f(x) and fa be the function values of f on a k . Then the following expressions hold for
k
a natural number n.
(1)
(1+ ) - r
a a r
x x m -1 -n
f <0>x dx n = Σ f <n + r>
x
n 1
r=0 1+ - r
(1+ ) - s x
s f a
-n + r
n -1 m -1 x
- ΣΣ <n -r+ s>
a dx r
r=0 s=0
a n-r 1+ - s n-r an n- r+1
n -1 r -1 r -1
t s m + n -1- r + tC m-1
+ (-1)m ΣΣΣ C
r =1 s =0 t=s
(1+ )
a
x x
- m- s
f am + n -r+ s an-r dx r
n- r 1+ - m - s an n- r+1
(1+ )
a a f
(-1) n -1 n -1C k
m x x
m- k
<m+ k >
(n , m) Σ
+ x - dx n (0.1)
k=0 m + k
n 1
1+ - m - k
Especially, when = m =0, 1, 2,
(1+m) m- r
a a r
x x m -n
f <0>x m dx n = Σ f <n + r>
x
n 1
r=0 1+m - r
(1+m) m- s x
f a
n -1 m -n + r x
- ΣΣ <n -r+ s>
a dx r (0.2)
r=0 s=0 s a n-r 1+m - s n-r an n- r+1
(1+ ) - r
x x m -1 -n
f <0>x dx n = Σ n r
f<+> x
a a r=0 r 1+ - r
(1+ )
(-1)m n -1 n -1C k x x <m+ k > m- k
+ Σ f x - dx n (0.3)
(n , m) k=0 m k a a
+ 1+ - m - k
Where, if = -1, -2, -3, , it shall read as follows.
(1+ ) (- + r)
(-1)-r r = r, s, m+ s , m+ k
1+ - r (- )
(2) When -1, -2, -3, & + n -1, -2, -3,
(1+ )
x x m -1 -n
x f(0)dx n = Σ x +n+ f
r (r)
an a1 r=0 r (1+ + n + r)
(1+ )
s (1+ + n -r + s) a a a
-n + r
n -1 m -1
+n-r+ s (s)
x x
- ΣΣ n-r fa dx r
r=0 s=0 n-r
n n- r+1
- 19 -
n -1 r -1 r -1
t s m + n -1- r + tC m-1
+ (-1)m ΣΣΣ C
r =1 s =0 t=s
(1+ )
a
x x
+ m + n-r+ s (m+ s)
an-r fa dx r
1+ + m + n - r + s n- r an n- r+1
(1+ )
a a
(-1) n -1 n -1C k
m x x
+m
m+ k
(n , m) Σ
+ x +k f ( )dx n (0.4)
k=0 m + k
n 1
(1+ + m + k)
a = 0 or f (a) = 0 (s =0, 1, ,m+ n -2)
(s)
Especially, when
(1+ )
x x m -1 -n
x f(0)dx n = Σ r (r)
x +n+ f
a a r=0 r (1+ + n + r)
(1+ )
(-1)m n -1 n -1C k x x +m
+ Σ m+ k
x +k f ( )dx n (0.5)
(n , m) k=0 m + k a a (1+ + m + k)
Proof (1)
When -1, -2, -3,
(r) (1+ ) - r (m+ k ) (1+ ) m- k
x = x , x = x -
1+ - r 1+ - m - k
Then, substituting these for (2.1) in Theorem 16.1.2,
(1+ ) - r
a a f
x x
<0>
m -1 -n
x dx = Σ n
f <n + r>
x
n 1
r=0 r 1+ - r
(1+ ) - s x x
n -1 m -1 -n + r
- ΣΣ fa<n -r+ s>
an-r dx r
r=0 s=0 s n-r 1+ - s a n a n- r+1
(1+ )
a
x x
- m- s
f am +s + n -r an-r dx r
n- r 1+ - m - s an n- r+1
n -1 r -1 r -1
t s m + n -1- r + tC m-1
+ (-1)m ΣΣΣ C
r =1 s =0 t=s
(1+ )
a a f
(-1) n -1 n -1C k
m x x
m- k
<m+ k >
(n , m) Σ
+ x - dx n (0.1)
k=0 m + k
n 1
1+ - m - k
When = m -1 m =0, 1, 2,
(1+ - m - s)= s =0, 1, 2, 3,
(1+ - m - k)= k =0, 1, 2, 3,
Hence, ΣΣΣ and the remainder term in (0.1) disappear, and is as follows.
(m) m-1- r
a a
x x m -1 -n
f <0>x
m-1
dx n = Σ f <n + r>
x
n 1
r=0 r m - r
(m) m-1- s x
a
n -1 m -1 -n + r x
- ΣΣ n s
f a< -r+ > a dx r
r=0 s=0 s n-r m - s n-r an n- r+1
Then, replacing m-1 with m, we obtain (0.2) .
- 20 -
When = -1, -2, -3, , from (5.5) in 1.1.5
(-z) (1+z + n)
= (-1)-n ( n is a nonnegative integer )
(-z - n) (1+ z)
Then replaceng -z = 1+ , n = r , we obtain the proviso.
Proof (2)
When -1, -2, -3, & + n -1, -2, -3,
n r (1+ ) n+ r m+ k (1+ ) m+ k
f<+> = x + , f< > = x +
1+ + n + r 1+ + m + k
(1+ ) + n-r+ s
f a -r+ =
n s
an-r
n- r 1+ + n - r + s
(1+ ) + m + n-r+ s
f a -r+ =
m+n s
an-r
n- r 1+ + m + n - r + s
Then, substituting these for (2.1) in Theorem 16.1.2 ,
(1+ )
a a r
x x m -1 -n n+ r (r)
x g(0)dx n = Σ x + g
n 1
r=0 1+ + n + r
(1+ )
s a
-n + r
n -1 m -1
+ n-r+ s (s)
x x
- ΣΣ an-r ga dx r
r=0 s=0 1+ + n - r + s n-r a
n n- r+1
n -1 r -1 r -1
+ (-1)m ΣΣΣ C t s m +n-1 -r + tC m-1
r =1 s =0 t=s
(1+ )
x x
+ m + n-r+ s m +s
an-r ga dx r
1+ + m + n - r + s n- r a n a n- r+1
(1+ )
m n -1
n -1C k
(-1) x x
m+ k m+ k
+ Σ x + g ( )dx n
(n , m) k=0 m + k an a1 1+ + m + k
Here, replacing g with f , we obtain (0.4) .
x x
Example 1 x log x dx 2
2 1
Let n =2, m =3 , f = log x , =1/2 , a1 =1 , a2 =2 . And integrating the left directly
a a
x x x x
Left: f <0>x dx 2 = x log x dx 2
1 2 1
2
5 5
4 log x 64 4 log 2 64 4
= - x2 - - 2 2 + (x -2)
15 225 15 225 9
Next ,
log x - (1+ r)- r
a dx = x -2
x x
<r>
f = (log x)<r> = x , dx 1 =
(1+ r) 2
2
- 21 -
s log 2 - (1+2+ s)- 2+ s s log 1 - (1+1+ s)- 1+ s
f a<2+ > = 2 , f a<1+ > = 1
2 (1+2+ s) 1 (1+1+ s)
Substituting these for the right side of (0.1) ,
r
2 -2 -r
Right: =Σ x x2
r=0 (1+2+ r) 1+1/2- r
log 2 - (1+2+ s)- 2+ s (1+1/2)
1
s
2 -2 -s
-Σ 2 22
s=0 (1+2+ s) 1+1/2- s
-1 log 1 - (1+1+ s) (1+1/2) 2 - s
1
s (1+1+ s) 1
2 - 1+ s
-Σ 1 (x -2)
s=0 1+1/2- s
log 1 - (1+4)- 4 (1+1/2)
1
-3
-3 1 1 2 (x -2)
(1+4) 1+1/2-3
log x - (1+3)- 3 (1+1/2)
1
x x -3
-4 x x 2 dx 2
2 1 (1+3) 1+1/2-3
log x - (1+4)- 3 (1+1/2)
1
x x -4
-3 x x 2 dx 2
2 1 (1+4) 1+1/2-4
5
x
1 3 1 11 1 25 2
= log x - - log x - - log x -
2 2 6 6 32 12
5
2
1 3 1 11 1 25 2
- log 2- - log 2- - log 2-
2 2 6 6 32 12
79 25
+ (x -2) + (x -2)
144 256
5 5
1 29 1 29 5
- log x - x2+ log 2- 22 - (x -2)
15 10 15 10 12
5 5
1 63 1 63 55
+ log x - x2- log 2- 22 + (x -2)
32 20 32 20 256
5 5
4 log x 64 4 log 2 64 4
= - x2 - - 2 2 + (x -2)
15 225 15 225 9
This is consistent with the left side.
x log x dx
x x
3 2
Example 2
2 1
Let n =2, m =3 , f = log x , =1/2 , a1 =1 , a2 =2 . And integrating the left directly
a a x log x dx
x x x x
Left: f <0>x dx 2 = 3 2
1 2 1
2
log x 9 log 2 9 1
= - x5 - - 25 + (x -2)
20 400 20 400 16
Next ,
- 22 -
log x - (1+ r)- r
x x
<r>
f = (log x)<r> = x , dx 1 = dx = x -2
(1+ r) a2 2
x
1 3 1 11 3 25 1 137 5
= log x - - log x - + log x - - log x -
2 2 1 6 4 12 5 60
log 2 9 1
- - 25 + (x -2)
20 400 16
=
400
x -
log x 9 5 log 2 9 1
- - 25 + (x -2)
20 20 400 16
x x
Example 3 x log x dx 2
0 0
Let n =2, m =3 , f = log x , =1/2 . And integrating the left side directly
5
f
x x
x x 4 log x 64
<0> 2 2
Left: x dx = x log x dx = - x2
0 0 0 0 15 225
Next,
<r> log x - (1+ r)- r
f = (log x)<r> = x
(1+ r)
Substituting these for the right side of (0.3) ,
r
2 -2 -r
Right: =Σ x x2
r=0 (1+2+ r) 1+1/2- r
log x - (1+3)- 3 (1+1/2)
1
x x -3
-4 x x 2 dx 2
0 0 (1+3) 1+1/2-3
log x - (1+4)- 3 (1+1/2)
1
x x -4
-3 x x 2 dx 2
0 0 (1+4) 1+1/2-4
5
x
1 3 1 11 1 25 2
= log x - - log x - - log x -
2 2 6 6 32 12
5 5
1 29 1 63
- log x - x2+ log x - x2
15 10 32 20
- 23 -
5
4 log x 64
= - x2
15 225
x x
Example 4 x sin x dx 2
2 1
Let f = sin x , then
r (m + k)
(r) m+ k )
f = sin x + , f( = sin x +
2 2
r (m + s)
(s)
fa = sin a + n-r ,
m+ s
f (a ) = sin an-r+
n-r 2 n- r 2
Substitute these for (0.4) ,
(1+ ) r
a a r
-n
x x m -1
x sin x dx n = Σ
r
x +n+ sin x +
n 1
r=0 (1+ + n + r) 2
(1+ ) r
s (1+ + n -r + s) a a a
-n + r
n -1 m -1 x x
+n-r+ s
-Σ Σ n-r sin an-r+ dx r
r=0 s=0 2
n n- r+1
n -1 r -1 r -1
t s m + n -1- r + tC m-1
+ (-1)m ΣΣΣ C
r =1 s =0 t=s
(1+ ) (m + s)
a a
x x
+ m + n-r+ s
an-r sin an-r+ dx r
1+ + m + n - r + s 2
n n- r+1
(1+) (m + k)
(-1) n -1 n -1C k x
m x
Σ x dx n
+m+k
+ sin x +
(n , m) k=0 m + k an a1 (1+ + m + k) 2
This right side is very complicated. However, fortunately, in the case of this function f = sin x , if m
ΣΣΣ and the remainder term converge to 0. And the following expression holds.
(1+ ) r
a a r
-n
x x
x sin x dx n = Σ
r
x +n+ sin x +
n 1
r=0 (1+ + n + r) 2
(1+ ) r
s (1+ + n -r + s) a a a
-n + r
n -1 x x
+n-r+ s
-Σ Σ n-r sin an-r+ dx r
r=0 s=0 2
n n- r+1
Substituting n =2, =1/2, a 1 =1, a 2 =2 for this,
(1+1/2) r
1
r (1+1/2+2+ r)
-2
x x +2+ r
x sin x dx = Σ 2
x 2
sin x +
2 1 r=0 2
0
1
(1+1/2)
s (1+1/2+2+ s)
-2 +2+ s
-Σ 2 2
sin 2+
s=0 2
1
1
(1+1/2)
s (1+1/2+1+ s)
-1 +2+ s
-Σ 1 2
sin 1+ (x -2)
s=0 2
When the both sides are illustrated by mathematical software, it is as follows. Both overlap exactly and
blue (left) can not be seen.
- 24 -
- 25 -
16.3 Higher Integral of x^a f (x) (particulars)
In this section, substituting various functions f for the Formula 16.2.0 in previous section, we obtain a various
formula. There are (1) and (2) in Formula 16.2.0, and we may also choose whichever. However, what we want
is the expression or approximation of higher integral of x f(x ) by the series. So, in the selection (1) or (2),
we choose the way where such a well-behaved series is obtained.
Moreover, also in which formula, if = -1, -2, -3, , it shall read as follows.
(1+ ) (- + r)
(-1)-r r = r, s, m+ s , m+ k
1+ - r (- )
Formula 16.3.1
The following expressions hold for p > 0 and a natural number n.
x x
p q n
(ax +b) (cx +d) dx
b b
- -
a a
(-1)m n -1 n -1C k m+ k
(1+p)(1+q)
c
(n , m) Σ
Rmn =
k=0 m + k a (1+p +m + k)(1+q -m - k)
x x
(ax +b)p+m+ k (cx +d)q-m- k dx n (1.1r)
b b
- -
a a
lim Rmn = 0
m
x x
(ax +b)p (cx +d)mdx n
b b
- -
a a
3
Example 1st order integral of x-2 3x+4
The zeros of this primitive function are x =-4/3 , 2 . If -4/3 is adopted, since x -2 is a complex
number, this higher integral becomes a complex function. It is inconvenient. Then if we adopt x =2 as
the lower limit of the integral, since a =1, b =-2, p =1/2 , c=3, d =4, q =1/3 , n =1 , substituting
these for (1.1), (1.1r) , we obtain
x
3
x -2 3x +4 dx
2
(3/2)(4/3)
3 1
r
m -1 -1 +r -r
=Σ 3r
(x -2) 2 (3x +4) 3 + Rm1
r=0 (5/2+ r)(4/3-r)
- 26 -
(3/2)(4/3)
1 1
(x -2)
x +m -m
m m
Rm1 = (-1) 3 2
(3x +4) 3
dx
(3/2+ m)(4/3-m) 2
1
This remainder becomes lim R =0 . And the convergence is quick. When m =5 , if both sides are
m m
illustrated, it is as follows. Both overlap exactly and blue (left) can not be seen. In addition, this integral is
not an elementary function.
+2
x ( +1)( +2)log x -2 -3
x log x
<2>
=
( +1) ( +2)
2 2
x +3( +1)( +2)( +3)log x -3 -12 -11
2
x log x
<3>
=
( +1) ( +2) ( +3)
2 2 2
And when > -n , the zeros of these are all x =0 .
Formula 16.3.2
When (z), (z) are the gamma function and the psi function respectively, the following expressions hold
for , n such that + n > 0 .
log x -(1+n+r) - (1+) +n
x x m -1 -n
x log x dx n = Σ x + Rmn (2.1)
0 0 r=0 r ( 1+n+r ) (1+ -r )
log x -(1+m+k) - (1+) x
m+ k
(-1) m
C n -1 n -1 k x x
Rmn = Σ dx n (2.1r)
(n , m) k=0 0 0 (1+m+k) (1+ - m- k)
lim Rmn = 0
m
- 27 -
log x -(1+n+r) - (1+ m) m+ n
r
x x m -n
x mlog x dx n = Σ x (2.1')
0 0 r=0 (1+n+r) (1+ m -r)
Complete Automorphism
Observing well the Formula 16.3.2, we notice that the integral of the completely same type as the left side
is included in the remainder. In such a case, we can take out the integral of the purpose by transposition.
Formula 16.3.2'
The following expression holds for ,n such that +n > 0.
x
x x
log xdx n
0 0
log x - (1+ n)- (1+ ) n -1 n -1C k (2+ k)+
(1+ n)
+n Σ
1+ + n k=0 (1+ k)2 (1+ k, - k) +n
= x
n -1 n -1C k 1
1 + nΣ 2 (1+ k, - k)
k=0 (1+ k)
(2.2)
Calculation
Let m =1 in (2.1), (2.1r), then
log x - (1+n+0)- (1+ ) +n
x x 1-1 -n
x log xdx n = Σ x + R1n
0 0 r=0 0 (1+n+0) (1+ -0)
log x - (1+1+k)- (1+ )x
(-1)1 n -1 n -1C k x x
(n , 1) Σ
R1n = dx n
k =0 1+ k 0 0 (1+1+ k) (1+ -1- k)
(1+ )
Ck
n -1 n -1 x x
= -nΣ x log xdx n
k =0 1+ k (2+ k)( - k) 0 0
n -1 n -1C k (1+ ) (2+ k)+
x x
+ nΣ x dx n
k=0 1+ k (2+ k)( - k) 0 0
Here,
(1+ ) +n
x dx
x x
(2+k) = (1+ k)(1+k) , n
= x
0 0 1+ + n
Substituting these for the remainder,
n -1 Ck 1 x x
x log xdx n
n -1
Rmn = -nΣ 2 (1+ k, - k)
k=0 (1+ k) 0 0
n -1 n -1C k (2+k)+ (1+ ) +n
+ nΣ 2 (1+ k, - k) 1+ + n x
k=0 (1+ k)
Thus
log x - (1+ n)- +n
x x
x log xdx n = x
0 0 (1+n)
n -1 n -1C k 1 x x
- nΣ 2 (1+ k, - k)
x log xdx n
k=0 (1+ k) 0 0
n -1 n -1C k (2+k)+ (1+ ) +n
+ nΣ 2 (1+ k, - k) 1+ + n x
k=0 (1+ k)
From this, we obtain (2.2).
- 28 -
Example When n =1 ,
log x - (2)- (1+ ) 0C 0 (2)+
+
(2) 2+ 12 (1 , ) +1
x
x log xdx = x
0 0C 0 1
1+
12 (1 , )
log x -1 + +1
1+
+1 x 1
= x = log x -
1 + +1 +1
But in this case, direct calculation is far easier !
<3>
x sin x = -9x 2 -60sin x + x 3-36xcosx
3
<4>
x 3 sin x = 12x 2 -120cosx + x 3-60xsin x
And the zeros of these are x =0 , 4.9762 , 0 , 3.1224 , 0 , respectively. Then the lower
3
limit of the higher integral of x sin x is variable. Similarly, the lower limit of the lineal higher integral of
Formula 16.3.3
(1) When m =0, 1, 2, .
(1+ m) m- r (n + r)
a a r
-n
x x m
x m sin x dx n = Σ x sin x -
n 1
r=0 1+ m - r 2
(1+ m) m- s (n -r + s)
a a
-n + r
n -1 m x x
- ΣΣ an-r sin an-r - dx r
r=0 s =0 s ( 1+ m - s ) 2
n n- r+1
(3.1)
Especially, when a 1 , a2 , , a n are the zeros of the lineal heigher primitive of x sin x m
(1+ m) m- r (n + r)
a a r
-n
x x m
x m sin x dx n = Σ x sin x - (3.1')
n 1
r=0 1+ m - r 2
(2) When
-1, -2, -3, & + n -1, -2, -3, .
(1+ ) r
m -1 -n
x x
x sin x dx n = Σ r
x +n+ sin x +
an a1 r=0 r (1+ + n + r) 2
(1+ ) s
a a
-n + r
n -1 m -1 x x
+n-r+ s
-Σ Σ an-r sin an-r+ dx r
r=0 s=0 s (1+ + n -r + s) 2
n n- r+1
+ Rmn (3.2)
- 29 -
n -1 r -1 r -1
t s m + n -1- r + C
Rmn = (-1)m ΣΣΣ C t m-1
r =1 s =0 t=s
(m + s)
a a
(1+)
x x
+ m + n-r+ s
an-r sin an-r+ dx r
1+ + m + n -r + s 2
n n- r+1
(1+) (m + k)
(-1) n -1 n -1C k x
m x
Σ x dx n
+m+k
+ sin x +
(n , m) k=0 m + k a a 1 (1+ + m + k) 2
n
lim Rmn = 0
m
Example 1
(2+ r)
(1+3) 3- r
-2
x x 3
x 3 sin x dx 2 = Σ sin x - x
2 1 r=0 r 1+3- r
2
(2+ s) (1+3) 3- s
-2
3
-Σ sin 2- 2
s=0 s 2 1+3- s
(1+ s) (1+3) 3- s
-1
3
-Σ sin 1- 1 (x -2)
s=0 s 2 1+3- s
= -x 3 -18 xsin x - 6x 2 -24cosx
+ (3x -6)sin 1 - (5x -10)cos 1 - 28 sin 2
Especially, when a 1=0 , a 2 =4.9762 ,
Example 2
See Example 4 in 16.2 .
<1> x 3 3x 2 x 3 3x 2
x sinh x
3
= e -x
+ +3x +3 +e x
- +3x -3
2 2 2 2
And the zeros of these are x =0 , 2.7085 , 0 , 3.1224 i , 0 , respectively. Then the lower
3
limit of the higher integral of x sinh x is variable. Similarly, the lower limit of the lineal higher integral of
x sinh x for any is variable.
- 30 -
Formula 16.3.4
(1) When m =0, 1, 2, .
(1+ m) m- r e x -(-1)n+ re -x
aa x r
x x m -n
m
sinh x dx = Σ n
x
n 1
r=0 1+ m - r 2
a n-r -a n-r
(1+ m) m- s e
a a
-(-1)n-r+ s e
n -1 m -n + r x x
- ΣΣ a dx r
r=0 s =0 s 1+ m - s n-r 2
n n- r+1
(4.1)
Especially, when a1 , a2 , , an are the zeros of the lineal heigher primitive of
m
x sinh x
(1+ m) m- r e x -(-1)n+ re -x
aa x r
x x m -n
m
sinh x dx = Σ n
x (4.1')
n 1
r=0 1+ m - r 2
(2) When
-1, -2, -3, & + n -1, -2, -3, .
(1+ )
x -r -x
x x m -1 -n
+n+ r e -(-1) e
x sinh x dx = Σ
n
x
an a1 r=0 r (1+ + n + r) 2
a n-r -a n-r
(1+)
a a
-(-1)-se
-n + r x x
n -1 m -1
+n-r+ s e
-Σ Σ an-r dx r
r=0 s=0 s (1+ + n -r + s) 2
n n- r+1
+ Rmn (4.2)
n -1 r -1 r -1
t s m + n -1- r + C
Rmn = (-1)m ΣΣΣ C t m-1
r =1 s =0 t=s
a n-r -a n-r
(1+)
a a
+ m + n-r+ s e -(-1)-m-se x x
an-r dx r
( 1+ + m + n -r + s ) 2
n n- r+1
a a
(-1) m
Ck n -1 n -1 x x (1+) +m+k e -(-1)
x -m-k -x
e
+ Σ x dx n
(n , m) k =0 m + k n 1
(1+ + m + k) 2
lim Rmn = 0
m
x x
Example x sinh x dx 2
2 1
From (4.2) ,
(1+1/2)
1
e x -(-1)-re -x
r (1+1/2+2+ r) x
x x -2 +2+ r
x sinh x dx = Σ 2 2
2 1 r=0 2
1 2
(1+1/2) e -(-1)-se -2
s (1+1/2+2+ s) 2
-2 +2+ s
-Σ 2
s=0 2
1 1
(1+1/2) e -(-1)-se -1
s (1+1/2+1+ s) 1
-1 +2+ s
-Σ 2
(x -2)
s=0 2
When the both sides are illustrated by mathematical software, it is as follows. Both overlap exactly and
blue (left) can not be seen.
- 31 -
Left:
a:=1/2:
Fl := int(int(t^a*sinh(t),t=1..u),u=2..x)
x u
t sinh(t) d t d u
2 1
Right:
fx := r-> (E^x-(-1)^-r*E^-x)/2:
gx := (r,s)-> gamma(1+a)/gamma(1+a+2-r+s)*x^(a+2-r+s):
m:=15:
f1 := sum(binomial(-2,s)*gx(0,s)*fx(s), s=0..m):
f2 := -sum(binomial(-2,s)*subs(gx(0,s)*fx(s),x=2), s=0..m):
f3 := -sum(binomial(-1,s)*subs(gx(1,s)*fx(s),x=1), s=0..m)*(x-2):
Fr := f1+f2+f3:
- 32 -
16.4 Higher Integral of log x f (x)
Formula 16.4.1
n 1
When Hn = Σ = (1+n)+ is a harmonic number, the following expression holds.
j=1 j
n
x x x
2 n
log xdx = log xlog x -Hn
0 0 n!
(r)
m -1 -n
+ Σ (-1)r-1 n
x log x -Hn+r+ Rmn (1.1)
r=1 r (1+ n +r)
n
1 x n -1 n -1C k
Rmn Σ (-1)k-1
(m + k)
= log x - Hn - Hm+ k (1.1r)
(n , m) n ! k=0 2
lim
m
R1 = 0
m
Calculation
f(x)= g(x)= log x , f (0) = 0 (r =1, 2, ,m+ n -1).
<r>
Since Then (2.3) in Theorem 16.1.2
is applicable.
log x - (1+ n + r)- n+ r r-1 (r)
(log x)<n + r> =
(r)
x , (log x) = (-1)
(1+ n + r) xr
Substituting these for (2.3) in Theorem 16.1.2 ,
log x - (1+ n)- log x
log xdx
x x
2 n
= x
n
0 0 (1+ n)
log x - (1+ n +r)-(r)
r
m -1 -n
+ Σ (-1)r-1 n
x + Rmn
r=1 (1+ n +r)
log x -Hn+r(r)
n
x m -1 -n
= log x log x -Hn + Σ (-1)r-1 n
x + Rmn
n! r=1 r (1+ n + r)
log x -(1+ m + k) - m+ k (m+ k) n
(-1)m n -1 n -1C k
x x
Rmn =
(n , m) Σ
x (-1)m+ k-1 dx
k=0 m + k 0 0 (1+ m + k) x m+ k
n -1C k
log x - (1+ m + k) dx
1 n -1 x x
= Σ (-1)k-1 -
n
(n , m) k =0 (m + k)
2
0 0
- 33 -
n -1C k log x -(1+ n) - (1+ m + k) + n
1 n -1
= Σ
(n , m) k =0
(-1)k-1
(m + k)
2
(1+ n)
-
(1+ n)
x
n
1 x n -1 n -1C k
Σ (-1)k-1
(m + k)
= log x - Hn - Hm+ k
(n , m) n ! k=0 2
2
Example 3rd order integral of log x
Let n =3, m =1000 in (1.1). And if both sides are illustrated, it is s follows. Since the convergence of
logarithm family is usually slow, both sides are not overlapping completely.
Left:
g := int(int(int(ln(x)^2,x),x),x)
x ln(x) 11 x ln(x) 85 x3
3 2 3
6 18 108
Right: Series
n:=3: m:=1000: MAXDEPTH:=1000:
f := ln(x)*(ln(x)-psi(1+n)-EULER)*x^n/gamma(1+n) +
sum((-1)^(r-1)*binomial(-n,r)*(ln(x)-psi(1+n+r)-EULER)
*gamma(r)/gamma(1+n+r)*x^n, r=1..m-1):
- 34 -
16.5 Higher Integral of e^x f (x)
Formula 16.5.1
t
When (a , x) = a-1 -t
e dt denots the incomplete gamma function, the following expressions hold
x
for -1,-2, -3, .
(1) When x 0
x n -1 C r (n -r + , -x)x r
x x
x
x e dx =
n -1
Σ
(-x) r=0
n
(n -1)!
(1.n- )
- -
(2) When x >0
x n -1 C r (n -r + , -x)x r
x x
x
x e dx =
n -1
Σ
(-x) r=0
n
(n -1)!
- -
C r (n -r + )x r
n -1 n -1
+ 2i sin Σ (1.n+ )
r=0 (n -1)!
Proof
As mentioned above, the lineal higher primitive function of exx was as follows.
- 35 -
And when -1, -2, -3, ,
x (-1) x 0
=
(-x) (-1)- x >0
Then, (1.n- ) holds as the limit of x -0 also at the time of x =0 .
When x >0 , we calculate separately for [-, 0] and [0 , x] . If the lineal higher integral is displayed
separately by Riemann-Liouville Integral, it is as follows..
(x -t )
1 x
n-1 t
t e dt
(n) -
(x -t )
1 0 1 x
= (x -t )n-1t e t dt + n-1 t
t e dt (w)
(n) - (n) 0
(x -t )
0
n-1 t
t e dt
-
<1> 0 0
= (x -t )n-1t e t -+ (n -1) (x -t )n-2t e t dt
<1>
-
0
t t 0 (1+ , -t)
(x -t )
0
t e
t <1>
= (x -t ) n-1
+ (n -1) n-2
dt
(-t) 0!
-
-
(1+ )
(x -t )
0
= (-1) x n-1 t e t dt
n-2 <1>
+ (n -1)
0! -
the 2nd,
(x -t )
0
t e t dt
n-2 <1>
-
<2> 0 0
= (x -t )n-2t e t -+ (n -2) (x -t )n-3t e t dt
<2>
-
(2+ )
+ (n -2) (x -t )
0
= (-1) x n-2 t e t dt
n-3 <2>
1! -
the 3rd,,
(x -t )
0
t e t dt
n-3 <2>
-
(3+ )
(x -t )
0
= (-1) x n-3 t e t dt
n-4 <3>
+ (n -3)
2! -
the last (n-1)th,
0
(x -t )1t e t dt
<n -1>
-
(n -1+ )
(x -t ) t e
0
= (-1) x 1
(n -2)!
+ 1 0 t <n -1>
dt
-
- 36 -
Substituting these for the former expression one by one, we obtain the following expression.
(x -t )
0 n -1 (n -1)!
n-1 t
t e dt = (-1) Σ (n -r + )x
r
- r=1 r !(n -1-r)!
(x -t ) t e
0
+ (n -1)! 0 t <n -1>
dt
-
Thus,
(n -r + )x r
(x -t )
1 0 n -1
n-1 t
(n)
t e dt = (-1) Σ
r=1
n -1C r
(n -1)!
-
0 <n -1>
+ t e t dt (w1)
-
Calculation of the 2nd term
Since x > 0 , then x /(-x) = 1/(-1) . So, the 1st,
(x -t )
x
n-1 t
t e dt
0
<1> x x
= (x -t )n-1t e t 0+ (n -1) (x -t )n-2t e t dt
<1>
0
x
t t (1+ , -t)
(x -t )
0 x
t e t
<1>
= (x -t ) n-1
0!
+ (n -1) n-2
dt
(-t) 0
0
(1+ )
n-1
(x -t )
x x
t e t dt
n-2 <1>
=-
+ (n -1)
(-1) 0! 0
2nd,
(x -t )
x
t e t dt
n-2 <1>
<2> x x
= (x -t )n-2t e t 0+ (n -2) (x -t )n-3t e t dt
<2>
0
n-2
(2+ )
(x -t )
x x
t e t dt
n-3 <2>
=- + (n -2)
(-1) 1! 0
the last (n-1)th,
x
(x -t )1t e t dt
<n -1>
x 1 (n -1+ )
x
(x -t )0t e t dt
<n -1>
=- + 1
(-1) (n -2)! 0
Substituting these for the former expression one by one, we obtain the following expression.
(x -t )
x 1 n -1 (n -1)!
n-1 t
t e dt = - Σ (n -r + )x
r
0 (-1) r=1 r !(n -1-r)!
(x -t ) t e
x
+ (n -1)! 0 t <n -1>
dt
0
Thus,
- 37 -
(n -r + )x r
(x -t )
1 x 1 n -1
n-1 t
(n) 0
t e dt = -
(-1)
Σ
r=1
n -1C r
(n -1)!
t e
x
t <n -1>
+ dt (w2)
0
Total
Substituting (w1) , (w2) for (w) ,
(n -r + )x r
1 x 1 n -1
(x -t )n-1t e t dt = (-1) - Σ Cr
(n) (-1)
n -1
- r=1 (n -1)!
t e
0 x
t e t
<n -1> t <n -1>
+ dt + dt
- 0
Here,
1 2 i
(-1) -
(-1)
= 2i sin
= ()(1-)
t e
0 x x <n>
t e t
<n -1>
t e t
<n -1> t <n -1>
dt + dt = dt = x e x
- 0 -
x n -1 C r (n -r + , -x)x r
n -1
= Σ
(-x) r=0 (n -1)!
Substituting these for the above, we obtain (1.n+ ) .
float(G(-1.7)); float(G(1.7))
0.3457294964 i
2.255551158 2.835926162 i
- 38 -
x>0
Fp := x-> Fm(x)+2*I*sin(a*PI)/(n-1)!
*sum(binomial(n-1,r)*gamma(n-r+a)*x^r, r=0..n-1)
n1
2 i sin( a) n 1 (n r a) xr
x Fm(x)
(n 1) r0
r
float(Fp(1.7))
2.255551158 2.835926161 i
When is a natural number, we can differentiate x completly. And the following formula holds.
Formula 16.5.1'
When m =0, 1, 2, ,
(1+ m) m- r
r
x x m -n
e x x mdx n = e x Σ x (1.n')
- - r=0 (1+ m - r)
Proof
Let f(x)= e x in the Theorem16.2.0 (0.2) . Then, since a1 = a2 == an = - ,
(1+m) m- r
x x m -n
e xx m dx n = e xΣ x
- - r=0 r 1+m - r
(1+m)
dx
n -1 m -n + r m- s x
x
- e -Σ Σ (-) r
r=0 s=0 s 1+m - s - -
Since this 2nd line is 0 clearly, we obtain (1.n') immediately.
4
Example 5th order integral of ex x
Substituting m =4 , n =5 for (1.n'),
(5) 4- r
r
x x 4 -5 4 4+ r 4!
e x x 4dx 5 = e xΣ x = e xΣ(-1)r 4 r
x -
- - r=0 5- r r=0 4 (4- r)!
When the both sides are calculated by mathematical software, it is as follows.
Le ft: Rie mann-Liouville integral
m:=4: n:=5:
g := 1/gamma(n)*int((x-t)^(n-1)*t^m*E^t, t=-infinity..x)
e x 20 x 180 x 840 x 1680
x 4 3 2
Right: Formula
f := E^x*sum(binomial(-n,r)*gamma(1+m)/gamma(1+m-r)
*x^(m-r), r=0..m)
e x 20 x 180 x 840 x 1680
x 4 3 2
- 39 -
Formula 16.5.1"
The following expressions hold for , n such that
-1, -2, -3, & + n >0.
(1+ )
x x m -1 -n
e x x dx n = e xΣ r
x +n+ + Rmn (1.n")
0 0 r=0 r (1+ + n r
+ )
(1+ )
n -1C k
m n -1 x x
(-1) m+k
Rmn = Σ e x x + dx n
(n , m) k =0 m + k (1+ + m +k) 0 0
lim Rmn = 0
m
3
Example Collateral the 2nd order integral of ex x
This function can be illustrated in a positive domain. Blue (Riemann-Liouville integral) hidden from red (Series)
cannot be seen. Although this turns into complex function in a negative domain, of course, both sides are
corresponding.
Le ft: Rie mann-Liouville integral
a:=1/3: n:=2:
g := x-> 1/gamma(n)*int((x-t)^(n-1)*t^a*E^t, t=0..x):
Right: Se rie s
m:=20:
f := x-> E^x*sum(binomial(-n,r)*gamma(1+a)/gamma(1+a+n+r)
*x^(a+n+r), r=0..m-1):
Formula 16.5.2
x et
When Ei(x) = dt is the Exponential Integral, the following expressions hold.
- t
e log x dx
x0 x1
x x 0! 0
x 2 x
=e log x + x - + Ei(x)
- - 1! 0! 1!
- 40 -
e log x dx
x x x 0! 1 0! 1! 0
x 3
= e x log x + x + + x
- - - 2! 1! 2!
x0 x1 x2
- + + Ei(x)
0! 1! 2!
x x
0! 0! 1! 0! 1! 2!
e x log x dx 4 = e x log x + 3! x 2+ 2! + 3! x 1+ 1! + 2! + 3! x 0
- -
0 1 2 3
x x x x
- + + + Ei(x)
0! 1! 2! 3!
Proof
x et
When Ei(x) = dt , lineal higher primitive functions of e x log x are as follows.
- t
<1>
e x log x = e xlog x - Ei(x)
<2>
e x log x = e x(log x +1) -(x +1)Ei(x)
x 2+2x +2
<3> x +3
e log x
x
=e x
log x + - Ei(x)
2 2
<4> x 2+4x +11 x 3+3x 2+6x +6
e log x
x
=e x
log x + - Ei(x)
3! 6
And the zeros of these are all x =- . Therefore, 1st ~ 4th order integrals can be written as mentioned above.
And hereafter, by induction, we obtain (2.n) .
-4 -2 2 4
-1
-2
- 41 -
Note
All polynomials obtained by applying Theorem 16.1.2 to e xlog x become the asymptotic expansions, and
they are hardly useful.
x x
x
e sin xdx = sin n
e xsin x - (3.0s)
- - 4 4
n
e cosxdx = sin 4 e cos x - 4
x x n
x n x
(3.0c)
- -
Proof
About the higher derivative of e xsin x , the following formula is known.(See 共立 数学公式 p187).
-n
n
(n)
e sin x
x
= sin e xsin x +
4 4
Replacing n with -n ,
n
n
(-n )
e sin x
x
= sin e xsin x -
4 4
Since a differentiation operator (-n) is equal to an integration operator <n > ,
n
n
<n>
e xsin x = sin e xsin x -
4 4
And since x = - is zero of this clearly, rewriting the left side, we obtain (3.0s).
The following formula holds also about e xcos x .
-n
n
(n)
e cosx
x
= sin e xcos x +
4 4
From this, we obtain (3.0c) in a similar way.
Example
e sin xdx = sin 4 e sin x - 4 =
x 2 x sin x - cosx
x x
e
- 2 2
1 x
= e (sin x - cosx)
2
2
2
x x 1 x
x
e sin xdx = sin 2
e xsin x - =- e cosx
-- 4 4 2
3
e cosxdx = sin 4 e cos x - 4 =
3
x x 2 x
x 3 x
e sin x -
- - 4 4
2 x sin x - cosx 1
= e = e x(sin x - cosx)
4 2 4
Higher Integral of e x sin x , e x cos x ends now. There is no necessity for Theorem 16.1.2.
However, daring use Theorem 16.1.2, we obtain an interesting result.
- 42 -
Formula 16.5.3'
(1)
r
r
-n
+R
x x m -1
e xsin xdx n = e xΣ sin x + n
m (3.1)
- - r=0 2
(m + k)
e sin x +
(-1)m n -1 n -1C k
x x
(n , m) Σ
Rmn = x
dx n (3.1r)
k =0 m + k - - 2
(2)
(n + r)
r
-n
+ R
x x m -1
e xsin xdx n = e xΣ sin x - n
m (3.2)
- - r=0 2
(m + k)
(-1)m n -1 n -1C k
x x
(n , m) Σ
Rmn = e xsin x - dx n (3.2r)
k =0 m + k - - 2
(3)
r
r
-n
+R
x x m -1
e xcosxdx n = e xΣ cos x + n
m (3.3)
- - r=0 2
(m + k)
(-1)m n -1 n -1C k
x x
(n , m) Σ
Rmn = e xcos x + dx n (3.3r)
k =0 m + k - - 2
(4)
(n + r)
r
-n
+ R
x x m -1
e xcosxdx n = e xΣ cos x - n
m (3.4)
- - r=0 2
(m + k)
e cosx -
(-1)m n -1 n -1C k
x x
(n , m) Σ
Rmn = x
dx n (3.4r)
k =0 m + k - - 2
Calculation
Let f(x)= e x , g(x)= sin x . Then
r
()
(sin x) r = sin x +
2
Substituting this for (2.3) in Theorem 16.1.2 , we obtain (3.1), (3.1r).
(n +r)
(sin x)<n + r> = sin x - 2
Substituting this for (2.3) in Theorem 16.1.2, we obtain (3.2), (3.2r).
- 43 -
Se rie s
m:=100:
f := E^x*sum(binomial(-n,r)*sin(x+r*PI/2), r=0..m-1):
float(subs(f, x=3))
1135.950099
Re mainde r
r0 := -1/(m+0)*binomial(n-1,0)*E^x/2*cos(x+(m+0)*PI/2):
r1 := -1/(m+1)*binomial(n-1,1)*E^x/2*cos(x+(m+1)*PI/2):
R := (-1)^m/beta(n,m)*(r0+r1):
float(subs(R, x=3))
1145.892364
Se rie s + Re mainde r
float(subs(f+R, x=3))
9.942265422
Complete Automorphism
As understood from this example, although (3.1), (3.1r) hold as an equation, it is not helpful at all. The same
is said for (3.2), (3.2r). However, if these are used in combination, the complete automorphism of the higher
Formula 16.5.3"
n n
e xsin x - cos
x x 4 4
e xsin xdx n = (3.5)
- - n -1 n -1C k (1+ k)
1+ nΣ cos
k=0 1+ k 2
n n
e xcos x -
sin
x x 4 4
e xcosx dx n = (3.6)
- - n -1 n -1C k (1+ k)
nΣ sin
k =0 1+ k 2
n n
e x
cos
x -
cos
x x 4 4
e xcosxdx n = (3.7)
- - n -1 n -1C k (1+ k)
1+ nΣ cos
k =0 1+ k 2
n n
e xsin x - sin
x x 4 4
e xsin x dx n = (3.8)
- - n -1 n -1C k (1+ k)
nΣ sin
k =0 1+ k 2
Calculation
Let m =1 in (3.1) ~ (3.2r) . Since (n , 1) =1/n ,
(1+ k)
1+ k
n -1 n -1C k
e sin x +
x x x x
e xsin xdx n = e xsin x - nΣ x
dx n
- - k=0 - - 2
- 44 -
n (1+k)
n -1 n -1C k
x x x x
e sin xdx = e xsin x -
x n
- nΣ e xsin x - dx n
- - 2 k =0 1+ k - - 2
n -1 Ck (1+ k)
Here, let Ck = , Bk = . Then
1+ k 2
e sinx +B dx
x x n -1 x x
e xsin xdx n = e xsin x - nΣCk x
k
n
- - k=0 - -
n
e sin xdx = e sin x -
2
- nΣC e sinx -B dx
x x n -1 x x
x n x x n
k k
- - k =0 - -
Using the sum and difference formulas,
n
e sin x cosB - cos x sinB dx
x x x x
n -1
e xsin x dx n = e xsin x - -n ΣCk x
k k
n
(b)
- - 2 k =0 - -
Adding (a) to (b),
n
e sin x cosB dx
x x n -1 x x
2 e xsin xdx n = e x sin x + sin x - -2nΣCk x
k
n
- - 2 k=0 - -
n n
- 2nΣC cosB e sin x dx
n -1 x x
= 2e xsin x - cos k k
x n
4 4 k =0 - -
From this,
n n
e xsin x - cos
x x 4 4
e xsin xdx n = n -1
- - 1 + nΣCk cosBk
k=0
Returning Ck , B k before,
n n
e xsin x - cos
x x 4 4
e xsin xdx n = (3.5)
- - n -1 n -1C k (1+ k)
1+ nΣ cos
k =0 1+ k 2
Next, subtract (b) from (a) . Then
n
e cosx sinB dx
n -1 x x
0 = e x sin xdx n - sin x - -2nΣCk x
k
n
2 k =0 - -
n n
-2nΣC sinB e cosx dx
n -1 x x
x n
= 2cos x - sin k k
4 4 k =0 - -
From this,
n n
e xcos x - sin
e cosx dx
x x 4 4
x n
= n -1
- - nΣCk sinBk
k =0
Returning Ck , B k before,
n n
e xcos x - sin
x x 4 4
e xcosx dx n = (3.6)
- - n -1 n -1C k (1+ k)
nΣ sin
k =0 1+ k 2
- 45 -
Also about e xcosx , in a similar way, we obtain (3.7), (3.8).
Formula 16.5.4
C k-1 k -n
n
n n -1 1
Σ
k =1 k
sin
2
=
n
sin
4
sin
4
(4.1)
n -1C k-1 k -n
n
n 1 1
Σ
k =1 k
cos
2
=
n
sin
4
cos
4
-
n
(4.2)
Proof
n
e cosxdx = sin 4 e cos x - 4
x x n
x n x
(3.0c)
- -
n n
e cosx -
4
x
sin
x x 4
e cosx dx = x n
(3.6)
- - C (1+ k) n -1 n -1 k
nΣ sin
k =0 1+ k 2
From these,
n
n n -1 n -1C k (1+ k)
sin
4 = sin
4
/ nΣ
k =0 1+ k
sin
2
Replacing k with k -1 , we obtain (4.1).
- 46 -
Next,
n
e sin xdx = sin 4 e sin x - 4
x x n
x n x
(3.0s)
- -
n n
e sin x - cos x
n
n n -1 n -1C k k
sin
4 = cos
4
/ 1+ nΣ
k =0 1+ k
cos
2
Replacing k with k -1 , we obtain (4.2).
Formula 16.5.4'
When , denote a floor function and a ceiling function respectivly,
n /2 (-1)k -n-1
(n +1)
1
Σ
k =0 2k +1 n
C 2k =
n +1
sin
4
sin
4
(4.3)
(n +1)
n /2 (-1)k -n-1
1
Σ
k =1 2k
nC 2k-1 =
n +1
sin
4
cos
4
-1 (4.4)
Proof
Replacing n,k with n +1, k +1 in (4.1) ,
Ck (k +1) -n-1
(n +1)
n n 1
Σ
k =0 k +1
sin
2
=
n +1
sin
4
sin
4
Since the even-numbered terms of the left side are all 0,
n Ck (k +1) nC 0 nC 2 nC 4 (-1)k
-+ + 2k n
n
Σ
k =0 k +1
sin
2
=
1
-
3
+
5 2k +1
nC 2k
n /2 (-1)k
=Σ nC 2k
k =0 2k +1
(-1)k
n /2 -n-1
(n +1)
1
Σ nC 2k = sin sin (4.3)
k =0 2k +1 n +1 4 4
Replacing n , k with n +1 , k +1 in (4.2) ,
(k +1) (n +1)
-n-1
Ck
n n 1
Σ
k =0 k +1
cos
2
=
n +1
sin
4
cos
4
-1
Since the odd-numbered terms of the left side are all 0,
n Ck (k +1) nC 1 nC 3 nC 5 (-1)k
+- +
n
Σ
k =0 k +1
cos
2
=-
2
+
4
-
6 2k
nC 2k-1 2k -1 n
- 47 -
n /2 (-1)k
=Σ nC 2k-1
k=1 2k
(n +1)
n /2 (-1)k -n-1
1
Σ nC 2k-1 = sin cos -1 (4.4)
k =1 2k n +1 4 4
Example
C0 C2 C4 C6 C8 -9-1
(9+1)
9 9 9 9 9 1 16
- + - + = sin sin =
1 3 5 7 9 9+1 4 4 5
C0 C2 C4 C6 C8 -8-1
(8+1)
8 8 8 8 8 1 16
- + - + = sin sin =
1 3 5 7 9 8+1 4 4 9
(7+1)
4
C1
7 C3
7 C5
7 C7
7 1 -7-1
15
- + - + = sin cos -1 =
2 4 6 8 7+1 4 8
(8+1)
8+1
- 1 =
C1 C3 C5 C7 -8-1
4
8 8 8 8 1 5
- + - + = sin cos
2 4 6 8 4 3
- 48 -
16.6 Higher Integral of f (x) / e^x
e x
-x
16.6.1 Higher Integral of
Formula 16.6.1
t
When (a , x) = a-1 -t
e dt is the incomplete gamma function,
x
x 1
e -x x dx (1+ , x) =-
0!
x x 1
e -x x dx 2 = {(2+ , x)- (1+ , x)x }
1!
x x x 1
e x dx = - {(3+ , x)- 2(2+ , x)x + (1+ , x)x 2}
-x 3
2!
e
x x 1
x dx = {(4+ , x) -3(3+ , x)x +3(2+ , x)x -(1+ , x)x }
-x 4 2 3
3!
x x
(-1)n n -1
(-1)rn -1C r (n -r + , x)x r
(n -1)! Σ
-x n
e x dx = (1.n)
r=0
Proof
If the integration is obediently repeated, they are led naturally. And these zeros are all x = . Then these
are lineal higher integrals.
-x
Example 2nd order integral of e x
Let substitute =1/2 , n =2 for (1.n), and calculate the value on thearbitrary point x = 2.3 .
Then, it is as follows. Both sides are corresponding exactry.
float(f(-2.3))
3.367662317 6.637309509 i
When is a natural number, we can differentiate x completly. And the following formula holds.
- 49 -
Formula 16.6.1'
When m =0, 1, 2, ,
(1+ m) m- r
(-1)n
r
x x m -n
-x m
e x dx = n
Σ (-1) r
x (1.n')
ex r=0 (1+ m - r)
-x 4
Example 5th order integral of e x
Substituting m =4 , n =5 for (1.n') ,
(1+4) 4- r
(-1)5
r
x x 4 -5
-x 4
e x dx = 5
Σ(-1)
r
x
ex r=0 (1+4- r)
When the both sides are calculated by mathematical software, it is as follows. Naturally, both sides are
corresponding.
Left: Riemann-Liouville inte gral
m:=4: n:=5:
g := 1/gamma(n)*int((x-t)^(n-1)*t^m/E^t,t=infinity..x):
expand(g)
840 x
180 x
20 x
2 3 4
x
1680
x x x x x
e e e e e
Right: Formula
f := (-1)^n/E^x*sum((-1)^r*binomial(-n,r)*gamma(1+m)
/gamma(1+m-r)*x^(m-r), r=0..m):
expand(f)
180 x
20 x 840 x
2 3 4
x
1680
x x x x x
e e e e e
e x
-x
Collateral higher integral of
The following formula holds about the collateral higher integral whose zero is 0.
Formula 16.6.1"
The following expressions hold for , n such that -1, -2, -3, & + n >0.
(1+ )
x x 1 m -1 -n
e -x x dx n = -r r
Σ (-1)
x +n+ + Rmn
0 0 e x r=0 r ( 1+ + +
n r)
+ m+k
(1+ )
n -1C k
1 n -1 x x x
Rmn = Σ
(n , m) k =0
(-1)-k
m + k (1+ + m +k) x dx n
0 0 e
lim R n
m m
=0
-x
Example Collateral the 2nd order integral of e x
This function can be illustrated in a positive domain. Blue ( Riemann-Liouville integral ) hidden from
red (Series) cannot be seen. Although this turns into complex function in a negative domain, of course,
both sides are corresponding.
- 50 -
Left: Riemann-Liouville inte gral
a := 1/2: n:=2:
g := x->1/gamma(n)*int((x-t)^(n-1)*t^a/E^t,t=0..x):
Right: Serie s
m:=20:
f := x->sum((-1)^-r*binomial(-n,r)*gamma(1+a)/gamma(1+a+n+r)
*x^(a+n+r)/E^x, r=0..m-1):
-x
16.6.2 Higher Integral of e log x
Formula 16.6.2
x et
When Ei(x) = dt is the Exponential Integral, the following expressions hold.
- t
x -
0! 1!
0 1
x x 0! x x
-x 2 -x 0
e log x dx = e log x + - Ei(-x)
1!
1! + 2! x
x x x 0! 0! 1!
-x 3 -x 1 0
e log x dx = -e log x - x +
2!
+
0! 1! 2!
0 1 2
x x x
- + Ei(-x)
-
0! 1! 2! 3!
0 1 2 3
x x x x
- + - Ei(-x)
x x n -2 n -2-r s!
e -x log x dx n = (-1)n e -x log x + Σ (-x)r Σ
r=0 s=0 (r + s +1)!
- 51 -
n -1 (-x)r
- (-1) Ei(-x)Σ
n
(2.n)
r=0 r!
Proof
Lineal higher primitive functions of e -x log x are as follows.
<1>
e -x log x = -e -x log x + Ei(-x)
<2>
e log x
-x
= e -x(log x +1) + (x -1)Ei(-x)
x 2-2x +2
<3> x 3
e -x log x = -e -x
log x - + + Ei(-x)
2 2 2!
<4> x 2-4x 11 x 3-3x 2+6x -6
e log x
-x
= e -x log x + + + Ei(-x)
3! 6 3!
And the zeros of these are all x = . Therefore, 1st ~ 4th order integrals can be written as mentioned above.
And hereafter, by induction, we obtain (2.n).
-x
Example 3rd order integral of e log x
Note
All polynomials obtained by applying Theorem 16.1.2 to e -xlog x become the asymptotic expansions,
and they are hardly useful.
-x -x
16.6.3 Higher Integrals of e sinx , e cosx
Formula 16.6.3
n
n
x x
e sin xdx = (-1)n sin
-x n
e -xsin x + (3.0s)
4 4
- 52 -
n
n
x x
-x
e cosxdx = (-1) n n
sin e -xcos x + (3.0c)
4 4
Proof
About the higher derivative of e xsin x , the following formula is known.(See 共立 数学公式 p187).
-n
n
(n)
e sin x
x
= sin e xsin x +
4 4
Replacing x with -x ,
-n
n
(n)
(-1) e sin(-x)
n -x
= sin e -xsin -x +
4 4
Using sin(-x) = -sin x
-n
n
(n)
e sin x
-x
= (-1)n sin e -xsin x -
4 4
Replacing n with -n ,
n
n
(-n )
e sin x
-x
= (-1)n sin e -xsin x +
4 4
Replacing (-n ) with <n > ,
n
n
<n>
e sin x
-x
= (-1)n sin e -xsin x +
4 4
And since x = is zero of this clearly, rewriting the left side, we obtain (3.0s).
The following formula holds also about e xcos x .
-n
n
(n)
e cosx
x
= sin e xcos x +
4 4
From this, we obtain (3.0c) in a similar way.
Example
e
x 2 -x sin x + cosx
-x
sin xdx = (-1)1 sin e -xsin x + =- e
4 4 2 2
1 -x
=- e (sin x + cosx)
2
2
2
x x 1 -x
-x 2 2
e sin xdx = (-1) sin e -xsin x + = e cosx
4 4 2
3
e cosxdx
3
x x 2 -x
x 3 3
= (-1) sin e -xcos x + = e sin x +
4 4 4 4
2 x cosx + sin x 1
= e = e -x(sin x + cosx)
4 2 4
- 53 -
16.7 Series Expansion of Higher Integral
By adopting easy function such as 1 or ex as one side of the functions f and g in Thorem 16.1.2, various
function series is obtained.
Theorem 16.7.1
r =0, 1, ,m +n
(r)
Let m,n be natural numbers, f be the r th derivative of f, a be an arbitrary
constant on the domain of f , (x , y) be the beta function. Then the following expressions hold.
f(x)dx
(x -a)n+ r (r)
r
x x m -1 -n
n
=Σ f (x) + Rmn (1.1)
a a r=0 (n + r)!
(-1)m n -1 n -1C k x t t (t -a)m+ k (m+ k )
(n , m) Σ
Rmn = f (t)dt n (1.1r)
k=0 m + k a a a (m + k)!
f(x)dx
x x m -1 (x -a )
n+ r
n
=Σ (r)
f (a) + Rmn (1.2)
a a r=0 (n + r)!
n -1 r -1 r -1 (x -a)m + n+s (m+s)
t s m +n -1- r + C
t m -1 m +n+sC r
n-r+s
Rmn = ΣΣ(-1) Σ C f (a)
r =1 s =0 t=s (m + n + s)!
(-1)m n -1 n -1C k x t t (t -x)m+ k (m+ k )
(n , m) Σ
+ f (t)dt n (1.2r)
k=0 m + k a a a (m + k)!
Where, there shall be no term of ΣΣ at the time of n <2 .
Proof
Exchanging f and g in Theorem 16.1.2 ,
r
x x m -1 -n
g <0>f(0)dx n = Σ g <n + r> (r)
f
a a r=0
(x -a)r
s g
-n + r
n -1 m -1
<n -r+ s> (s)
- ΣΣ a fa
r=0 s=0 r!
n -1 r -1 r -1 (x -a)r
t s m + n -1- r + tC m-1 g a
m + n -r+s m +s
+ (-1)m ΣΣΣ C fa
r =1 s =0 t=s r!
g
m
(-1) n -1 n -1C k x x
<m+ k > (m+ k )
(n , m) Σ
+ f dx n
k=0 m + k a a
r
(x -c)
( a c) for this,
<r>
Substituting g =
r!
1f
(x -c)n+ r (r)
r
x x m -1 -n
(0)
dx = Σ
n
f
a a r=0 (n + r)!
n-r+ s r
s (n -r + s)! r ! f
-n + r (a -c)
n -1 m -1 (x -a) (s)
- ΣΣ a
r=0 s=0
- 54 -
n -1 r -1 r -1 (a -c)m + n-r+s (x -a)r m +s
Σ ΣΣ C t s m + n -1- r + tC m-1
m
+ (-1) fa
r =1 s =0 t=s (m + n - r + s)! r !
(-1)m n -1 n -1C k x x (x -c)m+ k (m+ k ) n
(n , m) Σ
+ f dx
k=0 m + k a a (m + k)!
Change this to the definite higher integral from a to b ( c) , and rewrite the variable in the remainder term
from x to t . Then,
f
(b -c)n+ r (r)
r
b x x m -1 -n
(0)
dx = Σn
f
a a a r=0 (n + r)! b
n-r+ s r
s (n -r + s)! r !
-n + r (a -c)
n -1 m -1 (b -a) (s)
- ΣΣ fa
r=0 s=0
(-1)m n -1 n -1C k b t t (t -c)m+ k (m+ k ) n
(n , m) Σ
+ f dt (1.0)
k=0 m + k a a a (m + k)!
Here, put c=a , then
f
(b -a)n+ r (r)
r
b x x m -1 -n
(0)
dx = Σn
f
a a a r=0 (n + r)! b
n-r+ s
(b -a)r (s)
s (n -r + s)!
-n + r (a -a)
n -1 m -1
- ΣΣ fa
r=0 s=0 r!
n -1 r -1 r -1 (a -a)m + n-r+s (b -a)r m +s
Σ ΣΣ C t s m + n -1- r + tC m-1
m
+ (-1) fa
r =1 s =0 t=s (m + n - r + s)! r!
(-1)m n -1 n -1C k b t t (t -a)m+ k (m+ k ) n
(n , m) Σ
+ f dt
k=0 m + k a a a (m + k)!
(b -a)n+ r (r)
m -1 -n
=Σ f
r=0 r (n + r)! b
(-1)m n -1 n -1C k b t t (t -a)m+ k (m+ k ) n
(n , m) Σ
+ f dt
k=0 m + k a a a (m + k)!
Returning b to x , we obtain (1.1), (1.1r) .
Next, let c=b in (1.0) . Then,
(b -b)n+ r (r)
b x x m -1 -n
f dx = Σ
(0) n
f
a a a r=0 r (n + r)! b
(a -b)n-r+ s (b -a)r (s)
n -1 m -1 -n + r
- ΣΣ fa
r=0 s=0 s (n - r + s)! r!
n -1 r -1 r -1 (a -b)m + n-r+s (b -a)r m +s
+ (-1) ΣΣΣ C t s m + n -1- r + tC m-1
m
fa
r =1 s =0 t=s (m + n - r + s)! r!
(-1)m n -1 n -1C k b t t (t -b)m+ k (m+ k ) n
(n , m) Σ
+ f dt
k=0 m + k a a a (m + k)!
- 55 -
(b -a)n+ s
n -1 m -1
n-r+ s
-n + r (s)
= -ΣΣ (-1) fa
r=0 s=0 s r!(n - r + s)!
n -1 r -1 r -1 (b -a)m + n+s
t s m + n -1- r + tC m-1 f a +s
n-r+s m
+ ΣΣΣ(-1) C
r =1 s =0 t=s r!(m + n - r + s)!
m n -1 m+ k
n -1C k
(-1) b t t (t -b )
+ Σ m+ k
f ( )dt n
(n , m) k=0 m + k a a a (m + k)!
Here,
n+sC r
-n + r 1
(-1) s = n-r -1+sC s , =
s r!(n - r + s)! (n + s)!
Substituting these for the above,
b x x m -1 n -1 (b -a)n+ s (s)
f dx = -Σ Σ(-1) n-1-r +sC s n+sC r
(0) n n-r
f
a a a s=0 r=0 (n + s)! a
n -1 r -1 r -1 (b -a)m + n+s m +s
t s m + n -1- r + tC m-1 m +n+sC r
n-r+s
+ ΣΣ(-1) Σ C f
r =1 s =0 t=s (m + n + s)! a
(-1)m n -1 n -1C k b t t (t -b)m+ k (m+ k ) n
(n , m) Σ
+ f dt
k=0 m + k a a a (m + k)!
Here, the following equation holds for a natural number n and a non-negative integer s .
n -1
Σ (-1) n-r
n-1 -r +s C s n+sC r = -1
r=0
Because,
n -1 n -1 (n -1-r + s)! (n + s)!
Σ(-1) n-1-r +sC s n+sC r = Σ(-1)
n-r n-r
r=0 r=0 (n -1-r)! s ! (n -r + s)! r !
(n + s)! n -1 (n -1)! 1
= Σ (-1)n-r
(n -1)! s ! r=0 (n -1- r)! r ! n -r + s
(n + s)! n -1 n-1C n-1- r
=- Σ (-1)n-1-r
(n -1)! s ! r=0 s +1+ n -1-r
r
(n + s)! n -1 (-1)
(n -1)! s ! Σ n-1C r
=-
r=0 s +1+ r
According to 岩波数学公式Ⅱ p12,
m !(a)
r
m (-1)
Σ mC r = a 0, -1, -2,
r=0 a+ r (m +a +1)
Then,
n -1 (-1) r (n -1)! (s +1) (n -1)! s !
Σ
r=0 s +1+ r
n-1 Cr =
(n -1+ s +1+1)
=
(n + s)!
Therefore
f
b x x m -1 (b -a)n+ s (s)
(0)
dx = Σ
n
f
a a a s=0 (n + s)! a
n -1 r -1 r -1 (b -a)m + n+s m +s
t s m + n -1- r + C
t m-1 m +n+sC r
n-r+s
+ ΣΣ(-1) Σ C f
r =1 s =0 t=s (m + n + s)! a
(-1)m n -1 n -1C k b t t (t -b)m+ k (m+ k ) n
(n , m) Σ
+ f dt
k =0 m + k a a a (m + k)!
- 56 -
ΣΣ of the 2nd line cannot be simplified any more. Moreover, since this is a disturbance term, we combine this
with the 3rd line and make it a remainder. Needless to say, when this 2nd line doesn't exist when n is smaller
than 2. Thus, rewriting s to r in the 1st line and returning b to x , we obtain (1.2), (1.2r) .
Thorem 16.7.1'
n and an arbitrary number a on the domain of f .
The following formulas hold for a natural number
(x -a)n+ r (r)
1 x m -1 -n
n-1
(x -t) f(t)dt = Σ f (x) + Rmn (1.1')
(n) a r=0 r (n + r)!
x (x -t)n-1(t -a )m+ k
(-1)m n -1 n -1C k
m+ k
n
Rm = Σ
(n , m)(n) k =0 m k a
+ ( m k+ ) !
f ( )(t)dt n (1.1'r)
(x -a)n+ r (r)
(x -t)
1 x m -1
n-1
f(t)dt = Σ f (a) + Rmn (1.2')
(n) a r=0 (n + r)!
n -1 r -1 r -1 (x -a)m + n+s (m+s)
t s m +n -1- r + C
t m -1 m +n+sC r
n-r+s
Rmn = ΣΣ(-1) Σ C f (a)
r =1 s =0 t=s (m + n + s)!
(-1)m+ n x (t -x)m+n+ k
n -1 n -1C k
m+ k
- Σ
(n , m)(n) k =0 m + k a (m + k)!
f ( )(t)dt n (1.2'r)
r =1, 2, 3,
r () (r)
f(x)= e x , f (x)= e x , f (0)= 1
Substituting these for Theorem 16.7.1 ,
e dx
(x -a)n+ r
r
x x m -1 -n
a a
x n
=e x
Σ
r=0 (n + r)!
+ Rmn
(-1)m n -1 n -1C k x t t (t -a)m+ k t n
(n , m) Σ
Rmn = e dt
k=0 m + k a a a (m + k)!
x x m -1 (x -a)n+ r a
e dx = Σ
x n
e + Rmn
a a r=0 +
(n r) !
n -1 r -1 r -1 (x -a)m + n+s a
t s m + n -1- r + C
t m-1 m +n+sC r
n-r+s
Rmn = ΣΣ(-1) Σ C e
r =1 s =0 t=s (m + n + s)!
(-1)m n -1 n -1C k x t t (t -x)m+ k t n
(n , m) Σ
+ e dt
k=0 m + k a a a (m + k)!
Although the zero of the lineal higher integral of e x is a = - , these expressons can not have this zero.
That is, by the above formulas, the lineal higher integral of e x can not be expanded to the series. Then, we
- 57 -
put a =0 . and we expand the collateral higher integral of e x to the series. When n =2 ,
e dx
(x -0)2+ r
r
x x m -1 -2
x x
Σ
2
=e + Rm2
0 0 r=0 (2+ r)!
x t(t -0)m+ k
(m + k)! e dt
(-1)m 2-1 2-1C k
(2, m) Σ
t
Rm2 = 2
k =0 m + k 0 0
x x 2+ r
m -1 (x -0)
e xdx 2 = Σ e 0 + Rm2
0 0 r=0 (2+ r)!
(x -0)m +2 0
Rm2 = - mC m-1 m +2C 1 e
(m +2)!
x t(t -x)m+ k
(m + k)! e dt
(-1)m 2-1 2-1C k
(2, m) Σ
t 2
+
k=0 m + k 0 0
(x -a)n -n (x -a)n+ r (r -1)!
x x m -1
log x dx = log x -Σ (-1)
n r
r
+ Rmn
a a n! r=1 r (n + r)! x
n+ r
x x (x -a)n m -1
r (x -a ) (r -1)!
log x dx =
n
log a - Σ (-1) r
+ Rmn
a a n ! r=1 +
(n r) ! a
If a =0 then the first formula is lineal higher integral, else it is collateral higher integral. On the oter hand,
the second formula can not be lineal higher integral.
By the first formula, we calculate the lineal 3rd order integral and illustrate it as follows. Although the
convergence of this series was slow, as a result of taking m very greatly, both sids overlapped somehow.
Left: lineal highe r integral
g := n-> x^n/n!*(ln(x) - sum(1/k,k=1..n)):
Right: Series
m:=1000: a:=0:
f := n-> (x-a)^n/n!*ln(x)-sum((-1)^r*binomial(-n,r)
*((x-a)^(n+r)/(n+r)!)
*((r-1)!/x^r),r=1..m-1):
- 58 -
16.7.2 Series Expansion of a Function
Sliding the index of the integration operator in Theorem 16.7.1, we obtain the series expansion of a function.
Of course in these, Taylor expansion is also included.
Theorem 16.7.2
r =0, 1, ,m +n
(r)
Let m,n are natural numbers, f be the r th derivative of f , a be the arbitrary
constant on the domain of f , (x , y) be the beta function. Then the following expressions hold.
(x -a)r (r) (x -a)n+ r (n + r)
n -1 m -1 -n
f(x) = Σ f (a) + Σ f (x) + Rmn (2.1)
r=0 r! r=0 r (n + r)!
(-1)m n -1 n -1C k x t t (t -a)m+ k (n + m+ k )
(n , m) Σ
Rmn = f (t)dt n (2.1r)
k =0 m + k a a a (m + k)!
m + n -1 (x -a)r (r)
f(x) = Σ
r=0 r!
f (a) + Rmn (2.2)
(-1)m n -1 n -1C k x t t (t -x)m+ k (n + m+ k )
(n , m) Σ
+ f (t)dt n (2.2r)
k=0 m + k a a a (m + k)!
Where, there shall be no term of ΣΣ at the time of n <2 .
Proof
(0)
Adding n to the index of the differentiation operator of the function f in Theorem16.7.1 ,
n+ r
r
x x m -1 -n (x -a)
f
(n)
(x)dx n = Σ n r
f ( + )(x) + Rmn
a a r=0 (n + r)!
(-1)m n -1 n -1C k x t t (t -a)m+ k (n + m+ k )
(n , m) Σ
Rmn = f (t)dt n
k=0 m + k a a a (m + k)!
x x n+ r
m -1 (x -a)
(n)
f (x)dx n = Σ n r
f ( + )(a) + Rmn
a a r=0 (n + r)!
n -1 r -1 r -1 (x -a)m + n+s (n + m+s)
t s m +n -1- r + C
t m -1 m +n+sC r
n-r+s
Rmn = ΣΣ(-1) Σ C f (a)
r =1 s =0 t=s (m + n + s)!
(-1)m n -1 n -1C k x t t (t -x)m+ k (n + m+ k )
(n , m) Σ
+ f (t)dt n
k=0 m + k a a a (m + k)!
Here,
f f
x x x x (x -a)0 (n -1)
dx =
(n) n (n -1) n-1
- fa dx
a a a a 0!
= f
0
x
(x -a) x
(n -2) (n -2) (x -a)1 (n -1) n-2
- fa - fa dx
a 0! a 1!
= f
0
x
(x -a) x
(n -3) (n -3) (x -a)1 (n -2) (x -a)2 (n -1) n-3
- fa - fa - fa dx
a 0! a 1! 2!
- 59 -
n -1 (x -a)r (r)
=f (0)
-Σ fa
r=0 r!
And,
n+ r
n -1 (x -a)r (r) m -1 (x -a )
(n + r)
m + n -1 (x -a )r
(r)
Σ
r=0 r !
f a ( ) + Σ
r=0 ( +
n r )!
f (a ) = Σr=0 r !
f (a)
Substituting these for the above, we obtain the desired expressions.
Remark
When n =1 in (2.2), (2.2r) ,
m (x -a)r (r) x(t -x)m
m
f(x) = Σ f (a) + (-1)m f (1+ )(t)dt
r=0 r! a m!
Needless to say, this is the usual Taylor expansion. However, in fact, (2.2), (2.2r) is alredy the usual Taylor
expansion. Because, from Theorem 4.1.3 (1.2), this fearful remainder must be as follows.
x x 1 x
Rmn = f ( )(x)dx
m+ n m+ n
= (x -t)m+ n-1 f(t)dt
a a ( +
m n a )
(2.2), (2.2r) is a trivial expression.
After all, the clarification from this theorem is the following two.
(1) A series of functions of a function f(x) about a point a exists innumerably.
(2) The Taylor series of a function f(x) about a point a is unique.
r =1, 2, 3,
r
() (r)
f(x)= e x , f (x)= e x , f (0)= 1
Substituting these for (2.1) in the theorem,
n+ r
xr
n -1 m -1 -n x
e =Σx
+ e xΣ + Rmn
r=0 r! r=0 r (n + r)!
n
However, this is an automorphism and is not interesting. Then, using lim R = 0,
m m
n+ r
xr
r
-n x n -1
e x
1-Σ =Σ
r=0 (n + r)! r=0 r!
From this,
/Σ
(-x)n+ r (-x)r
r
-n n -1
e = 1 -Σ
x
r=0 (n + r)! r=0 r!
And giving n =1, 2, 3, to this, we obtain the following series.
-1 (-x)1+ r x
r
e = 1 -Σ
x
/1 =Σ
r=0 r (1+ r)! r=0 r !
-2 (-x)2+ r
x1
= 1 -Σ / 1-
r=0 r (2+ r)! 1!
r (3+ r)!
3+ r
/ 1 -
2!
-3 (-x) 1 2
x x
= 1 -Σ +
r=0 1!
- 60 -
1 -1 1 1
log x = -Σ -1 <x
r=0 1+ r r x 2
2+ r
(1-x)1
1 -2 1
= -Σ -1 -
r=0 2+ r r x 1
3+ r
(1-x)1 (1-x)2
3+ r r x
1 -3 1
= -Σ -1 - -
r=0 1 2
- 61 -
16.8 Pascal Type Triangle
When there is a number triangle, such that the numbers on two sides are given and the other numbers are
given by sum of the two numbers on the upper line, we will call it "Pascal Type Triangle".
Lemma 16.8.1
In the following number triangle, cr0 (r =0, 1, 2, ) are given, and let crr = c00 ,
crs = cr-1 s + cr-1 s-1 r, s 1 .
c00
c10 c11
c20 c21 c22
c30 c31 c32 c33
c40 c41 c42 c43 c44
Then, the following expressions hold.
r -s r -1
crs = Σ r -1-tC s-1 ct 0 = ΣC
t s-1 cr-1 -t 0
t=0 t=s -1
Proof
If we calculate one byone according to the condition crs = cr-1 s + cr-1 s-1 , it is as follows.
1 c10 c11
2 c20 c10 + c11 c22
3 c30 c10 + c11 + c20 c10 + c11 + c22 c33
4 c40 c10 + c11 + c20 + c30 2c10 + 2c11 + c20 + c22 c10 + c11 + c22 + c33 c44
- 62 -
That is,
0 c00 1 2 3 4
0
1 c10 Σ
t=0
0- tC 0 ct0
1 0
2 c20 Σ1-C
t 0 ct0 Σ1- C
t=0
t 1 ct0
t=0
2 1 0
3 c30 Σ
t=0
2- tC 0 ct0 Σ
t=0
2- tC 1 ct0 Σ2- tC 2 ct0
t=0
3 2 1 0
4 c40 Σ
t=0
C 0 ct0
3- t Σ
t=0
C 1 ct0 Σ3-tC 2 ct0 Σ3-tC 3 ct0
3- t
t=0 t=0
Thus, we obtain the following expression.
r -s
crs = Σr -1-tC s-1 ct 0 r, s 1
t=0
And,
r -s
crs = Σr -1-tC s-1 ct 0
t=0
= r -1-(r -s)C s-1 cr-s 0 + r -1-(r -s -1)C s-1 cr-s-10 + + r -1-1C s-1 c10 + r -1-0C s-1 c00
= s -1C s-1 cr-s 0 + sC s-1 cr-s-10 + + r -2C s-1 c10 + r -1C s-1 c00
r -1
= ΣC
t s-1 cr-1 -t 0
t=s -1
C2(3,0);C2(3,1);C2(3,2);C2(3,3)
Note
When ct0 = 1 t =0, 1, 2, , this number triangle reduces to Pascal's Triangle.
2009.02.18
K. Kono
Alien's Mathematics
- 63 -
17 Super Integral of the Product of Two Functions
Formula 17.1.1
r , p are positive numbers, f r be an arbitrary r th order primitive function of f(x) and g
<> (r )
Let
be the r th order derivative function of g(x ) . At this time, if there is a certain constant a and
f
<r>
(a ) =0 or g (r)(a) =0 for r [0, p], the following expression holds.
r f
x x p x x
<p>
f <0>g(0)dx p = f g(0) - Σ <r> (r)
g dx p (1.1)
a a r=1 a a
Proof
In Formula 16.1.1 (1.3) in 16.1.1 , let us analytically continue the index of the integration operator to
[0 , p] [1 , n] . Since Σ have to become an infinite series according to this, the upper limit n
from
is expanded to .
1
x x
3 6
Example Super Intagration by parts of x x dx
0 0
-1-
Note1
The above example is the one for the numerical proof of this formula. Actually, this formula is rarely used
directly.
Note2
The more general formulas of Theorem 16.1.1 in 16.1.1 are as follows.
s f a ga
x x p -1 r r x x
f <0> (0)
g
<p>
dx = f g(0) - ΣΣ
p <p -r+s> (s)
dx r
a(p) a(0) r=0 s=0 p-r p-r ap a p- r
r f g dx
p x x
<r> (r)
-Σ p
r=1 a(p) a(0)
f
r
s a
x x p -1 r r (x -a)
<0> (0)
g p <p>
dx = f g(0) - ΣΣ f g a
<p -r+s> (s)
a a r=0 s=0 r!
r f g dx
p x x
<r> (r)
-Σ p
a a r=1
These are notionally right. However, it is almost impossible to obtain the constant-of-integration function
r=0 s=0 s
p -1 r r
Σ Σ contained in these. So, regrettably below, we build a theory based on the easiest (1.1) .
Theorem 17.1.2
<r> (r)
Let r, p are positive numbers, f be an arbitrary r th order primitive function of f(x) , g be the r th
order derivative function of g(x) , (n , m),(p ) are the beta function and the gamma function respectively
At this time, if there is a number a such that
f (a) = 0 r [0 , m + p] g (a) = 0 s [0 , m + p -1,]
<r> s ()
or
then the following expression holds.
r
x x m -1 -p
f g dx p = Σ f<
p + r> (r)
g + Rmp (2.1)
a a r=0
k f
(-1)m 1 p -1 x x
<m+ k > (m+ k )
(p , m) Σ
Rmp = g dx p (2.1r)
k =0 m + k a a
Proof
In Formula 16.1.2 (2.3) in 16.1.2 , let us analytically continue the index of the integration operator to
[0 , p] [1 , n] .
from Since Σ have to become an infinite series according to this, the upper limit n
is expanded to .
-2-
Riemann-Liouville Integral Expression
According to Formula 7.2.3 in 7.2.3 , Super Integral is expressed by Riemann-Liouville Integral as follows.
Theorem 17.1.2'
<r> (r)
Let r, p are positive numbers, f be an arbitrary r th order primitive function of f(x) , g be the r th
order derivative function of g(x) , (n , m),(p ) are the beta function and the gamma function respectively
At this time, if there is a number a such that
f (a) = 0 r [0 , m + p] g (a) = 0 s [0 , m + p -1,]
<r> s()
or
then the following expression holds.
r
1 x m -1 -p
(x -t)p-1 f g dt = Σ f<
p + r> (r)
g + Rmp (2.1')
(p) a r=0
m+k k
m p -1 x
(-1) 1 p-1 <m+ k > (m+ k )
Rmp =
(p)(p , m) Σ k=0
(x -t)
a
f g dt (2.1'r)
-3-
17.2 Super Integral of x^a f (x) (general)
Formula 17.2.0
Let (z) be the gamma function, (n , m) be the beta function, f <r> be an arbitrary r th primitive
f ar
<> <r >
function of f(x) and be the function values of f on a . Then the following expressions hold
r [0 , m + p] or a = 0
<r>
(1) When f (a) = 0
(1+ ) - r
x x m -1 -p
<0>
f x dx p = Σ p r
f<+> x
a a r=0 r (1+ - r)
(1+ )
(-1)m 1 p -1 x x
m- k
+ Σ f <m+ k >
x - dx p (0.1)
(p , m) k =0 m + k k a a (1+ - -
m k )
Especially, when = m =0, 1, 2,
(1+m) m- r
x x m -p
f <0>x m dx p = Σ p r
f<+> x (0.1')
a a r=0 r (1+ m - r)
where, if = -1, -2, -3, , it shall read as follows.
(1+ ) (- + r)
(-1)-r r = r, s, m+ k
(1+ - r) (- )
f (a) = 0 s [0 , m + p -1]
s ()
(2) When a =0 or
(1+ )
x x m -1 -p
x f(0)dx p = Σ r (r)
x +p+ f
a a r=0 r ( 1+ + +
p r )
(1+ )
m x x
(-1) 1 p -1 +m
+ Σ m+ k
x +k f ( )dx p (0.2)
(p , m) k =0 m + k k a a (1+ + +
m k )
where -1, -2, -3, & + p -1, -2, -3, .
Proof
Analytically continuing the index of the integration operator in (0.3) and (0.5) in Formula 16.2.0 ( 16.2 ) to
[0 , p] [1 , n] , we obtain (0.1) and (0.2) in this theorem.
from
Since (1+ - m - k) = k =0, 1, 2, 3, when = m-1 m =0, 1, 2, , the remainder
term in (0.1) disappears, and is as follows.
(1+m -1) m-1- r
x x m -1 -p
f <0>x
m-1
dx p = Σ f <p + r>
x
a a r=0 r 1+m -1- r
Then, replacing m-1 with m, we obtain (0.1').
-4-
17.3 Super Integral of x^a f (x) (particulars)
In this section, substituting various functions f for Formula 17.2.0 in previous section, we obtain various
formulas. There are (1) and (2) in Formula 17.2.0, and we may also choose whichever. However, what we wont
is the expression or approximation of super integral of x f(x ) by the series. So, in the selection of (1) or (2),
we choose the way where such a well-behaved series is obtained.
Moreover, also in which formula, if = -1, -2, -3, , it shall read as follows.
(1+ ) (- + r)
(-1)-r r = r, s, m+ k
1+ - r (- )
Formula 17.3.1
The following expressions hold for p >0 and s >0 .
x x
(ax +b)p(cx +d)qdx s
b b
- -
a a
(-1)m 1 m+ k
(1+p)(1+q)
s -1 c
s
Rm = Σ
(s, m) k =0 m + k k a (1+p +m + k)(1+q -m - k)
x x
(ax +b)p+m+ k (cx +d)q-m- k dx s (1.1r)
b b
- -
a a
lim Rms = 0
m
x x
(ax +b)p (cx +d)mdx s
b b
- -
a a
3
Example1 The 1/2th order integral of x-2 3x+4
The zeros of this primitive function are x =-4/3 , 2 . If -4 / 3 is adopted, since x -2 is a complex
number, this higher integral becomes a complex function. It is inconvenient. Then if we assume x =2 as
the lower limit of the integral, since a =1, b =-2, p =1/2 , c=3, d =4, q =1/3 , n =1 , substituting
these for (1.1), (1.1r) , we obtain
1
x x
3 2
x -2 3x +4 dx
2 2
(3/2)(4/3)
1 1
m -1 -1/2 -r
=Σ 3r
(x -2)1+ r(3x +4) 3 + Rm2
r=0 r (2+ r)(4/3-r)
-5-
(-1)m (3/2)(4/3)
1 -1/2
Rms = Σ 3m+ k
(1/2, m) k=0 m + k k (3/2+m + k)(4/3-m - k)
1 4 1
(x -2)
x x +m+ k -m- k
2
(3x +4) 3
dx 2
2 2
When the left side is replaced with Riemann-Liouville Integral and given one arbitrary point x =3.3 ,
it is as follows.
a:=1: b:=-2: p:=1/2: c:=3: d:=4: q:=1/3: s:=1/2: m:=22:
Rie mann-Liouville Integral
g := x->1/gamma(s)*int((x-t)^(s-1)*(a*t+b)^p*(c*t+d)^q,t=-b/a..x)
x
1 s1 p q
x (x t) (a t b) (c t d) d t
(s)
b
a
Series
f := x-> sum(binomial(-s,r)*((1/a)^(s+r)/(1/c)^r)
*(gamma(1+p)*gamma(1+q)/(gamma(1+p+s+r)*gamma(1+q-r))
*((a*x+b)^(p+s+r)/(c*x+d)^(r-q)), r=0..m-1)
m1
(1 sr psr
s (1 p) (1 q)
a ) (a x b)
x 1 r (1 p s r) (1 q r) rq
r0
r (c) (c x d)
float(g(3.3)),float(f(3.3))
2.701947993, 2.701947994
Although this integral cannot be expressed with an elementary function, it can be expressed with a series.
Here is the meaning of this formula.
3 2
Example1' The 5 / 2th order integral of 3x+4 (x-2)
Since this is the example of application of (1.1') , we assume the noninteger root as the lower limit of the
integral. Since a =3, b =4, p =1/3 , c=1, d =-2, s =5/2 , substituting these for (1.1') , we obtain
5 17
5 +r +r
(4/3)(3)
x x 2 6
3
3 2
2 -5/2 1 (3x +4)
3x +4 x -22dx =Σ
4 4 r =0 r 23/6+ r(3-r) (x -2)r-2
- -
3 3
(4/3)(3)
3 2 4 5
3x +4 x -2 5 3 x +4 x -2 35 3 x +4
= - +
5 1 23/6 2! 23 29/6 1! 832 35/6 0!
2 6
3 3 x +4
When the left side is replaced with Riemann-Liouville Integral and given one arbitrary point x =2.1 ,
it is as follows.
a:=3: b:=4: p:=1/3: c:=1: d:=-2: m:=2: s:=5/2:
-6-
Series
f := x-> sum(binomial(-s,r)*((1/a)^(s+r)/(1/c)^r)*(gamma(1+p)
*gamma(1+m)/(gamma(1+p+s+r)*gamma(1+m-r)))
*((a*x+b)^(p+s+r)/(c*x+d)^(r-m)), r=0..m)
m
(1 sr psr
(1 p) (1 m)
a )
s (a x b)
x 1 r (1 p s r) (1 m r) (c x d)r m
r0
r (c)
float(g(2.1)),float(f(2.1))
44.54679172, 44.54679172
Formula 17.3.2
When (z), (z) are the gamma function and the psi function respectively, the following expressions
hold for , p such that + p >0 .
log x -(1+p+r) - (1+) +p
x r
x x
m -1 -p
log x dx p = Σ x + Rmp (2.1)
0 0 r=0 (1+p+r) (1+ -r)
log x -(1+m+k) - (1+) x
(p , m) Σ m + k k
m p -1 x x
(-1) 1
Rmp = dx p
k=0 0 0 (1+m+k) (1+ - m - k)
lim Rmp = 0
m
-7-
3
Example1' The 2.1th order integral of x log x
Substituting m =3 , p =2.1 for (2.1') , we obtain
Complete Automorphism
Observing well Formula 17.3.2 , we notice that the integral of the completely same type as the left side is
included in the remainder. In such a case, we can take out the integral of the purpose by transposition.
Formula 17.3.2'
The following expression holds for , p such that + p >0 .
x
x x
log xdx p
0 0
log x -(1+p) - (1+) p -1 (2+k) +
=
(1+p)
+p Σ
1+ + p k=0 k
(1+ k) (1+ k, - k) +p
2
x
p -1 1
1 + pΣ
(1+ k) (1+ k, - k)
2
k =0 k
(2.2)
Proof
Analytically continuing the index of the integration operator in Formula 16.3.2' ( 16.3 ) to [0 , p ] from
[1 , n] , we obtain the desired expression.
Formula 17.3.3
When a0 ap are the zeros of the lineal super primitive of x m sin x , the follwing expressions hold.
(1+ m) m- r (p + r)
a a r
-p
x x m
x m sin x dx p = Σ x sin x - (3.1s)
p 0
r=0 (1+ m - r) 2
(1+ m) (p + r)
r (1+ m - r) x
-p
x x m
x cosx dx = Σ m p m- r
cos x - (3.1c)
ap a0 r=0 2
-8-
Proof
Analytically continuing the index of the integration operator in Formula 16.3.3 (3.1') in 16.3.3 to [0 , p]
from[1 , n ] , we obtain (3.1s). Also, (3.1c) is obtained in a similar way.
3
Example The 7 / 2th order integral of x sinx
Substituting m =3 , p =7/2 for (3.1s) , we obtain
-7/2
x x 3
x sin xdx = Σ
3 2
x sin x -
ap a0 r=0 r (4- r) 2
This super integral can not be examined by Riemann-Liouville Integral. Then we draw this with the 3rd and
the 4th order integral on a figure side by side. It is as follows.
(1+ ) r
m -1 -p
x x
x sin xdx p = Σ r
x +p+ sin x + + Rmp (3.2s)
0 0 r=0 r 1+ + p + r 2
(1+) (m + k)
k
(-1)m 1 p -1 x x
Rmp = x dx
+m+ k p
(p , m) Σ
sin x +
k=0 m+ k 0 0 1+ +m +k 2
(1+ ) r
r
-p
+R
x x m -1
x cosxdx p = Σ
r p
x +p+ cos x + (3.2c)
0 0 r=0 1+ + p + r 2 m
1+ (m + k)
(-1)m 1
k 1+ +m +k
p -1 x x
dx
( )
Rmp = x
+m+ k p
(p , m) Σ
cos x +
k=0 m+ k 0 0 2
lim Rmp = 0
m
Proof
Substituting f(x)= sin x , a =0 , etc . for Formula 17.2.0 (2) , we obtain the desired expression.
3/2
Example Collateral the 1 / 2th order integral of x sinx
Substituting =3/2 , p =1/2 for (3.2s), we obtain
-9-
(5/2) 2+ r r
3 1
-1/2
+R
x x m -1
2
x sin xdx 2
=Σ x sin x + 1/2
0 0 r=0 r (3+ r) 2 m
When the left side is replaced with Riemann-Liouville Integral and given one arbitrary point x =3 ,
it is as follows.
(1+ m) m- r e x -(-1)p+ re -x
aa x r
x x m -p
m
sinh x dx = Σ
p
x (4.1s)
p 0
r=0 1+ m - r 2
(1+ m) m- r e x +(-1)p+ re -x
aa x r
x x m -p
m
cosh x dx = Σ
p
x (4.1c)
p 0
r=0 1+ m - r 2
Proof
Analytically continuing the index of the integration operator in Formula 16.3.4 (4.1') in 16.3.4 to [0 , p]
from[1 , n ] , we obtain (4.1s). Also, (4.1c) is obtained in a similar way.
3
Example The 0.999th order integral of x sinhx
Substituting m =3 , p =0.999 for (4.1s), we obtain
(1+3) 3- r e x -(-1)0.999+ re -x
aa x
x x m -0.999
3
sinh x dx = Σ
p
x
p 1
r=0 r 1+3- r 2
This super integral can not be examined by Riemann-Liouville Integral. Then we calculate on the arbitrary point
x =1 the value of the 0.999th order integral and the value of the 1st order integral. They are as follows.
Although the former becomes a complex number, both real number parts are very near.
- 10 -
Formula 17.3.4' (Collateral Super Integral)
, p such that -1, -2, -3, & + p >0 .
The following expressions hold for
(1+ )
x -r -x
x x m -1 -p
+p+ r e -(-1) e
x sinh x dx = Σ
p
x (4.2s)
0 0 r=0 r (1+ + p + r) 2
(1+)
x
k
-m-k -x
(-1)m 1 p -1 x x
+m+ k e -(-1) e
Rmp dx p
(p , m) Σ
= x
k=0 m+ k 0 0 1+ +m +k 2
1+
00 x cosh x dx = Σ r (1+( + p)+ r) x
e x +(-1)-re
-x
x x
-p p
m -1
+p+ r
+ Rmp (4.2c)
r=0 2
(1+) e x +(-1)-m-ke
m -x
m+ k k
(-1) 1 p -1 x x
p
x
+m+ k
R = Σ dx p
m
(p , m) k=0 1+ +m +k 0 0 2
lim Rmp = 0
m
Proof
Substituting f(x)= sinh x , a =0 , etc . for Formula 17.2.0 (2) , we obtain the desired expression.
(3/2) 2+ r e x-(-1)-re -x
3
x x m -1 -3/2
x sinh xdx =Σ 2
x + Rm3/2
0 0 r=0 r (3+ r) 2
When the left side is replaced with Riemann-Liouville Integral and given one arbitrary point x =2 ,
it is as follows.
- 11 -
17.4 Super Integral of log x f (x)
Formula 17.4.1
x p log x log x - (1+ p)-
x x
2 p
log xdx =
0 0 (1+ p)
-p log x - (1+ p +r)-(r)
m -1
- x p Σ (-1)r + Rmp (1.1)
r=1 r (1+ p +r)
xp
Rmp =
(p , m)(1+ p)
k-1
Proof
Analytically continuing the index of the integration operator in Formula 16.4.1 in 16.4.1 to [0 , p ] from
[1 , n ] , we obtain the desired expression.
The remainder in the super integral of product of two functions becomes a series of super integrals. So it is
2
very difficult to calculate this. However, the calculation of the remainder in the super integral of (log x) is
exceptionally easy. So we show it as follows.
2
Example The 3/2th order integral of (log x)
We substitute p =3/2 , m=3 for (1.1), (1.1r) and calculate the remainder 300 terms. The function values
on the arbitrary point x =0.3 are as follows. Although the difference of the left side and the series is large,
if the remainder is added, all digits in both sides are corresponding.
- 12 -
Incidentally, even if this series is calculated to m=300,000 , both are corresponding only to 4 digits below
the decimal point. From this, we know that the convergence of the series is extraordinarily slower than the
remainder.
- 13 -
17.5 Super Integral of e^x f (x)
Formula 17.5.1
t
When (a , x ) = x0
a-1 -t
e dt is the incomplete gamma function, the following expression holds for
x
(-1)
r (p -r + , -x)x
x x
x
p -1
x e dx = p
Σ r
(1.n)
- - (p) r=0
Proof
Analytically continuing the index of the integration operator in Formula 16.5.1 in 16.5.1 to [0 , p ] from
[1 , n] , we obtain the desired expression.
3
Example1 The 2.1th order integral of e x x
When =1/3 , p =2.1 , if the left side is replaced by Riemann-Liouville Integral and an arbitrary point
x = 1.7 is given to both sides, it is as follows.
When is a natural number, x can be differentiated completely and the following formula holds.
Formula 17.5.1'
When m =0, 1, 2, ,
(1+ m) m- r
r
x x m -p
e x x mdx p = e xΣ x (1.n')
- - r=0 (1+ m - r)
4
Example1' The 5.1th order integral of ex x
Substituting m=4 , n =5.1 for (1.n'), we obtain
(1+4) 4- r
x x 4 -5.1
e x x 4dx 5.1 = e xΣ x
- - r=0 r 1+4- r
If the values of the both sides on arbitrary point x =2 are calculated, it is as follows.
- 14 -
Collateral Super Integral of e x x
For the collateral super integral with the zero x=0, the following formula holds.
Formula 17.5.1"
The following expressions hold for , p such that
-1, -2, -3, & + p >0.
(1+ )
x x m -1 -p
e x x dx p = e xΣ r
x +p+ + Rmp (1.n")
0 0 r=0 r (1+ + p + r)
(1+ )
(-1)m 1
p -1 x x
p
Rm = Σ e x x + m+k dx p
(p , m) k =0 m + k k (1+ + m +k) 0 0
lim Rmp = 0
m
3
Example1" Collateral the 2.1th order integral of e x x
Assume =1/3 , p =2.1 and an arbitrary point x =1.7 are the same as Example1 ( Lineal 2.1 th
order integral ) . The values of the both sides are completely corresponding as follows. However, these are
considerably different from the value of Example1.
Formula 17.5.3
p
e sin xdx = sin 4 e sin x - 4
x x p
x p x
(3.0s)
- -
p
p
4 4
x x
e cosxdx x
= sin p
e cos x - x
(3.0c)
- -
- 15 -
Proof
Analytically continuing the index of the integration operator in Formula 16.5.3 in 16.5.3 to [0 , p] from
[1 , n] , we obtain the desired expressions.
Right: Formula
s := x-> sin(PI/4)^p*E^x*sin(x-p*PI/4)
p
p
x sin E sin x
x
4 4
Le ft , Right
float(g(2.3)), float(s(2.3))
7.916851572, 7.916851572
Trigonometric Series
If Formula 16.5.4 in 16.5.4 is extended to the real number, the following trigonometric series can be obtained
Formula 17.5.4
k -p
p
p -1
1 1
Σ
k =1 k k -1
sin
2
=
p
sin
4
sin
4
(4.1)
k -p
p
Σk
p -1
1 1 1
cos = sin cos - (4.2)
k =1 k -1 2 p 4 4 p
Formula 17.5.4'
(-1)k -p-1
(p +1)
p
1
Σ
k =0 2k +1 2k
=
p +1
sin
4
sin
4
(4.3)
(p +1)
p +1 4
(-1)k p 1 -p-1
Σ
k =1 2k
=
2k -1
sin cos
4
-1 (4.4)
If a horizontal axis is set as p and these are illustrated , it is as follows. Although the left side is blue and
right side is red, since both sides overlap exactly, the left side (blue) is not visible.
- 16 -
- 17 -
17.6 Super Integral of f (x) / e^x
e x
-x
17.6.1 Super Integral of
Since the zero of the lineal higher primitive of e -x x was , naturally, the zero of the lineal super
primitive is also .
Formula 17.6.1
t
When (a , x ) =
a-1 -t
e dt is the incomplete gamma function, the following expression holds.
x
(-1)p
x x
p -1
-x
e x dx = Σp
(-1)r (p -r + , x)x r (1.1)
(p) r=0 r
Proof
Analytically continuing the index of the integration operator in Formula 16.6.1 in 16.6.1 to [0 , p ] from
[1 , n] , we obtain the desired expression.
- 3
Example1 The 2.1th order integral of e x x
When =1/3 , p =2.1 , if the left side is replaced by Riemann-Liouville Integral and an arbitrary point
x = 1.7 is given to both sides, it is as follows.
When is a natural number, x can be differentiated completely. And the following formula holds.
Formula 17.6.1'
When m =0, 1, 2, ,
(1+ m) m- r
(-1)p
r
x x m -p
-x
e x dx = m p
Σ (-1) r
x (1.1')
ex r=0 (1+ m - r)
- 4
Example1' The 5 .1 th order integral of e x x
Substituting m =4 , p =5.1 for (1.1'), we obtain
(1+4) 4- r
(-1)5.1
x x 4 -5.1
-x
e x dx 4 5.1
= Σ(-1)
r
x
e x r=0 r (1+4- r)
If the values of the both sides on arbitrary point x =1.3 are calculated, it is as follows.
- 18 -
Collateral Super Integral of e x x
-
For the collateral super integral with the zero x=0, the following formula holds.
Formula 17.6.1"
The following expressions hold for , p such that -1, -2, -3, & + p >0.
(1+ )
x x
1 m -1 -p
-x
e x dx = p
Σ (-1)
-r r
x +p+ + Rmp (1.1")
0 0 e x r=0 r (1+ + p + r)
+ m+k
(1+ )
k (1+ + m +k)
-k p -1 x x
1 n -1 (-1) x
(p , m) Σ
Rmp = x dx p
k =0 m + k 0 0 e
lim R n = 0
m m
- 3
Example1" Collateral the 2.1th order integral of e x x
Assume =1/3 , p =2.1 and an arbitrary point x =1.7 are the same as Example1 ( Lineal 2.1 th order
integral ). The values of the both sides are completely corresponding as follows. However, these are different
from the value of Example1 considerably.
-x
17.6.2 Super integral of e log x
All the polynomials obtained by applying Theorem 17.1.2 to e -xlog x become asymptotic expansions, and
they are hardly helpful.
-x -x
17.6.3 Super integral of e sinx , e cosx
Formula 17.6.3
p
p
x x
e sin xdx = (-1)p sin
-x p
e -xsin x + (3.0s)
4 4
p
e
p
x x
-x
cosxdx p = (-1)p sin e -xcos x + (3.0c)
4 4
- 19 -
Proof
Analytically continuing the index of the integration operator in Formula 16.6.3 in 16.6.3 to [0 , p] from
[1 , n] , we obtain the desired expressions.
-
Example The 4 /3th order integral of e x cos x
If the values of the both sides on arbitrary point x =1.2 are calculated, it is as follows.
- 20 -
17.7 Series Expansion of Super Integral
Theorem 17.7.1
m be a natural number, f r =0, 1, 2, be the r th derivative of f , a be an arbitrary constant
(r)
Let
on the domain of f , (x, y) be the beta function. Then the following expressions hold for p >0 .
(x -a)p+ r (r)
x x m -1 -p
f(x)dx = Σ p
f (x) + Rmp (1.1)
a a r=0 r (1+ p + r)
(-1)m 1 p -1 x t (t -a )m+ k
m+ k
(p , m) Σ
p
Rm = f ( )(t)dt p (1.1r)
k =0 m + k k a a ( m + k) !
f(x)dx
x x m -1 (x -a)p+ r (r)
p
=Σ f (a) + Rmp (1.2)
a a r=0 (1+p + r)
r -1 (x -a ) m+p+s
Σ C
m +p +s r -1 m + p -1-r + t
Rmp = ΣΣ(-1) p-r+s
f m +s(a)
r =1 s =0 r t=s m -1
t s
1+m +p +s
(-1)m 1 p -1 x (t -x)m+ k (m+ k )
t
(p , m) Σ
+ f (t)dt p (1.2r)
k =0 m + k k a a (m + k)!
Where, there shall be no term of ΣΣ at the time of p <2 .
Proof
Analytically continuing the index of the integration operator in Formula 16.7.1 in 16.7.1 to [0 , p] from
[1 , n] , we obtain the desired expressions.
Thorem 17.7.1'
p and an arbitrary number a on the domain of f .
The following formulas hold for a positive number
(x -a)p+ r (r)
1 x m -1 -p
p-1
(x -t) f(t)dt = Σ f (x) + Rmp (1.1')
(p) a r=0 r (1+p + r)
x (x -t)p-1(t -a )m+ k
(-1)m 1 p -1 m+ k
p
Rm = Σ
(p , m)(p) k =0 m + k k a (1+m + k)
f ( )(t)dt p
(x -a)p+ r (r)
(x -t)
1 x m -1
p-1
f(t)dt = Σ f (a) + Rmp (1.2')
(p) a r=0 (1+p + r)
(x -a) m+p+s
t=s s
r -1 t
1+m +p +s
r -1 m + p -1-r + t m +p +s m+ s)
Rmp = ΣΣ(-1) p-r+s
Σ f( (a )
r =1 s =0 m -1 r
- 21 -
Example1 Series expansion of collateral the 2.1th order integral of ex
Let f(x)= e x , then
r =1, 2, 3,
r () (r)
f(x)= e x , f (x)= e x , f (0)= 1
Substituting these for Theorem 17.7.1 ,
e dx
(x -a)p+ r
x x m -1 -p
a a
x p
=e x
Σ
r=0 r (1+ p r
+ )
+ Rmp
(-1)m 1 p -1 x t (t -a )m+ k
(p , m) Σ
Rmp = e tdt p
k =0 m k
+ k a a ( m k
+ )!
x x m -1 (x -a)p+ r
e dx = e aΣ
x p
+ Rmp
a a r=0 (1+p + r)
(x -a) m+p+s
t=s s
r -1 t
1+m +p +s
r -1 m + p -1-r + t m +p +s
Rmp = ΣΣ(-1) Σ
p-r+s
ea
r =1 s =0 m -1 r
k
(-1)m 1 p -1 x t (t -x)m+ k t p
(p , m) Σ
+ e dt
k =0 m + k a a (m + k)!
x
Although the zero of the lineal super integral of e is a = - , these expressions can not have this zero.
That is, by the above formulas, the lineal super integral of e x can not be expanded to the series. Then, we put
a =0 and expand the collateral super integral of e x to the series. When p =2.1 , if these are illustrated,
it is as follows. The left side is calculated by Riemann-Liouville Integral. Both sides overlap exactly and blue (left)
can not be seen.
- 22 -
Substituting these for Theorem 17.7.1 ,
(x -a)p (x -a)p+ r (r -1)!
x x m -1 -p
log x dx = p
log x -Σ (-1)r
+ Rmp
a a (1+ p) r=1 r (1+ p + r) x r
p+ r
x x (x -a)p m -1
r (x -a ) (r -1)!
log x dx =
p
log a - Σ (-1) + Rmp
a a ( p)
1+ r=1 ( p r) a
1+ + r
If a =0 then the first formula is lineal super integral, else it is collateral super integral. On the other hand,
the second formula can not be lineal super integral.
Let us calculate the left side by the direct integration and compare it with the series by the first formula. Then
it is as follows. Although the convergence of this series is slow, as a result of taking very large m, both sids
overlapped somehow.
2010.07.14
K. Kono
Alien's Mathematics
- 23 -
18 Higher Derivative of the Product of Two Functions
r
(n) n n (n -r)
(x)g r (x)
()
f(x)g (x) =Σ f (1.1)
r=0
Proof
Theorem 16.1.2 (2.1) in 16.1.2 was as follows.
a a r
x x m -1 -n
f <0>g(0)dx n = Σ f <n + r> (r)
g
r=0
n 1
s f a a
-n + r
n -1 m -1 x x
- ΣΣ <n -r+ s>
a n-r ga
(s)
dx r
r=0 s=0 n-r
n n- r+1
a a
n -1 r -1 r -1 x x
t s m +n-1 -r + tC m-1 f a
m+ n -r+s (m+ s)
+ (-1)m ΣΣΣ C ga dx r
r =1 s =0 t=s n- r n- r
n n- r+1
a a f
m
(-1) Ck n -1 n -1 x x
<m+ k > (m+ k )
+ Σ
(n , m) k=0 m + k
g dx n
n 1
a a r
x x m -1 n
f <0>g(0)dx -n = Σ f <-n + r> (r)
g
r=0
n 1
f
(-1)m 1
-n -1 x x
<m+ k > (m+ k )
(-n , m) Σ
+ g dx -n
k=0 m + k k an a1
Since m may be arbitrary integer, when m = n +1 , it is as follows.
a a r f
x x n n
f <0>g(0)dx -n = Σ <-n + r> (r)
g
r=0
n 1
a a f
(-1)n+1
1 -n -1 x x
<n +1+ k > (n +1+ k )
+ Σ
(-n , n +1) k =0 n +1+ k
k
g dx -n
n 1
However, since (-n , n +1)= for n =0, 1, 2, , the 2nd line disappears. That is,
x x n n <-n + r> (r)
f g dx -n = Σ f g n =0, 1, 2,
an a1 r=0 r
-n
Then, replacing the integration operators dx , <-n + r> with the differentiation operators (n ),(n -r)
respectivly, we obtain the desired expression.
-1-
18.2 Higher Derivative of x^a f (x)
Formula 18.2.0
When (z) denotes the gamma function and f(x) is n times differentiable continuous function,
the following expressions hold for a natural number n.
(1)
(1+ ) - r (n - r)
r
(n) n n
x f(x) =Σ x f (x) (0.1)
r=0 (1+ -r)
Where, if = -1, -2, -3, , it shall read as follows.
(1+ ) (- + r)
(-1)-r
1+ - r (- )
(2) Especially, when = m =0, 1, 2,
(1+ m) m- r (n - r)
(n) m n
x m f(x) = Σ x f (x) (0.1')
r=0 r (1+ m -r)
(3) When -1, -2, -3, & -n -1, -2, -3,
(1+ )
(n) n n
x f(x) = Σ
r (r)
x -n+ f (x) (0.2)
r=0 r (1+ - n + r)
Proof
When g(x)= x in Theorem 18.1.1 , since
(r) (1+ ) - r
x = x
(1+ -r)
we obtain the following expression immediately.
(1+ ) - r (n - r)
r
n n
x f(x)
(n)
=Σ x f (x) (0.1)
r=0 (1+ -r)
Especially, when = m =0, 1, 2, , (0.1) is as follows.
(1+m) (1+ m)
r
(n) n n
m
x f(x) x m- r f n- r (x) = 0 for m < r n
( )
=Σ
r=0 (1+m -r) (1+ m -r)
(1+m)
r (1+m -r) x r =0
nm n
=Σ m- r (n- r )
f (x) for n < r m
r=0
We adopt the convenient latter for mathematical software.
When = -1, -2, -3, , from 1.1.5 ( Properties of the Gamma Function ) (5.5),
(-z) (1+z + n)
= (-1)-n (n is a non-negative integer )
(-z - n) (1+ z)
Then substituting -z = 1+ , n = r for this, we obtain the proviso.
Last, replacing r with n -r in (0.1), we obtain (0.2).
Below, substituting various functions f for Formula 18.2.0 , we obtain various formulas.
Although there are (1) and (2) in Formula 18.2.0, since (2) is almost meaningless in the case of higher
differentiation, we adopt (1) in principle.
-2-
18.2.1 Higher Derivative of (ax+b)p(cx+d)q
Formula 18.2.1
The following expressions hold for p >0 and n =1, 2, 3, .
q (n)
(ax +b)p(cx +d)
(1/a)-n+ r (1+p)(1+q) (ax +b)p-n+r
n n
=Σ (1.1)
r=0 r (1/c)r (1+p -n + r)(1+q -r) (cx +d)r-q
Especially, when m =0, 1, 2,
(n)
(ax +b) (cx +d)
p m
Proof
Let f(x)=(ax +b)p , g(x)=(cx +d)q , then
-n+r
(1+p)
n - r) (n -r) 1
f( = (ax +b)p = (ax +b)p-n+r
a (1+p -n + r)
-r
(1+q)
(r) 1
q 1, 2, 3,
(r)
g = (cx +d)q = (cx +d)q-r
c (1+q - r)
Substituting these for Theorem 18.1.1 , we obtain (1.1) .
And especially, when q = m = 0, 1, 2, , from (1.1)
(n)
(ax +b)p(cx +d)m
(1/a)-n+ r (1+p)(1+m) (ax +b)p-n+r
r
n n
=Σ
r=0 (1/c)r (1+p -n + r)(1+m -r) (cx +d)r-m
(1/a)-n+ r (1+p)(1+m) (ax +b)p-n+r
r
m n
=Σ
r=0 (1/c)r (1+p -n + r)(1+m -r) (cx +d)r-m
We adopt the latter expression as (1.1').
3
Example1 The 2nd order derivative of x-2 3x+4
a =1, b =-2, p =1/2 , c =3, d =4, q =1/3 , n =2 for (1.1) ,
Substituting
2 r (3/2)(4/3) 3 1
(2) 2 r- -r
x -2 3x +4 =Σ
3
3 (x -2) (3x +4) 3
2
r=0 r (r -1/2)(4/3-r)
1 1 2
= - x -2 3 3x +4 - +
4(x -2)2 (x -2)(3x +4) (3x +4)2
2
Example1' The 3rd order derivative of x-2 (3x+4)
Substituting a =1, b =-2, p =1/2 , c =3, d =4, m =2 , n =3 for (1.1') ,
-3-
5
- +r
(3/2)(3) 2
r
(3) 2 3 (x -2)
x -2 (3x +4)
2
=Σ 3r
r=0 (-3/2+r)(3-r) (3x +4)r-2
8(x -2)
1
3 2 32 33
= x -2 (3x +4) 3
- 2
+ 2
2(3x +4) (x -2) (3x +4) (x -2)
r
x -2 3 3
=Σ (-3)r
3x +4 r=0 (-3/2+r)(1) (3x +4)r+1
x -2
8(x -2)
3 9 27 162
= 3
+ 2
+ 2
- 3
3x +4 4(x -2) (3x +4) (x -2) (3x +4) (3x +4)
Formula 18.2.2
(n -r) (1+) -n (1+) -n
r
(n) n -1 n
x log x = -Σ(-1)n-r x + x log x
r=0 (1+ -r) (1+ -n) (2.1)
Especially, when m = 0, 1, 2,
(n -r) (1+m) m-n (1+m) m-n
r
m (n) n -1 n
x log x = -Σ(-1)n-r x + x log x
r=0 (1+m -r) (1+m -n)
(2.1')
Where, there shall be no 2nd term of the right side at the time of m< n .
Proof
Let f(x)= log x . Then
(log x) (n -r)
= -(-1)n-r(n -r) x -n+r r = 0, 1, , n -1
= log x r =n
Substituting these for (0.1) in Theorem 18.2.0 , we obtain (2.1).
When m = 0, 1, 2, , applying (0.1') , we obtain the following.
(1+m)
r (1+m -r) x (log x)
nm
m-n (n -r)
+Σ
r=n +1
Where, since r does not reach n at the time of m< n , the 2nd term does not exist.
-4-
Example1 The 3rd order derivaive of x log x
Substituting =1/2 , n =3 for (2.1) ,
5 5
(2-r) (3/2) - 2 (3/2) - 2
r
(3) 2 3
x log x = -Σ(-1) 3-r
x + x log x
r=0 (3/2-r) (-3/2)
5
(3/2) (3/2) (3/2)
3 3 3 -
2
=- - 2! + 1! - 0! x
0 (3/2) 1 (1/2) 2 (-1/2)
5
(3/2) -
2
+ x log x
(-3/2)
5 5 5
(3/2) - 2
1 3 -
3 3 -
= 2- - x +
2
x log x = - + log x x 2
2 4 (-3/2) 4 8
3
Example1' The 2nd order derivaive of x log x
Substituting m =3 , n =2 for (2.1') ,
(2-r)(4) 1 (4) 1
r
(2) 1 2
x log x
3
= -Σ(-1)2-r x + x log x
r=0 (4-r) (2)
0 3! 1 2! x
2 1! 3! 2
2 0! 3! 1 1 3! 1
= - (-1) + (-1) + x log x
1!
= (5 + 6 log x)x 1
2
Example1" The 3rd order derivaive of x log x
Substituting m =2 , n =3 for (2.1') ,
(3-r)(3) -1
r
(3) 2 3
x log x = -Σ
2
(-1)3-r x
r=0 (3-r)
0 1 2 (-1) 2! x
3 3 3
3 2 1 -1
=- (-1) 2! + (-1) 2! +
= 2 x -1
Example2 The 3rd order derivaive of log x /x
When = -1, -2, -3, , (2.1) can be read as follows.
(n -r) (- + r) -n (- + n) -n
r
(n) n -1 n
x log x = -(-1)nΣ x + (-1)n x log x
r=0 (-) (-)
Substituting =-1 , n =3 for this ,
(3)
(3-r)(1+ r) -4 (4) -4
r
3
log x 2
= -(-1)-3Σ x + (-1)-3 x log x
x r=0 (1) (1)
0 1 2 (1)(3)- 6 log x
3 3 3
= x -4 (3)(1)+ (2)(2)+
1
= (11 - 6 log x)
x4
-5-
18.2.3 Higher Derivatives of x sinx , x cosx
Formula 18.2.3
(1+ ) -r (n -r)
r
n
{n} n
x sin x =Σ x sin x + (3.1s)
r=0 (1+ -r) 2
(1+ ) -r (n -r)
r
n
{n} n
x cosx =Σ x cos x + (3.1c)
r=0 (1+ -r) 2
Especially, when m = 0, 1, 2,
(1+m) m-r (n -r)
n
{n} m
x m sin x =Σ x sin x + (3.1's)
r=0 r (1+m -r) 2
(1+m) m-r (n -r)
n
{n} m
x m
cosx =Σr=0 r (1+m -r)
x cos x +
2
(3.1'c)
3
Example1 The 2nd order derivative of x sinx
Substituting =1/3 , n =2 for (3.1s) ,
(4/3) 3 -r (2-r)
1
r
2
(2) 2
x sin x
3
=Σ x sin x +
r=0 4/3-r 2
1 2 5
4/3 3 4/3 - 3 4/3
0 2 -2/3
2 2 2 -
= x sinx ++ x sin x + + x 3
sin x
4/3 1 1/3 2
1 2 5
2 - 2 -
= - x sin x + x 3 cosx - x 3 sin x
3
3 9
2
Example1' The 3rd order derivative of x sinx
m =2 , n =3 for (3.1's) ,
Substituting
Formula 18.2.4
(1+ ) - r e x -(-1)-(n -r)e -x
r
(n) n n
x sinh x =Σ x (4.1s)
r=0 (1+ -r) 2
(1+ ) - r e +(-1)-(n -r)e -x
x
r
(n) n n
x cosh x =Σ x (4.1c)
r=0 (1+ -r) 2
-6-
Especially, when m = 0, 1, 2,
(1+ m) m- r e x -(-1)-(n -r)e -x
(n) m n
x sinh x = Σ
m
x (4.1's)
r=0 r (1+ m -r) 2
(1+ m) n -r) -x
e x +(-1)-(
(1+m -r) x
(n) m n m- r e
x m cosh x = Σ (4.1'c)
r=0 r 2
3
Example1 The 2nd order derivaive of x sinhx
Substituting =1/3 , n =2 for (4.1s) ,
(4/3) 3 -r e x -(-1)-(2-r)e -x
1
r
(2) 2 2
x sinh x =Σ
3
x
r=0 4/3- r 2
1 2 5
(4/3) 3 (4/3) - 3 (4/3) - 3
0
2 2 2
= x sinh x + x cosh x + x sinh x
4/3 1 1/3 2 -2/3
1 2 5
3 2 -3 2 -
= x sinh x + x cosh x - x 3 sinh x
3 9
2
Example1' The 3rd order derivative of x sinhx
m =2 , n =3 for (4.1's) ,
Substituting
-7-
18.3 Higher Derivative of log x f (x)
Formula 18.3.1
(-1)n-1
r (n - r)(r)
(n) n -1 n
log x
2
= 2(n)log x -Σ (1.1)
xn r=1
Proof
Let f(x)= g(x)= log x . Then
(n - r) (r)
n =1, 2,
(n- r) (r)
(log x) = (-1)n- r-1 , (log x) = (-1)r-1
x n- r xr
Substituting these for Theorem 18.1.1 ,
r
(n) n n
log x
2
=Σ (log x)(n - r)(log x)( r)
r=0
n (log x) (log x)
n (n - n ) (n)
+
n n -1
Σ r (n - r)(r)
2 (-1) n
= (-1)n-1 n
(n)log x + n
x x r=1
(-1)n-1
r (n - r)(r)
n n -1
= 2(n)log x -Σ
xn r=1
2
Example The 3rd order derivative of (log x)
(-1)3-1
r (3- r)(r)
(3) 2 3
log x
2
= 2(3)log x -Σ
x3 r=1
x 1 2(1)(2)
1 3 3
= 22log x -
3
(2)(1)-
1
= (4log x -6)
x3
Formula 18.3.2
n
(n)
(log xsin x) = log x sin x +
2
(r) (n -r)
r
n
n
+Σ(-1)r-1 sin x + (2.0s)
r=1 xr 2
-8-
n
(n)
(log xcos x) = log xcos x +
2
(r) (n -r)
r
n
n
+Σ(-1)r-1 cos x + (2.0c)
r=1 xr 2
3 (3) 0
+
3 x 2
sin x + 3
3 3 2
= -log x cosx - 1 sin x
- 2 cosx
+ sin x
x x x3
Formula 18.3.3
(n) e x -(-1)-ne -x
log xsinh x = log x
2
(r) e x -(-1)r-ne -x
r
n
r-1 n
+Σ(-1) (3.0s)
r=1 xr 2
(n) e x +(-1)-ne -x
log xcosh x = log x
2
(r) e x +(-1)r-ne -x
r
n
r-1 n
+Σ(-1) (3.0c)
r=1 xr 2
4 (3) 4 (4)
+
3 x sinh x -
4 x cosh x
3 4
4 6 8 6
= log x cosx + 1 sinh x
- 2 cosh x
+ 3 sinh x
- cosh x
x x x x4
-9-
18.4 Higher Derivative of e^x f (x)
Proof
(n-r )
Substite f(x)= e x for Theorem 18.2.0 . Then since e x = ex, we obtain the desired expression
immediately.
r
(2) 2 2
e x
x =e x
Σ
r=0 3/2- r
x
0
1 1 3
(3/2) 2 (3/2) - 2 (3/2) - 2
1 2
x
2 2 2
=e x + x + x
3/2 1/2 -1/2
= e x + x = e x 1+ x - 4 x
1 1 3
x 2
-
2 1 - x 1 1
- x 2 2
4
x x
e 2 2
= 1- + 2
x
- 10 -
(r)
(log x) r = (-1)r-1 r = 1, 2, 3,
()
xr
Substituting this for Theorem 18.1.1 ,
r
(n) n n (r) n n
(r)
e x log x =Σ e x (log x) = e x (log x)(0) +Σe x (log x)
r=0 0 r=1
n (r)
n
= e x log x + e xΣ(-1)r-1
r=1 r xr
x x x x
x 4 6x 8 6
= e log x + e - + - 1 2 3 4
Formula 18.4.3
-n
n
(n)
e sin x
x
= sin e xsin x + (3.0s)
4 4
-n
n
(n)
e cosx
x
= sin e xcos x + (3.0c)
4 4
Proof
"共立 数学公式" p187 was posted as it was.
Example
-2
2
= 2e cosx
(2)
e sin x
x
= sin e xsin x + x
4 4
3-3
= sin e cos x +
4
(3)
e cosx
x x
= -2e (sin x + cosx) x
4
Higher Derivatives of e xsin x, e xcos x end now. There is no necessity for Theorem 18.1.1.
However, daring use Theorem 18.1.1, we obtain an interesting result.
Trigonometric Polynomial
Formula 18.4.3'
r -n
n
r
n
n
Σ
r=0
sin x +
2
= sin
4
sin x +
4
(3.1s)
- 11 -
r -n n
n
n
Σr=0 r
cos x +
2
= sin
4
cos x +
4
(3.1c)
Especially, when x =0
r -n n
n
n
Σr=0 r
sin
2
= sin
4
sin
4
(3.1's)
r -n
n
Σ r cos 2 = sin 4
n
n
cos (3.1'c)
r=0 4
Proof
Substitutingf(x)= e x , g(x)= sin x, cosx for Theorem 18.1.1 ,
r
n
(n) n
e x sin x = e xΣ sin x +
r=0 r 2
r
n
(n) n
e x cos x = e xΣ cos x +
r=0 r 2
And comparing these with Formula 18.4.3 , we obtain the desired expressions.
When n =5 , if both sides of (3.1s) are illustrated, it is as follows. Both overlap exactly and blue (left) can
not be seen.
Formula 18.4.3"
When
denotes the floor function, the following expressions hold.
n
n
2r +1 = 2 sin 4
(n -1)/2 n
Σ r 2
(-1) (3.2s)
r=0
n
n
Σ (-1) 2r
n /2 n r 2
= 2 cos (3.2c)
r=0 4
- 12 -
Proof
Since the odd-numbered terms of the left side in (3.1's) are all 0,
r 0 1 2 n
r n
n n n n n n
Σ
r=0
sin
2
=
0
sin
2
+
1
sin
2
+
2
sin
2
++ sin
2
2
n-1
1 3 5 n
n n n
= sin + sin + sin + n-1 sin
2
1 2 3 2 5 2 2
n
n n n (n -1)/2 n
= - + - n-1 = Σ (-1)r
1 3 5 2 r=0 2r +1
Also, since sin /4 = 2n/2 in the right side in
-n
(3.1's) ,
n
n
(n -1)/2 n
Σ (-1)r = 2 2 sin (3.2s)
r=0 2r +1 4
Next, since the even-numbered terms of the left side in (3.1'c) are all 0,
r 0 1 2 n
r n
n n n n n n
Σ
r=0
cos
2
=
0
cos
2
+
1
cos
2
+
2
cos
2
++ cos
2
0 2 4 n /2
0 n /2
n n n n
= cos + cos + cos + cos
2 2 2 4 2 2
n n n n n /2 n
= - + - = Σ (-1)r
0 2 4 n /2 r=0 2r
Also, since sin /4 = 2n/2
-n
in the right side in (3.1'c) ,
n
n
n /2 n
Σ (-1) r
= 2 cos 2
(3.2c)
r=0 2r 4
In addition, this formula is known. (See "岩波 数学公式Ⅱ" p11)
Note
When n = 4k -3 , k =1, 2, 3,
(n -1)/2 n n /2 n
Σ
r=0
(-1)r
2r +1
= Σ
r=0
(-1)r
2r
Example
5
5
1 3 5
5 5 5 2
- + = 5 - 10 + 1 = 2 sin = -4
4
5
5
0 2 4 = 1 - 10 + 5 = 2
5 5 5 2
- + cos = -4
4
- 13 -
e x + (-1)-re -x
r
(n) n n
e cosh x
x
=e x
Σ
r=0 2
(4.0c)
Example
e x - (-1)-re -x
r
(0) 0 0
e sinh x
x
=e x
Σ
r=0 2
= e xsinh x
x -r -x
r
(3) 3 e + (-1) e
3
e cosh x
x
=e Σ x
r=0 2
0 1 2 3 cosh x
x
3 3 3 3
=e sinh x + cosh x + sinh x +
= 4e x(sinh x + cosh x)
Note
The following formula is known for a natural number n .
(n) (n) n-1 x
e sinh x = e cosh x = 2
x x
e (sinh x + cosh x)
However, this formula does not hold for n =0 . That is, in this formula, the natural number n is inextensible
to the real number p . So, this is insufficient as a general formula.
- 14 -
18.5 Higher Derivative of f (x) / e^x
e x
-x
18.5.1 Higher Derivative of
Formula 18.5.1
(1+)
r
-x
n n
x r for -1, -2, -3,
(n)
e x Σ
-x -
=e (-1)-(n-r )
(1.1)
r=0 (1+ -r)
n (- + r)
Σ r (-) x
n
- r
for = -1, -2, -3,
-x
= (-1)-ne (1.2)
r=0
Especially, when m = 0, 1, 2,
(1+ m) m- r
r
(n) m n
e -x x m = e -xΣ(-1)-(n -r) x (1.1')
r=0 (1+m -r)
Proof
-x (n-r )
e
-x -x
Substite f(x)= e for Theorem 18.2.0 . Then since = (-1)-(n-r)e , we obtain the
desired expression immediately.
-
Example1 The 3rd order derivative of e x/x
(3)
(1+r) -1-r
e -x
r
3 3
= (-1)-3e -xΣ x
x r=0 (1)
(1) -1 3 (2) -2 3 (3) -3 3 (4) -4
0
-x
3
= -e x + x + x + x
(1) 1 (1) 2 (1) 3 (1)
e -x
3 6 6
=- 1+ + 2 + 3
x x x x
- 7
Example1' The 3rd order derivative of e x x
(8) 7- r
r
(3) 7 3
e -x x 7 = e -xΣ(-1)-(3-r) x
r=0 (8-r)
(8) 7 3 (8) 6 3 (8) 5 3 (8) 4
0
3
= e -x - x + x - x + x
8 1 7 2 6 3 5
= e -xx 4-x 3 + 21 x 2 - 126 x 1 + 210
-x
18.5.2 Higher Derivative of e log x
Formula 18.5.2
(-1)- n (r)
r
(n) n n
e log x log x - Σ
-x
= (2.1)
ex r=1 xr
Proof
Let f(x)= e -x , g(x)= log x . Then
- 15 -
-x (n -r)
e = (-1)- n+re -x
(r)
(log x) r = (-1)r-1 r r = 1, 2, 3,
()
x
Substituting these for Theorem 18.1.1 ,
r
(n) n n
e -x log x (-1)- n+re -x(log x) r
()
=Σ
r=0
0 r (-1)
n n - n -x
n
- n+r -x r()
= (-1) e (log x) +Σ (0)
e (log x)
r=1
n (r)
-n
log x - Σ
x
(-1) n
= x r
e r r=1
-
Example The 4th order derivative of e xlog x
(-1)- 4 (r)
r
(4) 4 4
e log x
x
= x log x - Σ
e r=1 xr
(1) (2) (3) (4)
1 x 2 x 3 r
log x 1 4 4 4 4
= - + + +
ex ex
1 2 3 4
x x
x x x x
log x 1 4 6 8 6
= - + + +
ex ex 1 2 3 4
-x -x
18.5.3 Higher Derivatives of e sinx , e cosx
Formula 18.5.3
-n
n
(n)
e sin x
-x
= -sin e -xsin x - (3.0s)
4 4
-n
n
cos x -
4
(n)
e cosx
-x
= -sin e -x (3.0c)
4
Proof
Replacing x with -x in Formula 18.4.3 , we obtain the desired expressions.
Example
2
-2
= -2e
(2)
e sin x
-x
= -sin e -xsin x - -x
cosx
4 4
-3
3
= -2e
(3)
e cosx
-x
= -sin e -xcos x - -x
(sin x - cosx)
4 4
-x -x
Higher Derivatives of e sinh x , e cosh x end now. There is no necessity for Theorem 18.1.1.
Daring use Theorem 18.1.1, we obtain the following expression first.
r
r
n
(n) n
e -x sin x = e -xΣ(-1)n-r sin x +
r=0 2
- 16 -
And from this and (3.0s) , we obtain
r -n
n
r
n
n
Σ(-1)
-r
sin x + = sin sin x -
r=0 2 4 4
r r
A similar expression is obtained about e -xcosx too. Then removing sin
2
, cos from these, we obtain
2
the completely same results as Formula 18.4.3" .
-x -x
18.5.4 Higher Derivatives of e sinhx , e coshx
Formula 18.5.4
e x - (-1)-re -x
r
(n) n n
e sinh x
-n+ r
Σ
-x -x
=e (-1) (4.0s)
r=0 2
x -r -x
r 2
(n) n n e + (-1) e
e cosh x
-n+ r
Σ
-x -x
=e (-1) (4.0c)
r=0
Proof
-x
Substituting f(x)= e , g(x)= sinh x , cosh x for Theorem 18.1.1 , we obtain the dsired expressions.
Example
e x - (-1)-re -x
r
(0) 0 0
e sinh x
-0+ r
Σ
-x -x
=e (-1) = e -xsinh x
r=0 2
x -r -x
r 2
(3) 3 3 e + (-1) e
e cosh x
-x
=e -x
Σ (-1 ) -3+ r
r=0
Note
The following formula is known for a natural number n .
(n) (n)
e sinh x = e cosh x = (-2)n-1e x(cosh x - sinh x)
-x -x
However, this formula does not hold for n =0 . That is, in this formula, the natural number n is inextensible
to the real number p . So, this is insufficient as a general formula.
- 17 -
18.6 Higher Derivatives of sin x f (x), cos x f (x)
2 2
18.6.1 Higher Derivatives of sin x , cos x
Formula 18.6.1
n
(n)
sin x
2
= -2n-1cos 2x + (1.0s)
2
n
(n)
cos x
2
= 2n-1cos 2x + (1.0c)
2
Proof
From Formula 18.6.1' mentioned next ,
(n - r) r
r
n
(n) n
cos x =Σ
2
cos x + cos x +
r=0 2 2
Here
1
cosA cosB = cos(A +B )+cos(A -B)
21
Using this,
n n
1 n n
(n)
cos x
2
= Σ cos 2 x + + cos -r
21 r=0 r 2 2
n n n
n n
1 n 1
= 1 cos 2x +
2
Σ
2 r=0 r
+ 1 cos
2
Σ
2 r=0
(-1)r
r
And since
r r = 0
n n n n
Σ
r=0
= 2n , Σ
r=0
(-1)r
substituting these for the above, we obtain (1.0c). (1.0s) is also obtained in a similar way.
Example
2
(2)
sin x = 2 cos 2x = 2cos 2x - sin 2x
2
= -22-1cos 2x +
2
3
= 4 sin 2x = 8 sin x cosx
(3)
cos x
2
= 23-1cos 2x +
2
Formula 18.6.1'
(n - r) r
r
n
(n) n
sin x =Σ
2
sin x + sin x + (1.1s)
r=0 2 2
(n - r) r
r 2 2
(n) n
n
cos x
2
=Σ cos x + cos x + (1.1c)
r=0
Proof
Substituting f(x)= g(x)= sin x for Theorem 18.1.1 , we obtain (1.1s). (1.1c) is also obtained in a similar
way.
- 18 -
Formula 18.6.1"
When denotes the floor function, the following expressions hold.
n /2 n
Σ = 2n-1 (1.2e)
r=0 2r
Σ 2r +1 = 2
n
(n -1)/2
n-1
(1.2o)
r=0
Proof
(1.1s) is transformed as follows.
(n - r) r
r
n
(n) n
sin x
2
=Σ sin x + sin x +
r=0 2 2
(n -2r) 2r
2r 2 2
n
n /2
=Σ sin x + sin x +
r=0
(n -2r -1)
2r +1
n
cosx
(n -1)/2
+ Σ (-1) sin x + r
r=0 2
i.e.
n n
n n
(n) n /2 (n -1)/2
sin x
2
= sin x + sin xΣ - cos x + cosx Σ
2 r=0 2r 2 r=0 2r +1
On the other hand, (1.0s) is transformed as follows too.
n n
(n)
sin x
2
= 2n-1sin x + n-1
sin x - 2 cos x + cosx
2 2
From these, the following expression follows .
n n
Σ 2r +1 - 2
n /2 n (n -1)/2 n
sin x +
2 sin x
r=0 2r
- 2n-1 = cos x + 2 cos x Σ
r=0
n-1
In order to hold this equation for arbitrary x , the followings are necessary.
2r +1 - 2
n /2 n (n -1)/2 n
Σ
r=0 2r
- 2n-1 = 0 , Σ
r=0
n-1
=0
3 3
18.6.2 Higher Derivatives of sin x , cos x
Formula 18.6.2
n 3n n
(n) 3
sin x
3
= sin x + - sin 3x + (2.0s)
4 2 4 2
n n n
= cosx +
2 2
(n) 3 3
cos x
3
+ cos 3x + (2.0c)
4 4
- 19 -
Proof
From Formula 18.6.2' mentioned next, it is obtained in a similar way in the case of the 2nd degree.
However, it is not so easy as the case of the 2nd degree. ( See 20.1.3 )
Example
2 32 2
(2) 3 3 9
sin x
3
= sin x + - sin 3x + =- sin x + sin 3x
4 2 4 2 4 4
3 3 3
= cosx +
2 2=
(3) 3 3 3 27
cos x
3
+ cos 3x + sin x + sin 3x
4 4 4 4
Formula 18.6.2'
n
(n) (0)
sin x = sin 2x sin x +
3
2
r (n -r)
r
n
2 cos 2x + sin x +
n
r-1
-Σ (2.1s)
r=1 2 2
n
= cos x cos x +
2
(n) (0)
cos x
3 2
r (n -r)
r
n
2 cos 2x + cosx +
n
r-1
+Σ (2.1c)
r=1 2 2
Proof
2
Substituting f(x)= sin x , g(x)= sin x for Theorem 18.1.1, we obtain (2.1s). (2.1c) is also obtained in
a similar way.
Formula 18.6.2"
3n + (-1)n
n /2 n
Σ
2r-1
2 = (2.2e)
r=0 2r 4
n n
2r +1
(n -1)/2 n 3 - (-1)
Σ
2r
2 = (2.2o)
r=0 4
Proof
From (2.1s)
n r (n -r)
r
n
(n) (0) n
sin x = sin x -Σ
3 2
sin x + 2r-1cos 2x + sin x +
2 r=1 2 2
n
(0)
= sin 2x sin x +
2
(2r +1) (n -2r -1)
n
(n -1)/2
- Σ
r=0 2r +1
22rcos 2x +
2
sin x +
2
2r (n -2r)
n
n /2
- Σ 22r-1cos 2x + sin x +
r=1 2r 2 2
- 20 -
n
1
=- 2-1cos 2x sin x +
2 2
(n -2r -1)
n
(n -1)/2
+ Σ (-1)r 22rsin 2x sin x +
r=0 2r +1 2
(n -2r)
n
n /2
- Σ (-1)r 22r-1cos 2x sin x +
r=1 2r 2
n n
n
1 (n -1)/2
= sin x + - Σ 22rsin 2x cos x +
2 2 r=0 2r +1 2
n n
n n
n /2
- 2-1cos 2x sin x + -Σ 22r-1cos 2x sin x +
0 2 r=1 2r 2
n n
n
1 (n -1)/2
= sin x + - Σ 22rsin 2x cos x +
2 2 r=0 2r +1 2
n
n
n /2
-Σ 22r-1cos 2x sin x +
r=0 2r 2
When n =1 ,
(1)
sin x
3
1 1 1
0 1
1 1
1
= sin x + - cos 2x sin x + 2-1 - sin 2x cos x + 20
2 2 2 2
1
1 1
= sin x+ - cos 2x cosx + 1 sin 2x sin x
2 2 2
1
1 1 1
= sin x+ + sin 2x sin x - (cos 2x cosx - sin 2x sin x)
2 2 2 2
1
1 1 1
= sin x+ + sin 2x sin x - cos 3x
2 2 2 2
1
1 1 1 1
= sin x+ - (cos 3x - cosx) - cos 3x
2 2 2 2 2
1 1 1
= sin x +
2 2
3 3
- sin 3x +
4 4
31-1 31+(-1)1
2r
1/2 1 1
Σ 2 2r-1
= = =
r=0 4 2 4
1 1 1 1
2r +1 4 2 4
(1-1)/2 1 3 -1 1 3 +1 3 -(-1)
Σ 2 2r
= + = =1=
r=0 4
When n =2 ,
2 2
0 2
2 2
(2) 1
sin x
3
= sin x + - cos 2x sin x + 2-1 + 21
2 2 2
2
2 1
2
- sin 2x cosx + 2 0
- 21 -
2
1 5
= sin x + + cos 2x sin x + 2 sin 2x cosx
2 2 2
2
1 1 1
= sin x + + (sin 3x - sin x) + 2sin 3x
2 2 2 2
2
3 9
= sin x + + sin 3x
4 2 4
2 32 2
= sin x +
2
3
- sin 3x +
4 4 2
32-1 1 32+1 32+(-1)2
2r
2/2 2 5
Σ 2 2r-1
= + = = =
r=0 4 2 2 4 4
2 2 2
2r +1 4
(2-1)/2 2 3 -1 3 -(-1)
Σ 2 2r
= =2=
r=0 4
Hereafter, by induction, we obtain the desired expressions.
If both sides of Formula18.6.2" are illustrated, it is as follows. The left side is blue line and the right side is
red point.
- 22 -
18.6.3 Higher Derivatives of the product of trigonometric and hyperbolic functions
Formula 18.6.3
(n - r) e x -(-1)-re -x
n
(n) n
(sinx sinhx) =Σ sin x + (3.1)
r=0 r 2 2
(n - r) e x +(-1)-re -x
n
(n) n
(sinx coshx) =Σ sin x + (3.2)
r=0 r 2 2
(n - r) e -(-1) e x -r -x
r 2 2
(n) n n
(cosx sinhx) =Σ cos x + (3.3)
r=0
(n - r) e +(-1) e x -r -x
r 2 2
(n) n n
(cosx coshx) =Σ cos x + (3.4)
r=0
Proof
Substituting f(x)= sin x , g(x)= sinh x for Theorem 18.1.1, we obtain (3.1).
The others are also obtained in a similar way.
Example
(0- r) e x +(-1)-0e -x
0
0
(sinx coshx) = Σ = sinx coshx
(0)
sin x +
r=0 r 2 2
(2- r) e x +(-1)-re -x
2
2
(cosx coshx) = Σ
(2)
cos x +
r=0 r 2 2
2 1 0
= cos x +
2
coshx +2cos x +
2
sinhx + cos x +
2
coshx
= -cosx coshx - 2sin x sinhx + cosx coshx = -2sin x sinhx
- 23 -
18.7 Higher Derivatives of sinh x f (x), cosh x f (x)
2 2
18.7.1 Higher Derivatives of sinh x , cosh x
Formula 18.7.1
e x -(-1)-n+re -x e x -(-1)-re -x
r
(n) n n
sinh x =Σ
2
(1.1s)
r=0 2 2
x -n+r -x
e x +(-1)-re -x
r
(n) n n e +(-1) e
cosh x =Σ
2
(1.1c)
r=0 2 2
Proof
Substituting f(x)= g(x)= sinh x for Theorem 18.1.1, we obtain (1.1s). (1.1c) is also obtained in
a similar way.
Example
e x -(-1)-0+re -x e x -(-1)-re -x
r
(0) 0 0
sinh x =Σ
2
= sinh 2x
r=0 2 2
x -3+r -x
e x +(-1)-re -x
r
(3) 33 e +(-1) e
cosh x
2
=Σ
r=0 2 2
0 1 2 3 coshxsinhx
3 3 3 3
= sinhxcoshx + coshxsinhx + sinhxcoshx +
= 8sinhxcoshx = 4sinh(2x)
2007.05.06
K. Kono
Alien's Mathematics
- 24 -
19 Super Derivative of the Product of Two Functions
Theorem 19.1.1
Let (x,y) be the beta function and p be a positive number. And for r =0, 1, 2, , let f
<-p+ r >
be
(r )
arbitrary primitive function of f(x) and g be the r th order derivative function of g(x) .
Then the following expressions hold.
r
(p) m -1 p (p -r)
(x)g r (x) + Rmp
()
f(x)g(x) =Σ f (1.1)
r=0
(-1)m 1
-p -1 (p)
<m+ k > (m+ k )
(-p , m) Σ
Rmp
= f (x) g (x) (1.1r)
k =0 m + k k
Especially, when n =0, 1, 2,
(n) n n n (r)
f(x)g (x) =Σ f ( -r)(x)g (x) ( Leibniz ) (1.1')
r=0 r
Proof
Theorem 16.1.2 (2.1) in 16.1.2 was as follows.
a a r
x x m -1 -n
f <0>g(0)dx n = Σ f<
n + r> (r)
g
r=0
n 1
a a
n -1 m -1 -n + r x x
- ΣΣ f a<
n -r+ s>
ga
(s)
dx r
r=0 s=0 s n-r n-r
n n- r+1
a a
n -1 r -1 r -1 x x
t s m +n-1 -r + tC m-1 f a
m+ n -r+s (m+ s)
+ (-1)m ΣΣΣ C ga dx r
r =1 s =0 t=s n- r n- r
n n- r+1
a a f
(-1) n -1 n -1C k
m x x
<m+ k > (m+ k )
(n , m) Σ
+ g dx n
k=0 m + k 1
n
When n =1 ∑∑∑ of the 3rd line does not exist, when
n =0 ∑∑ of the 2nd line does not exist also. And the
upper limit n -1 of ∑ of the 4th line can be replaced by . Therefore, if the index n of the integration
operator is substituted for -n in consideration of these, it is as follows.
a a r
x x m -1 n
f <0>g(0)dx -n = Σ f <-n + r> (r)
g
r=0
n 1
f
(-1)m 1
-n -1 x x
<m+ k > (m+ k )
(-n , m) Σ
+ g dx -n
k=0 m + k k an a1
Analytically continuing the index of the integration operator to [0 , p ] from[1 , n ] ,
a(p) a(0) r
x x m -1 p
f <0>g(0)dx -p = Σ g f <-p + r> (r)
r=0
(-p , m) Σ m + k k
m -p -1 x x
(-1) 1 m+ k > (m+ k )
+ f< g dx -p
k=0 a (p ) a (0)
-1-
-p
Then, replacing the integration operators dx , <-p + r> with the differentiation operators (p ),(p -r)
respectivly, we obtain (1.1) and (1.1r) .
Especially, when p = m -1 , m =1, 2, 3, , since (-p , m) = (1-m , m) = , Rmp = 0 .
Then
(m-1)
m -1 m -1 m-1-r) (r)
f(x)g(x) = Σ
r=0 r
f( g
Furthermore, replacing m -1 with n , we obtain (1.1') .
Q.E.D
-2-
19.2 Super Derivative of x^a f (x)
Formula 19.2.0
Let (z) be the gamma function, (n , m) be the beta function, f <r> be arbitrary the r th primitive
<r> <r>
function of f(x) and f a be the function values of f on a . Then the following expressions hold for
a positive number p .
(1)
(1+ ) - r (p -r)
r
(p) m -1 p
x f(x) =Σ x f
r=0 (1+ - r)
(1+ )
(p)
(-1)m 1
-p -1 - m- k <m+ k >
+ Σ
(-p , m) k =0 m + k k (1+ - m - k)
x f (0.1)
(1+m) m- r (p -r)
(p) m p
x m f(x) = Σ x f (0.1')
r=0 r (1+ m - r)
Where, if = -1, -2, -3, , it shall read as follows.
(1+ ) (- + r)
(-1)-r
1+ - r (- )
(2) When -1, -2, -3, & - p -1, -2, -3,
(1+ )
m -1 p
(p) r (r)
x m f(x) = Σ x -p+ f
r=0 r (1+ - p + r)
(-1)m 1 (1+ ) (p)
-p -1 +m m+ k
(-p , m) Σ
+ x +k f ( ) (0.2)
k =0 m + k k (1+ + m + k)
Proof
Since differentiation is an inverse operation of integration, replacing the index p of the integration operator
with -p in Formula17.2.0 in 17.2 , we obtain the desired expressions.
Formula 19.2.1
The following expressions hold for p >0 , s >0 such that p - s -1, -2, -3,
q (s)
(ax +b)p(cx +d)
(1/a)-s+ r (1+p)(1+q) (ax +b)p-s+r
m -1 s
=Σ + Rms (1.1)
r=0 r (1/c) (1+p -s + r)(1+q -r) (cx +d)
r r-q
(-1)m 1 m+ k
(1+p)(1+q)
-s -1 c
s
Rm = Σ
(-s, m) k=0 m + k k a (1+p +m + k)(1+q -m - k)
( s)
(ax +b)p+m+ k (cx +d)q-m- k (1.1r)
lim R s = 0
m m
-3-
Especially, when m =0, 1, 2,
(s)
(ax +b)p (cx +d)m
(1/a)-s+ r (1+p)(1+ m) (ax +b)p-s+ r
r
m s
=Σ (1.1')
r=0 (1/c)r (1+p - s + r)(1+ m -r) (cx +d)r-m
Proof
Although it is an original way to substitute f(x)=(ax +b ) p , g(x)=(cx +d ) q for Theorem 19.1.1 ,
here, we reverse the sign of the index of the integration operator <s > in Formula 17.3.1 ( 17.3 ) and replace
it with the differentiation operator (s ) .
3
Example1 The 1/2th order derivative of x-2 3x+4
Substituting a =1, b =-2, p =1/2 , c =3, d =4, q =1/3 , s =1/2 for (1.1) ,
1
(3/2)(4/3)
1
r
2 1/2 -r
x -2 3x +4
3
=Σ 3r
(x -2)r(3x +4) 3
r=0 (1+ r)(4/3-r)
The left side is calculated by the expression which replaced the order of integration and differentiation in
Riemann-Liouville differintegral. The integration lower limit is taken as -b /a = 2 according to Example1
in 17.3 . When m=17, the values of the both sides on arbitrary point x =5 are as follows.
3
Example1' The 1/2th order derivative of 3x+4(x-2)2
Substituting a =3, b =4, p =1/3 , c =1, d =-2, m =2 , s =1/2 for (1.1') ,
1
1 - +r
(4/3)(3)
1 6
r
2 2 1/2 -r (3x +4)
3x +4(x -2)2
3
=Σ 3 2
r=0 (5/6+ r)(3-r) (x -2)r-2
The integration lower limit of Riemann-Liouville differintegral in the left side is taken as-b /a = -4/3
according to Example1' in 17.3.1 . The values of the both sides on arbitrary point x =3 are as follows.
-4-
(1+q) (-q + r)
= (-1)-r
1+q - r (-q)
Then (1.1) can be read as follows.
q (s)
(ax +b) (cx +d)
p
2
1
(3/2)(1+ r) (x -2) r
r
x -2 1/2
=Σ (-3) r
3x +4 r=0 (1+ r)(1) (3x +4)r+1
The integration lower limit of Riemann-Liouville differintegral in the left side is -b /a = 2 like Example1 .
When m=10, the values of the both sides on arbitrary point x =4 are as follows.
Formula 19.2.2
When (z), (z) denotes the gamma function and the digamma function respectively, the following
expressions hold.
(1)
log x -(1-p+r) - (1+) -p
r
(p) m -1 p
x log x =Σ x + Rmp (2.1)
r=0 (1-p+r) (1+ -r)
log x -(1+m+k) - (1+) x
(p)
(-1)m
1 -p -1
Rmp = Σ
(-p , m) k =0 m + k k (1+m+k) (1+ - m - k)
(2.1r)
lim R p = 0
m m
-5-
x -p (1+ ) (1+ , m + k)
k-1 -p -1
(-p , m) 1+ - p Σ
Rmp = (-1) (2.2r)
k =0 k m+k
lim R p = 0
m m
Proof
Substituting f(x)= log x for Formula 19.2.0 , we obtain the desired expressions.
3/2
Example1 The 4/5 th order derivative of x log x
Substituting =3/2 , p =4/5 for (2.1) ,
(4/5)
3 7 4
log x -(1/5 +r) - (5/2)
m -1 4/5
=Σ
2 10 5
x log x x + Rm
r=0 r (1/5 +r) (5/2-r)
The integration lower limit of Riemann-Liouville differintegral in the left side is x =0 .
When m=120, the values of both sides on arbitrary point x =3 are as follows.
3
Example1' The 1.3 th order derivative of x log x
Substituting m =3 , p =1.3 for (2.1') ,
The remainder in the super derivative of the product of two functions becomes a series of super derivatives.
So it is very difficult to calculate this. However, the calculation of the above (2.2r) is exceptionally easy.
So we show it as follows.
-6-
3
Example2 The 1/2th order derivative of x log x
Substituting =1/3 , p =1/2 for the above (2) ,
(4/3) - 6
1
(p)
x log x = x log x
5/6
(4/3)(r) - 6
1 1
m -1
r-1
-1/2
+ Σ (-1) x + Rm2
r=1 r 5/6+ r
1
-
(4/3) (4/3 , m + k)
1 6
x k-1
-3/2
(-1/2, m) 5/6 Σ
Rm2 = (-1)
k =0 k m+k
When m=10, the values of both sides on arbitrary point x =1 are as follows.
Complete Automorphism
Reversing the sign of the index of the integration operator <p> in Formula 17.3.2 in 17.3 , we obtain the
following expression without the remainder term. However, it is complicated and the convergence is also very
slow.
Formula 19.2.2'
(p)
x log x
log x -(1-p) - (1+) -p -1 (2+k) +
=
(1-p)
-p Σ
1+ - p k =0 k
(1+ k) (1+ k, - k) -p
2
x
-p -1 1
1 - pΣ
(1+ k) (1+ k, - k)
2
k=0 k
Formula 19.2.3
When m =0, 1, 2, , the following expressions hold for p >0 .
-7-
(1+ m) m- r (p -r)
r
p
(p) m
x m sin x =Σ x sin x + (3.1s)
r=0 (1+ m -r) 2
(1+ m) m- r (p -r)
r
p
(p) m
x m cosx =Σ x cos x + (3.1c)
r=0 (1+ m -r) 2
Proof
Reversing the sign of the index of the integration operator <p> in Formula 17.3.3 in 17.3 and replacing it
with the differentiation operator (p ), we obtain the desired expressions.
3
Example The 3/2th order derivative of x sinx
Substituting m =2 , p =3/2 for (3.1s) ,
2
3
(4) 3-r (3/2- r)
r
3/2
3
x 3 sin x =Σ x sin x +
r=0 (4- r) 2
This super derivative can not be examined by Riemann-Liouville differntegral. Then we draw this with the 1st and
the 2nd order derivative on a figure side by side. It is as follows.
(p)
(-1)m (1+ ) m + k
-p -1
1
Rmp = Σ x +m+ k sin x +
(-p , m) k =0 m + k k 1+ + m +k 2
(1+ ) r
r
p
+R
m -1
{p}
x cosx
r
=Σ x -p+ cos x + p
(3.2c)
r=0 (1+ -p + r) 2 m
(p)
(-1)m (1+ ) m + k
-p -1
1
Rmp = Σ x +m+ k cos x +
(-p , m) k =0 m + k k 1+ + m +k 2
lim Rmp = 0
m
-8-
Proof
Reversing the sign of the index of the integration operator <p> in Formula 17.3.3' in 17.3 and replacing it
with the differentiation operator (p ), we obtain the desired expressions.
3/2
Example Collateral the 1 / 2th order derivative of x sinx
Substituting =3/2 , p =1/2 for (3.2s), we obtain
2
1
(5/2) 1-r r
1
x sin x
3
r
1/2
+
m -1
2
=Σ x sin x + Rm2
r=0 (2- r) 2
The integration lower limit of Riemann-Liouville differintegral in the left side is x =0 .
When m=15, the values of the both sides on arbitrary point x =5 are as follows.
Formula 19.2.4
When m =0, 1, 2, , the following expressions hold for p >0 .
(1+ m) m- r e x -(-1)-p+ re -x
r
{p} m p
x sinh x
m
=Σ x (4.1s)
r=0 (1+ m - r) 2
(1+ m) e x +(-1)-p+ e -x
r
r (1+ m - r) x
{p} m p m- r
x cosh x
m
=Σ (4.1c)
r=0 2
3
Example The 0.999th order derivaive of x coshx
Substituting m =3 , p =0.999 for (4.1c), we obtain
(4) 3- r e x +(-1)-0.999+ re -x
{0.999} 3 0.999
x cosh x
3
=Σ x
r=0 r (4- r) 2
This super derivative can not be examined by Riemann-Liouville differntegral. Then we calculate the values on
arbitrary point x=2 for the 0.999th order derivative and the 1st order derivative respectively. Although the former
is a complex number, the real part is naturally near to the coefficient of the later.
-9-
Formula 19.2.4' (Collateral Super Derivative)
The following expressions hold for , p such that -1, -2, -3, & - p -1, -2, -3, .
(1+ ) x -r -x
r
{p} m -1 p -p+ r e -(-1) e
x sinh x =Σ x + Rmp (4.2s)
r=0 1+ -p + r 2
x (p)
-x
(1+ )
-m-k
(-1)m
1 -p -1 +m+ k e -(-1) e
Rmp = Σ x
(-p , m) k =0 m + k k 1+ + m + k 2
(1+ ) x -r -x
r
{p} m -1 p -p+ r e +(-1) e
x cosh x =Σ x + Rmp (4.2c)
r=0 1+ -p + r 2
x (p)
-x
(1+ )
-m-k
(-1)m
1 -p -1 +m+ k e +(-1) e
Rmp = Σ x
p , m k =0 m + k
(- ) k 1+ + m + k 2
lim Rmn = 0
m
3
Example Collateral the 3 / 2th order integral of x sinhx
Substituting =1/3 , p =3/2 for (4.2s), we obtain
3
(4/3)
1
e x -(-1)-re -x
r -1/6+ r
2 m -1 3/2 - +r
x sinh x
3
=Σ x 6
+ Rm3/2
r=0 2
The integration lower limit of Riemann-Liouville differintegral in the left side is x =0 .
When m=50, the values of the both sides on arbitrary point x =17 are as follows.
- 10 -
19.3 Super Derivative of log x f (x)
Formula 19.3.1
(p) x -p log x log x - 1- p -
log x
2
=
1-p
p log x - (1- p +r)-(r)
m -1
- x -p Σ (-1)r + Rmp (1.1)
r=1 r 1-p + r
-p
x
Rmp =
(-p , m) 1-p
(-1) k-1
k =0 (m+ k )
-p -1
Σ 2 log x - 1-p -(1+ m+ k ) -2 (1.1r)
k
lim Rmp = 0
m
Proof
Reversing the sign of the index of the integration operator <p> in Formula 17.4.1 in 17.4 and replacing it
with the differentiation operator (p ), we obtain the desired expressions.
- 11 -
19.4 Super Derivative of e^x f (x)
Formula 19.4.1
(1+ ) - r
(p) m -1 p
e x x = e xΣ x + Rmp (1.1)
r=0 r 1+ - r
(-1)m 1 (1+ )
-p -1 m-k (p)
p
Rm = Σ e x x - (1.1r)
(-p , m) k =0 m + k k 1+ - m -k
Especially, when m =0, 1, 2,
(1+m) m- r
(p) m p
e x x m = e xΣ x (1.1')
r=0 r (1+ m - r)
Proof
Let f(x)= e x , g(x)= x . Then
(r) (1+ ) - r
x = x ( -1, -2, -3, )
1+ - r
Substituting these for Theorem 19.1.1 , we obtain (1.1) and (1.1r).
Especially, when = m-1 m =1, 2, 3, , (1+ - m - k) = k =0, 1, 2, 3, .
Then, the remainder term disappears and (1.1) is as follows.
(1+m -1) m-1- r
r
m-1 (p)
m -1 p
e x x = e xΣ x
r=0 1+m -1- r
Thus, replacing m -1 with m , we obtain (1.1').
3/4
Example1 The 1/2.1th order integral of ex x
Substituting =3/4 , p =1/2 for (1.1) ,
2
1
(1+3/4) 4 - r
3
r
m -1 1/2
e x
x
x
Σ
3/4
=e x + Rm1/2
r=0 1+3/4- r
The left side is calculated by the expression which replaced an order of integration and differentiation in
Riemann-Liouville differintegral. When m=10, the values of the both sides on arbitrary point x =10 are
as follows.
7
Example1' The 5/2th order derivative of ex x
Substituting m =7 , p =5/2 for (1.1') ,
- 12 -
2
5
(8) 7- r
r
7 5/2
e x x
7
= e xΣ x
r=0 8- r
The values of the both sides on arbitrary point x = .5 are as follows.
3/4
Example1" Collateral the 1 / 2th order derivative of ex x
Substituting =3/4 , p =1/2 for (1.1") , we obtain
2
1
(3/2) 4 + r
3
r
m -1 1/2
e x
x
x
Σ
3/4
=e x + Rm1/2
r=0 3/4+ r
The values of both sides on the same point x =-10 as Example1 are as follows. Though the both sides are
corresponding, they are considerably different from the values of Example1 (Lineal the 1/2th order derivative.)
- 13 -
Formula 19.4.3
-p
p
(p)
e sin x
x
= sin e xsin x + (3.0s)
4 4
-p
p
(p)
e cosx
x
= sin e xcos x + (3.0c)
4 4
Proof
Analytically continuing the index of the differentiation operator in Formula 18.4.3 in 18.4 to [0 , p ] from
[1 , n] , we obtain the desired expressions.
Trigonometric Series
If Formula 18.4.3' in 18.4 is extended to the real number, the following trigonometric series are obtained.
Formula 19.4.3'
r -p p
p
Σr=0 r
sin x +
2
= sin
4
sin x +
4
(3.1s)
r -p p
p
Σr=0 r
cos x +
2
= sin
4
cos x +
4
(3.1c)
Especially, when x =0 ,
r -p p
p
Σr=0 r
sin
2
= sin
4
sin
4
(3.1's)
r -p
p
Σ r cos 2 = sin 4
p
n
cos (3.1'c)
r=0 4
Formula 19.4.3"
p
p
p
Σ k 2
(-1) = 2 sin (3.2s)
k=0 2k +1 4
- 14 -
p
p
p
Σ k 2
(-1) = 2 cos (3.2c)
k=0 2k 4
Proof
Since the odd-numbered terms of the left side in (3.1's) are all 0,
r
r
p p p p p p
Σ
r=0
sin
2
=
1
-
3
+
5
-
7
+- = Σ(-1)k
k=0 2k +1
p p
p
-p
= sin 4
p
Σ (-1) k
sin = 2 2 sin
k=0 2k +1 4 4
Next, since the even-numbered terms of the left side in (3.1'c) are all 0,
r
r
n p p p p p p
Σ
r=0
cos
2
=
0
-
2
+
4
-
6
+- = Σ(-1)k
k =0 2k
p p
p
-p
= sin 4
p
Σ (-1) k
cos = 2 2 cos
k=0 2k 4 4
If a horizontal axis is set as p and these are illustrated , it is as follows. Although the left side is blue and
right side is red, since both sides overlap exactly, the left side (blue) is not visible.
Formula 19.4.4
e x - (-1)-re -x
r
(p) p
e sinh x
x
=e x
Σ (4.0s)
r=0 2
x -r -x
r 2
(p) p e + (-1)
e
e cosh x
x
=e Σ x
(4.0c)
r=0
- 15 -
Example The 3/2th order derivative of e x coshx
When p =3/2 , the values of the both sides of (4.0c) on arbitrary point x =1.3 are as follows.
- 16 -
19.5 Super Derivative of f (x) / e^x
e x
-x
19.5.1 Super Derivative of
Formula 19.5.1
(-1)-p (1+ ) - r
r
(p)
m -1 p
e x
r
Σ + Rmp
-x
= (-1) x (1.1)
ex r=0 1+ - r
(p)
(1+) x
(-1)k - m-k
1 -p -1
Rmp
(-p , m) Σ
= (1.1r)
k=0 m+ k k 1+ - m -k ex
Especially, when m =0, 1, 2,
(-1)-p (1+ m) m- r
r
m (p)
m p
e x
r
Σ(-1)
-x
= x (1.1')
e x r=0 1+ m - r
-x
Example1 The 1/2th order derivative of e x
If we substitute =1/2 , p =1/2 for (1.1) and calculate the values of the both sides on arbitrary point
x =10 , it is as follows. In addition, this is an asymptotic expansion.
-x 7
Example1' The 5/2th order derivative of e x
If we substitute m =7 , p =5/2 for (1.1') and calculate the values of the both sides on arbitrary point
x =6 , it is as follows.
- 17 -
(p)
(1+) x
(-1)-k - m-k
1 -p -1
Rmp
(-p , m) Σ
=
k=0 m + k k 1+ + m +k ex
lim Rmp = 0
m
-x
Example1" Collateral the 1 / 2th order derivative of e x
If we substitute =1/2, p =1/2, m =30 for (1.1") and calculate the values of the both sides on arbitrary
point x =10 , it is as follows. Though the both sides are corresponding, they are completely different from
the values of Example1 (Lineal the 1/2th order derivative.)
-x
19.5.2 Super Derivative of e log x
All the polynomials obtained by applying Theorem 19.1.1 to e -xlog x become asymptotic expansions, and
they are hardly helpful.
-x -x
19.5.3 Super derivatives of e sinx , e cosx
Formula 19.5.3
-p
p
(p)
e sin x
-x -p
= (-1) sin e -xsin x - (3.0s)
4 4
-p
p
(p)
e cosx
-x
= (-1)-p sin e -xcos x - (3.0c)
4 4
-x
Example The 3/2th order derivative of e sinx
The values of the both sides of (3.0s) on arbitrary point x =1.7 are as follows.
-x -x
19.5.4 Super derivatives of e sinhx , e coshx
- 18 -
Formula 19.5.4
e x - (-1)-re -x
r
(p) -x
p
e sinh x
-x
=e Σ (-1) -p+ r
(4.0s)
r=0 2
x -r -x
r
(p) -x
p e + (-1) e
e cosh x
-p+ r
Σ
-x
=e (-1) (4.0c)
r=0 2
-x
Example The 3/2th order derivative of e coshx
The values of the both sides of (4.0c) on arbitrary point x =1.3 are as follows.
- 19 -
19.6 Super Derivatives of sin x f (x), cos x f (x)
2 2
19.6.1 Super Derivatives of sin x , cos x
Formula 19.6.1
p
(p)
sin x
2
= -2p-1cos 2x + (1.0s)
2
p
(p)
cos x
2
= 2p-1cos 2x + (1.0c)
2
Proof
Analytically continuing the index of the differentiation operator in Formula 18.6.1 in 18.6.1 to [0 , p ] from
[1 , n] , we obtain the desired expressions.
Example
2
1 1
-1 1
sin x
2 2
= -2 cos 2x + = (sin 2x + cos 2x)
4 2
2
3
3
3
Analytically continuing the index of the differentiation operator in Formula 18.6.1' and Formula 18.6.1" in 18.6
to [0 ,p ] from[1 ,n ] respectively, we obtain the following two formualas.
Formula 19.6.1'
(p - r) r
r
p
(p)
sin x =Σ
2
sin x + sin x + (1.1s)
r=0 2 2
(p - r) r
r 2 2
(p) p
cos x =Σ
2
cos x + cos x + (1.1c)
r=0
Formula 19.6.1"
=2
p p-1
Σ
r=0 2r
p >0 (1.2e)
Σ =2
p p-1
p >0 (1.2o)
r=0 2r +1
If a horizontal axis is set as p and these are illustrated , it is as follows. (1.2e) , (1.2o) and 2p-1 are blue,
p-1
red and green respectively. Since three curves overlap exactly, only red ( 2 ) is visible.
- 20 -
The following formula follows from Formula 19.6.1 immediately.
Formula 19.6.1"'
r = 2
p
Σ
r=0
p
p >0 (1.3)
Note
In fact, Formula 19.6.1" and Formula 19.6.1"' hold for p > -1 .
3 3
19.6.2 Super Derivatives of sin x , cos x
Formula 19.6.2
p 3p p
(p) 3
sin x
3
= sin x + - sin 3x + (2.0s)
4 2 4 2
p p p
= cosx +
2 2
(p) 3 3
cos x
3
+ cos 3x + (2.0c)
4 4
Example
2
1
3 3
sin x
3
= sin x + - sin 3x +
4 4 4 4
2
3
3 3
3 3 3
cos x
3
= cos x + + cos 3x +
4 4 4 4
If the later is drawn with the 1st and the 2nd order derivatives on a figure side by side, it is as follows.
- 21 -
19.6.3 Super Derivatives of the product of trigonometric and hyperbolic functions
Formula 19.6.3
(p - r) e x -(-1)-re -x
p
(p)
(sinx sinhx) =Σ sin x + (3.1)
r=0 r 2 2
(p - r) e x +(-1)-re -x
p
(p)
(sinx coshx) =Σ sin x + (3.2)
r=0 r 2 2
(p - r) e -(-1) e x -r -x
r 2 2
(p) p
(cosx sinhx) =Σ cos x + (3.3)
r=0
(p - r) e +(-1) e x -r -x
r 2 2
(p) p
(cosx coshx) =Σ cos x + (3.4)
r=0
If this is drawn with the 1st and the 2nd order derivatives on a figure side by side, it is as follows.
10
1 2 3 4
-5
-10
- 22 -
19.7 Super Derivatives of sinh x f (x), cosh x f (x)
2 2
19.7.1 Super Derivatives of sinh x , cosh x
Formula 19.7.1
e x -(-1)-p+re -x e x -(-1)-re -x
r
(p) p
sinh x
2
=Σ (1.1s)
r=0 2 2
x -p+r -x
e x +(-1)-re -x
r
(p) p e +(-1) e
cosh x =Σ
2
(1.1c)
r=0 2 2
Proof
Analytically continuing the index of the differentiation operator in Formula 18.7.1 in 18.7 to [0 ,p ] from
[1 , n], we obtain the desired expressions.
2
Example: The 0.01th and the 0.99th order derivatives of sinh x
According to (1.1s), if each differential coefficient of the 0th order, the 0.01th order, the 0.99th order and
the 1st order on arbitrary point x =2 are calculated , they are as follows. Naturally, the super differential
coefficients turn into complex numbers.
2010.11.11
K. Kono
Alien's Mathematics
- 23 -
20 Higher Calculus of the product of many functions
number n .
r x
n n n-r r
x1+ x2 = Σ
n
1 x2
r=0
On the other hand, according to the Leibniz Rule, the following expression holds for functions f1 , f2 of x and
natural number n .
r
(n) n n n -r) (r)
f1 f2 =Σ f1( f2
r=0
(2) Multinomial Theorem and Higher Derivative of the product of many functions
According to the multiomial theorem in 3.3 , the following expression holds for real numbers x 1 , x 2 , , x
and natural number n .
r r r
n r1 r-2 n r1 r-2 n-r1 r -r2 r
x1+ x2++ x =Σ Σ Σ x-1
n
x1 x21
r1=0 r 2=0 r -1=0 1 2 -1
Therefore, the following expression must hold for functions f1 , f2 , , f of x and natural number n .
r r r
n r1 r -2 n r1 r-2
f1
n - r1 r1 -r2
f -1
r
f1 f2 f = ΣΣ Σ
(n)
f2
r1=0 r 2=0 r -1=0 1 2 -1
Theorem 20.1.1
When fk
(r)
r th order derivative function of fk (x) (k =1, 2, , ) ,
denotes the
n r1 r-2 n r1 r-2
f1 1f2 1 2 f -1
n- r r -r r
(n)
1 2
f f f =Σ Σ Σ
r 1=0 r2=0 r-1=0 r1 r2 r-1
Proof
According to Theorem 18.1.1 (Leibniz) in 18.1 , the following expressions hold.
r
n n r1
f1
(n) n -r1
f1 f2 f3 f4 f =Σ
r1=0
f2 f3 f4 f (1)
1
r
r1
r1 r1 r2
f2 1
r -r2
f2 f3 f4 f =Σ
r2=0
f3 f4 f (2)
2
r f
r2
r r2 2 r3
r2-r3
f3 f4 f =Σ
r3=0
3 f4 f (3)
3
-1-
r
r-3 r-3
-3
f-2-3
r r - r-1 r-2
f-2 f-1 f = Σ f-1 f ( -2)
r =0
-2 -2
r f
r r-2 -2
-2 r r-2- r-1 r-1
f-1 f = Σ -1 f ( -1)
r-1=0 -1
Substituting (2), (3), , ( -2), ( -1) for (1) one by one, we obtain the desired expression.
Example
r s f f
n r n r (n - r) (r-s) (s)
f1 f2 f3
(n)
= ΣΣ 1 2 f3
r =0 s =0
r s t f
n n r r s s
(n - r) (r-s) (s -t) (t)
f1 f2 f3 f4
(n)
= ΣΣΣ 1 f2 f3 f4
r =0 s =0 t=0
(n -r) (1+)
x = x -n+ r
1+ - n + r
x (r-s)
e = ex
s
(s)
(sin x) = sin x +
2
Substituting these for Theorem 20.1.1 , we obtain
(1+) s
r s
n r n r
x (n)
x e sin x =Σ Σ x -n+ r e x sin x + (1.1)
r =0 s =0 1+ - n + r 2
Although this formula is troublesome in manual calculation, it is easy in mathematical software. When the 4th
order derivative of x e xsin x is calculated using the mathematical software MuPad, it is as follows.
-2-
fs := fs + binomial(n,r)*binomial(r,s)
*gamma(1+a)/gamma(1+a-n+r)*x^(a-n+r)
*E^x*sin(x+s*PI/2)
end_for:
fr:=fr+fs
end_for:
fr := expand(fr):
simplify(fr)
r r s f
n n n - r) (r-s) (0)
n r n r (n - r) (r-s) (s)
f1 f2 f3 = Σ Σ Σ
(n)
f1( f2 f3 + 1 f2 f3
r =0 r =1 s =1
Substituting the abave expressions for this, we obtain
(1+ )
n n
x e x log x
(n) r
= Σ
r =0 r 1+ - n + r
x -n+ e x log x
(1+ )(s) -n+ r-s x
n r n r
+ ΣΣ (-1)s-1 x e (1.2)
r =1 s =1 r s 1+ - n + r
When n =1
(1+ )
1 1
x e x log x
(1) r
= Σ
r =0 r 1+ -1+ r
x -1+ e x log x
-3-
(1+ )
2 2
x e x log x
(2) r
= Σ r 1+ -2+ r
r =0
x -2+ e x log x
Theorem 20.1.2
When f
(r)
r th order derivativefunction of f(x) and is a natural number,
denotes the
n r1 r-2 n r1 r-2
f (x) =Σ Σ Σ f 1 f 1 2 f -1
(n) n- r r -r r
r 1=0 r2=0 r-1=0 r1 r2 r-1
Example1 Higher Derivative of e x
x (r)
Since e = ex, immediately from the theorem,
r r
r1 r -2 n r1 r-2
x x
(n) n
e = e Σ Σ Σ
r1=0 r 2=0 r-1=0 r1 2 -1
Here, from (1.1") in 3.3 ,
n r3 r-2
x
(n)
n x
e = e (2.1)
3
Example2 Higher Derivative of log x
Let f = log x . Then
(r =1, 2, 3, )
(r)
f(0) = log x , f = (-1)r-1(r -1)! x -r
r =1, 2,
(0) (r ) (0)
Since f is different from f in function type, we have to separate the terms containing f
from
r s
(n) n r n r
f
3 n -r) (r-s) (s)
= ΣΣ f( f f
r =0 s =0
It is as follows.
-4-
0
n
r =0 r s =0 s s =0 s
n n n -r)
r r r-s) (s) (n)
0 0 (0-s) (s)
Σ f( Σ f( f = f Σ f f
n
n
r s f f s =0 s
n n-1 r (n -r)
r
(r-s) (s) n -n )
n n (n -s) (s)
+Σ f Σ + f( Σ f f
r =1 s =0
f + Σ f s f s f
n r
n-1 r n n
= f(n)f(0) (0) Σ
(n -r) (r-s) (s)
f + f(0)Σ (n -s) (s)
f (w)
r r =1 s =0 s =0
Here,
r =1 r s =0 s
n-1 n (n -r)
r r (r-s) (s)
Σ f Σ f f
1
n
s =0 s r s =0 s
n -1)
1 1 (1-s) (s)
n-1 n n -r)
r r (r-s) (s)
= f( Σ f f +Σ
r =2
f( Σ f f
= f
0 1 f f
n (n -1) 1 1
f(1-0)f(0)+ (1-1) (1)
1
r 0 + f
r
s
n n-1 r (n -r) r (r-0) (0)
r -1
(r-s) (s) (r-r) (r)
+Σ f f f +Σ f f f
r =2 r s =1
= 2f f
n
f + 2f Σ f f +Σ Σ f
(0) (n -1) (1)n n
(0) r
n-1
(n -r) (r)
n-1 r -1
(n -r) (r-s) (s)
f f
1 r r s
r =2 r =2 s =1
r r s f f f
n n-1 n r
(n -r) (r)
n-1 r -1
(n -r) (r-s) (s)
= 2f Σ (0)
f f + ΣΣ
r =1 r =2 s =1
0 Σs n f
n n
Σs
n n (n -s) (s) n (n -0) (0)
n-1
(n -s) (s) (n -n ) (n)
f(0) f f = f f f + (0)
f f + f
s =0 s =1
= f Σ f
n (0)
n-1
(n -s) (s) (0) (0) (n)
f + 2f f f
s s =1
Substitute these for (w) ,
r =0 s =0 r s
n r n r (n -r) (r-s) (s)
Σ Σ f f f
s f
n-1 n
= f(n)f(0)f(0) + f(0)Σ (n -s) (s)
f + 2f(0)f(0)f
(n)
s =1
Σr f r s
n-1 n (n -r) (r)
n-1 r -1 n r n -r) (r-s) (s)
+ 2f(0) f + ΣΣ f( f f
r =1 r =2 s =1
i.e.
r f
(n) n-1 n
f (x)
3
= 3f(0)f(0) f(n)+ 3f(0)Σ (n -r) (r)
f
r =1
r s f
n r
n-1 r -1
(n -r) (r-s) (s)
+ ΣΣ f f
r =2 s =1
-5-
3(n -1)! n-1 n
(n) n 3log x
log x
3
= (-1)n-1 log 2
x +( -1) Σ (n -r -1)!(r -1)!
xn x n r =1 r
(-1)n-1 n-1 r -1 n
r
+ Σ Σ (n -r -1)!(r -s -1)!(s -1)!
xn r =2 s =1 r s
Furthermore,
r
n n! n!
(n -r -1)!(r -1)! = (n -r -1)!(r -1)! =
(n -r)!r ! (n -r)r
r
n r n!
(n -r -1)!(r -s -1)!(s -1)! =
s (n -r)(r -s)s
Using these, we obtain
(n) 3(n -1)! n 3log x
n-1 n!
log x
3
= (-1)n-1 n log 2
x + (-1) n Σ
x x r =1 (n -r)r
(-1)n-1 n-1 r -1 n!
+ n Σ Σ
r =2 s =1 (n -r)(r -s)s
(2.2)
x
When n =1
(1) 3(1-1)! 3log 2x
log x
3 1-1 2
= (-1) log x =
x1 x1
When n =2
(2) 3(2-1)! 32(2-2)!
log x
3
= (-1)2-1 log 2x + (-1)2 log x
x2 x2
3log 2x 6log x
=- +
x2 x2
When n =3
(3) 3(3-1)! 3log x 3-1 3!
log x = (-1) Σ
3 3-1
log 2x + (-1)3
x3 x 3 r =1 (3-r)r
(-1)3-1 3-1 r -1 3!
+ 3 Σ Σ
r =2 s =1 (3-r)(r -s)s
x
6log 2x 18log x 6
= 3
- 3
+ 3
x x x
Lemma1
1 m
cos m x = ΣmC r cos(m -2r)x m =1, 2, 3, (3.m)
2m r=0
Proof
The following formulas are known about the power of cos x . (See "岩波 数学公式Ⅱ " p190. )
-6-
1 m -1 2m Cm
2m-1 Σ 2mC r cos{(2m -2r)x } +
cos 2mx =
2 r=0 22m
1 m
cos 2m+1x = Σ 2m +1C r cos{(2m +1-2r)x }
22m r=0
2m Σ
2m
= 2mC r cos{(2m -2r)x } + Σ 2mC r cos{(2m -2r)x }
2 r=0 r=m +1
1
+ C m cos{(2m -2m)x }
2m
22m
1 2m
1 m m
=
22m+1 Σ r=0
2m +1C r cos{(2m +1-2r)x }+Σ2m +1C r cos{(2m +1-2r)x }
r=0
1 m
2m+1 Σ
2m +1
= 2m +1C r cos{(2m +1-2r)x }+ Σ 2m +1C r cos(2m +1-2r)x
2 r=0 r=m +1
1 2m +1
= Σ 2m +1C r cos(2m +1-2r)x
22m+1 r=0
Formula 20.1.3
When denotes the floor function, the following expressions hold for natural number m .
n
Σ C r (m -2r) cos(m -2r)x +
(n) 1 m /2
cos mx = m-1 m
n
(3.mc)
2 r=0 2
n
Σ C r (m -2r) cos(m -2r)x - 2 +
(n) 1 m /2
sin mx = m-1 m
n
(3.ms)
2 r=0 2
-7-
Proof
Differentiating the both sides of (3.m) in Lemma1 with respect to x n times,
n
(n) 1 m
cos mx = mΣ
(m -2r)n mC r cos (m -2r)x +
2 r=0 2
When m is even,
n
n
m -2 2 x + 2 = 0
m m
m-2 C m cos
m
2 2
n n
C cos (m-2r) x +
2
m /2-1 m
Σ (m-2r) n m r = Σ (m -2r) n
m C r cos (m-2r) x +
r=0 r=m /2+1 2
Using these,
n
(n) 1 m
cos mx = Σ (m -2r)n
mC r cos (m -2r)x +
2m r=0 2
n
2 m /2-1
= m Σ (m -2r)n mC r cos (m -2r)x +
2 r=0 2
2 m n n
m
+ m m -2 mC m cos m -2 x+
2 2 2 2 2
n
1 m /2
= m-1 Σ (m -2r)n mC r cos (m -2r)x +
2 r=0 2
When m is odd,
n n
m /2 m
n n
Σ
r=0
(m-2r) m C r cos (m -2r) x +
2
= Σ (m -2r)
r=m /2+1
C r cos (m -2r) x +
m
2
Using this,
n
(n) 1 m
cos mx = mΣ
(m -2r)n mC r cos (m -2r)x +
2 r=0 2
n
2 m /2
= m Σ (m -2r)n mC r cos (m -2r)x +
2 r=0 2
Thus, we obtain (3.mc) . (3.ms) follows by the replacing x with x - /2 .
6
Example The 7th order derivative of cos x
-8-
20.1.4 Higher Derivatives of cos x , sin x
Formula 20.1.4
The following expressions hold for a positive number .
1 n
(n)
cos x Σ r
= ( -2r)n cos ( -2r)x + (4.c)
2 r=0 2
1 n
(n)
sin x = Σ ( -2r)n cos ( -2r) x - + (4.s)
2 r=0 r 2 2
Proof
Differentiating the both sides of (3.m) in Lemma1 with respect to x ,
n
(n) 1 m
cos mx = Σ (m -2r)n
mC r cos (m -2r)x +
2m r=0 2
Analytically continuing the power from the natural number m to the positive number , we obtain (4.c) .
(4.s) follows by the replacing x with x - /2 .
5.3
Example The 3rd order derivative of sin x
The differential quotient on arbitrary point x= 0.9 was calculated by the direct calculation and the formula.
Both are corresponding very well.
-9-
20.2 Higher Integral of the product of many functions
(1) Generalized binomial theorem and Higher integral of the product of 2 functions
According to generalized binomial theorem in 3.2 , the following expression holds for real numbers x 1, x2
such that x 1 >x 2 and natural number n .
r x
-n
x1+ x2 = Σ
-n -n-r r
1 x2
r=0
On the other hand, according to Formula 16.1.2 in 16.1 , the following expression holds for functions f1, f2
r f
x x m -1 -n
f1 f2 dx n= Σ <n +r> (r)
1 f2 + Rmn
a a r=0
f
(-1)m n -1 n -1C k x x
<m+ k > (m+ k ) n
(n , m) Σ
Rmn = 1 f2 dx
k=0 m + k a a
Reversing the sign of the index of the differentiation operator (n ) in the Leibniz Rule ,
r
-n
f1 f2
(-n )
=Σ f1(-n -r)f2(r)
r=0
Next, Replacing (-n) with the intagration operator <n > , and dividing the series into a polynomial and
a remainder, we obtain the above formula.
(2) Generalized multinomial theorem and Higher Integral of the product of many functions
According to the generalized multinomial theorem in 3.4 , the following expression holds for real numbers
x1 , x2 , , x such that x 1 >x 2 + x 3 ++ x and natural number n .
r r r
r1 r -2 -n r1 r-2 -n-r1 r -r2 r
x1+ x2++x =Σ Σ Σ x-1
-n
x1 x21
r1=0 r2=0 r-1=0 1 2 -1
Therefore, the following expression must hold for functions f1 , f2 , , f of x and natural number n .
r r r
x x m -1 r1 r-2 -n r1 r-2
f1
n - r1 r1-r2
f -1
r
f1 f2 fdx = Σ
n
Σ Σ f2
a a r =0 r =0 r =0 1 2 -1 1 2 -1
+ Rmn
Theorem 20.2.1
Let fk
(r)
be the r th order derivative function of fk (x) k =1, 2, , , fk<r> be the arbitrary r th order
primitive function of fk (x ) , m,n are natural numbers and (n , m) be the beta function. If there is a
r r r
x x m -1 r1 r-2 -n r1 r-2
f1
n + r1 r1-r2
f -1
r
f1 f2 f dx =Σ Σ Σ
n
f2
a a r1=0 r2=0 r-1=0 1 2 -1
+ Rmn
- 10 -
k k
n -1C k1 k-2
(-1)m n-1 m + k1 k2 k -2 m + k1 k2
Rmn = Σ Σ Σ Σ
(n , m) k 1=0 k 2=0 k 3=0 k -1=0 m + k1 k2 3 -1
f
x x
m+ k 1 m+k 1 -k 2 k 2-k 3
f -1dx n
k
1 f2 f3
a a
Proof
Let us integrate f1 f2 f3 f with respect to x from a to x n times. Then according to Theorem 16.1.2
in 16.1 , we obtain
r
x x m -1 -n r1
f1 f2 f3 f dx n = Σ f1 f2 f3 f
n + r1
+ Rmn
a a r1=0 1
(-1)m n -1 n -1C k1
f
x x
m+ k 1
f2 f3 f
m+ k 1
(n , m) Σ
Rmn = 1 dx n
k=0 m + k1 a a
According to Theorem20.1.1 ,
r r r
r1 r2 r-2 r1 r2 r-2
f2 1
r -r2 r2- r3
f -1
r1 r
f2 f3 f = ΣΣ Σ f3
r2=0 r3=0 r-1=0 2 3 -1
k k k f
m + k1 k2 m +k k k k -2 -2
k 1 -k 2 k 2- k 3
f -1
m+ k 1 1 2 k
f2 f3 f =Σ Σ Σ 2 f3
k2=0 k 3=0 k -1=0 2 3 -1
Substituting these for the above, we obtain the disired expression.
Example
r s f f f + R
x x m -1 r -n r
f1 f2 f3 dx =Σ Σ
n <n + r> (r-s) (s)
1 2 3
n
m
a a r =0 s =0
m + r s s
(-1) C m +r r m n -1 m + r n -1 x x
f
r <m+ r> (m+r-s) (s) n
R = n
(n , m)
m Σ Σ r =0 s =0 a
1 f2 f3 dx
a
f f f f dx ΣΣΣ r s t f f
x x -n r s m -1 r s
= 1 2 3 4
n <n + r> (r-s) (s-t) (t)
1 2 f3 f4 + Rmn
a a r =0 s =0 t=0
C m +r m
m + r s s t
(-1) r s n -1 m + r s n -1 r
R = n
(n , m)
m Σ Σ Σ r =0 s =0 t=0
f
x x
1
<m+ r> (m+r-s) (s-t) (t) n
f2 f3 f4 dx
a a
- 11 -
<n + r> (1+ ) <m+ r> (1+ )
x x +n+ , x
r r
= = x +m+
1+ + n + r 1+ + m + r
(r-s) (m+ r-s)
e x = e x = ex
s
(sin x) s = sin x +
()
2
Substituting these for Theorem 20.2.1 , we obtain
x
x x
x n
e sin xdx
- -
(1+ ) s
r s
-n r
+R
m -1 r
r
= ΣΣ x +n+ e x sin x + n
(1.1)
r =0 s =0 1+ + n + r 2 m
(-1)m n -1 m + r 1
r
n -1 m+r
(n , m) Σ Σ
Rmn =
r =0 s =0 m + r s
(1+ ) s
x x
r
x +m+ e xsin x + dx n (1.1r)
- -
1+ + m + r 2
And when n =2
( +1) +2+ r x
r s
s
x x m -1 r -2 r
x e x sin x dx = Σ Σ +R
2 2
x e sin x +
- - r =0 s =0 +3+ r 2 m
m+ r r s
(-1) 1 1
1 m+rm+ r
Σ Σ
2
Rm =
(2, m) r =0 s =0
(1+)
x s
x
x +m+ re xsin x + dx 2
1+ + m+ r - - 2
This remainder term is an automorphism. And the calculation is difficult when m is large. So let m =1 ,
then the above expression is simplified as follows.
( +1) +2 x
x
x x
e x sin x dx 2 = x e sin x + R12
- - +3
1+
(-1)1 1 1
0 s s
1 1 x x
( ) +1 x
(2, 1) Σ
R12 = x e sin x + dx 2
s =0 1 +2 - - 2
(1+)
s
1 2
1 1 x x
(-1) 2
+2 x
Σ
2
+ x e sin x + dx
(2, 1) 2 1 s
s =0 +3 - - 2
2( +1)
+2
x e sin xdx + x
x x x
x
+1 x 2 +1 x 2
= - e cos xdx
- - - -
2( +1)
+3
x x
+2 x 2
- x e cos xdx
- -
When =3 , if the values of the both sides on arbitrary point x =4 are calculated, it is as follows.
2
Although the remainder R1 is large because of m=1 , both sides are corresponding exactly.
- 12 -
When m
R = 0 at the time of m is asked, unfortunately, it is not so.
2
On the contrary, if whether to be
Surprisingly, when =3 , it becomes R = -3 regardless of the value of x . Then
2
(3+1) 3+2+ r x s
-3
x x
r s
r -2 r
x 3e xsin xdx 2 =Σ Σ x e sin x +
- - r =0 s =0 3+3+ r 2
As the result of much calculation, it turned out that the following formula holds.
Formula 20.2.1'
When = 0, 1, 2,
(1+ ) s
r s
-n r
x x r
x e xsin xdx n =Σ Σ
r
x +n+ e xsin x +
- - r =0 s =0 1+ + n + r 2
1 n -1
r
C r c +r x n-1-r
(n -1)! Σ
+ (-1) n -1 (1.1' )
r=0
Where
k-1
c4k+0 = -(-1)k 2 k !(2k -1)!!(4k -1)!!
k-1
c4k+1 = (-1)k 2 k !(2k -1)!!(4k +1)!!
k-1
c4k+2 = -(-1)k 2 k !(2k +1)!!(4k +1)!!
c4k+3 = 0
- 13 -
c3+0 = c40+3 = 0
c3+1 = c41+0 = -(-1)121-1 1!(21-1)!!(41-1)!! = 3
Substituting these for the above, we obtain R2 = -3 .
f f f dx r=0 r r s f
x x m -1 -n m -1 r -n r
1 2 3
n
= f3
(0)
Σ f1n+ r f2 r + Σ Σ
< > () <n+ r > (r -s) (s)
1 f2 f3 + Rmn
a a r=1 s=1
0 f f
(-1) n-1 n -1C r
m m+ r x x
<m + r> (m +r) (0) n
Rmn =
(n , m) Σ 1 2 f3 dx
r =0 m + r a a
(n , m) ΣΣ m+ r s
(-1) C m
m+r n-1 m + r n -1 r x x
<m + r > (m +r -s) (s) n
+ f 1 f2 f3 dx
r =0 s =1 a a
Substituting the above expressins for this,
(1+)
r
x x m -1 -n
x e x log x dx n = e x log xΣ x +n+ r
- - r=0 1+ + n + r
(1+) (s)
r s 1+ + n + r x
m -1 r -n r +n+ r-s x
+ Σ Σ(-1)s-1 e + Rmn (1.2)
r=1 s=1
- 14 -
(1+)
(-1)m n-1 n -1C r
x x
Rmn = x +m+ re x log x dx n
(n , m) Σ
r =0 m + r - - 1+ + m + r
(1+) (s) +m+ r-s x n
s
n -1C r
m m+ r x x
(-1) n-1 m + r
+ Σ Σ (-1)s-1 x e dx
(n , m) r =0 s =1 m+ r - - 1+ + m + r
(1.2r)
When =3/2 , n =3, m =80 , the values of the both sides on arbitrary point x=25 are as follows.
The right side of (1.2) seems to be asymptotic expansion.
Theorem 20.2.2
(r)
Let f be the r th order derivative function of f(x) , fk<r> be the arbitrary r th order primitive function of
f(x) , m , n are natural numbers and (n , m) be the beta function. At this time , if there is a number a
such that
r r r
x x m -1 r1 r-2 -n r1 r-2
f
n + r1 r1-r2
f -1 + Rmn
r
f dx =Σ Σ Σ
n
f
a a r1=0 r2=0 r-1=0 1 2 -1
(n , m) Σ Σ Σ Σ m + k k k k
(-1) m
C m +k k
n-1 m + k1 k2k k -2
n -1 k1 1 2 -2
Rmn =
k 1=0 k 2=0 k 3=0 k -1=0 1 2 3 -1
f
x x
m+ k 1 m+k 1-k 2 k 2 -k 3
f -1dx n
k
f f
a a
3
Example Higher Integral of log x
Let f = log x . Then
log x - 1+n + r - n+ r
<n+ r >
(log x) = x
1+n + r
(log x)
(r -s )
= log x (r =s) , (log x)
(r -s )
= (-1)r-s-1(r -s -1) ! x
-r+s
(r s)
(r 0)
(0) (s) -s
(log x) = log x (s =0) , (log x) = (-1)s-1(s -1) ! x
r =1, 2,
(0) (r ) (0)
Since f is different from f in function type, we have to separate the terms containing f
- 15 -
from
f r s f f f + R
x x m -1 r -n r
dx n = Σ Σ
3 <n+ r> (r -s ) (s ) n
m
a a r=0 s=0
(n , m) ΣΣ m + r r s
m n -1 m+ r
n-1 m + r x x
(-1) 1 <m + r> (m +r -s) (s)
Rmn = f f f dx n
r =0 s =0 a a
It is as follows.
r=0
r s=0 s
m -1 -n <n+ r >
r r (r -s) (s)
Σ f f Σ f
0 0 1 0 1 f f
-n 0 -n
<n+0>
1 1
(0) (0) <n+1> (1) (0) (0) (1)
= f f f + f f f +
r 0 s r f f
-n m -1 r r <n+ r> r (r) (0)
r-1
(r -s) (s) (0) (r)
+Σ f f f +Σ f f +
r=2 s=1
r r s f f
-n
<n> (0) (0) -n
m -1 r <n+ r> (r) (0)
m -1
<n+ r >
r-1
(r -s) (s)
= f f f + 2Σ f f f +Σ f Σ
r=1 r=2 s=1
Σ s f
m+ r
Σ r f
n -1
n-1
<m + r>
m +r (m +r -s) (s)
f
r =0 s =0
r s f f +f f
n -1 n-1 m+ r
<m + r> (m +r) (0) m + r-1 (m +r -s) (s ) (0) (m + r)
= Σ f f f + Σ
r =0 s =1
s f f
m+ r
r r
n -1 n-1 n -1 <m + r> (m +r) (0)
n-1
<m + r>
m + r-1 (m +r -s) (s)
= 2Σ f f f +Σ f Σ
r =0 r =0 s =1
Substituting thes for the above ,
f r f f f
x x m -1 -n
dx n = f
<n> (0) (0) <n+ r> (r) (0)
+ 2Σ
3
f f
a a r=1
r s f f f
-n r m -1 r-1
<n+ r> (r -s) (s)
+ ΣΣ
r=2 s=1
m+ r r
mn -1 n-1 x x
(-1) 2 <m + r> (m +r ) (0)
Rmn = Σ f f f dx n
(n , m) r =0 a a
m+ r r s
m n -1 m+ r
n-1 m + r-1 x x
(-1) 1 <m + r > (m +r -s) (s)
+ Σ Σ f f f dx n
(n , m) r =0 s =1 a a
Substituting the above expressions and the following expressions for this
- 16 -
+ Rmn (2.1)
r
1 n -1 m+ r
(n , m) Σ Σ
+
r =0 s =1 m+ r s
log x - 1+m + r -
x x
(m+ r -s) (s) dx n (2.1r)
0 0 1+m+ r
Although this formula is complicated, fortunately lim Rmn = 0 holds. However, the proof is difficult and the
m
convergence speed is very slow.
When n =2 , m=275 , the values of the both sides of (2.1) on arbitrary point x =0.3 are as follows.
The both sides are corresponding up to only 1 digit below the decimal point. If larger m is given, the accuracy
can be raised. However, it takes a lot of time.
Formula 20.2.3c
(1) Oddth power
n
x x x Cr
cos (2m-2r +1) x -
1 m 2m +1
cos x dx n =
2m+1
2m Σ n
(n-1) 1 0 2 r=0 (2m-2r +1) 2
2 2 2
(3.co)
(2) Eventh power
n
Cr
x x x 1 m -1
2m
cos 2m x dx n = Σ (2 n cos (2m -2r)x -
2
an a2 a1 22m-1 r=0 m -2r)
dx
Cm x x x
2m n
+ 2m
(3.ce)
2 an a2 a1
Cr k
=0
m -1
k =1, 2, ,n
2m
Σ cos (2m -2r)x -
r=0 (2 m -2r)k 2
- 17 -
Especially when m=1 ,
n
x x x 1 1 x x x
2 n
cos x dx = n+1 cos 2x - + dx n
(n -1) 1 0 2 2 2 (n -1) 1 0
4 4 4 4 4 4
(3.c2)
Proof
The following expression holds from Lemma1 .
1 m
cos 2m+1x = Σ 2m +1C r cos{(2m -2r +1)x }
22m r=0
Cr n
<n> 1 m 2m +1
cos 2m+1x = 2m Σ n cos (2m -2r +1)x -
2 r=0 (2 m -2r +1) 2
Here, let us consider the solutions of the following transcendental equations.
Cr k
=0
m
k =1, 2, ,n
2m +1
Σ cos (2m -2r +1)x -
r=0 (2 m -2r +1)k 2
Then
(k -1)
x = ak = k =1, 2, ,n
2
are the solutions of the transcendental equations. If it is why,
(k -1) k
cos (2m -2r +1)
2
-
2
= cos (m -r)(k -1) -
2
= (-1)(m-r)(k -1)cos =0
2
These solutions are not dependent on m. Thus we obtain (3.co) .
Next, the following expression holds from Lemma1 .
1 m -1 Cm
2m
cos 2mx = 2m Σ C r cos{(2m -2r)x } +
22m-1 r=0 22m
Integrating the both sides of this with respect to x n times,
Cr n Cm
<n> 1 m -1 2m 2m n -1
k
cos x Σ Σ
2m
= 2m-1 n
cos (2m-2r) x - + 2m
ck+1 x
2 r=0 (2m -2r ) 2 2 k=0
Here, let a k k =1, 2, ,n are the solutions of the following transcendental equations.
Cr k
=0
m -1
k =1, 2, ,n
2m
Σ cos (2m -2r)x -
r=0 (2 m -2r)k 2
Then, (3.ce) follows immediately.
Especially when m =1 , these transcendental equations are as follows.
2C 0 k
2k
cos 2 x -
2 =0
k =1, 2, ,n
Then
(k -1)
x = ak = k =1, 2, ,n
4
- 18 -
are solutions of the transcendental equations. If it is why,
C0 (k -1) k
1
k =1, 2, ,n
2
cos 2 - = k cos 2
=0
2k 4 2 2
These solutions are unrelated to m. Thus, (3.c2) holds.
Note1
According to Formula20.1.3 , the n th order derivative of cos 2m+1x is as follows.
n
(n) 1 m
cos 2m+1x = Σ 2m +1C r (2m +1-2r)
n
cos (2m +1-2r)x +
2 2m
r=0 2
Since an integration is the inverse operation of a differentiation, replacing n with -n ,
Cr n
(-n ) 1 m 2m +1
cos 2m+1x = Σ n cos (2m -2r +1)x -
2 2m
r=0 (2m -2r +1) 2
Replacing the differentiation operator with the integration operator, we can obtain (3.co) .
Note2
dx
x x
The general notation of n
is difficult. By reference, if these are shown for n =1, 2, 3 , it is as
an a1
follws. The 4th order or more are unmanageable.
x -a 1
x -a 2x -2a 1+ a2
x x x
1 2
dx = , dx =
a1 1! a2 a1 2!
x -a3 x + a3 x -3a 1 x + 6a 1 a2 -3a1 a3 -3a 2 + a3
2 2 2
dx
x x x
3
=
a3 a2 a1 3!
These resemble a part of constant-of-integration in 4.1.3 . However, these are completely another. That is,
these are undoubted parts of the lineal higher primitive function that originates in cos 2x = (cos 2x - 1)/2
etc. .
5
Example1 The 3rd order integral of cos x
2
Example2 The 3rd order integral of cos x
- 19 -
Formula 20.2.3s
(1) Oddth power
(-1)m-r2m +1C r n
x x x
1 m
sin
2m+1
x dx n = 2m Σ n
sin (2m -2r +1) x -
n 2 1 2 r=0 (2m -2r +1) 2
2 2 2
(3.so)
(2) Eventh power
n
sin
(-1)m-r2mC r
x x x 1 m -1
n
Σ
2m
x dx = n cos
(2m -2r)x -
an a2 a1 22m-1 r=0 (2m -2r) 2
dx
Cm x x x
2m n
+ 2m
(3.se)
2 an a2 a1
(-1)m-r2mC r k
=0
m -1
Σ k cos (2m -2r)x - k =1, 2, ,n
r=0 (2m -2r) 2
Especially when m =1 ,
n
x x x x x x
1 1
sin x dx n = -
2
n+1
cos 2x - + dx n
(n+1) 3 2 2 2 2 (n+1) 3 2
4 4 4 4 4 4
(3.s2)
Proof
Replacing x with x - /2 in (3.co) in Formula 20.2.3c ,
n
x x x Cr
cos (2m -2r +1) x - -
1 m 2m +1
sin
2m+1
x dx n = Σ
n 2 1 22m r=0 (2m-2r +1) n 2 2
2 2 2
Here
n (2m-2r +1) n
cos (2m-2r +1) x - 2 -
2
= cos (2m-2r +1) x -
2
-
2
n
= sin (2m -2r +1)x -
2
-(m -r)
n
= (-1)m-rsin (2m -2r +1)x -
2
Substituting this for the above, we obtain (3.so) . (3.se) and (3.s2) are obtained from Formul20.2.3c in a
similar way.
7
Example1 The 3rd order integral of sin x
- 20 -
2
Example2 The 4th order integral of sin x
Lemma2
1
cos x = Σ
2 r=0 r
cos( -2r)x >0 , |x|
2
(4. )
Formula 20.2.4
cos x dx = 2 Σ r -2r
x 11
sin ( -2r) x (4.c)
0 r=0
x 1 1
sin x dx =
Σ sin ( -2r) x - 2 (4.s)
2 r -2r r=0
2
Especially when = 2m ,
x 1 m -1 Cr
2m 2mCm
2m-1 Σ
cos 2m x dx = sin {(2m -2r)x } + x (4.ce)
0 2 r=0 2m -2r 22m
- 21 -
(-1)m-r2mC r
x Cm
1 m -1 2m
sin 2m x dx =
2 2m-1 Σ
r=0 2m-2r
sin(2m-2r) x +
2 2m
x-
2
2 (4.se)
Proof
Integrating both sides of (4. ) in Lemma2 with respect to x from 0 to x, we obtain
cos r=0 r
x 1 1
x dx = Σ sin( -2r) x (4.c)
0 2 -2r
Especially when = 2m ,
m -1 2m Cr 2m 2m Cr
Σ
r=0 2m -2r
sin (2m -2r)x = Σ
r=m +1 2m -2r
sin (2m -2r)x
Using this,
x 1 2m 2mCr
2m Σ
cos 2m x dx = sin (2m -2r)x
0 2 r=0 2m -2r
Cr Cr
2m Σ
sin (2m -2r)x
1 m -1 2m 2m 2m
= sin (2m -2r)x + Σ
2 r=0 2m -2r r=m +1 2m -2r
Cm
2m sin (2m -2r)x
+ lim
22m rm 2m -2r
2 m -1 2m Cr Cm
2m
= Σ 2m -2r sin (2m -2r)x + x (4.ce)
22m r=0 22m
(4.s), (4.se) are obtained by the replacing x with x - /2 .
Note
Even if the number of is non-even number, the following formula does not hold generally.
1 n
( -2r)
x x x 1
cos x dx n = Σ cos ( -2r)x -
an a2 a1 2 r=0 r n 2
It is because that the right side is always a real valued function although the left side is generally a complex
valued function for the real number . Although the left side can be a real valued function for the real number
, when all a 1 , a 2 , ,an is below /2 , it is difficult to find out such a case. This is the same also
about sin x .
3
Example1 The 1st order integral of sin x
- 22 -
Since both sides have overlapped, the left side (blue) can not be seen.
8
Example2 The 1st order integral of cos x
Since both sides have overlapped, the left side (blue) can not be seen.
2010.12.21
K. Kono
Alien's Mathematics
- 23 -
21 Super Calculus of the product of many functions
(1) Generalized binomial theorem and Super integral of the product of 2 functions
According to the generalized binomial theorem in 3.2 , the following expression holds for the real numbers
x1 , x2 such that x 1 >x 2 and a positive number p .
r
-p
x1+ x2 = Σ x1-p-rx2r
-p
r=0
On the other hand, according to Formula 17.1.2 in 17.1 , the following expression holds for the functions f1 , f2
of x and a positive number p .
r f
x x m -1 -p
f1 f2
<p>
dx 2= Σ <p +r> (r)
1 f2 + Rmp
a a r=0
k f
(-1)m 1 p -1 x x
<m+ k > (m+ k ) p
(p , m) Σ
Rmp = 1 f2 dx
k =0 m + k a a
Reversing the sign of the index of the differentiation operator (p ) in the Leibniz Rule about the Super
Differentiation in 19.1 ,
r
m -1 -p
f1 f2
(-p )
=Σ f1(-p -r)f2(r) + Rm-p
r=0
(-1)m 1
k
p -1 <m+ k > (m+ k ) (-p )
Rm-p =
(p , m) Σ
k =0 m + k
f1 f2
And, replacing (-p) with the intagration operator <p > , we obtain the above formula.
(2) Generalized multinomial theorem and Super Integral of the product of many functions
According to the generalized multinomial theorem in 3.4 , the following expression holds for real numbers
x1 , x2 , , x such that x 1 >x 2 + x 3 ++ x and a positive number p .
r r r
r1 r-2 -p r1 r-2 -p-r1 r -r2 r
x1+ x2++ x =Σ Σ Σ x-1
-p
x1 x21
r1=0 r2=0 r-1=0 1 2 -1
Therefore, the following expression must hold for functions f1, f2 , , f of x and a positive number p .
r r r
x x m -1 r1 r-2 -p r1 r-2
f1
p + r1 r1 -r2
f -1
r
f1 f2 f dx =Σ Σ Σ
p
f2
a a r1=0 r2=0 r-1=0 1 2 -1
+ Rmp
Theorem 21.1.1
m be a natural number, fk r
()
Let p,r are positive numbers, be the r th order derivative function of
fk (x) k =1, 2, , , fk r r th order primitive function of fk (x) and (p ,m) be
<>
be arbitrary
the beta function. At this time, if there is a number a such that
f1r (a ) = 0 r [0 , m+ p] or fk s (a ) = 0 s [0 ,m +p -1] for at least one k > 1 ,
<> ()
-1-
r r r
x x m -1 r1 r-2 -p r1 r-2
f1
p + r1 r1 -r2
f -1
r
f1 f2 f dx =Σ Σ Σ
p
f2
a a r1=0 r2=0 r-1=0 1 2 -1
+ Rmp
k k
k-2
k
(-1)m m + k1 k2 k -2
1 p -1 m + k1 k2
Rmp = Σ Σ Σ Σ
(p , m) k 1=0 k 2=0 k 3=0 k -1=0 m + k1 1 k2 3 -1
x x
f1
m+ k 1 m+k 1 -k 2 k 2 -k 3
f -1dx p
k
f2 f3
a a
Proof
Analytically continuing the index of the integration operator in Formula 20.2.1 in 20.2 to [0 ,p ] from[1 , n ]
we obtain the desired expression.
x
x x
x p
e sin xdx
- -
(1+ ) s
r s
-p r
+R
m -1 r
r
= ΣΣ x +p+ e x sin x + p
(1.1)
r =0 s =0 1+ + p + r 2 m
(-1)m m + r 1
r
p -1 m+r
(p , m) Σ Σ
Rmp =
r =0 s =0 m + r s
(1+ ) s
x x
r
x +m+ e xsin x + dx p (1.1r)
-
- 1+ + m r
+ 2
As seen in Theorem 20.2.1 in 20.2 , this Rmp is not converged on 0 at the time of m . That is, this
polynomial is an asymptotic expansion. When =3 , p =5/2 , the values of the both sides on arbitrary
point x = 20 are as follows.
-2-
Example2 Super Integral of x e x log x
Since the zero of the higher integral of x e x log x is x =- , the zero of the super integral is also
x
the same. Then let f 1 = x , f 2 = e , f 3 = log x
(1+) (1+)
x x r x x
<p+ r> +p+ <m + r> +m+ r
= , =
1+ + p + r 1+ + m + r
(r -s) (m + r -s)
e x = ex , e x = ex
(s 0)
(0) (s) -s
(log x) = log x (s =0) , (log x) = (-1)s-1(s -1) ! x
(0)
Separating the terms containing f3 from Theorem 21.1.1 and substituting these fot it,
(1+)
x r
x x
m -1 -p
e x log x dx p = e x log xΣ x r
+p+
- - r=0 1+ + p + r
(1+) (s)
r s 1+ + p + r x e + R
-pm -1 r r
+p+ r-s x p
- Σ Σ(-1) s
m (1.2)
r=1 s=1
(1+)
m
m+ r r 0
(-1) 1 p -1 m+ r x x
+m+ r x
Rmp = Σ x e log x dx p
(p , m) r =0 1+ + m + r - -
(1+) (s)
m m +r s-1
m+ r r s
(-1) (-1) p -1 m+r x x
+m+ r-s x p
+
(p , m) Σ Σ r =0 s =1 1+ + m+ r
x e dx
- -
(1.2r)
When =3/2 , p =2.7 , the values of the both sides on arbitrary point x = 17 are as follows. This also
seems to be asymptotic expansion.
Theorem 21.1.2
m be a natural number, f r be the r th order derivative function of f(x) ,
()
Let p,r are positive numbers,
f r be arbitrary r th order primitive function of f(x) and (p , m) be the beta function. At this time,
<>
-3-
the following expression holds for = 2, 3, 4, .
r r r
x x m -1 r1 r-2 -p r1 r-2
f
p + r1 r1 -r2
f dx p =Σ Σ Σ f -1 + Rmp
r
f
a a r1=0 r2=0 r-1=0 1 2 -1
m + k k k k k
(-1) m p -1
m + k1 k2
1 m +k k kk -2 1 2 -2
Rmp = Σ Σ Σ Σ
(p , m) k 1=0 k 2=0 k 3=0 k -1=0 1 1 2 3 -1
f
x x
m+ k 1 m+k 1 -k 2 k 1 -k 2
f -1dx p
k
f f
a a
3
Example Super Integral of log x
3
Since the zero of the higher integral of log x is x =0 , the zero of the super integral is also the same.
Then let f = log x
<p+ r> log x - 1+p + r - p+ r log x - 1+m + r - m+ r
, (log x) m r =
< + >
(log x) = x x
1+p + r 1+m + r
(log x)
(r -s)
= log x (r =s) , (log x)
(r -s)
= (-1)r-s-1(r -s -1) ! x
-r+s
(r s)
(r 0)
(0) (s) -s
(log x) = log x (s =0) , (log x) = (-1)s-1(s -1) ! x
(0)
f from Theorem 21.1.2 and substituting these fot it,
Separating the terms containing
log x - 1+p - p
x x
(log x) 3dx p = x (log x)
2
0 0 1+p
-p log x - 1+p + r -
m -1
- 2x p log xΣ (-1)r r
r=1 r 1+p + r
r log x - 1+p + r -
m -1 r-1 -p
+ x p Σ Σ(-1)r r -s (s)
r=2 s=1 r s 1+p + r
+ Rmp (2.1)
m + r (-1) r-s
r s
1 p -1 m+ r
Rmp = -
(p , m) Σ Σ
r =0 s =0 m + r
log x - 1+m + r -
x x
(m+r -s) x sdx p (2.1r)
0 0 1+m+ r
n
And lim R =0 holds although the proof is difficult. Then
m m
log x - 1+p - p
(log x) dx
x x
3 p
= x (log x)
2
0 0 1+p
-p log x - 1+p + r -
- 2x p log xΣ(-1)r r
r=1 r 1+p + r
r log x - 1+p + r -
r-1 -p
+ x pΣΣ(-1)r r -s (s)
r=2 s=1 r s 1+p + r
(2.1')
However, the convergence speed is very slow.
When p =3/2 , m =400 , the values of the both sides on arbitrary point x = 0.2 are as follows. Even if
calculated so far, both sides are corresponding only to 1 digit below the decimal point. Although (2.1') is not
suitable for the calculation, it is meaningful that the integral can be expressed by the double series.
-4-
21.1.3 Super Integrls of cos m x , sin m x
Formula 21.1.3
Let m be a natural number, p be a positive number and a(s) s [0 , p ] be the zero of the lineal super
2m+1
primitive function of cos x or sin 2m+1x . Then the following expressions hold.
p
2m +1C r
x x
1 m
cos
2m+1
x dx p = 2m Σ cos ( 2m -2r +1 ) x - (3.c)
a(p) a(0) 2 r=0 (2m -2r +1) p 2
(-1)m-r2m +1C r p
x x
1 m
x dx p = Σ
2m+1
sin 2m p
sin (2m-2r +1) x - (3.s)
a(p) a(0) 2 r=0 (2m -2r +1) 2
Proof
Analytically continuing the index of the integration operator in Formula 20.2.3c (1) and Formula 20.2.3s (1)
in 20.2 to [0 , p ] from[1 , n] , we obtain the desired expressions.
Note
For example, in the case of cos 2m+1x , a(s) is a solution of the following transcendental equation.
Cr s
s cos ( m =0
m
s [0 , p]
2m +1
Σ
r=0 (2m -2r +1)
2 -2r +1)x -
2
Although it is difficult to calculate this solution, the necessity does not exist in this formula.
3
Example The 1.5 th order intagral of cos x
-5-
-6-
21.2 Super Derivatives of the product of many functions
(1) Generalized binomial theorem and Leibniz Rule about Super Differentiation
According to the generalized binomial theorem in 3.2 , the following expression holds for the real numbers
x1 , x2 such that x 1 >x 2 and a positive number p .
r
p p-r
x1+ x2 = Σ
p
x1 x2r
r=0
On the other hand, according to Formula 19.1.1 in 19.1 , the following expression holds for the functions f1 , f2
rf
(p) m -1 p (p -r) (r)
f1 f2 =Σ 1 f2 + Rmp
r=0
(-1)m 1
-p -1 <m+ k > (m+ k ) (p)
Rmp =
(-p , m) Σ
k =0 m + k k
f1 f2
(2) Generalized multinomial theorem and Super Derivative of the product of many functions
According to the generalized multinomial theorem in 3.4 , the following expression holds for real numbers
x1 , x2 , , x such that x 1 >x 2 + x3 ++ x and a positive number p .
r r
r1 r-2 p r1 r-2 p-r1 r -r2 r
x-1
p
1 2
x + x ++ x =Σ Σ Σ
r1=0 r2 =0 r-1 =0 r1
x1 x21
2 -1
Therefore, the following expression must hold for functions f1 , f2 , , f of x and a positive number p .
r r r
m -1 r1 r-2 p r1 r-2
f1
p - r1 r1 -r2
f -1 +
r
f1 f2 f = ΣΣ Σ
(p)
f2 Rmp
r1=0 r2=0 r-1=0 1 2 -1
Theorem 21.2.1
m be a natural number, fk r
()
Let p,r are positive numbers, be the r th order derivative function of fk (x)
k =1, 2, , , fk r r th order primitive function of fk (x) and (p , m) be the beta function.
<>
be arbitrary
At this time, if there is a number a such that
f1r (a) = 0 r [0 , m - p] or fk s (a) = 0 s [0 , m-p -1]
<> ()
for at least one k >1 ,
the following expression holds
r r r
m -1 r1 r-2 p r1 r-2
f1
p - r1 r1 -r2
f -1
r
f1 f2 f =Σ Σ Σ
(p) p
+R f2 m
r1=0 r2=0 r-1=0 1 2 -1
(-p , m) Σ Σ Σ Σ m + k k k k k
(-1) m -p -1 m +k
m + k1 k2
1 k kk -2 1 2 -2
Rmp =
k 1=0 k 2=0 k 3=0 k -1=0 1 1 2 3 -1
(p)
m+ k 1 m+k 1-k 2 k 2-k 3
f1 f2 f3 k -1
f
Proof
Reversing the sign of the index of the integration operator <p> in Formula 21.1.1 ,
-7-
r r r
x x m -1 r1 r-2 p r1 r-2
f1
-p + r1 r1 -r2
f -1
r
f1 f2 f dx -p
=Σ Σ Σ f2
a a r1=0 r2=0 r-1=0 1 2 -1
+ Rm-p
k k
k-2
(-1)m m + k1 k2 m + k1 k2
k -2
1 -p -1
Rm-p = Σ Σ Σ Σ
(-p , m) k 1=0 k 2=0 k 3=0 k -1=0 m + k1 k1 k2 3 -1
f
x x
m+ k 1 m+k 1-k 2 k 2-k 3
f -1dx -p
k
1 f2 f3
a a
Then, replacing the integration operator <-p> with the differentiation operator (p) , we obtain the desired
expression.
(1.1)
m
m+ r
(-1) m+r 1 -p -1
Rmp = Σ Σ
(-p , m) r =0 s =0 m + r r s
(1+) s
(p)
x
+m+ r x
e sin x + (1.1r)
1+ + m + r 2
When =3/2 , p =1/2 , the values of the both sides on arbitrary point x = 1.4 are asfollows.
-8-
(s 0)
(0) (s) -s
(log x) = log x (s =0) , (log x) = (-1)s-1(s -1) ! x
(0)
Separating the terms containing f3 from Theorem21.2.1 and substituting these for it,
(1+)
r
x
m -1 p
x r
(p)
x e log x = e xlog xΣ
-p+
r =0 ( 1+ - p + r )
(1+) (s)
r s (1+ - p + r ) x
m -1 rp r -p+ r-s
- e Σ Σ (-1)
x s
+ Rmp (1.2)
r =1 s =1
-p -1
1
x
+m+ r x
Rmp = Σ e log x
(-p , m) r =0 m + r r ( 1+ + m + r )
(-1)m m + r (-1)s (1+) (s) +m+ r-s x (p)
s
-p -1 m+r
-
(-p , m) Σ Σ
r =0 s =1 m + r r ( 1+ + m+ r )
x e
(1.2r)
Polynomials in (1.2) seem to be asymptotic expansion. when =3/2 , p =1/2 , the values of the both sides
on arbitrary point x = 16 are as followas.
Theorem 21.2.2
m be a natural number, f r be the r th order derivative function of f(x) ,
()
Let p,r are positive numbers,
f r be arbitrary r th order primitive function of f(x) and (p , m) be the beta function. At this time,
<>
r r r
m -1 r1 r-2 p r1 r-2
f
p + r1 r1 -r2
(p)
f -1 + Rmp
r
f = ΣΣ Σ f
r1=0 r2=0 r-1=0 1 2 -1
(-p , m) Σ Σ Σ Σ m + k k k k k
(-1) m -p -1 m +k
m + k1 k2
1 k k k -2 1 2 -2
Rmp =
k 1=0 k 2=0 k 3=0 k -1=0 1 1 2 3 -1
(p)
f f -1
m+ k 1 m+k 1 -k 2 k 2 -k 3 k
f f
3
Example Super Derivative of log x
Let f = log x , then
-9-
(p- r) log x - 1-p + r - -p+ r log x - 1+m + r - m+ r
, (log x) m r =
< + >
(log x) = x x
1-p + r 1+m + r
(log x)
(r -s)
= log x (r =s) , (log x)
(r -s)
= (-1)r-s-1(r -s -1) ! x
-r+s
(r s)
(r 0)
(0) (s) -s
(log x) = log x (s =0) , (log x) = (-1)s-1(s -1) ! x
(0)
Separating the terms containing f3 from Theorem21.2.2 and substituting these for it , we obtain
(p) log x - 1-p - -p
(log x)
3 2
= x (log x)
1-p
p log x - ( 1-p + r ) -
m -1
- 2x log xΣ (-1)r
-p
(r)
r=1 r ( 1-p + r )
r log x - ( 1-p + r ) -
m -1 r-1 p
+ x Σ Σ(-1)r
-p
(r -s) (s)
r=2 s=1 r s ( 1-p + r )
+ Rmp (2.1)
(-1) r 2
1 -p -1
log x - 1+m + r -
(p)
Rmp = -
(-p , m) Σ 2
log x
r =0 (m + r ) r
m + r-1 (-1)r
m+ r
1 -p -1
(-p , m) Σ Σ
+
r =0 s =1 m + r r s
log x - 1+m + r -
(p)
1+m + r
(m + r -s) (s)
(2.1r)
n
And lim R =0 holds although the proof is difficult. Then
m m
- 10 -
21.2.3 Super Derivatives of cos m x , sin m x
Formula 21.2.3
When m is a natural number, p is a positive number and is the floor function,
p
Σ C r (m -2r) cos(m -2r)x +
(p) 1 m /2
cos mx = m-1 m
p
(3.c)
2 r=0 2
p
Σ C r (m -2r) cos(m -2r)x - 2 +
(p) 1 m /2
sin x
m p
= m-1 m (3.s)
2 r=0 2
Proof
Analytically continuing the index of the differentiation operator in Formula 20.1.3 in 20.1 to [0 , p ] from
[1 , n ] , we obtain the desired expressions.
4
Example The 2.5 th order derivative of cos x
- 11 -
3
For example, if we differenciate cos x with respect to x p times according to Theorem 21.2.2 , it is as
follows.
(p -r) (r -s) s
r s
p r
cos x + 2
(p) m -1 r
cos x =Σ Σ
3
cos x + cos x +
r=0 s=0 2 2
+ Rmp (4.c')
(-1)m
s
m +r m+r
1 -p -1
Rmp = Σ Σ
(-p , m) r =0 s =0 m + r r
(m +r) (m +r -s) s
p
cos x -
2
cos x +
2
cos x +
2 (4.cr')
Then, this is a collateral super derivative. If it is why, this theorem was drawn out as a reverse-operation of the
super integral of the product of many functions as seen during the proof of Theorem 21.2.1 . And the base was
the super integral with a fixed lower limit.
3
However, the lineal super integral of cos x is one with a variable lower limit. Therefore, (4.c') and (4.cr') derived
based on the fixed lower limit cannot be the lineal super derivative. Althogh this holds as a equation, the
polynomial of (4.c') is not well behaved.
By reference, let us compare (4.c') with the following lineal super derivative (4.c) derived from Formula 21.2.3 .
p 3p p
3 (p) 3
cos x = cos x + + cos 3x + (4.c)
4 2 4 2
Althogh (4.c') fits (4.c) most at the time of m=3 , a big difference is still seen by both.
n -r (r -s) s
r s
n r
n r
(n)
cos x =ΣΣ
3
cos x + cos x + cos x +
r=0 s=0 2 2 2
And this results in the following lineal super derivative. ( The proof is long, then omitted. )
n 3n n
+
3 (n) 3
cos x = cos x + cos 3x +
4 2 4 2
2010.12.25
K. Kono
Alien's Mathematics
- 12 -
22 Higher Derivative of Composition
(n =1, 2, 3, )
(n) (n) (n)
g = g ( f) , fn = f (x)
j1 j2 jn
n! f1 f2 fn
Bn,rf1 , f2 , , fn =Σ
j1! j2! jn! 1! 2! n!
(4)
Example gf(x)
First, j1 , j2 , j3 , j4 such that j1+j2+j3+j4 = 1 , j1+2j2+3j3+4j4 = 4 are
j1 , j2 , j3 , j4 = (0 , 0 , 0 , 1)
Then,
0 0 0 1
4! f1 f2 f3 f4
B4,1f1 f4 = = f41
0!0!0!1! 1! 2! 3! 4!
2nd, j1 , j2 , j3 , j4 such that j1+j2+j3+j4 = 2 , j1+2j2+3j3+4j4 = 4 are
j1 , j2 , j3 , j4 = (1 , 0 , 1 , 0)
j1 , j2 , j3 , j4 = (0 , 2 , 0 , 0)
Then,
1 0 1 0
4! f1 f2 f3 f4
B4,2f1 f4 =
1!0!1!0! 1! 2! 3! 4!
0 2 0 0
0!2!0!0! 1! 2! 3! 4!
4! f f f 1 f 2 3 4 1 1 2
+ = 4f 1 f3 +3f2
j1 , j2 , j3 , j4 = (2 , 1 , 0 , 0)
Then,
2 1 0 0
4! f1 f2 f3 f4
B4,3f1 f4 = = 6f12f21
2!1!0!0! 1! 2! 3! 4!
Last, j1 , j2 , j3 , j4 such that j1+j2+j3+j4 = 4 , j1+2j2+3j3+4j4 = 4 are
j1 , j2 , j3 , j4 = (4 , 0 , 0 , 0)
-1-
Then,
4 0 0 0
4! f1 f2 f3 f4
B4,4f1 f4 = = f14
4!0!0!0! 1! 2! 3! 4!
Thus,
(4)
gf(x) = g(1) B4,1f1 , f2 , f3 , f4 + g(2) B4,2f1 , f2 , f3 , f4
+ g(3)B4,3f1 , f2 , f3 , f4 + g(4) B4,4f1 , f2 , f3 , f4
= g(1)f41 + g(2)4f11 f31 + 3f22 + 6 g(3)f12 f21 + g(4)f14
Formula 22.1.1'
Let , , , and n are natural numbers, g
(n)
, fn are g
(n)
=g
(n)
(f ) , fn = f
(n)
(x)
Then, the higher derivative function with respect to x of the composition gf(x) is expressed as follows.
( + ++)
n! fp fq ft
p +q ++t = n
(n)
gf(x) = Σg
p! q! t! !!!
(1.1')
(1) Obtain all the products fp fq ft such that p +q ++t = n as follows.
Step1 Put n on the 1st row 1st column, and arrange 0 from the 2nd column to the
n th column .
Step2 Look for the first number that is smaller than the number of the 1st column as for 2 or more
sequentially from the 2 nd column.
I When such a number exists in the k th column, make a new sequence ( new row ) as follows.
i From the 2 nd column to the k th column, arrange the number that is 1 greater than
the number in the k th column of the old sequence ( the row just above ) .
ii From the k +1 th column to the n th column, copy the numbers of old sequence ( row just above )
iii To the 1 st column, write the complement that the total of each column becomes n .
II When such a number does not exist, finish calculating and go to Step4 .
Step3 Repeat Step2 for the new sequence ( new row ) .
Step4 Let the number in the each sequence be the order p and the repetition frequency
be the degree . And generate the product fp fq ft .
-2-
For example, the combination of the 7 th order is generated as follows.
Seq. 1 2 3 4 5 6 7
7 0 0 0 0 0 0 f71 Red is smaller than the 1st column by 2 or more.
5 2 0 0 0 0 0 f51 f21
4 3 0 0 0 0 0 f41 f31
5 1 1 0 0 0 0 f51 f12
4 2 1 0 0 0 0 f41 f21 f11
3 3 1 0 0 0 0 f32f11
3 2 2 0 0 0 0 f31 f22
4 1 1 1 0 0 0 f41 f13
3 2 1 1 0 0 0 f31 f21 f12
2 2 2 1 0 0 0 f23f11
3 1 1 1 1 0 0 f31 f14
2 2 1 1 1 0 0 f22f13
2 1 1 1 1 1 0 f21 f15
1 1 1 1 1 1 1 f17
(2) The coefficient of each term fp fq ft is calculated as follows, for example.
7!
Coefficient of f31 f21 f12 : = 210
3! 2! 1!21!1!2!
1 1
3 1 7!
Coefficient of f2 f1 : 3 1
= 105
2! 1! 3!1!
fp fq ft
(s)
(3) The multiplier g of each term is calculated as follows, for example..
(s)
Multiplier of f31 f21 f12 : g = g(1+1+2) = g(4)
6! 1! 1!1!
7! f71 (1)
f61 f11 f51 f21 f41 f31
z7 = 1 g + 7! 1 1
+ 1 1
+ 1 1 g(2)
7! 1! 5! 2! 1!1! 4! 3! 1!1!
-3-
3! 1! 1!4!
f31 f14 f22f13
+ 7! 1 1
+ 2 3 g(5)
2! 1! 2!3!
7! f21 f15 (6)
7! f17
+ 1 5 g + 1 g(7)
2! 1! 1!5! 1! 7!
= g(1)f7 + g(2)7 f6 f1 + 21 f5 f2 + 35 f4 f3
+ g(3)21 f5 f12 + 105 f4 f2 f1 + 70 f32f1 + 105 f3 f22
+ g(4)35 f4 f13 + 210 f3 f2 f12 + 105 f23f1
+ g(5)35 f3 f14 + 105 f22f13 + 21g(6)f2 f15 + g(7)f17
3
Example When z = gf(x) = sinx
-4-
Then,
z1 = cosx 33x 2
2
z2 = cosx 36x - sinx 33x 2
3
z3 = cosx 36 - 3 sinx 33x 26x - cosx 33x 2
s
(n) n g r r (n)
gf(x) =Σ Σ (-f)r-sf s (1.2)
r=0 r! s=0
Although this formula is beautiful, the calculation is not easy. When the differentiation is actually performed
according to (1.2), it is as follows.
(r)
s (-f)
2 g r r r-s s (2)
gf(x)
(2)
=Σ Σ f
r=0 r! s=0
() ()
g 0 g 1
0 s=0 s
0 0 (2)
1 1 s (2)
(-f) f Σ f
0 1-s
= + (-f)
0! 1!
( )
g 2
s=0 s
2 2 s (2)
Σ f
2-s
+ (-f)
2!
()
g 1
0 1
1 (2) 1 (2)
(-f) f (-f) f
1 0 0 1
= +
1!
(2)
2! 0 1 2 (-f) f
g 2 2 0 (2)
2 1 (2) 2 (2)
(-f) f + (-f) f +
2 1 0
+
(1) (2)
1! 1 2! 1 2 2f f + f f
g 1 g 2
1
2 1 1
= f + 2 (-f) f + 2 1 1 2
g(1)f21 g(2)
2!
= + -2f f2 + 2f1 f1 + f f2 = g(1)f2 + g(2)f12
1!
-5-
Further, replacing the table {x1 , x2 , , xn-r+1} with a function, we can generate Formula 22.1.1 itself.
For example, let
Of course, if a specific function is given toor fk , a desired higher order derivative function can be obtained.
gr
The examples are shown in the following chapters.
22.1.4 Higher Derivative in case the core function is the 1st degree
Although the higher differentiation of a general composition is complicated in this way, If the core function
f(x ) is the 1st degree, it becomes remarkably easy.
Formula 22.1.4
When g
(n)
=g
(n)
( f ) , fn = f (x) (n =1, 2, 3, ),
(n)
if f(x) is the 1st degree, the higher derivative
Proof
In Formula 22.1.1 , when f(x) is the 1st degree, f2 = f3 = f4 = = 0 . Then, the Bell polynomials
Bn,rf1 , f2 , are as follows.
j1 j2 jn
=0
n! f1 f2 fn
Bn,rf1 , f2 , , fn =Σ
j1! j2! jn! 1! 2! n!
n! f1 f2 fn
Bn,nf1 , f2 , , fn =Σ = f1n
n! 0! 0! 1! 2! n!
Therefore,
n
(n)
gf(x) = Σg(r) Bn,rf1 , f2 , , fn
r=1
n -1
= Σg(r) Bn,rf1 , f2 , , fn + g(n) Bn,nf1 , f2 , , fn
r=1
= g(n)f1n
-6-
Example When z = sin(ax+b )
1
z1 = cos(ax +b)a 1 = a 1 sin ax +b +
2
2
z2 = -sin(ax +b)a 2 = a 2 sin ax +b +
2
n
zn =
a n sin ax +b +
2
The last expression is consistent with linear form in Formula 9.2.1 in " 9.2 Higher Derivative ".
-7-
22.2 Higher Derivative of Some Composition
There are quite a lot of combinations even if the composition is limited to the elementary function. We cannot
calculate these one by one. So, in this section, we calculate only an easy and interesting thing.
e f
(x)
22.2.1 Higher Derivative of
g = e f(x
)
Placing with in Formula 22.1.1 , we obtain the following formulas immediately.
n
f(x) (n)
e = e f(x)ΣBn,kf1 , f2 , , fn (2.1+)
k =1
n
-f(x) (n)
e = e -f(x)Σ(-1)rBn,kf1 , f2 , , fn (2.1-)
k =1
If (2.1+) is written down to the 5th order, it is as follows.
( ) (1)
e f x = e f(x)f1
( ) (2)
e f x = e f(x)f2 + f12
( ) (3)
e f x = e f(x)f3 + 3f2 f1 + f13
( ) (4)
e f x = e f(x)f4 + 4f3 f1 +3f22 + 6f2 f12 + f14
( ) (5)
e f x = e f(x)f5 + 5f4 f1 +10f3 f2 + 10f3 f12+15f22f1 + 10f2 f13 + f15
2 (3)
Example1 e -x
f = -x 2 , f1 = -2x , f2 = -2 , f3 =0
Then,
2 (3) 2 2
e -x = e -x 0 + 3(-2)(-2x) + (-2x)3 = 4e -x x3 - 2x 2
(4)
Example2 e sin x
f = sin x , f1 = cosx , f2 = -sin x , f3 = -cosx , f4 = sin x
Then,
sin x (4)
e = e sin xsin x + -4cos 2x +3sin 2x - 6 sin xcos 2x + cos 4x
-cos x (8)
Example3 e
We calculate using a formula manipulation software Mathematica according to (2.1-) .
n
f = cosx , fn = cos x + 2 n =1, 2, 3,
So,
-8-
22.2.2 Higher Derivative of ge x
Placing with f = e x in Formula 22.1.1 , we obtain the following formulas.
(n) n
ge x = Σg(r) Bn,re x , , e x
r=1
(n) n
ge = (-1)nΣg(r) Bn,re -x , , e -x
-x
r=1
If the first expression is written down to the 5th order, it is as follows.
x (1)
ge
(1) x
=g e
(2)
ge x = g(1)e x + g(2)e 2x
(3)
ge x = g(1)e x + 3g(2)e 2x + g(3)e 3x
(4)
ge x = g(1)e x + 7g(2)e 2x + 6g(3)e 3x + g(4)e 4x
(5)
ge x = g(1)e x + 15g(2)e 2x +25g(3)e 3x +10g(4)e 4x + g(5)e 5x
Coefficients (1), (1, 1) , (1, 3, 1) , (1, 7, 6, 1) , (1, 15, 25, 10, 1) , of these right sides are called
Stirling numbers of the 2nd kind and is given by the following expression.
1 r r
S(n,r) = r Σ(-1)s (r -s) n (2.s)
! s=0 s
Using this notation, the above formulas are more briefly expressed as follows
n
x (n) (r) rx
ge = ΣS(n,r)g e (2.1+)
r=1
(n) n
ge = (-1)nΣS(n,r)g(r)e -rx
-x
(2.1-)
r=1
(3)
Example1 tane x
- x (5)
Example2 e e
-x
(n =1, 2, 3, )
(n)
g = ef , g = ef = ee
So,
-9-
(5) -x
e e
-x
= -e e e -x + 15e -2x +25e -3x +10e -4x + e -5x
Note
The horizontal total of Stirling numbers of the 2nd kind is called Bell Number . That is
n
Bn = ΣS(n,r)
r=1
The first few Bell numbers for n =1, 2, 3, are 1, 2, 5, 15, 52, 203, 877, .
This Bell number is given by the following expression that is called Dobinski's formula , too.
n n n -r s
1 r n r (-1)
Bn = Σ
e r=0 r !
=Σ Σ
r=1 r ! s=0 s!
(3)
Example (log sinx)
f = sin x , f1 = cosx , f2 = -sin x , f3 = -cosx ,
So,
3
cos x
(3) sin x cos x (cos x) 3
(log sin x) = -0! + 1!3 2
+ 2! 2
= 2cot x + 2(cot x)
sin x (sin x ) (sin x)
0! 1! n-1 (n -1)!
n
(n) (r)
g(log x) = Σ g Bn,r , - 2 , , (-1) (2.4)
r=1 x x xn
If these are written down to the 4th order, it is as follows.
(1) 1 (1)
g(log x) = g 0!
x
1 (1)
g(log x) = - 2 g 1!- g (0!)
(2) (2) 2
x
- 10 -
1
g 2!- 3g(2)1!0!+ g(3)(0!)3
(3) (1)
g(log x) = 3
x
(4) 1
g (0!)4
(1) (2) (3) (4)
42! 0!+3(1!) + 6g
2
g(log x) = - 4
3! - g 1!(0!)2 - g
x
(4)
Example1 (cos log x)
1 1 21! 32! 3!
(1/log x)(3) = 2g - 3g(2) + g(3) =
(1)
3 3
- 2
- 3
-
x x f f f4
(log x)
1 2 6 6
=- + +
x3 2
(log x)3 (log x)4
(3)
Example3 (log log x)
0! 1! 2!
Since g(1) = , g(2) = - 2 , g(3) = 3 ,
f f f
1 1 20! 31! 2!
(log log x)(3) = 3 2g(1) - 3g(2) + g(3) = 3 + + 3
x x f f2 f
1 2 3 2
= 3 + +
x log x (log x)2 (log x)3
- 11 -
22.3 Higher Derivative of Gamma Function & the Reciprocal
dn 1 1 n
n (z)
= Σ (-1)k Bn,k 0(z) , 1(z) , , n-1(z) (3.1-)
dz (z) k=1
Proof
When f(z)= log(z) ,
d
f1 = log(z) = 0(z)
dz
d
f2 = (z) = 1(z)
dz 0
d
fn = (z) = n-1(z)
dz n-2
Substituting these for (2.1+) , (2.1-) in Formula 22.2.1 , we obtain
n
( ) (n)
e log z = e log(z)ΣBn,r 0(z) , 1(z) , , n-1(z) (3.1+)
r=1
n
(z) (n)
e -log = e -log(z)Σ(-1)rBn,r 0(z) , 1(z) , , n-1(z) (3.1-)
r=1
Example1 (z)
(4)
We calculate using a formula manipulation software Mathematica according to (3.1+) . When this table
definition is adopted, the result is slightly hard to see, but the differential coefficient is calculable.
Example2 1/(z)
(4)
We calculate using a formula manipulation software Mathematica according to (3.1-) . When this table
definition is adopted, the result is easy to see, but the differential coefficient is incalculable.
- 12 -
Note
On December 9, 2016, I received a mail from Mr. Ui living in Yokohama city. In the mail, it was written that
the coefficients of the Bell polynomials appear in the higher order derivative of the gamma function. I was vary
surprised. Because, it means that the gamma function is a composite function. In fact, it was a too simple
composite function. ( Mr. Ui seems to have noticed it soon, but I needed 3 days to notice it. )
As far as I get to know, the discoverer of Formula 22.3.1 is Mr. Ui. This section is what I added a simple proof
with the consent of Mr. Ui.
- 13 -
22.4 Possibility to super differentiation of composition
Faà di Bruno's Formula was as follows.
n
(n) (r)
gf(x) = Σ
r=1
g Bn,rf1 , f2 , , fn (1.1)
s
(n) n g r r (n)
gf(x) =Σ Σ (-f)r-sf s (1.2)
r=0 r! s=0
We cannot make the upper limit n of ∑ in both formulas. It is because the number of terms in the right
side cannot become larger than the number of orders in the left side. So, it is hopeless to extend the domain
of (1.1) or (1.2) from the natural number n to the real number p . That is, the super differentiation of the
general composition gf(x ) is impossible now.
However, when the core function f(x) is the 1st degree, according to Formula 22.1.4 ,
(n) (n) n
gf(x) = g f1 (1.4)
So, there is no obstacle in extending this domain from the natural number n to real number p .
Thus, the following expression holds for a real number p >0 .
(p) (p)
gf(x) = g f1p (4.4)
This is the grounds for which we have used "Linear form" since " 12 Super Derivative " as a fait accompli .
2011.01.06
2016.12.26 updated.
K. Kono
Alien's Mathematics
- 14 -
23 Higher Integral of Composition
Formula 23.1.1
<n> (k)
Let f = f(x), g be the lineal higher primitive function of g(f ) and h , h are the functions of f
such that
dx 1 d kh
k = 1, 2, 3,
(k)
h= = (1) , h =
df f df k
Let Sk , Mk , rk are the polynomials such that
r Σ r Σ r
m k -1 -1 rk1 rk1 rk2 rk 2 rk k-1 rk k -1
Sk = Σ Σ k =1, 2, ,n
rk1=0 k1 rk2=0 k2 rk3=0 k3 r =0
kk rkk
And let a , fa are the zeros of the lineal higher primitive functions of gf(x) , g h respectively. Then,
the lineal higher integral with respect to x of the composition gf(x) is expressed as follows.
x x R
h
-Rnn
n+ Rn1 Rn1 -Rn2 Rnn
gf(x )dx
n
= S1S2Sn g h n n-1
h + Rmn 1 (n.1)
a a
g
f f f
+S M g
f f f
1+ R11 +m2 R11 -R22 +m2 R22
1 2 h h df hdf hdf
fa fa fa
+S S M g
f f
2+ R21 + m3 R21 -R32+m3 R32 -R33 R33
1 2 3 h h h df hdf
fa fa
f n-1+ Rn-1 1 + mn Rn-1 1 -Rn2 +mn Rn2 -Rn3 Rnn
+ S1S2Sn-1Mn g h h h df (n.r)
fa
Proof
df
dx = from g = g ( f ) , f = f(x) .
f(1)
Then, considering the transformation of variabl such that x f , [a , x] fa , f ,
x f g(f)
gf(x)dx = df
a fa f(1)
Here, let
f df
r r
1 dx
r = 0, 1, 2,
(r)
h = =
f r (1)
f r
Then,
-1-
g( f)h( f)df
x f
gf(x)dx =
a fa
According to Formula 16.1.2 in 16.1 ,
r g
f m 1-1 -1
1+ r11 r11
g h df = Σ h + Rm11
fa r11=0 11
= (-1) g
f
m1 m1 m1
Rm11 h df
fa
Using this,
r
x m 1-1 -1
g
1+ r11 r11
gf(x)dx =Σ h + Rm11 (1)
a r11=0 11
= (-1) g
f
m1 m1 m1
Rm11 h df (1r)
fa
Next, integrating both sides of (1) and (1r) with respect to x from a to x ,
r g R
x x m 1-1 -1 x x
1+ r11 r11
gf(x)dx = Σ
2 1
h dx + m1dx
a a r11=0 11 a a
R dx = (-1) g
x x f
1 m1 m1 m1
m1 h df dx
a a fa
Substituting dx = h df for this and rewriting the 2nd term of the right side with Rm21 ,
r g
x x m 1-1 -1 f
1+ r11 r11
gf(x)dx = Σ
2
h hdf + Rm21
a a r11=0 11 fa
= (-1) g
f f
m1 m1 m1
Rm21 h df hdf
fa fa
Here, according to Formula 20.2.1 in 20.2 ,
f f f dx = ΣΣ r s f
x m -1 r -1 r <1+ r> (r-s) (s)
1 2 3 1 f2 f3 + Rm1
a r=0 s=0
R = (-1) Σ f
m m x
1 m <m> (m-s) (s)
m 1 f2 f3 dx
s s=0 a
Then, replacing m with m2 and using this,
r r
f m 2-1 r21 -1 r21
g
2+ r11 +r21 r11 +r21 -r22 r22
g
1+ r11 r11
h hdf = Σ Σ h h + Rm12
fa r21=0 r22=0 21 22
Σ r g
m
m2 f
m2 2
1+ r11 + m2 r11 -r22 + m2 r22
Rm12 = (-1) h h df
r22=0 22 fa
Substituting this for the above,
x x
2
gf(x)dx
a a
r r
r21
r
m 1-1 -1 m 2-1 r21 -1
g
2+ r11 +r21 r11 +r21 -r22 r22
=Σ ΣΣ h h
r11=0 11 r =0 r =0
21 22 21 22
r R + R
-1
m 0-1
+ Σ 1
m1
2
m0
r11=0 11
-2-
r g
m2 m2 f
m2 1+ r11 + m2 r11 -r22 + m2 r22
Rm12 = (-1) Σ
r =0
h
fa
h df
22 22
x x
2
gf(x)dx
a a
r r
r21
r
m 1-1 -1 m 2-1 r21 -1
g
2+ r11 +r21 r11 +r21 -r22 r22
=Σ ΣΣ h h + Rm21 (2)
r11=0 11 r =0 r =0
21 22 21 22
r g
m
r
-1
m 1-1
m2
m2 2 f
1+ r11 + m2 r11 -r22 + m2 r22
Rm21 =Σ (-1) Σ h h df
r11=0 11 r22=0 22 fa
+ (-1) g
f f
m1 m1 m1
h df hdf (2r)
fa fa
Here, let
r Σ r r
r21 m2
r
m 1-1 -1 m 2-1 -1 r21
m2
m2
S1 =Σ , S2 =Σ , M2 =(-1) Σ
r11=0 11 r21=0 21 r22=0 22 r22=0 22
x x
2 2+ R21 R21 -R22 R22
gf(x)dx = S 1 S 2 g h h + Rm21
a a
f 1+ R + m
Rm21 = S1 M2 g 11 2 h 11 22 2h 22df
R -R + m R
fa
g
f f
m1 m1 m1
+ (-1) h df hdf
fa fa
Next, integrating both sides of (2) and (2r) with respect to x from a to x ,
r Σ Σ r r
x x x m 1-1 -1 m 2-1 r21 -1 r21
=Σ
3
gf(x )dx
a a a r11=0 11 r21=0 r22=0 21 22
R
x x
g
2+ r11 +r21 r11 +r21 -r22 r22
2
h h dx + m1dx
a a
r g
m2
r
x m 1-1 -1 m2 x f
m2 1+ r11 + m2 r11 -r22 + m2 r22
Rm21dx = Σ (-1) Σ h h df dx
a r11=0 11 r =022 22 a fa
g
f
x f
m1 m 1 m 1
+ (-1) h df hdf dx
a fa fa
Substituting dx = h df for this and rewriting the 2nd term of the right side with Rm31 ,
x x x
3
gf(x )dx
a a a
r r g
r21
r
m 1-1 -1 m 2-1 r21 -1 f
2+ r11 +r21 r11 + r21 -r22 r22
=Σ ΣΣ h h hdf + Rm31
r11=0 11 r =0 r =0
21 22 21 22 fa
-3-
r
m2
r
m 1-1 -1 m2 f f
g
m2 1+ r11 + m2 r11 -r22 + m2 r22
Rm31 = Σ (-1) Σ h h df hdf
r11=0 11 r =022 22 fa fa
g
f
f f
m1 m1 m1
+ (-1) h df hdf hdf
fa fa fa
Here again, according to Formula 20.2.1 in 20.2 ,
r s t f
x m -1 r s -1 r s <1+ r> (r-s) (s-t) (t)
f1 f2 f3 f4dx = Σ ΣΣ 1 f2 f3 f4 + Rm1
a r=0 s=0 t=0
s t
= (-1) ΣΣ
m s m s x
m <m> (m-s) (s-t) (t)
Rm1 f 1 f2 f3 f4 dx
s=0 t=0 a
Then, replacing m with m3 and using this,
r r r
f m 3-1 r31 r32 -1 r31 r32
g
2+ r11+ r21 r11+r21-r22 r22
h h hdf = Σ Σ Σ
fa r31=0 r32=0 r33=0 31 32 33
3+ r11+ r21+ r31 r11+ r21+ r31-r22-r32 r22+ r32-r33 r33
g h h h + Rm13
r r
m3
m3 r32 m3 r32
Rm13 = (-1) ΣΣ
r32=0 r33=0 32 33
f
g
2+ r11+ r21+m3 r11+r21-r22-r32+m3 r22+r32-r33 r33
h h h df
fa
Substituting this for the above,
x x x
3
gf(x)dx
a a a
r Σ Σ r r Σ Σ Σ r r r
m 1-1 -1 m 2-1 r21 -1 r21 m 3-1 r31 r32 -1 r31 r32
=Σ
r11=0 11 r21=0 r22=0 21 22 r31=0 r32=0 r33=0 31 32 33
g
3+ r11+ r21+ r31 r11+ r21+ r31- r22-r32 r22+ r32-r33 r33
h h h
r Σ Σ r r R
m 1-1 -1 m 2-1 r21 -1 r21
+Σ 1
m3 + Rm31
r11=0 11 r21=0 r22=0 21 22
r r
m3
m3 r32 m2 r32
Rm13 = (-1) ΣΣ
r =0 r =0
32 33 32 33
f
g
2+ r11+ r21+m3 r11+r21-r22-r32+m3 r22+r32-r33 r33
h h h df
fa
r g
m2
r
m 1-1 -1 m2 f f
m2 1+ r11 + m2 r11 - r22 + m2 r22
Rm31 = Σ (-1) Σ h h df hdf
r11=0 11 r =0
22 22 fa fa
g
f
f f
m1 m 1 m 1
+ (-1) h df hdf hdf
fa fa fa
Redefining Rm31 ,
x x x
3
gf(x)dx
a a a
-4-
r r r r r
r21 r31 r32
r
m 1-1 -1 m 2-1 r21 -1 m 3-1 r31 r32 -1
=Σ ΣΣ ΣΣΣ
r11=0 11 r =0 r =0
21 22 21 22 r =0 r =0 r =0
31 32 33 31 32 33
3+ r11+ r21+ r31 r11+ r21+ r31-r22-r32 r22+ r32-r33 r33
g h h h + Rm31 (3)
r r
m3
r Σ Σ r r
m 1-1 -1 m 2-1 r21 -1 r21 m3
m3 r32 r32
Rm31 = Σ (-1) ΣΣ
r11=0 11 r21=0 r22=0 21 22 r32=0 r33=0 32 33
g
f
2+ r11+ r21+m3 r11+r21-r22-r32+m3 r22+r32-r33 r33
h h h df
fa
r
m2
r
m 1-1 -1 m2 f f
1+ r11 + m2 r11 -r22 + m2 r 22
g
m2
+Σ (-1) Σ h h df hdf
r11=0 11 r =022 22 fa fa
x x x
= S 1S 2S 3 g
3 3+ R31 R31-R32 R32-R33 R33
gf(x)dx h h h + Rm31
a a a
f
Rm31 = S1S2M3 g
2+ R21+m3 R21-R32+m3 R32-R33 R33
h h h df
fa
+ S M g
f f
1 + R11+m2 R11-R22+m2 R22
1 2 h h df hdf
fa fa
+ (-1) g
f f f
m1 m1 m1
h df hdf hdf
fa fa fa
Calculating the 4th order integral in a similar way,
x x 4+ R41 R41 -R42 R42 -R43 R43 -R44 R44
= S1S2S3S4 g
4 4
gf(x )dx h h h h + Rm1
a a
f
= S 1S 2S 3M4 g
3+ R31+m4 R31-R42+m4 R42-R43 R43-R44 R44
Rm41 h h h h df
fa
+ S S M g
hdf
f f
2 + R21+m3 R21-R32+m3 R32-R33 R33
1 2 3 h h h df
fa fa
+ S M g
f f f
1 + R11+m2 R11-R22+m2 R22
1 2 h h df hdf hdf
fa fa fa
+ (-1) g
f f f f
m1 m1 m1
h df hdf hdf hdf
fa fa fa fa
Hereafter, by induction, we obtained the desired expression.
Formula 23.1.2
When f(x)=cx +d ,
-5-
x x n f f
1
gf(x)dx
n
= g( f )df n
a a c fa fa
Proof
When f(x)= cx +d in Formula 23.1.1 ,
dx 1 1 d kh
k = 1, 2, 3,
(k)
h= = (1) = , h = =0
df f c df k
Then, the remainder term
g
f f f
+S M g
f f f
1+ R11 +m2 R11 -R22 +m2 R22
1 2 h h df hdf hdf
fa fa fa
+S S M g
f f
2+ R21 + m3 R21 -R32+m3 R32 -R33 R33
1 2 3 h h h df hdf
fa fa
f n-1+ Rn-1 1 + mn Rn-1 1 -Rn2 +mn Rn2 -Rn3 Rnn
+ S1S2Sn-1Mn g h h h df (nr)
fa
becomes as follows.
m1 > 0 , h
m1
1st line: Since = 0. Then the 1st line = 0.
R22 > 0 , h
R22
2nd line: When = 0. Then, the 2nd line = 0.
R22 = 0 , h
R11 - R22 + m2
When = 0. Then, the 2nd line = 0.
R33 > 0 , h
R33
3rd line: When = 0. Then, the 3rd line = 0.
h
Rnn
n th line: When Rnn > 0 , = 0. Then, the n th line = 0.
When Rnn = 0 , if at least one is positive in Rn n-1 Rn2 , Then, the n th line = 0.
x x
= S 1S 2 S n g
n + Rn1 Rn1 -Rn2
h
Rn n-1 -Rnn Rnn
gf(x)dx
n
h h (n)
a a
Next, observing (n), we find out Rnn = Rn n-1 = = Rn1 = 0 immediately. If it is why, in order for (n)
not to be 0, Rnn = 0 is required first. The 2nd, Rnn-1 = 0 is required. The 3rd, . Hereafter, it advances
one after another, finaly Rn1 = 0 is required. Then,
n
Rn1 -Rn2 Rn n-1 -Rnn Rnn (0) n 1
h h h = h =
c
S1 S2 Sn = 1
-6-
Thus, we obtain
n n
x x 1 1 f f
gf(x)dx =
n
g
<n>
= g( f )df n
a a c c fa fa
Q.E.D
n
x (ax +b)n
x 1 n 1
log(ax +b)dx = n
log(ax +b) -Σ
b b a n! k =1 k
- -
a a
This is consistent with linear form in Formula 4.3.3 in " 4 HigherIntegral " .
g n ( f )= c . If it is not so,
<>
of the enclosing function g( f ) must have the property such as nlim
not the
series but the remainder term becomes the prime part of the integration.
Considering these, there are not many compositions that Formula 23.1.1 is applicable. This decreases the
usefulness of Formula 23.1.1. However, about some compositions that Formula 23.1.1 was applicable, a quite
interesting result was obtained. I describe it in the following sections.
-7-
23.2 Higher Integral of (x^ - c)^
Formula 23.2.1
Let (z) be the gamma function. Let Sk , rk and gf(x) are as follows respectively.
gf(x) = x - c , >0 & 1/k (k =1, 2, ) , , , c > 0
r Σ r Σ r
-1 rk1 rk1 rk2 rk 2 rk k-1 rk k -1
Sk =Σ Σ k =1, 2, ,n
rk1=0 k1 rk2=0 k2 rk3=0 k3 r =0
kk rkk
j
Rj k =Σrik j,k = 1, 2, ,n
i=k
Then, the following expressions hold for x> c .
(1+) (1/ )
1+ r
r=0
x -1
x c
x - c dx = x - c
Σ r 1+ +1+ r 1/ - r
1-
x
c
(1.1)
Σ r ΣΣ s
(1+ )
t 1+ + 2+ r +s
2 -1 -1
x x
x s s
x - c dx = x - c
2
c
c 2
r =0 s =0 t =0
(1/ ) (1/ )
2+ r+s
1- x
c
(1.2)
1/ - r - s + t 1/ - t
3
x x
x
3
x - c dx = x - c
c
c 3
r r r r r
r21 r31 r32
r
-1 r21 -1 r31 r32 -1
Σ ΣΣ ΣΣΣ
r11=0 11 r =0 r =0
21 22 21 22 r =0 r =0 r =0
31 32 33 31 32 33
(1+) (1/ )
1+ + 3+ r11+ r21+ r31 1/ - r11- r21- r31+ r22+ r32
(1/ ) (1/ ) 3+ r11+ r21+ r31
c
1- (1.3)
1/ - r22- r32+ r33 1/ - r33 x
n + Rn1
(1+)
xn
x x
1
x - c dx = x - c
n
S S Sn 1-
c
c n 1 2 1+ + n + Rn1 x
(1/ ) (1/ ) (1/ )
(1.n)
1/ - Rn1+Rn2 1/ - Rn n-1+Rnn 1/ - Rn n
Proof
1
dx 1 -1
gf(x ) = x - c , h= = (f + c)
df
From these,
n+ Rn1 n+ Rn1 (1+) + n+ Rn1
g = f = f
1+ +n + Rn1
-8-
Rn1 -Rn2
(f + c)
1
1 -1
Rn1 -Rn2
h =
1
1(1/ ) -1-Rn1 + Rn2
= (f + c)
1/ - Rn1+Rn2
1
Rn n-1-Rnn 1(1/ ) -1-R +R
(f + c)
n n-1 nn
h =
1/ - Rn n-1+Rnn
1
(1/ )
1 -1-Rnn
h m 1
= (f + c)
1/ - Rnn
Substituting these for Formula 23.1.1 (n.1) ,
(1+)
x x
+ n+ Rn1
x - c dx = S1S2Sn
n
f
c
c 1+ +n + Rn1
1
1(1/ ) -1-Rn1 + Rn2
(f + c)
1/ - Rn1+Rn2
1
1(1/ ) -1-R +R
(f + c)
n n-1 nn
1/ - Rn n-1+Rnn
1
(1/ )
1 -1-Rnn
(f + c) + Rmn 1
1/ - Rnn
n
1 (1+)
+ n+ Rn1
-n-Rn1
= n S1S2S n x - c
x
1+ +n + Rn1
(1/ ) (1/ ) (1/ )
+ Rmn 1
1/ - Rn1+Rn2 1/ - Rn n-1 +R nn 1/ - Rnn
i.e.
n+ R
(1+)
xn n1
x x
c
x - c dx = x - c S n
S S 1-
c
c n
1 2 n
1+ +n + Rn1 x
(1/ ) (1/ ) (1/ )
+ Rmn 1 (n)
1/ - Rn1+Rn2 1/ - Rn n-1+Rnn 1/ - Rnn
Since Rmn 1 is too long, we omit it. If (n) is written about the 1st order the 3rd order, it is as follows.
x
x - c dx
c
(1+) (1/ ) 1+ r
r=0 r
m 1-1 -1
x c
= x - c Σ
1
1- + Rm1 (1)
1+ +1+ r 1/ - r x
m1 1
(1+) (1/ )
(-1) f +m -1-m1
Rm11 = f 1
(f + c) df (1r)
1+ + m1 1/ - m1 0
x x
x - c dx
2
c c
-9-
2 m -1
(1+)
2 r=0 Σ
s=0 t=0 t
x -1 m 2-1 s -1 s
= x - c
1
Σ r
Σ s 1+ + 2+ r +s
(1/ ) (1/ ) 2+ r+s
c
2
1- + Rm1 (2)
1/ - r - s +t 1/ - t x
(1+) (1/ )
m2 m -1
t
m2
r=0 Σ
(-1) 1 -1 m2
Σ
2
Rm1 =
2
r t=0 1+ + 1+ r + m2 1/ - r + t - m2
2
(1/ )
f
f +1+ r+ m -2-r- m2
2
(f + c) df
1/ - t 0
(1+) (1/ )
m1
(-1)
+
2 1+ + m1 1/ - m1
1 1
f f -1-m1 -1
+m 1
f (f + c) df (f + c) df (2r)
0 0
x x
x3
3
x - c dx = x - c
c
c 3
r Σ Σ r r Σ Σ Σ r r r
m 1-1 -1 m 2-1 r21 -1 r21 m 3-1 r31 r32 -1 r31 r32
Σ
r11=0 11 r21=0 r22=0 21 22 r31=0 r32=0 r33=0 31 32 33
(1+) (1/ )
1+ + 3+ r11+ r21+ r31 1/ - r11- r21- r31+r22+r32
(1/ ) (1/ ) 3+ r11+ r21+ r31
c
1- + Rm31 (3)
1/ - r22- r32+r33 1/ - r33 x
m3 m -1
r r
m3
r r
r21 r32
r
(-1) 1 -1 m 2-1 r21 -1 m3 r32
Rm31 =
3
Σ
r =0 11
ΣΣ
r =0 r =0 21
ΣΣ
r =0 r =0
11 21 22 22 32 33 32 33
(1+) (1/ )
1+ +2+ r11+ r21+m3 1/ - r11-r21+r22+r32-m3
(1/ ) (1/ )
1/ - r22- r32+r33 1/ - r33
3
f + 2+ r + r +m -3- r11 -r21-m3
f 11 21 3
(f + c) df
0
m2 m -1
r Σ r
(-1) 1 -1 m2 m2
+
3
Σ
r11=0 11 r22=0 22
(1+) (1/ ) (1/ )
1+ +1+ r11+m2 1/ - r11+r22- m2 1/ - r22
2 1
f f + 1+ r +m -2-r11 - m2 -1
f 11 2
(f + c) df (f + c) df
0 0
m1
(-1) (1+) (1/ )
+
1+ +m1 1/ - m1
- 10 -
1 1 1
f f f -1-m1 -1 -1
+ m1
f (f + c) df (f + c) df (f + c) df (3r)
0 0 0
Rmn 1 also roughly looks like (3r). The integration included in these is more difficult than the original one (the left
side), and solving these is unthinkable. However, If >0, >0 , Rmn 1 0 at the time mi i =1, 2, .
That is shown as follows.
Now, cnsider the function S(m) such that
1
S(m) =
(1+ m)(1/b - m)
Giving the values to b of this function parametrically and illustrating this, we find out the following. (Refer to the
figure.)
b <0 : S(m) diverges to .
b =0 : S(m) oscillates around 0.
b >0 : S(m) converges to .
Although the description was omitted previously, the last term of the remainder Rmn 1 is as follows.
m1
(-1) (1+) (1/ )
1+ +m1 1/ - m1
n n n
f f f + n-1+ R -n- Rn-11 - mn -1 -1
n-11 m n
+
f (f + c) df (f + c) df (f + c) df
0 0 0
And when >0 , >0 ,
1 1 1 1
lim < lim =0
m 1 1+ +m1 1/ - m1 m 1 1+m1 1/ - m1
Moreover, the integrand is as follows.
n
+ n-1+ Rn-11 + mn -n- Rn-11 - m n
f (f + c)
n
+ n-1+ Rn-11 + mn
-n- Rn-11 - mn
= x - c
x
- 11 -
n mn
+ n-1+ Rn-11
-n- Rn-11
1- x
c
= x - c x
By assumption, x> c , c >0 . Then,
n mn
+ n-1+ Rn-11
-n- Rn-11
1- x
c
lim x - c
m n
x
=0
Together with these, if mi i =1, 2, 3, , the last term of the remainder Rmn 1 have to converge to 0 .
. Next, terms except the last term of the remainder Rmn 1 contain the products as follows.
(1+) (1/ )
1+ +k -1+ Rk-1 1+ mk 1/ - Rk-11+Rk2-mk
(1/ ) (1/ )
1/ - Rk2+Rk3 1/ - Rkn
k k k
f f f + k-1+ R -k- Rk-11 - mk -1 -1
k-11 m k
+
f (f + c) df (f + c) df (f + c) df
0 0 0
The term that does not include mk among these can be regarded as a constant. Since k -10 ,
Together with these, if mi i =1, 2, 3, , the non-last term of the remainder Rmn 1 have to also converge
to 0 . Therefore, mlim R n = 0 i =1, 2, 3, .
m 1
Thus, (1),(2),(3), ,(n) reduce to (1.1),(1.2),(1.3), ,(1.n)
i
respectively.
(1+1/3) (1/3)
1+ r
1+1/3+1+ r
x3 x -1
3 1
3Σ
x 3-1 dx = x 3-1 1-
1 r=0 r 1/3- r x3
3 3 1 1
The function y= x -1 is a part of the third hyperbola 3 - = 1 in the 1st quadran.
x y3
Therefore, this integration means to ask for the function expressing the area between the hyperbola and x-axis in
the 1st quadrant. Although even the elliptic integral is impossible to express this area, this example can express
it with a series. In fact, when both sides are drawn in figure, it is as follows. As the result of the calculation until
the 100th term in the right side, both sides almost overlapp.
- 12 -
log2
x -
e
Example2 The 2nd integral of
Let =log 2 , =e , c = in (1,2) . Then,
x2
e r=0 Σ
s=0 t=0 t
x x log2 log2 -1 s -1 s
x - dx 2 = x e-
e
2Σ Σ
e
e
r s
(1+log 2) 1/e 1/e 2+ r+s
1+log 2+ 2+ r +s 1/e - r - s + t
1/e - t
1- e
x
The values of the both sides on arbitrary point x = 3.5 are as follows.
- 13 -
23.3 Higher Integral of g (log x)
f = log x , the inverse function is x = e f . Then h = e f r =0, 1, .
(r )
When the core function is
Thus, the n th order integral of g (log x ) is expressed with n-fold series.
Formula 23.3.1
<n>
When g is the lineal higher primitive function of g(f ) and a , fa are zeros of the lineal higher primitive
f
functions of g(log x) , g ( f)e respectively, the lineal higher integral of g(log x)is expressed as follows.
n n
Σ rk n +Σ rk
+ Rn
x x m 1-1 m 2-1 m n -1 n
= x n Σ Σ Σ (-1)k=1
rk
a a
g(log x)dx
n
r1=0 r2=0 rn =0
Π
k=1
k g k=1
m1 (1.n)
g e df e df e df e df
f f f f
m1 f
m1
Rmn1 = (-1)m11 f f f
(n-fold nest)
fa fa fa fa
1 2 g
m 1-1 f f f
1
r1+ m2 m1 m2 + r1+m2 2f
+ Σ (-1) e df e df e df f f
r1=0 fa fa fa
1 2 3 g
m 1-1 m 2-1 f f
2 r1+ r2+m3 m1 m2 m3 + r1+ r2+m3 3f
+ Σ Σ (-1) e df e df f
r1=0 r2=0 fa fa
n -1 n -1
g e n fdf
m 1-1 m 2-1 m n-1-1 Σ rk+ mn n mk f n -1 + Σ rk+ mn
+ Σ Σ Σ (-1)k =1
Π k k=1
(1.nr)
r1=0 r2=0 r =0n-1 k=1 fa
Proof
Since f = log x , x = e f
dx
(r =1, 2, 3, ,n)
(r)
h= = ef = x , h = ef = x
df
Substituting these for (1), (1,r) in Formula 23.1.1 ,
r
x m 1-1 -1
g
1+ r11
g(log x)dx = xΣ + Rm11
a r11=0 11
= (-1) g
f
m1 m1 f
Rm11 e df
fa
r
-1 r r m1
Replacing with (-1) 11 and adding 1 11 , 1 to this, we obtain
11
x m 1-1
= xΣ (-1) 11 1 11 g1+
r r r11
g(log x)dx + Rm11 (1.1)
a r11=0
g
f
m m1 m1 f
Rm11 = (-1) 1 1 e df (1.1r)
fa
r r
r21
r
x x m 1-1 -1 m 2-1 r21 -1
g
2+ r11+r21
Σ ΣΣ
2 2
g (log x)dx =x + Rm21
a a r =0
11 11 r =0 r =0
21 22 21 22
- 14 -
r g
m2
r
m 1-1 -1 m2
m2 f
1+ r11 + m2 2f
Rm21 = Σ (-1) Σ e df
r11=0 11 r =0
22 22 fa
+ (-1) g
f f
m1 m1 f
e df e fdf
fa fa
Here,
r r
-1 r
-1 r
= (-1) 11 , = (-1) 21
11 21
r
m2
r
r21 r21 r21
m2
m2
Σ
r =0
=2 , Σ
r =0
=2
22 22 22 22
r m1
Substituting these for the above, and adding 1 11 , 1 to this, we obtain
x x m 1-1 m 2-1
r +r21 r11 r21 2+ r11+r21
g (log x)dx
2
= x 2Σ Σ (-1) 11 1 2 g + Rm21 (1.2)
a a r11 =0 r =0 21
g
m 1-1 f
r + m2 m1 m2 1+ r11 + m2 2f
Rm21 = Σ (-1) 11 1 2 e df
r11=0 fa
1 g
f f
m1 m1 m1 f
+ (-1) e df e fdf (1.2r)
fa fa
x x x
3
g(log x)dx
a a a
r r r r r
r21 r31 r32
r
m 1-1 -1 m 2-1 r21 -1 m 3-1 r31 r32 -1
= x 3Σ ΣΣ ΣΣΣ
r11=0 11 r =0 r =0
21 22 21 22 r =0 r =0 r =0
31 32 33 31 32 33
3+ r11+ r21+ r31
g + Rm31
r r
m3
r r
r21 r32
r
m 1-1 -1 m 2-1 r21 -1 m3
m3 r32
Rm31 =Σ ΣΣ (-1) ΣΣ
r11=0 11 r =0 r =0
21 22 21 22 r =0 r =0
32 33 32 33
g
f
2+ r11+ r21+m3 3f
e df
fa
r g
m2
r
-1
m 1-1 m2 f f
m2 1+ r11 + m2 2f
+Σ (-1) Σ e df e fdf
r11=0 11 r =0
22 22 fa fa
+ (-1) g
f f f
m1 m1 f f f
e df e df e df
fa fa fa
Here,
r r r
-1 r
-1 r
-1 r
= (-1) 11 , = (-1) 21 , = (-1) 31
11 21 31
r r r
r21 r21 r
r31 r32 r31 r32 r
Σ
r =0
= 2 21 , ΣΣ
r =0 r =0
= 3 31
22 22 32 33 32 33
Σ r = 2 Σ Σ r r = 3
m2m 2 m2
m
m3 r r32 3 32 m3
,
r22=0 22 r32=0 r33=0 32 33
- 15 -
r m1
Substituting these for the above, and adding 1 11 , 1 to this, we obtain
x x x
3
g (log x)dx
a a a
m 1-1 m 2-1 m 3-1
1 2 21 3 31 g
r + r21+ r31 r11 r r 3+ r11+ r21+ r31
Σ Σ Σ
3
=x (-1) 11 + Rm31 (1.3)
r =0 r =0 r =0
11 21 31
g
m 1-1 m 2-1 f
r + r21+m3 m1 m2 m3 2+ r11+ r21+m3 3f
Rm31 = Σ Σ (-1) 11 1 2 3 e df
r11=0 r21=0 fa
1 2 g
m 1-1 f f
r11+ m2 m1 m2 1+ r11 + m2 2f
+ Σ (-1) e df e fdf
r11=0 fa fa
+ (-1) 1 g
f f f
m1 m1 m1 f
e df e fdf e fdf (1.3r)
fa fa fa
Hereafter, by induction, we obtain
n n
Σ rk1 n +Σ + Rn
x x m 1-1 m 2-1 m n -1 n rk1
rk 1
g (log x)dx
n
= x n Σ Σ Σ (-1)k=1 Π k g k=1
m1 (1.n)
a a r11=0 r21=0 rn 1=0 k=1
g e df e df e df e df
f f f f
m1 f
Rmn1 = (-1) 1 m1 m1
f f f
(n-fold nest)
fa fa fa fa
1 2 g
m 1-1 f f f
1
r11+ m2 m1 m2 + r11+m2 2f
+ Σ (-1) e df e df e df f f
r11=0 fa fa fa
1 2 3 g
m 1-1 m 2-1 f f
2r11 + r21+m3 m1 m2 m3 + r11+ r21+m3 3f
+ Σ Σ (-1) e df e df f
r11=0 r21=0 fa fa
n -1 n -1
g e n fdf
m 1-1 m 2-1 m n-1-1 Σ rk1+ mn n mk f n -1 + Σ rk1+ mn
+ Σ Σ Σ (-1)k=1 Π k k=1
(1.nr)
r11=0 r21=0 r =0
n-1 1 k=1 fa
And replacing rk1 with rk , we obtain the desired expression.
Q.E.D.
Although fa is determined by the function g(f ) in Formula 23.3.1 , in many cases, fa = - that is a zero
f
of e . The calculation at the time of applying a concrete function to g(f ) should be a part of this section.
However, since these are very long, these are described in the following independent sections.
- 16 -
23.4 Higher Integrals of (log x) ^ , (log x)^m
Lemma1
t
When (a , x ) = >0
a-1 -t
e dt is the incomplete gamma function, the following expressions hold for
x
and -1, -2, -3, .
(1) When x 0
Proof
Let g(f ) = f e f , f = x in Formula 23.1.2 . Then,
f
x x n f f
x 1
( x) e dx = n
e df n
- - - -
Here, substitute Formula 16.5.1 in 16.5 for this right side. Then, when x >0 ,
f n -1 C r (n -r + , -f)f r
n
x x 1
( x) e x dx n =
n -1
Σ
(-f) r=0 (n -1)!
- -
n
C r (n -r + )f r
2i sin Σ
1 n -1 n -1
+
r=0 (n -1)!
n
( x) C r (n -r + , - x)(x)r
1 n -1 n -1
= Σ
(- x) r=0 (n -1)!
n
C r (n -r + )(x)r
2i sin Σ
1 n -1 n -1
+
r=0 (n -1)!
i.e.
x n -1 C r (n -r + , - x)(x)r
x x
x
x e dx =
n -1
Σ
(-x) r=0
n
- - n(n -1)!
n -1 n -1C r (n -r + )( x)
r
+ 2i sin Σ
r=0 n(n -1)!
From this, we obtain the desired expressions.
- 17 -
Formula 23.4.1
When (z) is the gamma function, the following expressions hold for a real number -1, -2, -3, .
(1+ )
x
(log x) dx = x 1Σ(-1)r1r (log x)+1+ r
0 r=0 (1+ +1+ r)
(1+ ) x0
+ (-1) 1C 1 (1.1)
1! 1
(1+ )
x x
(log x) dx 2 = x 2ΣΣ(-1)r+ s 1r2s (log x)+2+ r+s
0 0 r=0 s=0 1+ +2+ r +s
(1+ )
x1 x0
+ (-1) 2C 1 - 2C 2 (1.2)
2! 1 2
x x x
(log x) dx 3 = x 3ΣΣΣ(-1)r+ s+ t 1r2s3t
0 0 0 r=0 s=0 t=0
(1+ )
(log x)+3+ r+ s+ t
1+ +3+ r + s + t
(1+ )
x2 x1 x0
+ (-1) 3C 1 - 3C 2 + 3C 3 (1.3)
3! 1 2 3
n
x x
Σ rk n rk
n
(log x) dx = x n
Σ Σ Σ (-1) k =1
Π k
0 0 r =0 r =0
1 2 r =0 n k=1
(1+ )
n
+ n+ Σ rk
n (log x) k =1
1+ + n + Σrk
k=1
(1+ ) n x
n-k
+ (-1) k-1
Σ(-1) nC k (1.n)
n! k=1 k
Proof
When the 1st order, substituting g =f , f = log x for (1.1), (1.1r) in Formula 23.3.1 ,
(1+ )
x m -1
(log x) dx = xΣ (-1)r1r (log x)+1+r + Rm1 (1)
0 r=0 1+ +1+r
(1+ )
f
Rm1 = (-1)m1m f +me fdf (1r)
1+ + m -
If f 0 , from Lemma1,
f
f
+m f f +m
e df = (1+ +m , -f)
- (-f)+m
Substituting this for (1r),
(1+ ) f +m
Rm1 = (-1) m1m (1+ +m , -f)
1+ + m (-f)+m
- 18 -
(1+ )
f
= (-1)m(-1)+m1m (1+ + m , -f) = (-1)
1+ + m (-f )
i.e.
(1+ +m , -f)
Rm1 = (-1)(1+ )
1+ + m
Let m . Then,
(1+ + m , -f)
lim =1
m 1+ + m
Therefore,
R1 = (-1)(1+ )
If f >0 , from Lemma1,
f
f
+m f f +m
e df = (1+ +m , -f)+ 2i sin( +m)(1+ + m)
- (-f)+m
Substituting this for (1r) ,
(1+ )
f
f
+m f
Rm1 = (-1)m1m e df
1+ + m -
(1+) f + m
= (-1) m
1+ + m (-f) + m
(1+ + m, -f) + 2i sin( +m )(1+ + m)
(1+)
= (-1)m (-1)--m(1+ + m, -f) + 2i sin( +m )(1+ + m)
1+ + m
(1+ ) (1+ + m, -f)
= + 2i (1+ )(-1)msin( + m )
(-1) 1+ + m
i.e.
(1+ ) (1+ + m, -f)
Rm1 = + 2i (1+ )sin
(-1) 1+ + m
Let m . Then,
1
R1 = (1+ ) + 2i sin
(-1)
Here,
1
+ 2i sin = (-1)
(-1)
Therefore R1 = (-1)(1+) . This is the same result as the time of f 0.
Thus, (1) becomes as follows.
(1+ )
(log x) dx = x Σ(-1) 1 (1+ +1+ r) (log x)
x
1 r r +1+ r
0 r=0
(1+ ) x0
+ (-1) 1C 1 (1.1)
1! 1
When the 2nd order, substituting g =f , f = log x for (1.2), (1.2r) in Formula 23.3.1 ,
(1+)
x x m 1-1 m 2-1
(log x) dx = x Σ Σ (-1)r+s 1r 2s (log x)
2 2 +2+ r+s 2
+ Rm1 (2)
0 0 r=0 s=0 1+ +2+r + s
- 19 -
(1+)
f
f f
m m1 +m1 f
Rmn 1 = (-1) 11 e df e fdf
1+ + m1 - -
(1+)
f
m 1-1 r+ m 2 m 1 m 2 f +1+ r+ m
+ Σ (-1) 1 2 2 2f
e df (2r)
r=0 1+ +1+ r + m2 -
If f 0 , from Lemma1,
+m1
f
f f
+m1 f
e df = 1+ + m1 , -f
+m1
- (-f)
+1+ r+ m2
1+ +1+ r + m2, -2f
x
f +1+ r+ m2 f f
e df = +1+ r+ m2 +1+ r+ m2+1
- (-f) 2
Substituting these for (2r),
+m1
(-f)
(1+) f f
Rm21 = (-1) 1
m1 m1
1+ + m1 , -f e fdf
1+ + m1 - +m1
+1+ r+ m
m 1-1 (1+)
r+ m2 m1 m2 f 2
1+ +1+ r + m2, -2f
+ Σ (-1) 1 2
r=0 1+ +1+ r + m2 (-f) +1+ r+ m 2
2
+1+ r+ m2 +1
1
R2 = (-1)(1+) e f - (-1)(1+) Σ +2+ r
r=0 2
1
= (-1)(1+) e log x -
2+1
(1+ )
2x 1 x0
= (-1) -
2! 1 2
If f >0 , from Lemma1,
- 20 -
+m1
f
f x
+m 1 f
e df = +m 1
1+ +m1 , -f + 2i sin +m1 1+ +m1
- (-f)
+1+ r+ m2
1+ +1+ r + m2, -2f
f +1+ r+ m2 f f
x e df = +1+ r+ m2 +1+ r+ m2+1
- (-f) 2
1+ +1+ r + m2
+ 2i sin +1+ r + m2 +1+ r+ m2+1
2
Substituting these for (2r) ,
+m1
(-f)
(1+) f f
Rm21 = (-1) 1
m1 m1
1+ + m1 , -f e fdf
1+ + m1 - +m1
(1+)
2i sin +m 1+ +m e df
f
m m1 f
+ (-1) 11
1+ + m1 -
1 1
+1+ r+ m
m 1-1 (1+)
r+ m 2 m 1 m 2 f 2
1+ +1+ r + m2, -2f
+ Σ (-1) 1 2
r=0 1+ +1+ r + m2 (-f) +1+ r+ m2 2
+1+ r+ m2+1
(1+) e f
+ 2i sin (1+) e f
2
R =
(-1)
1 (1+) 1 (1+) 1
+ Σ + 2i sin( +1) Σ
(-1)+1 2+2 r=0 2r 2+2 r=0 2r
- 21 -
(-1)
1
=
+ 2i sin (1+) e log x
(1+)
(-1)
1
+ +1
+ 2i sin( +1)
2+1
Here,
1
+ 2i sin = (-1)
(-1)
Therefore,
(1+)
+1 (1+) 2x 1 x0
R2 = (-1)(1+) x + (-1) = (-1) -
2+1 2 1 2
This is the same result as the time of f 0.
Thus, (2) becomes as follows.
(1+ )
(log x) dx
x x
2
= x 2ΣΣ(-1)r+ s 1r2s (log x)+2+ r+s
0 0 r=0 s=0 1+ +2+ r +s
(1+ )
x1 x0
+ (-1) 2C 1 - 2C 2 (1.2)
2! 1 2
Hereafter, by induction, we obtain the general expression.
Lemma2
Let be a positive number, (z) be the gamma function and m,n are non-negative integers.
e x (1+ m)
r
x x
x m
m -n
(x)
m- r
m+n Σ
n
e x dx =
- - r=0 (1+ m - r)
Proof
Let g(f ) = f me f , f = x in Formula 23.1.2 . Then,
f
x x n f f
m x 1
( x) e dx = n
m f
e df n
- - - -
- 22 -
Here, substitute Formula 16.5.1' in 16.5 for this right side. Then,
(1+ m) m- r
(x) e
1 n f m -n
r
x x
m x n
dx = e Σ
r=0 (1+ m - r)
f
- -
i.e.
e x (1+ m)
r
x x
x m
m -n
m
e x dx = n
Σ ( x)
m- r
- - n r=0 (1+ m - r)
From this, we obtain the desired expression.
Formula 23.4.2
When m =1, 2, 3, ,
m- r
x m! 1 m (-1)r (log x)
(log x) dx = m
xΣ (2.1)
0 1 r=0 1r m - r !
r+s
(log x)m+1- r- s
x x m! 2 m +1 m +1- r (-1)
(log x) dx = - x Σ Σ
m 2
(2.2)
0 0 2 r=1 s=0 1r 2s m +1- r - s !
r+ s+ t
(log x)m+2- r- s- t
x x x m! 3 m +2 m +2- r m +2- r- s (-1)
(log x) dx = m
x Σ Σ Σ 3
0 0 0 3 r=1 s=1 t=0 1r 2s 3t m +2- r - s - t !
(2.3)
n -2 n -1
m + n -1- Σ rk m + n -1- Σ rk
x x m! n m + n -1 m + n -1- r1 k =1 k =1
(log x)m dx n = (-1)n x Σ Σ Σ Σ
0 0 n r =1
1 r =1
2 r =1
n-1 r =0n
n n
Σ rk m+ n-1-Σ rk
(-1) k =1 (log x) k =1
n n (2.n)
m + n -1-Σrk !
rk
Π
k=1
k
k=1
Proof
When the 1st order, substituting g =f , f = log x for (1.1), (1.1r) in Formula 23.3.1 ,
(1+m)
x m 1-1
(log x)m dx = xΣ (-1)r1r (log x)m+1+r + Rm11
0 r=0 1+m +1+r
(1+m)
1+m + m
f
m1 m+m1 f
Rm11 = (-1)m11 f e df
1 -
On the other hand, from Lemma2 ,
1+ m +m1 m+m1- r
ex r
f
f m+m1
m +m 1 -1
dx = e f
n
Σ
r=0 1+ m +m1- r
f
-
(1+ m)
r 1+ m +m - r (log x)
m1 m1 log x
m +m 1 -1 m+m1- r
= (-1) 1 e Σ
r=0 1
- 23 -
Therefore,
(1+m)
x m 1-1
(log x) dx = xΣ (-1)r1r
m
(log x)m+1+r
0 r=0 1+m +1+r
m
m +m (1+ m) 1
m+m - r
+ (-1)m11 1 x Σ (-1)r (log x) 1
r=0 1+ m +m1- r
Here, let m1 = 1 . Then,
(1+m)
(log x)
x
m
dx = x (log x)m+1
0 1+m +1
m +1 (1+ m)
- xΣ (-1)r (log x)m+1- r
r=0 (1+ m +1- r)
(1+m) (1+ m)
=x (log x)m+1 - x (log x)m+1
1+m +1 (1+ m +1)
m +1 (1+ m)
- xΣ (-1)r (log x)m+1- r
r=1 (1+ m +1- r)
m (1+ m)
= xΣ(-1)r (log x)m- r
r=0 (1+ m - r)
i.e.
m- r
x m! 1 m (-1)r (log x)
m
(log x) dx = xΣ (2.1)
0 1 r=0 1r m - r !
When the 2nd order, substituting g =f , f = log x for (1.2), (1.2r) in Formula 23.3.1 ,
(1+m)
x x m 1-1 m 2-1
m m+2+ r+s
(log x) dx = x Σ Σ (-1)r+s 1r 2s
2 2 2
(log x) + Rm1
0 0 r=0 s=0 1+m +2+r + s
(1+m)
1+m + m
f f m+m
m m1 1 f
Rm21 = (-1) 11 f e df e fdf
1 - -
(1+m)
1+m +1+ r + m
m 1-1 r+ m2 m1 m2 f m+1+ r+ m
+ Σ (-1) 1 2 f 2 2f
e df
r=0 2 -
On the other hand, from Lemma2 ,
1+ m +m1 m+m1- r
r
f
f m+m1
m +m 1 -1
ef dx = e n f
Σ
r=0 1+ m +m1- r
f
-
e
f
2f m+1+ r+ m2 e 2f
f dx = m+1+ r+ m2 +1
- 2
1+ m +1+ r + m2
m +1+ r+ m 2 -1
m+1+ r+ m2- s
Σ (2f)
s=0 s 1+ m +1+ r + m2- s
1+ m +m1- r
f
2f m+m - r e 2f m +m - r -1 1
m+m - r- s
e f df = m+m - r+1 Σ
1
(2f) 1
- 2 1 s=0 s 1+ m +m1- r - s
Substituting these for Rm21 ,
(1+m) 1+ m+m1 m+m1- r f
r=0
m m1
f m +m 1 -1
Rm21 = (-1) 11 ef Σ f e df
1+m + m1 - r 1+ m +m1- r
- 24 -
m 1-1 r+ m 2 m1 m2 (1+m) e 2f
+ Σ (-1) 1 2
r=0 1+m +1+ r + m2 2m+1+ r+ m2+1
1+ m +1+ r + m2
s
m +1+ r+ m 2 -1 m+1+ r+ m2- s
Σ (2f)
s=0 1+ m +1+ r + m2- s
(1+ m)
r=0 r 1+ m +m - r
m m1
m +m 1 -1 f m+m1- r
Σ
2f
= (-1) 11 e f df
1 -
m+1+ r+ m2 - s
e 2f m +1+ r + m2
(1+ m)
s 1+ m +1+ r + m - s
m 1-1 r+ m 2 m1 m2 -1 f
+ Σ (-1) 1 2 Σ
r=0 21 s=0 2 2s
x
(1+ m) e 2log
r=0 1+ m +m - r 2
m1 m1
m +m 1 -1
= (-1) 1 Σ r 1
m+m1- r+1
1+ m+m1- r
s
m +m 1- r -1
m+m 1- r- s
Σ
s=0 1+ m +m1- r - s
(2f)
- 25 -
x 2 m +1 m +1- r (1+ m) (log x) m+1- r- s
2Σ Σ (-1) (1+ m +1- r - s)
r+s
-
r=1 s=0 2s
m +2 (1+ m) (log x) m+2- s
Σ
2 s+1
+x (-1)
s=1 (1+ m +2- s) 2s
x 2 m +1 m +1- r (1+ m) (log x) m+1- r- s
2 Σ Σ (-1) (1+ m+1- r - s)
r+s
=-
r=1 s=0 2s
i.e.
r+s
(log x)m+1- r- s
x x m! 2 m +1 m +1- r (-1)
m 2
(log x) dx = - x Σ Σ (2.2)
0 0 2 r=1 s=0 1r 2s m +1- r - s !
Hereafter, by induction, we obtain the general expression.
- 26 -
23.5 Higher Integral of 1 / log x
Formula 23.5.1
When (z), (z ), denote the gamma function, the digamma function, Euler-Mascheroni-Constant
(= 0.57721566...) respectively, the following expressions hold for x 0 .
x x x 1
r+ s+ t r s t
dx 3 = x 3ΣΣΣ(-1) 12 3
0 0 0 log x r=0 s=0 t=0
x x 1 Σ rk n r
dx x Σ Σ Σ
n
= n
(-1) k =1
Π kk
0 0 log x r1=0 r2=0 rn=0 k=1
n
log|log x|- n +Σrk - k =1 (log x)
n-1+Σ rk
k =1
n
n +Σrk
k =1
1 n -1
+ Σ (-1)rn -1C r + log(r +1)x n-1-r
(1.n)
(n -1)! r=0
Calculation
Let g = 1/f . Then, for f > 0,
log f - (k) - k-1
k =1, 2, 3,
<k>
g = (log f)<k -1> = f
(k)
Therefore,
n
g
n + Σ
k=1
rk
n
=
log f - n + Σrk -
k =1 f
n
n+ Σ rk-1
k=1
n
n + Σrk
k=1
log f - m1 - m1-1
g
m1
= f
m1
log f - 1+ r1+ m2 - 1+ r1+ m2-1
g
1+ r1+ m2
= f
1+ r1+ m2
- 27 -
n -1
g
n -1
n -1 + Σ rk+ mn
k =1 =
log f - n -1+ Σrk+ mn -
k =1 f
n -1
n-1 + Σ rk+ mn -1
k=1
n -1
n -1+ Σrk+ mn
k=1
Substitute these for Formula 23.3.1 and returning f to x in the non-remainder terms,
since [- , f ] [0, x ] ,
n
Σ rk n r
x x 1 m 1-1 m 2-1 m n -1
dx = x Σ Σ Σ (-1)k=1 Πk k
n n
0 0 log x r1=0 r2=0 rn =0 k=1
n
log|log x| - n + Σrk - k=1 (log x) k=1
n
n+ Σ rk-1
+ Rmn1 (n)
n
n + Σrk
k =1
log f - m1 -
f f f f
m m m 1-1 f
Rmn 1 = (-1) 11 1 f e df e fdf e fdf e fdf
- - - - m1
m 1-1 r + m2 m1 m2
+ Σ (-1) 1 1 2
r1=0
log f -1+ r1+ m2 -
e df e df
f f f r 1 + m2 2f
f e df f f
- - - 1+ r1+ m2
n -1
m 1-1 m 2-1 m n-1-1 Σ r k+ m n n mk
+ Σ Σ Σ (-1)k =1 Π k
r1=0 r2=0 r =0
n-1 k =1
n -1
log f - n -1 + Σrk+ mn -
n -1
Σ rk+ mn -1
f n-1 +
k=1
n -1
f k=1
e nf df (n,r)
n -1 + Σrk+ mn
-
k=1
When the 1st order,
log|log x|- 1+ r -
x 1 m 1-1
dx = x Σ (-1)r 1r (log x) r + Rm11
0 log x r=0 1+ r
f log f - m1 - m -1
m m
Rm11 = (-1) 1 1 1 f 1 e fdf
-
m1
Here, surprisingly, when m1 ,
log f - m1 - m1-1 f
f
e df =
1 m m1
lim R = mlim (-1) 1 1 f (1r)
m
1
m1 1
-
m1
Therefore,
log|log x|- (1+ r)-
x 1
dx = xΣ(-1)r 1r (log x) r + (1.1)
0 log x r=0 (1+ r)
When the 2nd order,
log|log x| - 2+ r +s -
x x 1 m 1-1 m 2-1
dx 2 = x 2 Σ Σ (-1)r+s1r2s (log x) r+s-1
2+
0 0 log x r=0 s=0 2+ r +s
2
+ Rm1
- 28 -
log f - 1+ r + m2 - 1+ r+ m2-1 2f
m 1-1
r+ m2 m1 m2 f
Rm21 = Σ (-1) 1 2 f e df
r=0 - 1+ r + m2
f log f - m1 -
f
m m1 m -1
+ (-1) 11 f 1 e fdf e fdf
- - m1
Here, surprisingly again, when m1 , m2 ,
lim Rm21 = x - ( + log 2)
m i
i =1, 2 (2r)
Therefore,
log|log x|- 2+ r +s -
x x 1
r+s
dx 2= x 2Σ Σ(-1) 1r2s (log x) 1+ r+s
0 0 log x r=0 s=0 2+ r + s
+ ( + log 1) x - ( + log 2) (1.2)
Also, when the 3rd order,
x x x 1 m 1-1 m 2-1 m 3-1 r + r2+ r3 r1 r r
dx 3 = x 3 Σ Σ Σ (-1) 1 1 2 23 3
0 0 0 log x r1=0 r2=0 r3=0
log f - m -
+ (-1) 1
f f f 1 m1-1 f
m1 m1 f f
f e df e df e df
m
- - - 1
And
+ log 2 + log 3
lim R 3 =
m m1
x2 - x+ i =1, 2, 3 (3r)
i 2! 1! 2!
Therefore,
x x x 1
r+ s+ t r s t
dx 3 = x 3ΣΣΣ(-1) 12 3
0 0 0 log x r=0 s=0 t=0
Although the problems are (1r), (2r), (3r), , these proofs are difficult. However, surely they hold on
the numerical computation
- 29 -
Note
x
1
Therefore, the higher integral of the logarithmic integral li(x) = dt can be expressed
0 log t
as follows.
n
li(x)dx
x x
n+1 Σ rk n +1 r
0 0
n
=x Σ Σ Σ
r =0 r =0 r =0
(-1) k =1
Π k
k =1
k
1 2 n
n +1
log|log x|- 1+ n +Σrk - k =1 n+Σ rk
(log x)
n +1
k =1
n +1
1+ n +Σrk
k =1
1 n
+ Σ (-1)rnC r + log(r +1)x n-r
(1.n)
n! r=0
Formula 14.4.3 ( 14.4 ) using exponential integral Ei(x) is easier as the formula of the higher integral of
li(x) , In contrast, Formula 23.5.1 looks truly complicated and not useful. However, both have a decisive
difference. That is well understood if (1.n) is rewritten using a harmonic number Hn and a factorial ! as follows.
n log|log x| -H n +1 n +1
n+Σ rk
li(x)dx
x x Σ rk n +1 n+Σ rk
rk
Σ Σ(-1)
n n+1
Σ Π
=
k 1 k =1 k =1
=x k n +1
(log x)
n +Σr !
0 0 r =0 r =0 r =0
1 2 n k =1
k
k =1
1 n
+ Σ (-1)rnC r + log(r +1)x n-r
(1.n)
n! r=0
That is, (1.n) is expressed by the elementary functions while Formula 14.4.3 is expressed by non-elementary
functions. Here is the meaning of Formula 23.5.1 .
Now, Formula 23.5.1 can be simplified a little more. For the purpose, the following two lemmas are necessary.
Lemma3
1r r 1 1
Σ
r=0
(-1)
(1+ r)
x =r
0! e x
- 30 -
1r2s
e
1 1 1
r+s r+s
Σ Σ(-1) (2+ r +s)
x 1+ =
1! x
-
r=0 s=0 e 2x
1r2s3t
e
1 1 2 1
r+s+t r+s+t
Σ ΣΣ(-1) (3+ r + s +t)
x 2+ =
2! x
- 2x
+
r=0 s=0 t=0 e e 3x
n
rk
n
Σ rk Π k n
n-1+Σ rk
1 n -1 n -1C r
Σ (-1)k =1
k=1
Σ Σ n x k=1
= Σ (-1)r 1+ r
(n -1)! r=0
r =0 r =0
1 r =0
2 n n +Σrk e
k=1
Proof
By induction. Details are discussed in one chapter of "À la carte".
Lemma3'
1 1r r x0
x Σ(-1) (log x) = r
r=0 (1+ r) 0!
2 r+s 1r2s (x -1)1
x ΣΣ(-1) (log x)1+ r+s =
r=0 s=0 (2+ r +s) 1!
3
r+s+t 1r2s3t 2+ r+s+t (x -1)2
x Σ ΣΣ(-1)
r=0 s=0 t=0 (3+ r +s + t)
(log x) =
2!
n
rk
n
Σ rk Π k n-1+Σ rk
n
(x -1)n-1
Σ Σ
k=1
x n
Σ (-1) k=1
n (log x) k =1
= (3'.n)
(n -1)!
r =0 r =0
1 r =0
2 n n +Σrk
k=1
Formula 23.5.1'
When (z), (z ) denote the gamma function and the digamma function respectively, the following
expressions hold for x 0 .
(1+ r)
x 1
dx = -xΣ(-1)r1r (log x) r
0 log x r=0 (1+ r)
1
(x -1) log|log x|+ x log 1
0 0
+ (1.1')
0!
(2+ r + s)
x x 1
r+s
dx 2 = -x 2ΣΣ(-1) 1r2s (log x) 1+ r+s
0 0 log x r=0 s=0 ( 2+ r s
+ )
1
+ (x -1)1 log|log x|+ x 1log 1 - x 0log 2 (1.2')
1!
- 31 -
x x x 1
r+ s+ t r s t
dx 3 = -x 3ΣΣΣ(-1) 1 23
0 0 0 log x r=0 s=0 t=0
(3+ r + s + t)
(log x)2+ r+ s+ t
(3+ r + s + t)
1
+ (x -1)2 log|log x|+ x 2log 1-2x 1log 2+ x 0log 3 (1.3)
2!
n
n
n +Σrk
(log x) n
x x Σ rk n r n-1+Σ rk
1 k =1
dx n = -x nΣ Σ Σ(-1) k =1 Πk k n
k =1
log x
0 0 r1=0 r2=0 rn=0 k=1
n +Σr k
k =1
1 n -1
+
(n -1)!
(x -1) n-1 log|log x| -Σ(-1)n-rn -1C r x r log(n -r)
r=0
(1.n')
Proof
Substituting Lemma3' (3'.n) for Formula 23.5.1 (1.n) ,
n
n
n +Σrk
(log x) n
x x Σ rk n r n-1+Σ rk
1 k =1
dx n = -x nΣ Σ Σ(-1) k =1 Πk k n
k =1
log x
0 0 r1=0 r2=0 rn=0 k=1
n +Σr k
k =1
n rk
n
Σ rk Π k n
n-1+Σ rk
+ log|log x| - x nΣ Σ Σ(-1) k =1
k=1 k =1
n
(log x)
n +Σrk
r1=0 r2=0 rn=0
k =1
1 n -1
+ Σ (-1)rn -1C r + log(r +1)x n-1-r
(n -1)! r=0
n
= -x nΣ Σ Σ(-1)
Σ
n
rk n rk
n +Σrk
(log x)
k =1
n
n-1+Σ rk
Π
=
k 1 k =1
k n
n +Σr
r1=0 r2=0 rn=0 k=1
k
k =1
log|log x|
+ (x -1) n-1 - (x -1) n-1
(n -1) ! (n -1) !
n -1
n-1-r 1 n -1
+ Σ (-1)r
n C
-1 rx + Σ (-1)rn -1C r x n-1-rlog(r +1)
(n -1)! r=0 (n -1)! r=0
n
n Σ
n
rk n rk
n +Σrk
(log x)
k =1
n
n-1+Σ rk
Σ Σ Σ(-1) Π
=
k 1 k =1
= -x k n
n +Σr
r =0 r =0 r =0
1 2 n k=1
k
k =1
1 n -1
+
(n -1)!
(x -1) n-1 log log x - Σ(-1)n-rn -1C r x rlog(n -r)
r=0
n -1 n -1
- 32 -
Example The 3rd order integral of 1/log x (The 2nd order integral of li(x) )
When both sides of (1.3') are illustrated, it is as follows. Since both sides overlapp exactly, the left side (blue)
is not visible.
- 33 -
23.6 Higher Integral of log|log x|
Formula 23.6.1
When (z), (z ), denote the gamma function, the digamma function, Euler-Mascheroni Constant
(= 0.57721566...) respectively, the following expressions hold for x 0 .
log|log x| - 2+ r -
log|log x|dx = xΣ(-1) 1
x
(log x) r -
r r 1+
(1.1)
0 r 2+ r
r+s r s log|log x | - 3+ r +s -
log|log x|dx = x ΣΣ(-1)
x x
(log x) r+s
2 2 2+
12
0 0 r=0 s=0 3+ r +s
1
- 2( + log 1) x - ( + log 2) (1.2)
2!
3 log|log x| - 4+ r +s +t -
Σ Σ Σ (log x ) r+s+t
3+
=x (-1)r+ s+ t1r 2s 3t
r=0 s=0 t=0 4+ r +s +t
1
3 + log 1x 2 -3 + log 2x + + log 3
3!
- (1.3)
= x nΣ Σ Σ(-1)
Σ
n
rk n rk
log|log x| - 1+ n +Σrk -
k =1 n
n+Σ r k
Π
=
k 1 k =1
k n
(log x)
1+ n +Σrk
r1=0 r2=0 rn=0 k =1
k =1
1 n
(-1)rnC r( + logr) x n-r
n! Σ
+ (1.n)
r=1
Calculation
Let g = log f , f = log x . Then,
n
g
n
n +Σ rk
k =1 = log f - 1+ n +Σrk - k=1 f
n
n + Σ rk
k=1
n
1+n +Σrk
k =1
log f - 1+ m 1 -
g 1 =
m m
f 1
1+m1
log f - 1+1+ r1+ m2 - 1+ r1+m2
g 1 2 =
1+ r +m
f
1+1+ r1+m2
log f - 1+2+ r1+ r2+m3 - 2+ r1+ r2+m3
g 1 2 3 =
2+ r + r +m
f
1+2+ r1+ r2+ m3
n -1
g
n -1
n -1 + Σ rk+ mn
k =1 =
logf - 1+ n -1+Σrk+ mn -
k=1 f
n -1
n -1 + Σ rk+ mn
k =1
n -1
1+n -1+Σrk+ mn
k =1
- 34 -
Substitute these for Formula 23.3.1 and returning f to x in the non-remainder terms,
since [- , f ] [0, x ] ,
n
Σ rk
x x m 1-1 m 2-1 m n -1 n
log|log x|dx n = x n Σ Σ Σ (-1)k=1
rk
0 0 r1=0 r2=0 rn =0
Π
k=1
k
n
log|log x|- 1+ n +Σrk - k=1 (log x)
n +Σ rk
k =1
n
+ Rmn1 (n)
n
1+n +Σrk
k =1
Rmn1 =
log f - 1+m1-
f f f f m1 f
m m1
(-1) 11 f e df e fdf e fdf e fdf
- - - - 1+m1
m 1-1
r + m2 m1 m2
+ Σ (-1) 1 1 2
r1=0
n -1
logf - n + Σrk + mn - n -1 + Σ rk+ mn
f
k =1
n-1
f k=1
e nfdf (nr)
n + Σrk + mn
-
k =1
When the 1st order,
log|log x| - 2+ r -
x m 1-1
log|log x| dx = x Σ (-1)r 1r (log x) r + Rm1
1+ 1
0 r 2+ r
log|f|- 1+m1- m1 f
1 m m1 f
Rm1 = (-1) 1 1 f e df
- 1+m1
Here, surprisingly, when m1 ,
log|f|- 1+ m1 - m1 f
f
f e df = -
1 m m1
lim R = mlim (-1) 1 1 (1r)
m
1
m1 1
-
1+m1
Therefore,
log|log x| - 2+ r -
log|log x|dx = xΣ(-1) 1
x
(log x) r -
r r 1+
(1.1)
0 r 2+ r
When the 2nd order,
- 35 -
log|log x|dx
x x
2
0 0
m 1-1 m 2-1 log|log x| - 3+ r +s -
Σ Σ (log x) r+s + Rm1
2 2+ 2
=x (-1)r+s1r2s
r=0 s=0 3+ r +s
f log |f|- 2+ r + m2 - 1+ r+ m
m 1-1 r+ m m m
Rm21 = Σ (-1) 21 12 2 f 2 2f
e df
r=0 - 2+ r + m2
f log |f|- 1+m1-
m m f m
+ (-1) 11 1 f 1e fdf e fdf
- - 1+m1
Here, surprisingly again, when m1 , m2 ,
1
lim R 2 = - 2( + log 1)x - ( + log 2)
2!
m m1
(i =1, 2) (2r)
i
Therefore,
log|log x| - 3+ r +s -
log|log x|dx 2
x x
= x Σ Σ(-1)r+s1r2s (log x) r+s
2 2+
0 0 r=0 s=0 3+ r +s
1
- 2( + log 1) x - ( + log 2) (1.2)
2!
Also, when the 3rd order,
3 log|log x | - 4+ r +s +t -
Σ Σ Σ (log x) r+s+t
3+
=x (-1)r+ s+ t1r 2s 3t
r=0 s=0 t=0 4+ +s +t
r
1
- 3 + log 1x 2 -3 + log 2x + + log 3 (1.3)
3!
Hereafter, by induction, we obtain the general expression.
Although the problems are (1r), (2r), (3r), , these proofs are difficult. However, surely they hold on
the numerical computation.
- 36 -
Example The 2nd integral of loglog x
When the one arbitrary point x =4.7 is given in (1.2) , the values of the both sides are as follows.
Note
Formula 14.5.1 (14.5 ) using Ei(x) is easier as the formula of the higher integral of log|log(x)| . In contrast,
Formula 23.6.1 looks truly complicated and not useful. However, both have a decisive difference. For example,
if (1.2) is rewritten using a harmonic number Hn and a factorial ! , it is as follows.
x x log|log x| -H2+ r+s
log|log x| dx 2 = x 2Σ Σ(-1)r+s1r2s (log x) r+s
2+
0 0 r=0 s=0 2+ r +s !
1
- 2( + log 1) x - ( + log 2) (1.2)
2!
That is, (1.2) is expressed by the elementary functions while Formula 14.5.1 is expressed by non-elementary
functions. Here is the meaning of Formula 23.6.1.
Now, Formula 23.6.1 can be simplified a little more. For the purpose, the following two lemmas are necessary.
Lemma4
1r
1 1
r
Σ (-1) r
(2+ r)
x 1+ =
1!
1- x
r=0 e
r s
12 1 2 1
Σ Σ (-1)r+s r+s
x 2+ = 1- x + 2x
r=0 s=0 (3+ r +s) 2! e e
r s t
123 1 3 3 1
Σ Σ Σ (-1)r+s+t x 3+
r+s+t
= 1- x + 2x - 3x
r=0 s=0 t=0 (4+ r + s +t) 3! e e e
n
rk
n
Σ rk Π k n
n+Σ rk
1 n nC r
Σ Σ Σ (-1)k =1
k =1
n x k=1
= Σ (-1)r r
r1=0 r2=0 rn =0
1+ n +Σrk n ! r=0 e
k =1
Proof
By induction. Details are discussed in one chapter of "À la carte". In addition, this can be drawn also from
Lemma3.
- 37 -
Lemma4'
1r
1 1
Σ (-1) (log x)1+ r =
r
1-
r=0 (2+ r) 1! x
1r2s 1 2
1
r+s 2+ r+s
Σ Σ(-1) (3+ r +s) (log x)
r=0 s=0
=
2!
1-
x
r s t 3
123 1 1
Σ Σ Σ (-1)r+s+t (log x)3+ r+s+t = 1-
r=0 s=0 t=0 (4+ r + s +t) 3! x
n
rk
Σ
n
rk Π k n+Σ rk
n
n
1 1
Σ Σ Σ (-1)k =1 k =1 k=1
n (log x) = 1- (4'.n)
r1=0 r2=0 rn =0
1+ n +Σrk n! x
k =1
Formula 23.6.1'
When (z), (z ), denote the gamma function, the digamma function, Euler-Mascheroni-Constant
(= 0.57721566...) respectively, the following expressions hold for x 0 .
2+ r
log|log x|dx = -xΣ(-1) 1
x
(log x ) r
r r 1+
0 r 2+ r
1
log|log x| - (x -1) -( + log1) x
1 0
+ (1.1')
1!
3+ r +s
x x
log|log x | dx 2 = -x 2ΣΣ(-1)r+s1r2s (log x) r+s
2+
0 0 r=0 s=0 3+ r +s
1
2
log|log x | - (x -1) +Σ(-1) 2C r( + logr) x
2 r 2-r
+ (1.2')
2! r=1
4+ r +s +t
x x x
log|log x| dx 3 = -x 3ΣΣΣ(-1)r+ s+ t1r 2s 3t (log x) r+s+t
3+
0 0 0 r=0 s=0 t=0 4+ r +s +t
1
3
log|log x | - (x -1) +Σ(-1) 3C r( + logr) x
3 r 3-r
+ (1.3')
3! r=1
n
x x Σ rk n rk k =1
+ k
log|log x | dx = -x ΣΣ Σ ( n
-1 ) Π k n k 1=
n
k =1
1+n +Σr
0 0 r1=0 r2=0 rn=0 k=1
k
k =1
1 n
n! (1.n')
log|log x | - (x -1) +Σ(-1) C ( + logr) x
n r n-r
+ n r
r=1
Proof
Substituting Lemma4' (4'.n) for Formula 23.6.1 (1.n) ,
log|log x|dx
x x
n
0 0
- 38 -
n
n Σ
n
rk n rk
1+ n +Σrk
(log x)
k =1
n
n+Σ rk
Σ Σ Σ(-1) Π
=
k 1 k =1
= -x k n
1+ n +Σr
r =0 r =0 r =0
1 2 n k =1
k
k =1
n rk
Σ
n
rk Π k n
n+Σ rk
+ log|log x| - x nΣ Σ Σ(-1) =
k 1 k=1 k =1
n
(log x)
1+ n +Σrk
r1=0 r2=0 rn=0
k =1
1 n
(-1)rnC r( + logr) x n-r
n! Σ
+
r=1
n
Σ
n
rk n rk
1+ n +Σrk (log x)n+Σ r n
k
Σ Σ
k =1
= -x n
Σ (-1) k =1
Π k n
k =1
r =0 r =0
1 2 r =0 n k =1
1+ n +Σrk
k =1
n n
x 1 1 n
+ log|log x|- 1- + Σ (-1)rnC r( + logr)x n-r
n! x n! r=1
n
= -x nΣ Σ Σ(-1)
Σ
n
rk n rk
1+ n +Σrk (log x)n+Σ r
k =1
n
k
k =1
Πk n
k =1
1 n
log |log x|- (x -1) +Σ(-1) nC r( + logr)x
n r n-r
+
n! r=1
- 39 -
23.7 Possibility to super integral of composition
As understood from Formula23.1.1 , the n th order integral of composision g{f(x )} becomes n(n+1)/2 - fold
series in general. Therefore, for example, the 1.5 th order integral of composition g{f(x )} will be expressed by
1.875-fold series. However, such a multiple series does not exist. i.e. it is impossible to make Formula23.1.1
Super Integral.
However, when the core function f(x) is the 1st degree, according to Formula 23.1.2 ,
x x n f f
1
gf(x)dx
n
= g(f)df n f(x)= cx +d
a a c fa fa
So, there is no obstacle in extending this domain from the natural number n to real number p .
Thus, the following expression holds for a real number p >0 .
x x p f f
1
gf(x)dx
p
= g(f)df p f(x)= cx +d
a a c fa fa
(1.3p)
This is the grounds for which we have used "Linear form" since " 07 Super Integral " as a fait accompli .
2011.01.15
K. Kono
Alien's Mathematics
- 40 -
24 Sugioka's Theorem on the Series of Higher (Repeated) Integrals
Mikio Sugioka discovered that the series of the higher integral of arbitrary functions results in one integral in
March, 2003. ( http://www5b.biglobe.ne.jp/~sugi_m/page030.htm ). I was introducing these theorems in the
one section in my work " 16 Higher Integral of the product of two functions ". Since Mr. Sugioka discovered
many theorems after that, they can no longer be housed in one section. Then, I decided to prepare a chapter
independent here and to introduce his wonderful theorems.
Although Mr. Sugioka named the thorem "An operator's theorem", in this paper, I named like the title so that
the contents might be understood well.
f(x)dx r = e f(x)e
r=1
x x x
Σ x -x
dx (1.1')
a a a
Proof
Let
f(x)dx
x x
r =1, 2, ,m
<r> r
f (x) = (1.r)
a a
f f
x x x x
f(x)e -xdx = f <1>(x)e -xa + <1>
(x)e -x = f <1>(x)e -x + <1>
(x)e -x
a a a
f
x x
= f <1>(x)e -x + f <2>(x)e -xa + <2>
(x)e -x
a
f
x
= f <1>(x)e -x + f <2>(x)e -x + <2>
(x)e -x
a
f
m x
= e -x Σf < r>
(x) +
<m> -x
e dx
r=1 a
Substituting (1.r) for this,
f(x)dx e
x m x x x x x
f(x)e dx = e -x Σ
-x
f(x)dx r + m -x
dx
a r=1 a a a a a
f(x)dx
m -1 (x -a)r x x
f
<m>
(x) = Σ f <m-r>
(a) + m
(1.m)
r=0 r! a a
-1-
On the other hand, since f(x)is analytic function on I , this can be expanded to Taylor series as follows.
(r) (x -a)r (x -a)r
f(x) = Σf (a) = Σf <-r>
(a)
r=0 r! r=0 r!
Shifting by m the index of the integration operator of f(x)
<m>
<m-r> (x -a)r
f (x) = Σf (a) (1.t)
r=0 r!
Comparing the right-hand sides of (1.m) and (1.t) , we obtain
f(x)dx
x x
m
lim
m
=0
a a
Thus, (1.1') holds.
f(x)e f(x)e
x x
ex -x x
dx = e-x
dx
a a
y1(x) e f(x)e
x
= e xe -xy1(x)a x
dx
-x
(-1)dx (-1)dxdx
-
y1(x)= e x f(x)e -xdx = e f(x) e
Then, we notice that this is a particular solution of the 1st order differential equation y'(x)- y(x)= f(x).
dx
x x x x x x
Example 1 dx + dx 2 + 3
+
a a a a a a
dx = m !
x x m
(x -a) m
f(x)=1 ,
a a
Substituting these for (1.1) and using the incomplete gamma function (x, y ) ,
dx e
r=1
m x x x x (x -a)m -x
Σ r
=e x -x
dx - e x
m!
e dx
a a a a
x
x e x
= e x-e -xa - -e -a(m +1, x -a)a
m!
That is
dx
x-a
r=1
m x x e
Σ (m +1, x -a) - (m +1, 0)
r x-a
=e -1 + (1.2)
a a m!
x-a
e
And, since lim (m +1, x -a ) - (m +1, 0) = 0 ,
m m !
dx r = e x-a -1
r=1
x x
Σ (1.2')
a a
The higher order integral of the left side of (1.2) is as follows.
-2-
a dx
m x x m (x - a)r
Σ
r=1 a
r
=Σ
r=1 r!
Using this as the left side of (1.2) and (1.2') ,
r x-a
m (x - a ) e
(m +1)-(m +1 , x -a)
x-a
Σ
r=1 r!
=e -1 -
m!
(1.3)
(x - a)r
Σ = e x-a - 1 (1.3')
r=1 r!
When x =2, a =1, m =4 , Both sides of (1.3) are calculated as follows.
e dx +
x x x x x x
Example 2 e x dx + e x dx 2 + x 3
a a a a a a
e dx = e - e Σ r !
x x r
x (x -a)
x m x a
m -1
f(x)= e ,
a a r=0
Substituting these for (1.1) and using the incomplete gamma function (x, y ) ,
r=1
e dx e dx
m x x x x x x
Σ e xdx r = e x e xe -xdx - e x x m -x
a a a a a a
= e dx - e e - e Σ
r!
x x r
x (x -a) x x a
m -1
-x
e dx
a a r=0
= e (x -a) - e e e dx + e Σ
r!
1 x m -1 x
x x x -x x+ a r -x
(x -a) e dx
a r=0 a
= e (x -a) - e e e dx - e Σ
x m -1 -a
e
(r +1, x -a)- (r +1, 0)
x x x -x x+ a
r! a r=0
m -1 (r +1, x -a)- (r +1)
= e x(x -a) - e x(x -a) - e x Σ
r=0 r!
i.e.
(r +1, x -a)
e xdx r = e x(x -a)- e x(x -a)- e x Σ
r=1
m x x m -1
Σ r=0 r!
-1 (1.4)
a a
Next,
-3-
n -1
xs
(n , x) = (n -1)! e Σ -x
s=0 s!
From this,
(r +1, x -a) r (x -a)s
r!
=e -(x-a )
Σ
s=0 s!
Using this,
(r +1, x -a)
e
r=1 r=0
m x x m -1
Σ x r
dx = -e x
Σ r!
-1
a a
Σ
m -1 r (x -a)s
= - e xe -(x-a)Σ - e (x-a)
r=0 s=0 s!
Here,
m -1 r (x -a)s (m -r)(x -a)r
m -1
Σ
r=0
Σ
s=0 s !
- e (x-a)
r=0
=Σ
r !
- e (x-a)
m -1 r(x -a )r m -1 m(x -a )r
= -Σ +Σ - m e (x-a)
r=0 r! r=0 r!
m -1 (x -a )r-1 m -1 (x -a )r
= -(x -a)Σ + mΣ - m e (x-a)
r=1 (r -1)! r=0 r!
Substituting this for the above,
(r +1, x -a)
e
r=1
m x x m -1
Σ x r
dx = -e x Σ
r=0 r!
-1
a a
m -1 (x -a )
r-1 m -1 (x -a )r
= e xe -(x-a) (x -a)Σ - mΣ + m e (x-a)
r=1 (r -1) ! r=0 r !
When m ,
Note
e dx = e r=1
x x x x
If we adopt the lineal higher integral
x r x
then Σ e x dx r = .
- - - -
Therefore, it is for a collateral higher integral that this theorem is significant.
A by-product
Σ = lim Σ
m -1 r (x -a)s m -1 (m -r)(x -a)r
lim Σ
m r=0
- e (x-a)
m r=0
- e (x-a)
s=0 s! r!
= -e (x-a)(x -a)
-4-
log xdx
x x x x x x
Example 3 log xdx + log xdx 2 + 3
+
0 0 0 0 0 0
Substituting f(x)= log x , a =0 for (1.1') ,
x x
log xe -xdx = -e -x log|x|+ Ei(-x)0
0
-x
= -e log|x|+ Ei(-x) -
x et
Where, Ei(x) = dt , = 0.5772 (Euler-Mascheroni Constant ).
- t
This integral is not lineal as understood from this constant. Multiplying by e x the both sides, we obtain
r=1
log xdx
x x
Σ r
= -log|x|+ e xEi(-x) - (1.6')
0 0
The higher integral of the left side is the following lineal integral.
= -log|x|+ e Ei(-x)-
x r 1
Σ
r=1 r!
log|x| -Σ
s=1 s
x
When the first 10 terms of ∑ are calculated and both sides are illustrated, it is as follows.
Both sides overlap exactly and blue (left) can not be seen.
-5-
Theorem 24.1.2 ( Sugioka's Theorem 4 )
When f(x )is analytic function defined on a closed interval I, the following expressions hold for a ,x I .
a a
Σ
r=1 a a
r -x
a
x -x
a
m x
(2.1)
Proof
In a similar way to the proof of Theorem 24.1.1 , we obtain the desired expressions.
x x
sin x e dx = sin 4 e sin x - 4
x
x
a a
2
= e xsin x - - e asin a -
2 4 4
Here, 2 sin x - 4 = sin x -cosx . Then
n (n -s)
sin x dx
n -1 (x -a )s
-Σ
x x
n
= sin x - sin a -
a a 2 s=0 s! 2
Then, the left side is
r (r -s)
sin x - 2 -Σ
(x -a)s
r-1
Σ
r=1
(-1) r-1
s=0 s!
sin a -
2
When a =3/2 , the first 25 terms of ∑ are calculated and both sides are illustrated, it is as follows.
Both sides overlap exactly and blue (left) can not be seen.
-6-
-7-
24.2 Series of the odd-th order Integrals
In this section, we ask for the sum of the following series of higher integrals.
f(x)dx
x x x x x x x x
f(x)dx f(x)dx + f(x)dx +
3 5 7
a a a a a a a a
Theorem 24.2.1
When f(x )is analytic function defined on a closed interval I, the following expressions hold for a ,x I .
e x +(-1)-2m+1e
x x x -x
2m-1
- (-1)2m-1cosh x f(x)dx dx
a a a 2
f(x)e dx
x x 1 x x x
Σ f(x)dx 2r-1 = e f(x)e -xdx + e -x x
(1.1')
r=1 a a 2 a a
Proof
Formula of repeated integration by parts was as follows. (" 01 Generalized Taylor's Theorem " ( A la Carte ) )
x m x x
f(x)g(x)dx = (-1) f <r>(x)g ( )(x)a + (-1)m f <m>(x)g (x)dx
r-1 r-1 (m)
a r=1
Σ a
When g(x)= cosh x , sinh x ,
(r-1) e x +(-1)-r+1e -x (m) e x +(-1)-me -x
(cosh x) = , (cosh x) =
2 2
(r-1) e x -(-1)-r+1e -x (m) e x -(-1)-me -x
(sinh x) = , (sinh x) =
2 2
Substituting these for the above,
x
f
m
x
<m> e x +(-1)-me -x
+ (-1) (x) dx ,
a 2
x
f
m
x
<m> e x -(-1)-me -x
+ (-1) (x) dx ,
a 2
Here, let
f(x)dx
x x
<r>
f (x) = r
r =1, 2, ,m
a a
-8-
f(x)dx
x m e x +(-1)-r+1e -x x x
f(x)cosh xdx = Σ(-1)
r-1
r
a r=1 2 a a
f(x)dx
m
x x x
m e x +(-1)-me -x
+ (-1) dx
a a a 2
f(x)dx
x x -r+1 -x x x
r-1
m e +(-1) e
f(x)cosh xdx = Σ(-1) r
a r=1 2 a a
f(x)dx
m
x x x
m e x -(-1)-me -x
+ (-1) dx
a a a 2
Expanding a part of the 1st term of the right side,
x -r+1 -x
r-1 e +(-1) e
m
Σ
r=1
(-1)
2
= cosh x - sinh x + cosh x - sinh x +-
m
r-1 e x-(-1)-r+1e -x
Σ
r=1
(-1)
2
= sinh x - cosh x + sinh x - cosh x +-
Multiplying both sides by cosh x, sinh x respectively,
e x+(-1)-r+1e
m
-x
cosh xΣ(-1) r-1
2 2
= cosh x - cosh x sinh x + cosh x - cosh x sinh x +-
r=1 2
e x-(-1)-r+1e
m
-x
sinh xΣ(-1) r-1 2 2
= sinh x - sinh x cosh x + sinh x - sinh x cosh x +-
r=1 2
The 2nd term of the right side is as follows.
e x +(-1)-me
x x x -x
m
(-1) cosh x
a a a
f(x)dx m
2
dx
m e -(-1) m e
dx
a 2 a a
2 2
Since cosh x - sinh x = 1 ,
f(x)dx
x x m x x
cosh x f(x)cosh xdx - sinh x f(x)sinh xdx = Σ 2r-1
a a r=1 a a
e x +(-1)-2m+1e
x x x -x
2m-1
+ (-1)2m-1cosh x f(x)dx dx
a a a 2
f(x)sinh xdx
x x
cosh x f(x)cosh xdx - sinh x
a a
e x+e -x x e x+e -x e x-e -x x e x-e -x
= f(x) dx - f(x) dx
2 a 2 2 a 2
-9-
Using this, we obtain
e x +(-1)-2m+1e
x x x -x
2m-1
- (-1)2m-1cosh x f(x)dx dx
a a a 2
f(x)dx
x x
m
And since lim =0 holds by assumption ( See proof of Theorem 24.1.1 ), we obtain (1.1').
m a a
dx
x x x x x x x x
dx dx + dx
3 5 7
Example1
a a a a a a a a
x x
m (x -a)m
f(x)=1 , dx =
a a m!
Substituting these for (1.1) ,
a dx e e dx
m x x 1 x x x
Σ x
2r-1 -x -x
= e dx + e
r=1 a 2 a a
2m-1
x (x -a)2m-1 e x +(-1)-2m+1e -x
- (-1) cosh x dx
a (2m -1)! 2
2m-1
x (x -a)2m-1 e x -(-1)-2m+1e -x
+ (-1) sinh x dx
a (2m -1)! 2
Here,
e e dx
1 x x x ex -x x ex x x
e -x
dx + e -x x
= -e a + e a
2 a a 2 2
x-a
e - e -(x-a)
= = sinh(x -a)
2
Then,
dx
m x x
Σ 2r-1
= sinh(x -a)
r=1 a a
2m-1
x (x -a)2m-1 e x +(-1)-2m+1e -x
- (-1) cosh x dx
a (2m -1)! 2
2m-1
x (x -a)2m-1 e x -(-1)-2m+1e -x
+ (-1) sinh x dx
a (2m -1)! 2
(1.2)
dx
x x
Σ 2r-1
= sinh(x -a) (1.2')
r=1 a a
The left side is
a dx
m x x m (x - a)2r-1
Σ =Σ
2r-1
r=1 a r=1 (2r -1)!
- 10 -
Using this as the left side of (1.2) and (1.2') ,
2r-1
m (x -a)
Σ
r=1 (2r -1)!
= sinh(x -a)
2m-1
x (x -a)2m-1 e x +(-1)-2m+1e -x
- (-1) cosh x dx
a (2m -1)! 2
x (x -a)2m-1 e x -(-1)-2m+1e -x
+ (-1)2m-1sinh x dx
a (2m -1)! 2
(1.3)
2r-1
(x -a)
Σ
r=1 (2r -1)!
= sinh(x -a) (1.3')
Also, (1.3') is correct. Because the left side is Taylor expansion of the right side.
f(x)sin xdx
m x x x x
Σ(-1)
r-1
2r-1
f(x)dx = cosx f(x)cosxdx + sin x
r=1 a a a a
2m -1
f(x)dx cos x + ( 2 ) dx
x x x
2m-1
- (-1)2m-1cos x
a a a
(2m -1)
sin x f(x)dx
x x x
sin x + dx
2m-1
- (-1)2m-1
a a a 2
(2.1)
f(x)sin xdx
x x x x
Σ (-1) r-1
f(x)dx 2r-1 = cosx f(x)cosxdx + sin x (2.1')
r=1 a a a a
Proof
Formula of repeated integration by parts was as follows (" 1 Generalized Taylor's Theorem ") .
x m x x
f(x)g(x)dx = (-1) f <r>(x)g ( )(x)a + (-1)m f <m>(x)g (x)dx
r-1 r-1 (m)
a r=1
Σ a
When g(x)= cosx , sin x ,
- 11 -
(r -1) m
(cosx)(r-1) = cos x + (cosx) m = cos x +
( )
,
2 2
(r -1) m
(sin x)(r-1) (sin x) m = sin x +
( )
= sin x + ,
2 2
Substituting these for the above,
(r -1) x
f(x)cosxdx = Σ(-1)
x m
r-1 <r>
f (x)cos x +
a r=1 2 a
m
x
+ (-1)m f <m>(x)cos x + dx
a 2
(r -1) x
f(x)dx
x x
<r>
f (x) = r
r =1, 2, ,m
a a
(r -1)
f(x)dx
x m x x
r-1
f(x)cosxdx = Σ(-1) cos x + r
a r=1 2 a a
m
+ (-1) f(x)dx cos x +
2
x x x
m m
dx
a a a
(r -1)
f(x)sin xdx = Σ(-1) 2
sin x +
x m x x
r-1
f(x)dx r
a r=1 a a
m
+ (-1) f(x)dx sin x +
2
x x x
m m
dx
a a a
Expanding a part of the 1st term of the right side,
(r -1)
= cosx + sin x - cosx - sin x ++--
m
r-1
Σ
r=1
(-1) cos x +
2
(r -1)
= sin x - cosx - sin x + cosx ++--
m
Σ (-1) r-1 sin x +
r=1 2
Multiplying both sides by
cos x, sin x respectively,
(r -1)
m
cos xΣ(-1) r-1 cos x +
2 2
= cos x + sin x cos x - cos x - sin x cos x ++--
r=1 2
(r -1)
sin x + = sin x - sin x cos x - sin x + sin x cos x ++--
m
sin xΣ(-1) r-1
2 2
r=1 2
The 2nd term of the right side is as follows.
2
m m
dx
a a a
- 12 -
2 2
Since cos x + sin x = 1 ,
f(x)dx
x x m x x
cosx f(x)cosxdx + sin x f(x)sin xdx = Σ(-1)
r-1
2r-1
a a r=1 a a
2m -1
f(x)dx cos x + ( 2 ) dx
x x x
2m-1
+ (-1)2m-1cos x
a a a
(2m -1)
sin x f(x)dx
x x x
sin x + 2 dx
2m-1
+ (-1)2m-1
a a a
Here, m in the 2nd term and the 3rd term on the right side has to be an odd number corresponding to the 1st
term. Then, transposing the remainder terms , we obtain (2.1) .
f(x)dx
x x
And since lim
m a
m
=0 holds by assumption ( See proof of Theorem 24.1.1 ), we obtain (2.1').
a
e dx
x x x x x x x x
e xdx e xdx + e xdx x
+-
3 5 7
Example2
a a a a a a a a
e sin xdx
x x x x
Σ(-1)
r-1
e xdx 2r-1 = cosx e xcosxdx + sin x x
r=1 a a a a
The right side is as follow.
x 1 x
x
e cosxdx = e (cosx + sin x)-e a (cosa + sin a)
a 2
x 1
e xsin xdx = - e x(cosx - sin x)-e a(cosa - sin a)
a 2
From these,
x
x
x
x ex ea
cosx e cosxdx + sin x e sin xdx = - cos(x -a)- sin(x -a)
a a 2 2
Therefore,
e dx
r-1
x x ex ea
Σ x 2r-1
(-1) = - cos(x -a)- sin(x -a)
2
(2.2')
r=1 a a 2
The higher integral of the left side is as follows according to Theorem 4.1.3 ( 4.1 )
x x (x -a)s n -1
e dx = e - Σe
x n x a
a a s=0 s!
Then, (2.2') becomes
s
e
2r-2 (x -a ) ex ea
Σ
r=1
(-1) r-1 x
-e a
Σ
s=0 s!
=
2
-
2
cos(x -a ) - sin(x -a ) (2.3')
When a =0.3 , the first 20 terms of ∑ are calculated and both sides are illustrated, it is as follows.
Both sides overlap exactly and blue (left) can not be seen.
- 13 -
- 14 -
24.3 Series of the even-th order Integrals
In this section, we ask for the sum of the following series of higher integrals.
f(x)dx
x x x x x x x x
f(x)dx f(x)dx + f(x)dx +
2 4 6 8
a a a a a a a a
Theorem 24.3.1
When f(x )is analytic function defined on a closed interval I, the following expressions hold for a ,x I .
Proof
The following expressions were obtained during the proof of Theorem 24.2.1.
f(x)dx
m
x x x
m e x +(-1)-me -x
+ (-1) dx
a a a 2
f(x)dx
x x -r+1 -x x x
r-1
m e -(-1) e
f(x)sinh xdx = Σ(-1) r
a r=1 2 a a
f(x)dx
x x x e x -(-1)-me -x
+ (-1)m m
dx
a a a 2
Expanding a part of the 1st term of the right side,
x -r+1 -x
r-1 e +(-1) e
m
Σ
r=1
(-1)
2
= cosh x - sinh x + cosh x - sinh x +-
m
r-1 e x-(-1)-r+1e -x
Σ
r=1
(-1)
2
= sinh x - cosh x + sinh x - cosh x +-
Multiplying both sides by sinh x, cosh x respectively,
e x+(-1)-r+1e
m
-x
sinh xΣ(-1) r-1
2 2
= sinh x cosh x - sinh x + sinh x cosh x - sinh x +-
r=1 2
e x-(-1)-r+1e
m
-x
cosh xΣ(-1) r-1 2 2
= cosh x sinh x - cosh x + cosh x sinh x - cosh x +-
r=1 2
The 2nd term of the right side is as follows.
f(x)dx
e x +(-1)-me
x x x -x
m m
(-1) sinh x dx
a a a 2
- 15 -
e x -(-1)-me
x x x -x
(-1)mcosh x
a a
a
f(x)dx m
2
dx
2 2
Since cosh cx - sinh cx = 1
f(x)dx
x x m x x
sinh x f(x)cosh xdx - cosh x f(x)sinh xdx = Σ 2r
a a r=1 a a
f(x)dx
e x +(-1)-2me
x x x -x
2m 2m
+ (-1) sinh x dx
a a a 2
e -(-1) 2m e
- (-1)2m dx
a 2 a a
Here, m in the 2nd term and the 3rd term on the right side has to be an even number corresponding to the 1st
term. Furthermore,
f(x)sinh xdx
x x
sinh x f(x)cosh xdx - cosh x
a a
e x-e -x x e x+e -x e x+e -x x e x-e -x
= f(x) dx - f(x) dx
2 a 2 2 a 2
1 x x x
= e f(x)e -xdx - e -x f(x)e xdx
2 a a
Using this, we obtain
f(x)e dx
m x x 1 x x x
Σ f(x)dx 2r = e f(x)e -xdx - e -x x
r=1 a a 2 a a
f(x)dx
x x
m
And since lim =0 holds by assumption ( See proof of Theorem 24.1.1 ), we obtain (1.1').
m a a
Q.E.D.
dx
x x x x x x x x
dx + dx + dx +
2 4 6 8
Example1 +
a a a a a a a a
Substituting f(x)= 1 for (1.1') ,
x
x 1 x x x 1 x
e -e -xa - e -xe xa
x x
Σ dx =2r
e e dx - e -x
-x
e xdx =
r=1 a a 2 a a 2
x-a a-x
1 e +e
= -1+ e x-a
-1-e =
a-x
-1
2 2
i.e.
dx
x x
Σ 2r
= cosh(x -a) - 1
r=1 a a
- 16 -
Theorem 24.3.2
When f(x )is analytic function defined on a closed interval I, the following expressions hold for a ,x I .
f(x)sin xdx
m x x x x
Σ (-1) r-1 2r
f(x)dx = sin x f(x)cosxdx - cosx
r=1 a a a a
Proof
The following expressions were obtained during the proof of Theorem 24.2.2.
(r -1)
f(x)dx
x m x x
r-1
f(x)cosxdx = Σ(-1) cos x + r
a r=1 2 a a
m
+ (-1) f(x)dx cos x +
2
x x x
m m
dx
a a a
(r -1)
2
sin x +
x m x x
f(x)sin xdx = Σ(-1)
r-1
f(x)dx r
a r=1 a a
m
+ (-1) f(x)dx sin x +
2
x x x
m m
dx
a a a
Expanding a part of the 1st term of the right side,
(r -1)
= cosx + sin x - cosx - sin x ++--
m
r-1
Σ
r=1
(-1)
2
cos x +
(r -1)
= sin x - cosx - sin x + cosx ++--
m
Σ (-1) r-1 sin x +
r=1 2
Multiplying both sides by
sin x, cos x respectively,
(r -1)
m
sin xΣ(-1) r-1 cos x +
2 2
= sin x cos x + sin x - sin x cos x - sin x ++--
r=1 2
(r -1)
m
cos xΣ(-1) r-1 sin x +
2 2
= cos x sin x - cos x - cos x sin x + cos x ++--
r=1 2
The 2nd term of the right side is as follows.
2
m m
dx
a a a
2 2
Since cos cx + sin cx = 1 ,
f(x)dx
x x m x x
sin x f(x)cosxdx - cosx f(x)sin xdx = Σ(-1)
r-1
2r
a a r=1 a a
- 17 -
f(x)dx cos (x +m )dx
x x x
+ (-1)2msin x 2m
a a a
f(x)dx
x x
And since lim m
=0 holds by assumption ( See proof of Theorem 24.1.1 ), we obtain (2.1').
m a a
x dx
x x x x x x x x
x dx x dx + x dx +
2 2 2 4 2 6 2 8
Example2
0 0 0 0 0 0 0 0
x sin xdx
m x x x x
Σ (-1) r-1
x 2dx 2r = sin x x 2cosxdx - cosx 2
r=1 0 0 0 0
- sin x x dx
cos (x +m )dx
x x x
2 2m
0 0 0
+ cosx x dx
sin (x +m )dx
x x x
2 2m
0 0 0
Here,
x
x x 2!
2
dx 2m = x 2+2m
0 0 (2+2m )!
Substituting these for the above,
x dx
m x x
r-1
Σ
2 2r
(-1) = x 2-2 + 2cosx
r=1 0 0
x 2!
- sin x x 2+2mcos (x +m )dx
0 (2+2m )!
+ cosx
x 2!
x 2+2msin (x +m )dx (2.2)
0 (2+2m )!
x dx
x x
Σ(-1)
r-1
2 2r
= x 2-2 + 2cosx (2.2')
r=1 0 0
The higher integral of the left side is
x
x x 2!
2
dx n = x
2+n
0 0 ( 2+ n )!
Using this as the left side of (2.2) and (2.2') ,
m
r-1 2!
Σ
r=1
(-1)
(2+2r )!
x 2+2r = x 2-2 + 2cosx
x 2!
- sin x x 2+2mcos (x +m )dx
0 (2+2m )!
- 18 -
x 2!
+ cosx x 2+2msin (x +m )dx (2.3)
0 (2+2m )!
2!
r-1
Σ (-1) x
2+2r
= x 2-2 + 2cosx (2.3')
r=1 ( 2+ 2r )!
When x =5, m =6 , Both sides of (2.3) are calculated as follows.
- 19 -
24.4 Integrals Series Expansion (exp x)
Mikio Sugioka discovered that the series of the higher integral of arbitrary functions is expanded into
the series of the higher integrals of exponential functions or trigonometric functions in March, 2012.
( http://www5b.biglobe.ne.jp/~sugi_m/page216.htm ). The expressions by the higher order integral series are
beautiful. However, more importantly, the expressions by a double series are useful for numerical calculations.
Then I introduce them in the following three sections.
Theorem 24.4.1
When f(x )is analytic function defined on a closed interval I , the following expressions hold for a ,x I .
f(x)dx
s+r
x x (x -a)
Σ r
= Σ f (r)(a)Σ (1.1)
r=0 a a r=0 s=0 (s + r)!
e
x x
= Σ f (a) (r) x-a
dx r (1.1')
r=0 a a
(-1)
x x (x -a)s+ r
(r)
Σ
r=0 a a
r
f(x)dx = Σ f (a)Σ(-1)
r=0
r
s=0 (s + r)!
s
(1.2)
e
x x
= Σ f (r)(a) -(x-a )
dx r (1.2')
r=0 a a
Proof
r
()
Let f (a)= far and expand f(x) into Taylor series around a. Then
(x -a)0 1 (x -a )
1
2 (x -a )
2
3 (x -a )
3
f(x)= fa0 + fa + fa + fa + (0)
0! 1! 2! 3!
Integrating both sides of this with respect to x from a to x one by one,
x 1 2 3 4
0 (x -a ) 1 (x -a ) 2 (x -a ) 3 (x -a )
f(x)dx = fa + fa + fa + fa +
a 1! 2! 3! 4!
2 3 4 5
0 (x -a ) 1 (x -a ) 2 (x -a ) 3 (x -a )
x x
f(x)dx = fa + fa + fa + fa
2
+
a a 2! 3! 4! 5!
(x -a)3 4 5 6
f(x)dx 1 (x -a ) 2 (x -a ) 3 (x -a )
x x x
= fa + fa + fa + fa
3 0
+
a a a 3! 4! 5! 6!
Adding these also including (0) perpendicularly.
f(x)dx
x x x x x x
+
2 3
f(x)+ f(x)dx + f(x)dx +
a a a a a a
(x -a)0 (x -a)1 (x -a)2 (x -a)3
= fa0 + + + +
0! 1! 2! 3!
+f +
1
(x -a)
1 (x -a)2 (x -a)3 (x -a)4
a + + +
1! 2! 3! 4!
+f +
2
(x -a)
2 (x -a)3 (x -a)4 (x -a)5
a + + +
2! 3! 4! 5!
- 20 -
(x -a)s+ r
=Σ far Σ
r=0 s=0 (s + r)!
i.e.
a f(x)dx
x x (r) (x -a)s+ r
Σ
r=0 a
r
= Σ f (a)Σ
r=0 s=0 (s + r)!
(1.1)
Furthermore,
e
(x -a)s+ r x x
x-a
Σ
s=0 (s + r)!
=
a a
dx r
Then
e
x x x x
Σ f(x)dx = Σ f (a)
(r) r x-a
dx r (1.1')
r=0 a a r=0 a a
Next,
(x -a)0 1 (x -a )
1
2 (x -a )
2
3 (x -a )
3
f(x)= fa0 + fa + fa + fa +
0! 1! 2! 3!
x (x -a)1 (x -a)2 (x -a)3 (x -a)4
- f(x)dx = -fa0 - fa1 - fa2 - fa3 -
a 1! 2! 3! 4!
(x -a)2 (x -a)3 (x -a)4 (x -a)5
x x
f(x)dx = fa + fa1 + fa2 + fa3
2 0
+
a a 2! 3! 4! 5!
(x -a)3 4 5 6
f(x)dx 1 (x -a ) 2 (x -a ) 3 (x -a )
x x x
-fa - fa - fa - fa
3 0
- = -
a a a 3! 4! 5! 6!
Adding these perpendicularly.
f(x)dx
x x x x x x
+-
2 3
f(x)- f(x)dx + f(x)dx -
a a a a a a
(x -a)0 (x -a)1 (x -a)2 (x -a)3
= fa0 - + - +-
0! 1! 2! 3!
+f +-
1
(x -a)
1 (x -a)2 (x -a)3 (x -a)4
a - + -
1! 2! 3! 4!
+f +-
2
(x -a)
2 (x -a)3 (x -a)4 (x -a)5
a - + -
2! 3! 4! 5!
(x -a)s+ r
=Σ (-1)rfar Σ (-1) s
r=0 s=0 (s + r)!
i.e.
x x (x -a)s+ r (r)
Σ
r=0
(-1) r
a a
f(x)dx = Σ f (a)Σ(-1)
r=0 s=0
r
(s + r)!
s
(1.2)
Furthermore,
e
(x -a)s+ r x x
Σ (-1) =s
-(x-a )
dx r
s=0 (s + r)! a a
Then
- 21 -
e
x x x x
Σ(-1)r f(x)dx r = Σ f (a)
(r) -(x-a )
dx r (1.2')
r=0 a a r=0 a a
n =1, 2, 3,
(n ) n-1
f(x) = log x , (log x) =(-1) (n -1)! x -n
Substituting these for (1.1), (1.1') ,
e
x x
= Σ(-1)r-1(r -1)! x-1
dx r
r=1 1 1
e
x x x x x x x x
e x-1dx -1! e x-1dx +2! e x-1dx -3! x-1
2 3 4
= 0! dx +-
1 1 1 1 1 1 1 1
When higher integrals are replaced with Riemann-Liouville Integral and one arbitrary point x =0.8 is given,
it is as follows.
x x x x x x
Example2 x - x dx + x dx 2 - x dx 3 +-
1 1 1 1 1 1
f(x) = x
(1+1/2)
1
(n ) -n (2n -3)!!
f (1) = 1 2
= (-1)n-1
1+1/2- n 2n
Substituting these for (1.2), (1.2') ,
(x -1) s+r
x x (2r -3) !!
Σ(-1)r
r=0
1 1
x dx r = e
2r
-(x-1)
s=0
+ Σ (-1)r-1
r=1
Σ(-1)s (s + r) !
(2r -3) !! x x -(x-1) r
= e x-1 + Σ (-1)r-1
-( )
e dx
r=1 2r 1 1
x 1!! x x x-1 2 3!! x x x x-1 3
-( ) (-1)!!
= e x-1 + e x-1
dx - e dx + 3 e dx -+
21 1 22 1 1 2 1 1 1
During the proof of the previous theorem, if the calculation is done without including (0) ,
we obtain the following theorem.
- 22 -
Theorem 24.4.1'
When f(x )is analytic function defined on a closed interval I , the following expressions hold for a ,x I .
a f(x)dx
x s+ r
x () (x -a )
r
Σ
r=1 a
r
= Σ f (a)Σ
r=0 s=1 (s + r)!
f(x)dx
r-1
x x (r)
s-1 (x -a)s+ r
Σ
r=1
(-1)
a a
r
= Σ f (a)Σ(-1)
r=0 s=1 (s + r)!
Combining Theorem 24.4.1' and 24.1 , we obtain the following formula which gives the collateral integral of
the product of an exponential function and arbitrary functions.
Formula 24.4.2
When f(x )is analytic function defined on a closed interval I, the following expressions hold for a ,x I .
x s+ r
(r) (x -a)
s-1
e x f(x)dx = e xΣ f (a)Σ(-1) (2.1)
a r=0 s=1 (s + r)!
e
x
-x (r) (x -a)s+ r
Σ (a)Σ
-x
f(x)dx = e f (2.2)
a r=0 s=1 (s + r)!
e
x
-x
Example cosxdx
a
r (x -a)
s+1
x
-x
e cosxdx = e Σ cos a +
2 Σ
-x
a r=0 s=r (s +1)!
When a =2 and the first 50 terms of ∑ are calculated and both sides are illustrated, it is as follows.
Both sides overlap exactly and blue (left) can not be seen.
- 23 -
24.5 Integrals Series Expansion (sin x)
Theorem 24.5.1
When f(x )is analytic function defined on a closed interval I, the following expressions hold for a ,x I .
2s+1+ r
x x (x -a)
Σ (r)
f(x)dx 2r-1 = Σ f (a)Σ (1.1)
r=1 a a r=0 s=0 (2s +1+ r)!
sinh(x -a)dx
x x
= Σ f (r)(a) r
(1.1')
r=0 a a
f(x)dx
x x (x -a)2s+1+ r
Σ(-1)
r-1 2r-1
= Σ f (r)(a)Σ(-1)s (1.2)
r=1 a a r=0 s=0 (2s +1+ r)!
sin(x -a)dx
x x
= Σ f (a) (r) r
(1.2')
r=0 a a
Proof
r
()
Let f (a)= far and expand f(x) into Taylor series around a. Then
(x -a)0 1 (x -a )
1
2 (x -a )
2
3 (x -a )
3
f(x)= fa0 + fa + fa + fa +
0! 1! 2! 3!
Integrating both sides of this with respect to x from a to x alternately,
x 1 2 3 4
0 (x -a ) 1 (x -a ) 2 (x -a ) 3 (x -a )
f(x)dx fa= + fa + fa + fa +
a 1! 2! 3! 4!
3
(x -a)4 (x -a)5 (x -a)6
0 (x -a )
x x x
f(x)dx 3 = fa + fa1 + fa2 + fa3 +
a a a 3! 4! 5! 6!
(x -a)5 6 7 8
f(x)dx 1 (x -a ) 2 (x -a ) 3 (x -a )
x x
= fa + fa + fa + fa
5 0
+
a a 5! 6! 7! 8!
Adding these perpendicularly.
f(x)dx
x x x x x x
f(x)dx + f(x)dx 3 + 5
+
a a a a a a
(x -a)1 (x -a)3 (x -a)5 (x -a)7
= fa0 + + + +
1! 3! 5! 7!
+f +
2
(x -a)
1 (x -a)4 (x -a)6 (x -a)8
a + + +
2! 4! 6! 8!
+f +
3
(x -a)
2 (x -a)5 (x -a)7 (x -a)9
a + + +
3! 5! 7! 9!
(x -a)2s+1+ r
=Σ far Σ
r=0 s=0 (2s +1+ r)!
i.e.
- 24 -
(x -a)2s+1+ r
a f(x)dx
x x
Σ = Σ f (a)Σ
2r-1 (r)
(1.1)
r=1 a r=0 s=0 (2s +1+ r)!
Furthermore,
sinh(x -a)dx
(x -a)2s+1+ r x x
Σ = r
s=0 (2s +1+ r)! a a
Then,
a sinh(x -a)dx
x x x x
Σ f(x)dx 2r-1 = Σ f r(a) r
()
(1.1')
r=1 a r=0 a a
Next,
f(x)dx
x (x -a)1 1 (x -a )
2
2 (x -a )
3
3 (x -a )
4
= fa0 + fa + fa + fa +
a 1! 2! 3! 4!
(x -a)3 4 5 6
f(x)dx 1 (x -a ) 2 (x -a ) 3 (x -a )
x x x
-fa - fa - fa - fa
3 0
- = -
a a a 3! 4! 5! 6!
(x -a)5 6 7 8
f(x)dx 1 (x -a ) 2 (x -a ) 3 (x -a )
x x
5
= fa0 + fa + fa + fa +
a a 5! 6! 7! 8!
(x -a)7 (x -a)8 (x -a)9 (x -a)10
x x
- f(x )dx = -fa - fa1 - fa2 - fa3
7 0
-
a a 7! 8! 9! 10!
Adding these perpendicularly.
f(x)dx
x x x x x x x x
f(x)dx - f(x)dx 3 + f(x)dx 5 - 7
+-
a a a a a a a a
(x -a)1 (x -a)3 (x -a)5 (x -a)7
= fa0 - + - +-
1! 3! 5! 7!
+f +-
2
(x -a)
1 (x -a)4 (x -a)6 (x -a)8
a - + -
2! 4! 6! 8!
+f +-
3
(x -a)
2 (x -a)5 (x -a)7 (x -a)9
a - + -
3! 5! 7! 9!
+f +-
4
(x -a)
3 (x -a)6 (x -a)8 (x -a)10
a - + -
4! 6! 8! 10!
(x -a)2s+1+ r
=Σ far Σ (-1) s
r=0 s=0 (2s +1+ r)!
i.e.
Then,
- 25 -
sin(x -a)dx
x x x x
Σ(-1)
r-1
f(x)dx 2r-1 = Σ f (a)
(r) r
(1.2')
r=1 a a r=0 a a
tan x dx
x x x x x x x x
Example1 tan x dx tan x dx 3 + tan x dx 5 + 7
0 0 0 0 0 0 0 0
The higher differential quotient of tan x on x =0 is as follows according to Theorem 9.2.6 ( 9.2 )
(2n -1)
22n22n-1B2n
(tan x) |x=0 = T2n-1 = , (tan x)(2n)|x=0 = 0
2n
Where, T 2n-1 is the tangent number and B2n is the Bernoulli number. Thus, from (1.1), (1.1') ,
0 sinh xdx
2s+2r
x x x x x
Σ
r=1 0
tan x dx 2r-1
= ΣT2r-1Σ
r=1 s=0 (2s +2r) !
= ΣT2r-1
r=1 0 0
2r-1
secx dx
x x x x x x x x
Example2 secx dx secx dx 3 + secx dx 5 - 7
-
0 0 0 0 0 0 0 0
The higher differential quotient of secx on x =0 is as follows according to Theorem 9.2.8 ( 9.2 )
secxdx
x x x 2s+1+ 2r
Σ
r=1
(-1) r-1
0 0
2r-1
= ΣE2rΣ(-1)
r=0 s=0 (2s +1+ 2r) !
s
sin xdx
x x
= ΣE2r
2r
r=0 0 0
- 26 -
24.6 Integrals Series Expansion (cos x)
Theorem 24.6.1
When f(x )is analytic function defined on a closed interval I , the following expressions hold for a ,x I .
f(x)dx
2s+ r
x x (x -a)
Σ 2r
= Σf (r)(a)Σ (1.1)
r=0 a a r=0 s=0 (2s + r)!
cosh(x -a)dx
x x
= Σf (a) (r) 2r
(1.1')
r=0 a a
(-1)
x x (x -a)2s+ r
(r)
Σ r
f(x)dx =Σf (a)Σ(-1) s
2r
(1.2)
r=0 a a r=0 s=0 (2s + r)!
cos(x -a)dx
x x
= Σf (r)(a) r
(1.2')
r=0 a a
Proof
r
()
Let f (a)= far and expand f(x) into Taylor series around a. Then
(x -a)0 1 (x -a )
1
2 (x -a )
2
3 (x -a )
3
f(x)= fa0 + fa + fa + fa + (0)
0! 1! 2! 3!
Integrating both sides of this with respect to x from a to x alternately,
2 3 4
(x -a)5
0 (x -a ) 1 (x -a ) 2 (x -a )
x x
) 2
f(x dx = fa + fa + fa + fa3 +
a a 2! 3! 4! 5!
(x -a)4 5 6 7
f(x)dx 1 (x -a ) 2 (x -a ) 3 (x -a )
x x
= fa + fa + fa + fa
4 0
+
a a 4! 5! 6! 7!
(x -a)6 7 8 9
f(x)dx 1 (x -a ) 2 (x -a ) 3 (x -a )
x x
= fa + fa + fa + fa
6 0
+
a a 6! 7! 8! 9!
Adding these also including (0) perpendicularly.
0 2
(x -a) (x -a) (x -a)4 (x -a)6
= fa0 + + + +
0! 2! 4! 6!
+f +
1
(x -a)
1 (x -a)3 (x -a)5 (x -a)7
a + + +
1! 3! 5! 7!
+f +
2
(x -a)
2 (x -a)4 (x -a)6 (x -a)8
a + + +
2! 4! 6! 8!
(x -a)2s+ r
=Σ far Σ
r=0 s=0 (2s + r)!
i.e.
a f(x)dx
x x (r) (x -a)2s+ r
Σ = Σf (a)Σ
2r
(1.1)
r=0 a r=0 s=0 (2s + r)!
- 27 -
Furthermore,
cosh(x -a)dx
(x -a)2s+r x x
Σ = r
s=0 (2s + r)! a a
Then,
f(x)dx 1 (x -a ) 2 (x -a ) 3 (x -a )
x x
-fa - fa - fa - fa
2 0
- = -
a a 2! 3! 4! 5!
(x -a)4 5 6 7
f(x)dx 1 (x -a ) 2 (x -a ) 3 (x -a )
x x
4
= fa0 + fa + fa + fa +
a a 4! 5! 6! 7!
(x -a)6 7 8 9
f(x)dx 1 (x -a ) 2 (x -a ) 3 (x -a )
x x
- -fa - fa - fa - fa
6 0
= -
a a 6! 7! 8! 9!
Adding these perpendicularly and doing calculation similar to Theorem 24.5.1, we obtain (1.2) and (1.2') .
tan
x x x x x x
tan x dx tan x dx x dx +
-1 -1 2 -1 4 -1 6
Example1 tan x+
1 1 1 1 1 1
According to " 岩波数学公式Ⅰ" p39, the following expression holds for a natural number n .
(n )
tan x = (n -1)!cos n(tan -1 x )sin n tan -1 x +
-1
2
From this,
3n
(n ) (n -1)!
tan x
-1
|x=1 = sin
2n/2 4
Substituting this for (1.1), (1.1') ,
3r (x -1)
r
2s 2s+
x x (x -1)
(r -1)!
Σ tan x dx =
4Σ
+Σ
4 Σ
-1 2r
sin
r=0 1 1 s=0 (2s ) ! r=1 2r/2 s=0 (2s + r) !
3r
(n -1)! x x
=
4
cosh(x -1) +Σ
r=1 2r/2
sin
4
1 1
cosh(x -1) dx r
1 x 1 x x
2
= cosh(x -1) + cosh(x -1) dx - cosh(x -1) dx +
4 2 1 2 1 1
When higher integrals are replaced with Riemann-Liouville Integral and one arbitrary point x =1.8 is given,
it is as follows.
- 28 -
sin
x x x x x x
sin x dx sin x dx - x dx +
-1 -1 2 -1 4 -1 6
Example2 sin x-
0 0 0 0 0 0
-1
The higher differential quotient of sin x on x =0 is as follows according to Theorem 9.3.2 ( 9.3 )
( 2n +1 )
sin x |x=0 = 2nC 0(2n -1)!!(2n -1)!!00 = (2n -1)!!2
-1
cos xdx
x x
= Σ(2r -1)!!2
2r+1
r=0 0 0
During the proof of the previous theorem, if the calculation is done without including (0) ,
we obtain the following theorem.
Theorem 24.6.1'
When f(x )is analytic function defined on a closed interval I , the following expressions hold for a ,x I .
2s+ r
x x (x -a)
Σ (r)
f(x)dx 2r = Σf (a)Σ (1.3)
r=1 a a r=0 s=1 (2s + r)!
f(x)dx
r-1
x x (r)
s-1 (x -a)2s+ r
Σ = Σf (a)Σ(-1)
2r
(-1) (1.4)
r=1 a a r=0 s=1 (2s + r)!
Note
This theorem is not so beautiful. However, this is useful for the calculation.
sin
x x x x x x x x
sin x dx sin x dx + sin x dx x dx +
-1 2 -1 4 -1 6 -1 8
Example
0 0 0 0 0 0 0 0
-1
Since the higher differential quotient of sin x on x =0 is same as Example2, from (1.4),
sin
r-1
x x
2 s-1 x 2s+2r+1
Σ = Σ(2r -1)!! Σ
-1 2r
(-1) xdx (-1)
r=1 0 0 r=0 s=1 (2s +2r +1)!
If the left side is replaced with Riemann-Liouville Integral and one arbitrary point x =0.6 is given, then
- 29 -
24.7 Series of Higher Integrals with coefficients
Theorem 24.1.1 is extensible to the higher integral series with coefficients n -1+ rC n-1 . Although these
coefficients are numbers on the hypotenuse of Pascal's triangle, coefficients such as n 3 is not so
interesting. Then, in this section, we ask for the sum of the higher integral series at n=2 i.e.
f(x)dx
x x x x x x x x
f(x)dx 2 f(x)dx 4 +
2 3 4
1 f(x)dx + 3
a a a a a a a a
Theorem 24.7.1
When f(x )is analytic function defined on a closed interval I , the following expressions hold for a ,x I .
+ m e f(x)dx e dx
x x x x
x m -x 2
(1.1)
a a a a
Proof
f (a) = 0 (r =1, 2, ,m+ n -1), Theorem 16.1.2 ( 16.1 ) was as follows.
r
<>
When
x x m -1 -n
f <0>g (0)dx n = Σ n r (r)
f < + >g
a a r=0 r
(-1)m n -1 n -1C k x x <m+ k > (m+ k ) n
f
(n , m) Σ
+ g dx
k=0 m + k a a
(r)
Here, let g = e -x , n =2 . Then, since g =(-1)re -x ,
x x m -1 -2
r
f <0>e -xdx 2 = Σ f <2+ >(-1)re -x
a a r=0 r
(-1)m 1C 0 x x <m+0>
+ f (-1)m+0e -xdx 2
(2, m) m +0 a a
(-1)m 1C 1 x x <m+1>
+ f (-1)m+1e -xdx 2
(2, m) m +1 a a
i.e.
f
x x m -1 x x x x
<0> -x <2+ r > -x <m > -x <m +1> -x
e dx = Σ (1+ r) f
2 2 2
f e + (m +1) f e dx - m e dx
a a r=0 a a a a
r = 1+ r ,
-2 1
(-1) r
= m(m +1)
(2, m)
Here, subtracting 1 from the index <r> of the integration operator of the function f,
f
x x m -1
<1+ r> -x
<-1> -x 2
e dx = Σ (1+ r)f e
a a r=0
f f
x x x x
<m-1> -x 2 <m> -x
+ (m +1) e dx - m e dx 2
a a a a
- 30 -
Left side becomes as follows.
f
x x
x x
<0> -x x
<-1> -x
f2
e dx = f e a + <0> -x
e dx dx
a a a a
= f dx + f
x x x
<0> -x <0> -x
e e dx 2
a a a
Substituting this for the above,
x x x m -1
r
f <0> -x
e dx + f <0>e -xdx 2 = Σ (1+ r)f <1+ >e -x
a a a r=0
f
x x x x
m-1> -x <m> -x
+ (m +1) f< e dx 2 - m e dx 2
a a a a
From this,
f(x)e dx
m x x x
r <>
e -xΣr f (x)= f(x)e -xdx + -x 2
r=1 a a a
- (m +1) f f
x x x x
<m-1> -x 2 <m> -x
e dx +m e dx 2
a a a a
Here, let
f(x)dx
x x x x x x
f
<r>
= f(x)dx r , f <
m-1>
= f(x)dx
m-1
, f
<m>
= m
a a a a a a
f(x)e dx
m x x x x x
Σr r x
f(x)e -xdx + e x -x 2
f(x)dx = e
r=1 a a a a a
+ m e f(x)dx e
x x x x
x m -x
dx 2 (1.1)
a a a a
f(x)dx
x x
And since lim m
=0 holds by assumption ( See proof of Theorem 24.1.1 ), we obtain (1.1').
m a a
dx
x x x x x x
Example 1 1 dx + 2 dx 2 + 3 3
+
a a a a a a
dx
m-1
x x (x -a) x x (x -a)m
f(x)=1 ,
m-1
= , dx m =
a a (m -1)! a a m!
Substituting these for (1.1) ,
r dx e (m -1)! e
x x (x -a )m-1
e
m x x x x x
Σ r x x x
-x -x 2 -x
=e dx + e dx -(m +1)e dx 2
r=1 a a a a a a a
+ m e
x x m
(x -a) x -x 2
e dx
m! a a
The 1st term and the 2nd term of the right side are as follows.
- 31 -
e - e dx
x x x x
ex e -xdx + e x e -xdx 2 = e xe a- e x + e x a x
a a a a
x-a
=e (x - a)
After the long calculation, the 3rd term and the 4th term on the right side become as follows.
x x (x -a )m-1 x x (x -a )m
x -x 2 x -x 2
-(m +1)e e dx + me e dx
a a (m -1)! a a m!
x-a
= -e (x -a)
m +1
e (x -a)(m , x -a)- (1+ m , x -a)
x-a
+
(m -1)!
1
e (x -a)(1+ m , x -a)- (2+ m , x -a)
x-a
-
(m -1)!
Substituting these for the above,
r dx
m x x
x-a
Σ r
=e (x -a) - e x-a(x -a)
r=1 a a
m +1
e (x -a)(m , x -a)- (1+ m , x -a)
x-a
+
(m -1)!
1
e (x -a)(1+ m , x -a)- (2+ m , x -a)
x-a
-
(m -1)!
(1.2)
r dx r = e
x x
x-a
Σ
r=1 a a
(x -a) (1.2')
r dx
m x x m r(x -a)r
Σ
r=1 a a
r
=Σ
r=1 r!
Using this as the left side of (1.2) and (1.2') ,
m r(x -a)r x-a x-a
Σ
r=1 r!
=e (x -a) - e (x -a)
m +1
e (x -a)(m , x -a)- (1+ m , x -a)
x-a
+
(m -1)!
1
e (x -a)(1+ m , x -a)- (2+ m , x -a)
x-a
-
(m -1)!
(1.3)
r(x -a)r
Σ = e x-a(x -a) (1.3')
r=1 r!
When x =4, a =3, m =7 , Both sides of (1.3) are calculated as follows.
- 32 -
Further, when a =3 , the first 50 terms of ∑ are calculated and both sides of (1.3') are illustrated, it is as
follows. Both sides overlap exactly and blue (left) can not be seen.
cos x dx
x x x x x x
Example 2 1 cos x dx + 2 2
cos x dx + 3 3
+
a a a a a a
Substituting f(x)= cosx for (1.1') ,
cosx e
x x x x x
Σr cosx dx = e x
r
cosx e dx + e x
-x -x
dx 2
r=1 a a a a a
Integrating the right side with respect to x from a to x by force,
r cosx dx
x x-a x-a
x e (x -a)(cos a - sin a) e cos a - cos x
Σ
r=1 a a
r
=
2
+
2
The higher integral of the left side is as follows according to Theorem 4.1.3 ( 4.1 )
n (n -s)
cosx dx
n -1 (x -a )s
- Σ
x x
n
= cos x - cos a -
a a 2 s=0 s! 2
Then, the left side is
r (r -s)
(x -a)s
r-1
Σr cos x -
r=1 2
-Σ
s=0 s!
cos a -
2
When a =3/2 , the first 18 terms of ∑ are calculated and both sides are illustrated, it is as follows.
Both sides overlap exactly and blue (left) can not be seen.
- 33 -
Theorem 24.7.2
When f(x )is analytic function defined on a closed interval I , the following expressions hold for a ,x I .
Proof
In a way similar to Theorem 24.7.1, we obtain the desired expressioons
log x dx
x x x x x x x x
log x dx 4
2 3 4
Example3 1 log x dx -2 log x dx +3 +-
0 0 0 0 0 0 0 0
- 34 -
log xe dx = e log|x| - Ei(x) = e log|x|- Ei(x)+
x x
x x x
0
0
x et
Where, Ei(x) = dt , = 0.5772
(Euler-Mascheroni Constant ).
- t
Therefore, the 1st term and the 2nd term on the right side are
log xe dx - e log xe dx
x x x
-x
e x -x x 2 -x
= e xEi(x) - + e
-x
-1
0 a a
Next,
x xm
x m 1
log x dx = m
log|x|-Σ
0 0 m! s=1 s
Using thes for the 3rd term and the 4th term on the right side, the right side becomes
e xEi(x)- + e
-x -x
-1
x x m-1
x m -1 1
m
- (-1) (m +1)e x
log|x|-Σ e xdx 2
a a (m -1)! s=1 s
x x x m
m 1
+ (-1)m+1 m e x log|x|-Σ e xdx 2
a a m! s=1 s
i.e.
e -x xEi(x)- + e -x - 1
x x
m +1 m-1
m -1 1
- (-1)m ex x log|x|-Σ e xdx 2
(m -1)! a a s=1 s
x
1 x m 1
+ (-1)m+1 ex x m log|x|-Σ e xdx 2
(m -1)! a a s=1 s
Thus, we obtain
log x dx
m x x
Σ(-1) r
r-1 r
= e -x xEi(x) - + e -x - 1
r=1 a a
x x
m +1 m-1
m -1 1
- (-1)m ex x log|x|-Σ e xdx 2
(m -1)! a a s=1 s
x x
1 m 1
+ (-1)m+1 ex x m log|x|-Σ e xdx 2 (2.2)
(m -1)! a a s=1 s
log x dx
x x
Σ (-1) r-1
r r
= e -x xEi(x) - + e -x - 1 (2.2')
r=1 a a
The left side is
xn
x x
n 1
n
log x dx = log|x|-Σ
0 0 n! s=1 s
Using this,
- 35 -
m rx r
=e
r 1
xEi(x)- + e -x - 1
r-1
Σ
-x
(-1) log x -Σ
r=1 r! s=1 s
x x
m +1 m-1
m -1 1
- (-1)m ex x log|x|-Σ e xdx 2
(m -1)! a a s=1 s
x x
1 m 1
+ (-1)m+1 ex x m log|x|-Σ e xdx 2 (2.3)
(m -1)! a a s=1 s
rx r
r 1
Σ(-1)
r-1
log|x|-Σ = e -x xEi(x) - + e -x - 1 (2.3')
r=1 r! s=1 s
When x =-0.2, m =4 , Both sides of (2.3) are calculated as follows.
Further, when the first 30 terms of ∑ are calculated and both sides of (2.3') are illustrated, it is as follows.
Both sides overlap exactly and blue (left) can not be seen.
c.f.
Mr. Sugioka showed the following theorem ( Sugioka's Theorem 3-1) just like Theorem24.7.1 .
f(x)e
x x x x x x x x x
f(x)dx 3 f(x)dx + = e x
2 3 4 -x 2
1 f(x)dx + 2 dx
a a a a a a a a a
Adding this to Theorem 24.1.1 (1.1') , we obtain Theorem 24.7.1 (1.1') .
- 36 -
2012.09.01
2018.06.05 Updated
Kano Kono
Alien's Mathematics
- 37 -
25 Series of Higher Integral with Geometric Coefficients
This chapter is a generalization of " 24 Sugioka's Theorem on the Series of Higher Integral ". The origin of
theses paper is " e ^x に関する公式の発見 " ( Mikio Sugioka 2003 ) .
f(x)dx
x x x x x x x x
c
1
f(x)dx c
2
f(x)dx + c
2 3
f(x)dx c
3 4
4
+
a a a a a a a a
Theorem 25.1.1
When c is a positive number and a is a real number on the domain of analytic function f(x) ,
the following expressions hold
c f(x)dx f(x)dx e
r=1
m x x x x x x
Σ r r
= ce cx m+1
f(x)e -cxdx - c e cx m -cx
dx
a a a a a a
(1.1)
f(x)dx = 0
x x
Especially, when lim c m
m
,
m a
a
Proof
Let
f(x)dx
x x
<r>
f (x) = r
r =1, 2, ,m (1.r)
a a
f
x x x
f(x)e -cxdx = f <1>(x)e -cxa + c <1>
(x)e -cxdx
a a
f
x
= f <1>(x)e -cx + c <1>
(x)e -cxdx
a
f
x x
= f <1>(x)e -cx + c f <2>(x)e -cxa + c 2 <2>
(x)e -cxdx
a
f
x
= f <1>(x)e -cx + c 1f <2>(x)e -cx + c 2 <2>
(x)e -cxdx
a
f
m x
= e -cx Σc r-1f < r>
(x) + c m
<m> -cx
e dx
r=1 a
Substituting (1.r) for this,
f(x)dx e
x m x x x x x
-cx
f(x)e dx = e -cxΣc
r-1
f(x)dx r + c m m -cx
dx
a r=1 a a a a a
Multiplying by e cx the both sides,
f(x)dx e
x m x x x x x
e cx
f(x)e -cxdx = Σc
r-1
f(x)dx r + c me cx m -cx
dx
a r=1 a a a a a
Multiplying by c the both sides and transposing them, we obtain (1.1) .
-1-
dx
x x x x x x
Example 1 c 1
dx + c 2 dx + c 3
2 3
+
a a a a a a
dx
x x m
(x -a)
f(x)=1 , m
=
a a m!
Substituting these for (1.1) ,
c dx
r=1 e
m x x x x (x -a)m -cx
Σ r r
= ce cx -cx
dx - c
m+1 cx
e
m!
e dx
a a a a
The 1st term of the right sideis
x
e
x e -cx
ce cx -cx
dx = ce cx - = e c(x-a)-1
a c a
The 2ndt term of the right sideis is expressed as follows using the incomplete gamma function (x, y ) .
x (x -a)m -cx e -ac
e dx = m+1 (m +1)-m +1 , c(x -a )
a m! c m!
Thus,
c dx
x x c(x-a )
m e
(m +1)-m +1 , c(x -a)
c(x-a )
Σ
r=1 a
r
a
r
=e -1 -
m!
(1.2)
c(x-a )
e
And, since lim (m +1)-m +1 ,c(x -a) = 0,
m m!
c dx r = e
x x
c(x-a )
Σ
r=1 a
r
a
-1 (1.2')
dx
x x
r (x - a)r
=
a a r!
Then, the left side is
c
m x x (x - a)r
m
Σ
r=1 a
r
a
dx = Σc
r
r=1 r!
r
(x - a)r
Σ c r
= e c(x-a)-1 (1.3')
r=1 r!
When x =2, a =0, c =3.3, m =4 , both sides of (1.3) are calculated as follows.
-2-
Also, it is clear from the following that (1.3') is correct.
r c(x - a )
r 0
r (x - a ) c(x - a )
Σ c =Σ - = e c(x-a)-1
r=1 r! r=0 r! 0!
e dx
x x x x x x
Example 2 c 1 x
e dx + c 2 x
e dx + c 3 2 x 3
+
a a a a a a
e dx
x x m -1 (x -a)r
f(x)= e x , x m
= e x- e a Σ
a a r=0 r!
Substituting these for (1.1) and using the incomplete gamma function (x, y ) ,
c ee e dx e
m x x x x x x
Σ r
e dx = ce cx
x r x -cx m+1
dx - c e cx x m -cx
dx
r=1 a a a a a a
= ce e
x
(1-c)x m+1
x m -1 (x -a)r -cx
cx
dx - c e cx e - ea Σ
x
e dx
a a r=0 r!
= ce e e
x
(1-c)x m+1 cx
x
(1-c)x m+1 cx+a
m -1 x (x -a)r -cx
cx
a
dx - c e
a
dx + c e Σ
r=0 a r!
e dx
-3-
e dx
x x
x r
= ex r =1, 2, 3, : Lineal Higher Integral
- -
c
ce cx e(1-c) - e (1-c)
x x a
x
Σ
r=1 a a
r x
e dx = r
1- c
(1.4')
e dx
x x r-1 (x -a )s
a a
x r x a
= e -e Σ
s=0 s!
Then, the left side is
c dx e - e Σ
m x x m r-1 (x -a)s
Σ
r=1 a a
r r
= Σc
r=1
r x a
s=0 s!
Using this as the left side of (1.4) and (1.4') ,
m r-1 (x -a )s 1- c m
ce cx e (1-c) - e(1-c)
x a
Σ
r=1
c r x a
e -e Σ
s=0 s!
=
1- c
1+r, c(x -a)
m -1 1
- c me cx+(1-c)a Σ r
-1 (1.5)
r=0 c r!
ce cx e (1-c) - e (1-c)
x a
e - e Σ
r-1 (x -a)s
Σ
r=1
c r x a
s=0 s!
=
1- c
(1.5')
When x =4, a =1, c =2, m =3 , both sides of (1.5) are calculated as follows. Both sides coincide exactly.
So, (1.4) is confirmed to be correct.
-4-
log x dx
x x x x x x
Example 3 c1 log x dx + c 2 log x dx 2 + c 3 3
+
0 0 0 0 0 0
Substituting f(x)= log x , a =0 for (1.1') ,
log x e
x x x
Σc r
r=1
log x dx r = ce cx -cx
dx
a a a
The integral of the right side is as follows.
x
log x e
x
-cx -e -cx log|x|+ Ei(-cx)
dx =
0 c 0
-e -cx
log|x|+ Ei(-cx) log c +
= -
c c
x et
Where, Ei(x) = dt , = 0.5772 (Euler-Mascheroni Constant ).
- t
Multiplying by e cx the both sides,
log x e
x
ce cx -cx
dx = -log|x|+ e cxEi(-cx) - - log c
0
Substituting this for the right side of the above,
c log x dx
x x
Σ r r
= -log|x|+ e cxEi(-cx) - - log c (1.6)
r=1 a a
The higher order integral of the left side becomes lineal higher integral as follows.
log x dx
x xn
log|x|-Σ s
x n 1
n
=
0 0 n! s=1
Then, (1.6) is
r r
= -log|x|+ e
cx r 1
Σ log|x|-Σ Ei(-cx) -
cx
- log c
r=1 r! s=1 s
When c =1.7 , the first 10 terms of ∑ are calculated and both sides are illustrated, it is as follows.
Both sides overlap exactly and blue (left) can not be seen.
-5-
Theorem 25.1.2
When c is a positive number and a is a real number on the domain of analytic function f(x) ,
the following expressions hold
f(x)e
m x x x
Σ (-1) r-1 r
c f(x)dx = ce -cx r cx
dx
r=1 a a a
f(x)dx e
x x x
- (-1)mc m+1e -cx m cx
dx (2.1)
a a a
f(x)dx = 0
x x
Especially, when lim c m
m
m
,
a a
Proof
In a similar way to the proof of Theorem 25.1.1 , we obtain the desired expressions.
Example 4
sin xe
x x x
Σ(-1)r-1c r
r=1
r
sin xdx = ce -cx cx
dx
a a a
The right side is as follows.
sin x e
cx
x
cx csin x -cosx e - csin a -cosa e ca
dx =
a 1+ c 2
-cx
Multiplying by ce the both sides,
-6-
sin x e
x c
2 csin x cosx
ce -cx cx
dx = - - csin a -cosa e c(a -x)
a 1+ c
Thus, we obtain
x x c
Σ(-1) c r 2 csin x cosx
- csin a -cosa e c(a -x)
r-1
sin xdx r = -
r=1 a a 1+ c
(2.2')
The higher integral of the left side is as follows according to Theorem 4.1.3 ( 4.1 )
n (n -s)
sin x dx
n -1 (x -a )s
-Σ
x x
n
= sin x - sin a -
a a 2 s=0 s! 2
Then, (2.2') is
r (r -s)
sin x - 2 -Σ
(x -a)s
r-1
Σ
r=1
(-1) r-1 r
c
s=0 s!
sin a -
2
c
2 csin x cosx
= - - csin a -cosa e c(a -x)
1+ c
When a =3/2 , c=1.3 , the first 30 terms of ∑ are calculated and both sides are illustrated, it is as follows.
Both sides overlap exactly and blue (left) can not be seen.
-7-
25.2 Series of the odd-th order Integrals
In this section, we ask for the sum of series of higher integral with geometric coefficients, such as
f(x)dx
x x x x x x x x
c 1 f(x)dx c 3 +
5 7
f(x)dx c
3 5 7
f(x)dx + c
a a a a a a a a
Theorem 25.2.1
When c is a positive number and a is a real number on the domain of analytic function f(x) ,
the following expressions hold
f(x)dx 2r-1 =
f(x)e dx
m x x c x x
Σ
r=1
c
2r-1
a a 2
e cx
a
f(x)e
-cx
dx + e
-cx
a
cx
2m-1 2m e +(-1) 2m-1 e
- (-1) c dx
a a 2 a
2m e -(-1) 2m-1 e
+ (-1)2m-1c dx
a a 2 a
(1.1)
f(x)dx
x x
Especially, when lim c 2m
m
2m-1
= 0,
a a
f(x)dx 2r-1 =
f(x)e dx
x x c x x
Σ
r=1
c
2r-1
a a 2
e cx
a
f(x)e
-cx
dx + e
-cx
a
cx
(1.1')
Proof
Formula of repeated integration by parts was as follows. (" 01 Generalized Taylor's Theorem " ( A la Carte ) )
x m x x
f(x)g(x)dx = (-1) f <r>(x)g ( )(x)a + (-1)m f <m>(x)g (x)dx
r-1 r-1 (m)
a r=1
Σ a
When g(x)=cosh cx , sinh cx ,
(r-1) r-1 e cx +(-1)-r+1e -cx (m)
cx -m -cx
m e +(-1) e
(cosh cx) =c , (cosh cx) = c
2 2
(r-1) r-1 e cx -(-1)-r+1e -cx (m)
cx -m -cx
m e -(-1) e
(sinh cx) =c , (sinh cx) = c
2 2
Substituting these for the above,
x
f
m
x
<m> e cx +(-1)-me -cx
m
+ (-1) (x)c dx ,
a 2
x
f
m
x
<m> e cx -(-1)-me -cx
m
+ (-1) (x)c dx ,
a 2
Here, let
f(x)dx
x x
<r>
f (x) = r
r =1, 2, ,m
a a
-8-
f (a) = 0 (r =1, 2, ,m) ,
<>r
Since
f(x)dx
x m e cx +(-1)-r+1e -cx x x
r-1 r-1
f(x)cosh cxdx = Σ(-1) c r
a r=1 2 a a
f(x)dx
m
x x x
m me cx +(-1)-me -cx
+ (-1) c dx
a a a 2
f(x)dx
x m e cx -(-1)-r+1e -cx x x
f(x)sinh xdx = Σ(-1)
r-1
c
r-1
r
a r=1 2 a a
f(x)dx
m
x x x
m me cx -(-1)-me -cx
+ (-1) c dx
a a a 2
Expanding a part of the 1st term of the right side,
e cx +(-1)-r+1e
m -cx
Σ
r=1
(-1) r-1 r-1
c
2
0 1
= c cosh cx - c sinh cx + c cosh cx - c sinh cx +-
2 3
e cx -(-1)-r+1e
m -cx
Σ(-1) r-1 c r-1
r=1 2
0 1
= c sinh cx - c cosh cx + c sinh cx - c cosh cx +-
2 3
f(x)dx
e cx +(-1)-me
x x x -cx
m m m
(-1) cosh cx c dx
a a a 2
m m m
dx
a a a 2
2 2
Since cosh cx - sinh cx = 1 ,
f(x)dx
x x m x x
cosh cx f(x)cosh cx dx - sinh cx f(x)sinh cx dx = Σc
2r-2
2r-1
a a r=1 a a
2m-1 e +(-1) 2m-1 2m-1 e
+ (-1) c dx
a 2
a a
2m-1 e -(-1) 2m-1 2m-1 e
- (-1) c dx
a a 2 a
Here, m in the 2nd term and the 3rd term on the right side has to be an odd number corresponding to the 1st
term. Furthermore,
f(x)sinh cx dx
x x
cosh cx f(x)cosh cx dx - sinh cx
a a
-9-
e cx+e e cx+e e cx-e e cx-e
-cx x -cx -cx x -cx
= f(x) dx - f(x) dx
2 a 2 2 a 2
f(x)e dx
1 cx x x
cx
= e f(x)e -cxdx + e -cx
2 a a
Using this, we obtain
f(x)dx
e cx +(-1)-2m+1e
x x x -cx
2m-1 2m-1 2m-1
- (-1) c cosh cx dx
a a a 2
dx
x x x x x x x x
c 1 dx c 3
5 7
3 5 7
Example 1 dx + c dx + c
a a a a a a a a
x x
m (x -a)m
f(x)=1 , dx =
a a m!
Substituting these for (1.1) ,
e dx
m x x c x x
Σc
r=1
2r-1
a
a
dx
2r-1
=
2
e cx
a
e
-cx
dx + e
-cx
a
cx
cosh cx
2m-1 cx -cx
2m-1 2m (x -a) e x +(-1)-2m+1e
- (-1) c dx
(2m -1) ! a 2
sinh cx
2m-1 cx -cx
2m-1 2m (x -a ) e x -(-1)-2m+1e
+ (-1) c dx
(2m -1) ! a 2
Here,
x x
e e dx
c x x ce cx e -cx ce cx e cx
e cx -cx
dx + e -cx cx
= - +
2 a a 2 c a 2 c a
c(x-a ) -c(x-a )
e -e
= = sinhc(x -a)
2
Then,
dx
m x x
Σc
2r-1
2r-1
= sinhc(x -a)
r=1 a a
2m-1
e cx +(-1)-2m+1e
-cx
2m-1 2m
x (x -a)
- (-1) c cosh cx dx
a (2m -1) ! 2
sinh cx
2m-1
e cx -(-1)-2m+1e
-cx
2m-1 2m (x -a ) x
+ (-1) c dx
(2m -1) ! a 2
(1.2)
aa dx
x x
Σ 2r-1 2r-1
c = sinhc(x -a) (1.2')
r=1
- 10 -
The left side is
dx
m x x m (x - a)2r-1
Σ = Σc
2r-1 2r-1 2r-1
c
r=1 a a r=1 (2r -1)!
Using this as the left side of (1.2) and (1.2') ,
2r-1
m (x - a)
Σ
r=1
c
2r-1
(2r -1)!
= sinhc(x -a)
2m-1
e cx +(-1)-2m+1e
-cx
2m-1 2m
x (x -a)
- (-1) c cosh cx dx
a (2m -1) ! 2
sinh cx
2m-1
e cx -(-1)-2m+1e
-cx
2m-1 2m (x -a ) x
+ (-1) c dx
(2m -1) ! a 2
(1.3)
(x - a)2r-1
Σc
r=1
2r-1
(2r -1)!
= sinhc(x -a) (1.3')
When x =3, a =1, c =2.1, m =5 , Both sides of (1.3) are calculated as follows.
Also, (1.3') is correct. Because the left side is Taylor expansion of the right side.
Theorem 25.2.2
When c is a positive number and a is a real number on the domain of analytic function f(x) ,
the following expressions hold
f(x)sin cx dx
m x x x x
Σ(-1) r-1c
r=1
2r-1
a a
f(x)dx
2r-1
= c cos cx
a
f(x)cos cx dx + c sin cx
a
(2m -1)
cos cx f(x)dx
x x x
cos cx + dx
-(-1)2m-1c 2m
2m-1
a a 2 a
(2m -1)
sin cx f(x)dx
x x x
sin cx + dx
-(-1)2m-1c 2m
2m-1
a a 2 a
(2.1)
f(x)dx
x x
Especially, when lim c 2m
m
2m-1
= 0,
a a
f(x)sin cx dx
x x x x
Σ(-1) r-1c
r=1
2r-1
a a
f(x)dx
2r-1
= c cos cx
a
f(x)cos cx dx + c sin cx
a
(2.1')
- 11 -
Proof
Formula of repeated integration by parts was as follows (" 1 Generalized Taylor's Theorem ") .
x m x x
f(x)g(x)dx = (-1) f <r>(x)g ( )(x)a + (-1)m f <m>(x)g (x)dx
r-1 r-1 (m)
a r=1
Σ a
When g(x)= sin cx , coscx ,
(r -1) m
(coscx)(r-1) = c r-1cos cx +
(m)
, (coscx) = c mcos cx +
2 2
(r -1) m
(sin cx)(r-1) = c r-1sin cx +
(m)
, (sin cx) = c msin cx +
2 2
Substituting these for the above,
(r -1) x
f(x)coscxdx = Σ(-1) f
x m
r-1 <r> r-1
(x)c cos cx +
a r=1 2 a
m
+ (-1) f
x
m <m>
(x)c mcos cx + dx
a 2
(r -1) x
f(x)dx
x x
r =1, 2, ,m
<r> r
f (x) =
a a
(r -1)
f(x)dx
x m x x
r-1 r-1
f(x)coscxdx = Σ(-1) c cos cx + r
a r=1 2 a a
m
+ (-1) f(x)dx c cos cx +
2
x x x
m m m
dx
a a a
(r -1)
2
c sin cx +
x m x x
f(x)sin xdx = Σ(-1)
r-1 r-1
f(x)dx r
a r=1 a a
m
+ (-1) f(x)dx c sin cx +
2
x x x
m m m
dx
a a a
Expanding a part of the 1st term of the right side,
(r -1)
(-1) r-1c r-1cos cx +
m
Σ 0 1 2 3
= c cos cx + c sincx - c cos cx - c sin cx ++--
r=1 2
(r -1)
m
Σr=1
r-1
(-1) r-1c sin cx +
2
0 1 2 3
= c sin cx - c cos cx - c sin cx + c cos cx ++--
- 12 -
(r -1)
m
sin cxΣ(-1) r-1c r-1sin cx +
r=1 2
0 2 1 2 2 3
= c sin cx - c sincx cos cx - c sin cx + c sin cx cos cx ++--
2
m m m
dx
a a a
2 2
Since cos cx + sin cx = 1 ,
f(x)dx
x x m x x
cos cx f(x)cos cx dx + sin cx f(x)sin cx dx = Σ(-1) r-1c
2r-2
2r-1
a a r=1 a a
2m -1
f(x)dx cos cx + ( 2 ) dx
x x x
+ (-1)2m-1c 2m-1cos cx 2m-1
a a a
(2m -1)
sin cx f(x)dx
x x x
sin cx + dx
+ (-1)2m-1c 2m-1
2m-1
a a 2 a
Here, m in the 2nd term and the 3rd term on the right side has to be an odd number corresponding to the 1st
term. Then, multiplying by c the both sides and transposing them, we obtain (2.1) .
e dx
x x x x x x x x
c 1 e xdx c 3 +-
5 7
e xdx + c e xdx c x
3 5 7
Example 2
a a a a a a a a
e sin cx dx
x x x x
Σ(-1) r-1c
r=1
2r-1
a
a
e xdx
2r-1
= c cos cx
a
e xcos cx dx + c sin cx
a
x
x 1
2 e (csin cx coscx) e (csin ca cosca )
x x
e coscxdx = + - a +
a 1+ c
x 1
e xsin cxdx = 2 e (sin cx ccoscx) e (sin ca ccosca )
x
- - a -
a 1+ c
From these,
x x ex ea
cos x e xcos x dx + sin x e xsin x dx = 2
- 2
[cos{c(x -a )}- c sin{c(x -a )}]
a a 1+ c 1+ c
Therefore,
e dx
x x ce x ce a
Σ
r=1
(-1) r-1 2r-1
c
a a
x 2r-1
=
1+ c
2
-
1+ c
2
[cos{c(x -a )}- csin{c(x -a )}]
(2.2')
The higher integral of the left side is as follows according to Theorem 4.1.3 ( 4.1 )
x x (x -a)s n -1
e dx = e - Σe
x n x a
a a s=0 s!
Then, (2.2') becomes
- 13 -
s
c ex ce a
= 1+ c
(x -a )
2r-2
Σ
r=1
(-1) r-1 2r-1
c x a
e -e Σ
s=0 s! 2
-
1+ c
[cos{c(x -a )}- csin{c(x -a )}]
2
(2.3')
When a =2 , c=0.8 , the first 10 terms of ∑ are calculated and both sides are illustrated, it is as follows.
Both sides overlap exactly and blue (left) can not be seen.
- 14 -
25.3 Series of the even-th order Integrals
In this section, we ask for the sum of series of higher integral with geometric coefficients, such as
f(x)dx
x x x x x x x x
c
2
f(x)dx c
2 4
f(x)dx + c
4 6
f(x)dx c
6 8
8
+
a a a a a a a a
Theorem 25.3.1
When c is a positive number and a is a real number on the domain of analytic function f(x) ,
the following expressions hold
c f(x)dx =
f(x)e dx
m x x c x x
Σ
r=1
2r
a a
2r
2
e cx
a
f(x)e
-cx
dx - e
-cx
a
cx
cosh cx dx
2m
- c 2m+1
a a a
sinh cx dx
2m
+ c 2m+1 (1.1)
a a a
f(x)dx = 0
x x
Especially, when lim c 2m+1 2m
,
m a a
Proof
The following expressions were obtained during the proof of Theorem 25.2.1 .
f(x)dx
x cx -r+1 -cx x x
r-1 r-1 e +(-1) e
m
f(x)cosh cxdx = Σ(-1) c r
a r=1 2 a a
f(x)dx
m
x x x
m e cx +(-1)-me -cx
m
+ (-1) c dx
a a a 2
f(x)dx
x m e cx -(-1)-r+1e -cx x x
f(x)sinh cxdx = Σ(-1)
r-1
c
r-1
r
a r=1 2 a a
f(x)dx
cx
m
x x x
m e -(-1)-me -cx
m
+ (-1) c dx
a a a 2
Expanding a part of the 1st term of the right side,
e cx +(-1)-r+1e
m -cx
Σ(-1) r-1 c r-1
r=1 2
0 1
= c cosh cx - c sinhcx + c cosh cx - c sinh cx +-
2 3
e cx -(-1)-r+1e
m -cx
Σ
r=1
(-1) r-1 r-1
c
2
0 1
= c sinh cx - c cosh cx + c sinh cx - c cosh cx +-
2 3
- 15 -
The 2nd term of the right side is as follows.
f(x)dx
e cx +(-1)-me
x x x -cx
m m m
(-1) sinh cx c dx
a a a 2
m m m
dx
a a a 2
2 2
Since cosh cx - sinh cx = 1
f(x)dx
x x m x x
sinh cx f(x)coshcx dx - cosh cx f(x)sinh cx dx = Σc
2r-1
2r
a a r=1 a a
f(x)dx c
e cx +(-1)-2me
x x x -cx
2m 2m 2m
+(-1) sinh cx dx
a a a 2
c
2m 2m
-(-1)2m dx
a a a 2
Here, m in the 2nd term and the 3rd term on the right side has to be an even number corresponding to the 1st
term. Furthermore,
f(x)sinh cx dx
x x
sinh cx f(x)cosh cx dx - cosh cx
a a
cx cx cx cx
-cx -cx -cx -cx
e -e e +e e +e e
x x -e
= f(x ) dx - f(x ) dx
2 2 2 a a 2
2
f(x)e dx - e f(x)e dx
x x
1 cx -cx -cx cx
= e
a a
Using this, we obtain
f(x)e dx
m x x 1 x x
Σ c
2r-1
f(x)dx
2r
= e cx f(x)e
-cx
dx - e
-cx cx
r=1 a a 2 a a
dx
x x x x x x x x
c2
2 4 4 6 6 8 8
Example 1 dx + c dx + c dx + c +
a a a a a a a a
x x (x -a)m
m
f(x)=1 , dx =
a a m!
Substituting these for (1.1) ,
e dx
m x x c x x
Σ
2r
c 2r
dx = e cx e -cxdx - e -cx cx
r=1 a a 2 a a
sinh cx
2m+1
x (x -a)2m
-c cosh cx dx
a (2m)!
2m+1
x (x -a)2m
+c cosh cx sinh cx dx
a (2m)!
- 16 -
Here,
x x
e e dx
c x x ce cx e -cx ce cx e cx
e cx -cx
dx - e -cx cx
= - -
2 a a 2 c a 2 c a
c(x-a ) -c(x-a )
e -1 -e +1
= -
2 2
= coshc(x -a) - 1
Then,
dx
m x x x (x -a)2m
Σ
2r 2r 2m+1
c = coshc(x -a) - 1 - c sinh cx cosh cx dx
r=1 a a a (2m)!
2m+1
x (x -a)2m
+c cosh cx sinh cx dx
a (2m)!
(1.2)
dx
x x
Σc
2r 2r
= coshc(x -a) - 1 (1.2')
r=1 a a
The left side is
dx
m x x m (x - a)2r
Σ c
2r
2r
= Σc 2r
r=1 a a r=1 (2r)!
Using this as the left side of (1.2) and (1.2') ,
2r
m (x - a) x (x -a)2m
Σ
r=1
c
2r
(2r)!
= coshc(x -a) - 1 - c 2m+1sinh cx
a (2m)!
cosh cx dx
2m+1
x (x -a)2m
+c cosh cx sinh cx dx
a (2m)!
(1.3)
(x - a)2r
Σ
r=1
c
2r
(2r)!
= coshc(x -a) - 1 (1.3')
When x =3, a =1, c =2.3, m =4 , Both sides of (1.3) are calculated as follows.
Also, it is clear that (1.3') is correct if coshc(x -a) is expanded to Taylor series.
- 17 -
Theorem 25.3.2
When c is a positive number and a is a real number on the domain of analytic function f(x) ,
the following expressions hold
f(x)sin cx dx
m x x x x
Σ(-1) r-1c
r=1
2r
a a
f(x)dx
2r
= c sin cx
a
f(x)cos cx dx - c cos cx
a
f(x)dx = 0
x x
Especially, when lim c 2m+1 2m
,
m a a
Proof
The following expressions were obtained during the proof of Theorem 25.2.2.
(r -1)
f(x)dx
x m x x
r-1 r-1
f(x)coscxdx = Σ(-1) c cos cx + r
a r=1 2 a a
m
+ (-1) f(x)dx c cos cx +
2
x x x
m m m
dx
a a a
(r -1)
2
x m x x
f(x)sin cxdx = Σ (-1) c sin cx + f(x)dx r-1 r-1 r
a r=1 a a
m
+ (-1) f(x)dx c sin cx +
2
x x x
m m m
dx
a a a
Expanding a part of the 1st term of the right side,
(r -1)
(-1) r-1c r-1cos cx +
m
Σ 0 1 2 3
= c cos cx + c sincx - c cos cx - c sin cx ++--
r=1 2
(r -1)
= c sin cx - c cos cx - c sin cx + c cos cx ++--
m
Σ
r=1
(-1) r-1c r-1sin cx +
2
0 1 2 3
2
m m m
dx
a a a
- 18 -
2 2
Since cos cx + sin cx = 1 ,
f(x)dx
x x m x x
sin cx f(x)cos cx dx - cos cx f(x)sin cx dx = Σ(-1) r-1c
2r-1
2r
a a r=1 a a
x dx
x x x x x x x x
Example 2 c
2
x dx c
2 2 4
2
x dx + c
4 6
x dx c
2 6 8
2 8
+
0 0 0 0 0 0 0 0
x dx
x x x x x x x x
c
2
x dx c
2 2 4
2
x dx + c
4 6
x dx c
2 6 8
2 8
+
0 0 0 0 0 0 0 0
The right side is,
x cos cx dx =
2cx cos cx + c x -2sin cx
x 2 2
2
3
0 c
x sin cx dx =
2cx sin cx - c x -2cos cx - 2
x 2 2
2
3
0 c
So,
x dx
x x 2 2
c x -2 + 2cos cx
Σ
r=1
(-1) r-1 2r
c
0 0
2 2r
=
c
2
(2.2')
x x 2!
x dx n =
2
x
2+n
0 0 ( 2+ n )!
Then, (2.2') becoms
2 2
2! c x -2 + 2cos cx
Σ
r=1
(-1) r-1 2r
c
( 2+2r ) !
x
2+2r
=
c
2
(2.3')
When c=3.1 , the first 25 terms of ∑ are calculated and both sides are illustrated, it is as follows.
Both sides overlap exactly and blue (left) can not be seen.
- 19 -
25.4 Series of Higher Integrals with coefficients
In this section, we ask for the sum of series of higher integral with arithmetic coefficients and geometric
coefficients, such as
f(x)dx
x x x x x x x x
1c
1
f(x)dx 2c
2
f(x)dx + 3c
2 3
f(x)dx 4c
3 4
4
+
a a a a a a a a
Theorem 25.4.1
When c is a positive number and a is a real number on the domain of analytic function f(x) ,
the following expressions hold
f(x)dx e
x x x x
- (m +1) c m+1 e cx m-1 -cx
dx
2
a a a a
e f(x)dx e dx
x x x x
+ m c m+2
cx m -cx 2
(1.1)
a a a a
f(x)dx = 0
x x
Especially, when lim m c m+2 m
,
m a a
Proof
f (a) = 0 (r =1, 2, ,m+ n -1), Theorem 16.1.2 ( 16.1 ) was as follows.
r
<>
When
x x m -1 -n
f <0>g (0)dx n = Σ n r (r)
f < + >g
a a r=0 r
(-1)m n -1 n -1C k x x <m+ k > (m+ k ) n
f
(n , m) Σ
+ g dx
k=0 m + k a a
r ()
Here, let g = e -cx , n =2 . Then, since g =(-1)rc re -cx ,
x x m -1 -2
r
f <0>e -cxdx 2 = Σ f <2+ >(-1)rc re -cx
a a r=0 r
(-1)m 1C 0 x x <m+0>
+ f (-1)m+0c m+0e -cxdx 2
(2, m) m +0 a a
(2, m) m +1
(-1) C m
1 1 x x
<m+1>
+ f (-1)m+1c m+1e -cxdx 2
a a
i.e.
x x m -1
r
f <0> -cx
e dx 2 = Σ (1+ r)f <2+ >c re -cx
a a r=0
f
x x x x
<m> -cx m+1 <m+1> -cx
+ (m +1)c m f e dx 2 - m c e dx 2
a a a a
r
-2 1
(-1)r = 1+ r , = m(m +1)
(2, m)
Here, subtracting 1 from the index <r> of the integration operator of the function f,
f
x x m -1
<1+ r> r -cx
<-1> -cx 2
e dx = Σ (1+ r)f ce
a a r=0
- 20 -
f
x x x x
m-1> -cx m+1 <m> -cx
+ (m +1)c m f< e dx 2 - m c e dx 2
a a a a
Left side becomes as follows.
f
x x
x x
<0> -cx x
<-1> -cx
f 2
e dx = f e a +c <0> -cx
e dx dx
a a a a
= f dx + c f
x x x
<0> -cx <0> -cx
e e dx 2
a a a
Substituting this for the above,
x x x m -1
<1+ r> r -cx
f <0>e -cxdx + c f <0>e -cxdx 2 = Σ (1+ r)f ce
a a a r=0
f f
x x x x
m <m-1> -cx 2 m+1 <m> -cx
+ (m +1)c e dx - m c e dx 2
a a a a
From this,
f f
m x x x
r-1 <r>
Σ
-cx <0> -cx <0> -cx
e rc f (x)= e dx + c e dx 2
r=1 a a a
f f
x x x x
m <m-1> -cx 2 m+1 <m> -cx
- (m +1)c e dx + m c e dx 2
a a a a
Here, let
f(x)e dx
m x x x x x
Σrc r
r=1 a
a
f(x)dx r = ce cx
a
f(x)e
-cx
dx + c e cx
2
a a
-cx 2
e f(x)dx
x x x x
- (m +1) c m+1
e dx
cx m-1 -cx 2
a a a a
e f(x)dx e dx
x x x x
+ m c m+2
cx m -cx 2
(1.1)
a a a a
dx
x x x x x x
+
1 2 2 3 3
Example 1 1c dx + 2c dx + 3c
a a a a a a
dx dx
x x m-1 x x m
m-1 (x -a ) m (x -a )
f(x) = 1 , = , =
a a (m -1) ! a a m!
Substituting these for (1.1) ,
x x m-1
m+1 cx (x -a ) -cx 2
- (m +1) c e e dx
a a (m -1) !
(x -a) m -cx 2
x x
m+2 cx
+ mc e e dx
a a m!
The 1st term and the 2nd term of the right side are as follows.
- 21 -
e e
cx
x
cx
x x
cx e ca- e cx x e ca- e cx
+ ce cx
-cx -cx 2
ce dx + ce dx = ce dx
a a a c a c
c(x-a)
= ce (x - a )
After the long calculation, the 3rd term and the 4th term on the right side become as follows.
x x m-1 x x m
(x -a ) (x -a) -cx 2
-(m +1) c m+1 e cx e dx + mc m+2 e cx
-cx 2
e dx
a a (m -1) ! a a m!
c(x-a)
= -ce (x -a)
m +1
e c(x-a c(x -a) m , c(x -a) - 1+ m , c(x -a )
)
+
(m -1) !
1
e c(x-a c(x -a) 1+ m, c(x -a) - 2+ m , c(x -a )
)
-
(m -1) !
Substituting these for the above,
dx
m x x
Σrc r r
= ce c(x-a (x -a ) - ce c(x-a (x -a )
) )
r=1 a a
m +1
e c(x-a c(x -a) m , c(x -a) - 1+ m , c(x -a )
)
+
(m -1) !
1
e c(x-a c(x -a) 1+ m, c(x -a ) - 2+ m , c(x -a )
)
-
(m -1) !
(1.2)
rc r dx r = c e
x x
c(x-a )
Σ
r=1 a a
(x -a) (1.2')
rc dx
m x x m rc r(x -a ) r
Σ
r=1 a a
r
r
=Σ
r=1 r!
Using this as the left side of (1.2) and (1.2') ,
m rc r(x -a) r
Σ = ce c(x-a (x -a) - ce c(x-a (x -a)
) )
r=1 r!
m +1
e c(x-a c(x -a) m , c(x -a) - 1+ m , c(x -a )
)
+
(m -1) !
1
e c(x-a c(x -a) 1+ m, c(x -a ) - 2+ m , c(x -a )
)
-
(m -1) !
(1.3)
r r
rc (x -a)
Σ = c e c(x-a)(x -a) (1.3')
r=1 r!
When x =4, a =3, c=2, m =7 , both sides of (1.3) are calculated as follows.
- 22 -
Further, when a =3, c=2 , the first 50 terms of ∑ are calculated and both sides of (1.3') are illustrated,
it is as follows. Both sides overlap exactly and blue (left) can not be seen.
Theorem 25.4.2
When c is a positive number and a is a real number on the domain of analytic function f(x) ,
the following expressions hold
f(x)e dx
m x x x x x
r-1
Σ(-1) rc r
r=1 a a
f(x)dx r = ce
-cx
a
f(x)e cxdx - c e
2 -cx
a a
cx 2
e f(x)dx
x x x x
- (-1)m(m +1) c m+1
e
-cx m-1 cx 2
dx
a a a a
f(x)dx e dx
x x x x
+ (-1)m+1 m c m+2 e m cx
-cx 2
(2.1)
a a a a
f(x)dx
x x
Especially, when lim m c m+2 m
= 0,
m a a
f(x)e
x x x x x
Σ(-1)r-1 rc r
r=1 a a
f(x)dx r = ce
-cx
a
f(x)e cxdx - c e
2 -cx
a a
cx
dx
2
(2.1')
Proof
In a way similar to Theorem 25.4.1, we obtain the desired expressioons
Example 2
log x dx
x x x x x x x x
1c log x dx - 2c
2
log x dx + 3c
2 3
log x dx 4c
3 4
4
+-
0 0 0 0 0 0 0 0
- 23 -
log x e dx
m x x x x x
r-1
Σ(-1) rc r
r=1 0 0
log x dx r = ce
-cx
0
log x e cxdx - c e
2 -cx
0 0
cx 2
e log x dx
x x x x
e
m
- (-1) (m +1) c m+1 m-1 cx
-cx 2
dx
0 0 0 0
log x dx e dx
x x x x
m+1
m c m+2 e m cx
-cx 2
+ (-1)
0 0 0 0
The 1st term and the 2nd term of the right side are as follows.
x 1 cx x 1
e log|x| - Ei(cx)0 = e cx log|x|- Ei(cx)+ + log c
log xe cxdx =
0 c c
x x 1 x cx
log xe cxdx = e log|x|- Ei(cx)+ + log cdx
0 0 c 0
1 x
= 2 e cxlog|x|+1-(cx +1)Ei(cx)+ ( + log c)cx0
c
1 1
= 2 e cxlog|x|+1-(cx +1)Ei(cx)+( + log c)cx - 2(1- - log c)
c c
1 cx
= 2 e log|x|+1-(cx +1)Ei(cx)+ cx( + log c)+ + log c -1
c
x et
Where, Ei(x) = dt , = 0.5772 (Euler-Mascheroni Constant ).
- t
From this,
log x e
x x x -cx
ce
e log|x|- Ei(cx)+ + log c
-cx cx 2 -cx cx 2 cx
ce dx - c e log x e dx =
0 0 0 c
2 -cx
c e
e log|x|+1-(cx +1)Ei(cx)+ cx( + log c) + + log c -1
cx
- 2
c
- cx e
-cx -cx -cx -cx
= -1 + e + cx e Ei(cx)- cx e log c
Ei(cx)- log c -
-cx -cx
= cx e +e -1
i.e.
log x e
x x x
cx
ce
-cx
log x e cxdx - c e
2 -cx
dx 2 = cx e -cxEi(cx)- logc - + e -cx - 1
0 0 0
Next,
x xm
x m 1
log x dx = m
log|x|-Σ
0 0 m! s=1 s
Using thes for the 3rd term and the 4th term on the right side, the right side becomes
Ei(cx )- logc - + e
-cx -cx
cx e -1
x m-1
x x
m -1 1
m
- (-1) (m+1) c m+1e log|x| -Σ e cxdx
-cx 2
0 0 (m -1) ! s=1 s
mx !log|x| -Σ s e
x x m
m+1 1 m
m c m+2e cx
-cx 2
+ (-1) dx
0 0 s=1
i.e.
- 24 -
Ei(cx )- logc - + e
-cx -cx
cx e -1
m +1 x x
m m -1 1
c m+1e x m-1 log|x| -Σ e cxdx
-cx 2
- (-1)
(m -1) ! 0 0 s=1 s
x x
m+1 1 m 1
c m+2e x m log|x| -Σ e cxdx
-cx 2
+ (-1)
(m-1) ! 0 0 s=1 s
Thus, we obtain
log x dx
m x x
r-1
Σ (-1) rc r r
= cx e -cxEi(cx)- logc - + e -cx - 1
r=1 0 0
m +1 x x
m m -1 1
c m+1e x m-1 log|x| -Σ e cxdx
-cx 2
- (-1)
(m -1) ! 0 0 s=1 s
x x
m+1 1 m 1
c m+2e x m log|x| -Σ e cxdx
-cx 2
+ (-1) (2.2)
(m -1) ! 0 0 s=1 s
log x dx
r-1
x x
Σ (-1) rc r r
= cx e -cxEi(cx)- logc - + e -cx - 1 (2.2')
r=1 0 0
xn
x x
n 1
n
log x dx = log|x|-Σ
0 0 n! s=1 s
Using this,
r c rx r
m r 1
r-1
Σ(-1) log|x|-Σ = cx e -cxEi(cx)- logc - + e -cx - 1
r=1 r! s=1 s
m +1 x x
m -1 1
m
c m+1e x m-1 log|x| -Σ e cxdx
-cx 2
- (-1)
(m -1) ! 0 0 s=1 s
x x
m+1 1 m 1
c m+2e x m log|x| -Σ e cxdx
-cx 2
+ (-1) (2.3)
(m -1) ! 0 0 s=1 s
rx r
=e
r 1
xEi(x) - + e -x - 1
r-1
Σ(-1)
-x
log|x|-Σ (2.3')
r=1 r! s=1 s
When x =-0.2, c =1.7, m =3 , Both sides of (2.3) are calculated as follows.
Further, when c=1.7 , the first 30 terms of ∑ are calculated and both sides of (2.3') are illustrated, it is as
- 25 -
follows. Both sides overlap exactly and blue (left) can not be seen.
- 26 -
25.5 Calculation by Double Series
The sum of series of higher integral with geometric coefficients can also be calculated by a double series.
That way is almost the same as 24.4 ~ 24.6 .
Theorem 25.5.1
When c is a positive number and a is a real number on the domain of analytic function f(x) ,
the following expressions hold
c f(x)dx
s+r
x x (r) c s(x -a )
Σ
r=0 a a
r r
= Σ f (a)Σ
r=0 s=0 (s + r)!
(1.1)
e
x x
= Σ f (r)(a) c(x-a )
dx r (1.1')
r=0 a a
e
x x
= Σ f (a) (r) -c(x-a )
dx r (1.2')
r=0 a a
Proof
r
()
Let f (a)= far and expand f(x) into Taylor series around a. Then
(x -a)0 1 (x -a )
1
2 (x -a )
2
3 (x -a )
3
f(x)= fa0 + fa + fa + fa + (0)
0! 1! 2! 3!
Integrating both sides of this with respect to x from a to x one by one,
x 1 2 3 4
0 (x -a ) 1 (x -a ) 2 (x -a ) 3 (x -a )
f(x dx fa
) = + fa + fa + fa +
a 1! 2! 3! 4!
2
(x -a)3 (x -a)4 (x -a)5
0 (x -a )
x x
f(x)dx = fa + fa1 + fa2 + fa3
2
+
a a 2! 3! 4! 5!
(x -a)3 4 5 6
f(x)dx 1 (x -a ) 2 (x -a ) 3 (x -a )
x x x
= fa + fa + fa + fa
3 0
+
a a a 3! 4! 5! 6!
c , c , c ,
0 1 2
Multiplying these also including (0) by , respectively,
0 0 0 1 0 2 0 3
c (x -a ) 1 c (x -a ) 2 c (x -a ) 3 c (x -a )
fa + fa + fa + fa
0 0
c f(x)= +
0! 1! 2! 3!
f(x)dx =
1 1 1 2 1 3 1 4
x c (x -a) 1 c (x -a ) 2 c (x -a ) 3 c (x -a )
fa + fa + fa + fa
1 0
c +
a 1! 2! 3! 4!
c f(x)dx
2 2 2 3 2 4 2 5
x x c (x -a) 1 c (x -a ) 2 c (x -a ) 3 c (x -a )
fa + fa + fa + fa
2 2 0
= +
a a 2! 3! 4! 5!
f(x)dx
3 3 3 4 3 5 3 6
x x x c (x -a) 1 c (x -a ) 2 c (x -a ) 3 c (x -a )
fa + fa + fa + fa
3 3 0
c = +
a a a 3! 4! 5! 6!
Adding these perpendicularly.
- 27 -
f(x)dx
x x x x x x
0
c f(x)+ c
1
f(x)dx + c
2
f(x)dx + c
2 3 3
a a a a a a
0 0 1 1 2 2 3 3
0 c (x -a ) c (x -a) c (x -a ) c (x -a)
= fa + + + +
0! 1! 2! 3!
1! +
0 1 1 2 2 3 3 4
1 c (x -a ) c (x -a) c (x -a ) c (x -a)
+ fa + + +
2! 3! 4!
2! +
0 2 1 3 2 4 3 5
2 c (x -a ) c (x -a) c (x -a ) c (x -a)
+ fa + + +
3! 4! 5!
s+ r
c s(x -a)
=Σ far Σ
r=0 s=0 (s + r)!
i.e.
c f(x)dx
s+r
x x (r) c s(x -a )
Σ
r=0 a a
r r
= Σ f (a)Σ
r=0 s=0 (s + r)!
(1.1)
Furthermore,
e
s+ r
c s(x -a) x x
c(x-a )
Σ
s=0 (s + r)!
=
a a
dx r
Then
e
x x x x
Σc f(x)dx = Σ f (a)
(r) r c(x-a )
r
dx r (1.1')
r=0 a a r=0 a a
(1.2) and (1.2') are also proved in a similar way.
Note
If (1.1') and (1.2') are written down, they are as follows.
f(x)dx
x x x x x x
c f(x ) c f(x)dx c +
0 1 2 2 3 3
f(x)dx + c
a a a a a a
e
x x x x x x
= f (0()a) f (1()a) e c(x-a)dx + f (2()a) e c x-a dx f (3(
( ) ) 2
a) c(x-a)
dx +
3
a a a a a a
These are similar to the Taylor series and are beautiful, so these are very good for viewing.
However, these are not so useful.
log x dx
x x x x x x
1 2 2 3 3
Example 1 log x + c log x dx + c log x dx + c +
1 1 1 1 1 1
n =1, 2, 3,
(n ) n-1
f(x) = log x , (log x) =(-1) (n -1)! x -n
Substituting these for (1.1), (1.1') ,
e
x x
= Σ(-1)r-1(r -1)! c(x-1)
dx r
r=1 1 1
When higher integrals are replaced with Riemann-Liouville Integral and x =0.8 , c=1.7 , m =10 are given,
each is calculated as follows.
- 28 -
x x x x x x
Example 2 x - c1 x dx + c 2 x dx 2 - c 3 x dx 3 +-
1 1 1 1 1 1
f(x) = x
(1+1/2)
1
(n ) -n (2n -3)!!
f (1) = 1 2
= (-1)n-1
1+1/2- n 2n
Substituting these for (1.2),
(-1) c
s+ r
x x (2r -3) !! c s(x -a)
Σ
r=0 a a
r r r
x dx = e
-c(x-1)
2r s=0
+Σ(-1)
(s + r)!
r=0
r-1
Σ(-1)
s
When higher integrals are replaced with Riemann-Liouville Integral and x =0.9 , c=1.3 , m =15 are given,
both sides are calculated as follows.
During the proof of the previous theorem , if the calculation is done without including (0) ,
we obtain the following theorem.
Theorem 25.5.1'
When c is a positive number and a is a real number on the domain of analytic function f(x) ,
the following expressions hold
s+r
x x c s(x -a )
Σc r
(r)
f(x)dx r = Σ f (a)Σ (1.3)
r=1 a a r=0 s=1 (s + r)!
f(x)dx
s+ r
r-1 r
x x (r)
s-1 c s(x -a )
Σ
r=1
(-1) c
a a
r
= Σ f (a)Σ(-1)
r=0 s=1 (s + r)!
(1.3')
- 29 -
Combining this theorem and 25.1 , we obtain the following theorem which gives the collateral integral
of the product of an exponential function and arbitrary functions.
Theorem 25.5.1"
When c is a positive number and a is a real number on the domain of analytic function f(x) ,
the following expressions hold
s+ r
x
(r) s-1 c s-1(x -a)
e f(x)dx = e cxΣ f (a)Σ(-1)
cx
(1.4)
a r=0 s=1 (s + r)!
e
s+r
x
(r) c s-1(x -a )
-cx
f(x)dx = e -cxΣ f (a)Σ (1.4')
a r=0 s=1 (s + r)!
e
x
cx
Example 2" sin x dx
a
r
s+ r
c s-1(x -a )
Σ(-1)
x
s-1
e sin xdx = e Σ sin a +
cx cx
a r=0 2 s=1 (s + r)!
When a =2 , c=1.3 and the first 50 terms of ∑ are calculated and both sides are illustrated, it is as follows.
Both sides overlap exactly and blue (left) can not be seen.
Theorem 25.5.2
When c is a positive number and a is a real number on the domain of analytic function f(x) ,
the following expressions hold
2s+1+ r
f(x)dx
2s+1
x x c (x -a)
Σ
(r )
= Σ f (a )Σ
2r-1 2r-1
c (2.1)
r=1 a a r=0 s=0 (2s +1+ r) !
f(x)dx
2s+1
x x c (x -a )
Σ
(r)
r-1 2r-1
= Σ f (a )Σ(-1)
2r-1 s
(-1) c (2.2)
r=1 a a r=0 s=0 (2s +1+ r) !
- 30 -
Proof
In a way similar to Theorem 25.5.1, we obtain the desired expressioons
tan x dx
x x x x x x x x
Example 3 c
1
tan x dx c
3
tan x dx + c
3 5
5
tan x dx + c
7
7
0 0 0 0 0 0 0 0
The higher differential quotient of tan x on x =0 is as follows according to Theorem 9.2.6 ( 9.2 )
(2n -1)
22n22n-1B2n
(tan x) |x=0 = T2n-1 = , (tan x)(2n)|x=0 = 0
2n
Where, T 2n-1 is the tangent number and B 2n is the Bernoulli number. Thus, from (2.1) ,
2s+1 2s+2r
x x c x
Σc
r=1
2r-1
0
0
tan x dx
2r-1
= ΣT2r-1Σ
r=1 s=0 (2s +2r) !
When higher integrals are replaced with Riemann-Liouville Integral and x =0.9 , c=1.2 , m =12 are given,
both sides are calculated as follows.
secx dx
x x x x x x x x
Example 4 c
1
secx dx c
3
secx dx + c
3 5
secx dx - c
5 7
7
-
0 0 0 0 0 0 0 0
The higher differential quotient of secx on x =0 is as follows according to Theorem 9.2.8 ( 9.2 )
2s+1 2s+1+ 2r
x x c x
Σ(-1)r-1c
r=1
2r-1
0 0
secxdx
2r-1
= ΣE2rΣ(-1)s
r=0 s=0 (2s +1+ 2r) !
When higher integrals are replaced with Riemann-Liouville Integral and x =0.3 , c=1.1 , m =12 are given,
both sides are calculated as follows.
- 31 -
Theorem 25.5.3
When c is a positive number and a is a real number on the domain of analytic function f(x) ,
the following expressions hold
f(x)dx
x 2s 2s+ r
x (r) c (x -a)
Σ = Σf (a)Σ
2r 2r
c (3.1)
r=0 a a r=0 s=0 (2s + r)!
x x 2s 2s+ r
c (x -a)
Σ(-1)rc 2r
(r)
f(x)dx =Σf (a)Σ(-1)s
2r
(3.2)
r=0 a a r=0 s=0 (2s + r)!
Proof
In a way similar to Theorem 25.5.1, we obtain the desired expressioons
tan
x x x x x x
Example 5 tan
-1
x +c
2
tan x dx c
-1 2 4
tan x dx c
-1 4 6
-1
x dx +
6
1 1 1 1 1 1
According to " 岩波数学公式Ⅰ" p39, the following expression holds for a natural number n .
(n )
tan x = (n -1)!cos n(tan -1 x )sin n tan -1 x +
-1
2
From this,
3n
(n ) (n -1)!
tan x
-1
|x=1 = sin
2n/2 4
Substituting this for (3.1) ,
tan
2s 2s
x x c (x -1)
Σ 4Σ
2r -1 2r
c x dx =
r=0 1 1 s=0 (2s) !
3r
2s 2s+ r
(r -1)! c (x -1)
+Σ
r=1 2r/2
sin 4 Σ s=0 (2s + r) !
When higher integrals are replaced with Riemann-Liouville Integral and x =1.8 , c=2.1 , m =17 are given,
both sides are calculated as follows.
During the proof of the previous theorem , if the calculation is done without including (0) ,
we obtain the following theorem.
- 32 -
Theorem 25.5.3'
When c is a positive number and a is a real number on the domain of analytic function f(x) ,
the following expressions hold
Example 6
sin
x x x x x x x x
c
2
sin x dx c
-1 2 4
-1
sin x dx + c
4 6
sin x dx c
-1 6 8
x dx +
-1 8
0 0 0 0 0 0 0 0
-1
The higher differential quotient of sin x on x =0 is as follows according to Theorem 9.3.2 ( 9.3 )
( 2n +1 )
sin x |x=0 = 2nC 0(2n -1)!!(2n -1)!!00 = (2n -1)!!2
-1
sin
2s 2s+2r+1
x x c x
Σ
r=1
(-1) r-1 2r
c
0 0
-1
x dx
2r
= Σ(2r -1)!!
r=0
2
Σ
s=1
(-1) s-1
(2s +2r +1) !
When higher integrals are replaced with Riemann-Liouville Integral and x =0.6 , c=2.3 , m =10 are given,
both sides are calculated as follows.
2018.06.09
Kano Kono
Alien's Mathematics
- 33 -
26 Higher and Super Calculus of Zeta Function etc
a harmonic number ( where H0 = 0 ) , the following expression holds on whole complex plane.
n-1 r+n
z -1 z -1
n(z) = logz -1 - Hn-1 Σ
+ (-1) r
r n =1, 2, 3,
n -1! r=0 r + n!
(1.1h)
Where, r is Stieltjes constant defined by the following expression.
n (log k)r (log n)r+1
r = lim Σ -
n k =1 k r +1
Proof
It is known that the Riemann zeta function z is expanded to Laurent series around 1 as follows.
1 (z -1)r
(z) = + Σ(-1) r r
z -1 r=0 r!
Then, integrating the both sides with respect to z without considering the constant of the integration,
r+1
z -1
<1>
(z) = logz -1 + Σ(-1) r r
r=0 r +1!
Integrating this once more without considering the constant of the integration,
1 r+2
z -1 z -1
<2>(z) = logz -1-1 + Σ(-1) r
r
1! r=0 r +2!
Integrating this once more without considering the constant of the integration,
2 r+3
z -1 1 z -1
<3>(z) = logz -1- 1+ +Σ(-1) r r
2! 2 r=0 r +3!
Hereafter, by induction,
n-1 r+n
+Σ(-1)
z -1 1 1 z -1
n
(z) = logz -1- 1+ ++ r
r
n -1! 2 n -1 r=0 r +n!
Rewriting the harmonic number as Hn-1 , we obtain the desired expression.
- 1-
Note
If the left side of (1.1h) is represented by integral symbols, it is as follows.
z n-1 r+n
z z -1 z -1
(z)dz = n-1 Σ
log z -1 - H + (-1) r
r
an a1 n -1! r=0 r + n!
Here, the lower limits of the integral are as follows.
cf.
The higher integral with variable lower limits a1 = a2 = = an = 0 is as folows.
z n-1 r+n
z
z -1 z -1
(z)dz n = n-1 Σ
log z -1 -H + (-1) r
r
0 0 n -1! r=0 r + n!
z -1
n-1 n -1 (-1) r Hr z n-1- r n (-1) sr z n-s
- i +Σ -ΣΣ
n -1! r=1 r!n -1- r! r=0 s=1 r + s!n - s!
As seen from the existence of constant-of-integration polynomials, this is a collateral higher integral.
Proof
From (1.1h) ,
- 2-
n-1 r+n
z -1 z -1
n(z) = logz -1 - Hn-1 + Σ(-1) r
r
n =1, 2, 3,
n -1! r=0 r + n!
At first,
figure is an imaginary part. In both figures, the orange is z and the blue is <0.1>z . Since p is neer
0 , both curved surfaces look double.
(-1)-n n ! (z -1)r-n
n
(z) = + Σ(-1) r r
(1.2h)
(z -1)n+1 r=0 1+ r - n
Proof
It is known that the Riemann zeta function z is expanded to Laurent series around 1 as follows.
- 3-
1 (z -1)r
(z) = + Σ(-1) r
r
z -1 r=0 r!
Differentiating the both sides n times with respect to z,
n
r
1
n (n )
(z) = + Σ(-1)r (z -1)r
z -1 r=0 r!
According to Formula 9.2.1 in " 09 Higher Derivative ",
n 1+ -n
x = x 0
1+ - n
- + n -n
= (-1)-n x < 0
-
Applying this,
(n ) (1+ r) r!
(z -1)r = (z -1)r-n = (z -1)r-n
1+ r - n 1+ r - n
n
(1+ n)
1 n!
= (-1)-n (z -1)-1-n = (-1)-n
z -1 (1) (z -1)n+1
Substituting these for the above,
(-1)-n n ! (z -1)r-n
n
(z) = + Σ(-1) r r
(1.2h)
(z -1)n+1 r=0 1+ r - n
When these are calculated by formula manipulation soft Mathematica , it is as follows. We can see that this
formula is numerically right.
- 4-
Proof
From Formula 26.1.2h ,
(-1)-n n ! (z -1)r-n
n(z) = + Σ(-1)r r (1.2h)
(z -1)n+1 r=0 r - n !
According to Formula 1.3.1 in " 01 Gamma Function & Digamma Function " ,
(-n)
(-1)-n n ! = - , n =0, 1, 2, 3,
(-n)
Using this, (1.2h) is rewritten as follows.
logz -1--n - 0 z -1
r-n
n(z)= z -1
-n-1
+ Σ(-1) r r
-n r=0 1+ r - p
Because,
logz -1 - 0 logz -1 - 0
= =0 for n =0, 1, 2, 3,
-n
So, replacing the natural number n with a complex number p , we obtain the desired expression.
Note
If the sign of p is inverted in (1.2s), it becomes as follows.
right figure is an imaginary part. In both figures, the orange is (2)z and the blue is (1.9)z . Since
p is near 2 , both curved surfaces look double.
- 5-
26.2 Higher and Super Calculus of Dirichlet Lambda Function
log r+12 r-1 r
r+n
1 r-s z -1
+ Σ (-1) r
r + -Σ slog 2 n =1, 2, 3,
2 r=0 r +1 s=0 s r + n!
(2.1h)
Where, r is Stieltjes constant defined by the following expression.
n (log k)r (log n)r+1
r = lim Σ -
n k =1 k r +1
Proof
According to Formula 3.1.3 in " 03 Complementary Series of Dirichlet Series " ( Dirichlet Series ), Dirichlet
lambda function is expanded to Laurent series around 1 as follows.
log r+12 r-1 r
r
1 1 z -1
(z) = + Σ(-1) r r + -Σ s log 2r-s
2z -1 2 r=0 r +1 s=0 s r!
Where, in { } is absent for r =0 .
Hereafter, in a way similar to the proof of Formula 26.1.1h , the desired expression is obtained.
Note
This is also lineal higher integral with variable lower limits. The first few of the integral lower limits are as follows.
- 6-
a1 = 1.509052 , a2 = 2.203125 , a3 = 2.891846 ,
1+ r + p
log r+12 r-1 r
r+p
1 z -1
+ Σ (-1) r
r + -Σ s log 2r-s (2.1s)
2 r=0 r +1 s=0 s
Proof
In a way similar to the proof of Formula 26.1.1s , the desired expression is obtained.
(-1) -n n !
(n)z =
2z -1n+1
log r+12 r-1 r
r-n
1 z -1
+ Σ(-1) r +
r
-Σ s log 2 r-s
n =0, 1, 2,
2 r=0 r +1 s=0 s 1+ r - n
(2.2h)
Proof
As seen in the proof of Formula 26.2.1h , the Dirichlet lambda function z is expanded to Laurent series
- 7-
around 1 as follows.
log r+12 r-1 r
r
1 1 r-s z -1
(z) = + Σ(-1) r + r
-Σ s log 2
2z -1 2 r=0 r +1 s=0 s r!
Where, in { } is absent for r =0 .
Hereafter, in a way similar to the proof of Formula 26.1.2h , the desired expression is obtained.
When these are calculated by formula manipulation soft Mathematica , it is as follows. We can see that this
formula is numerically right.
1+ r - p
log r+12 r-1 r (z -1)r-p
1
+ Σ(-1) r r + -Σ s log 2r-s (2.2s)
2 r=0 r +1 s=0 s
Proof
In a way similar to the proof of Formula 26.1.2s , the desired expression is obtained.
Note
If the sign of p is inverted in (2.2s), it becomes as follows.
1+ r + p
log r+12 r-1 r
r+p
1 z -1
+ Σ(-1) r +
r
-Σ s log 2r-s
2 r=0 r +1 s=0 s
This results in Formula 26.2.1s . That is, the lineal super calculus of Dirichlet lambda function is seamless.
- 8-
Example The 0.9th order derivative
z = x +i y , (0.9)z
(1)
When z and are illustrated as follows. The left figure is a real part and the
right figure is an imaginary part. In both figures, the orange is (1)z and the blue is (0.9)z . Since
p is near 1 , both curved surfaces look double.
- 9-
26.3 Higher and Super Calculus of Dirichlet Eta Function
zn -n
r-1 log r
n
(z) = + (-1) Σ(-1)
n
n =0, 1, 2, (3.1h)
n! r=2 rz
(2) The following expression holds on whole complex plane.
r
r-1
zn k (-1) k log -nr
n
(z) = + (-1) ΣΣ
n
n =0, 1, 2, (3.1h')
n! k =2 r=2 2k+1 rz
Proof
At Rez > 0 , Dirichlet eta function z is expressed with the following series which is called DIrichlet
eta series.
1 1 1
(z) = 1 + Σ(-1)r-1e -z log r = 1 - + - +-
r=2 2z 3z 4z
So, integrating the both sides n times with respect to z without considering the constant of the integration,
zn log -n r
n(z) = + (-1) nΣ(-1) r-1 n =0, 1, 2, (3.1h)
n! r=2 rz
Applying Euler transformation to this second term, we obtain (3.1h') . By this transformation, (3.1h) is
analytically continued from Rez > 0 to the whole complex plane. In addition, about Euler transformation,
see " 10 Convergence Acceleration & Summation Method by Double Series of Functions " ( A la carte ).
- 10 -
Note
If the left side of (3.1h) is represented by integral symbols, it is as follows.
z z
zn -n
r-1 log r
(z)dz = + (-1) Σ(-1)
n
an a1 n! r=2 rz
Here, the lower limits of the integral are as follows.
zp p i
r-1 log
-p
r
p
(z) = + e Σ(-1) (3.1s)
(1+p) r=2 rz
(2) The following expression holds on whole complex plane.
r
r-1
zp p i
k (-1) k log -p r
p
(z) = + e ΣΣ (3.1s')
(1+p) k =2 r=2 2k+1 rz
Proof
In Formula 26.3.1h , replacing n! with 1+n , replacing (-1) -n with e -n i and replacing the natural
- 11 -
26.3.2 Higher and Super Derivative of Dirichlet Eta Function
z -n n
r-1 log r
n
(z) = + (-1) Σ(-1)
-n
n =0, 1, 2, (3.2h)
(1-n) r=1 rz
(2) The following expression holds on whole complex plane.
r-1
z -n k (-1) k log nr
n
(z) = +(-1) ΣΣ
-n
n =0, 1, 2, (3.2h')
(1-n) k =2 r=2 2k+1 r rz
Proof
As seen in the proof of Formula 26.3.1h ,
1 1 1
(z) = 1 + Σ(-1)r-1e -z log r = 1 - + - +-
r=2 2z 3z 4z
0
So, assuming 1 = z /0! and differentiating the both sides n times with respect to z,
z -n n
r-1 log r
n
(z) = + (-1) Σ(-1)
-n
n =0, 1, 2,
(-n)! r=1 rz
Replacing -n! with 1-n , we obtain (3.2h) . In addition,
0
z -n 1 for n =0
=
(1-n) for n =1, 2, 3,
And applying Euler transformation to this second term, we obtain (3.2h') . By this transformation, (3.2h) is
analytically continued from Rez > 0 to the whole complex plane.
If these are calculated according to (3,2h') by formula manipulation soft Mathematica , it is as follows. We can
see that this formula is numerically right.
- 12 -
and p is gamma function,
(1) The following expression holds for z s.t. Rez > 0 .
z -p log p r
p
(z) = + e -p iΣ(-1) r-1 (3.2s)
(1-p) r=1 rz
(2) The following expression holds on whole complex plane.
r-1
z -p -p i
k (-1) k log p r
p
(z) = +e Σ Σ (3.2s')
(1-p) k =2 r=2 2k+1 r rz
Proof
In Formula 26.3.2h , replacing (-1) -n with e -n i and replacing the natural number n with a complex
Note
If the sign of p is inverted in (3.2s), it becomes as follows.
zp log -p r
-p(z) = + e p iΣ(-1) r-1
(1+p) r=1 rz
This results in Formula 26.3.1s . That is, the lineal super calculus of Dirichlet eta function is seamless.
- 13 -
26.4 Higher and Super Calculus of Dirichlet Beta Function
zn -n
r-1 log 2r -1
n
(z) = + (-1) Σ(-1)
n
n =0, 1, 2, (4.1h)
n! r=2 2r -1
z
zn k (-1) r-1 k log -n2r -1
<n>
(z) = + (-1) ΣΣ
-n
n =0, 1, 2, (4.1h')
n! k =2 r=2 2k+1 r 2r -1
z
Proof
At Rez > 0 , Dirichlet beta function z is expressed with the following series which is called DIrichlet
beta series.
1 1 1
(z) = 1 + Σ(-1)r-1e -z log2r-1 = 1 - + - +-
r=2 3z 5z 7z
So, integrating the both sides n times with respect to z without considering the constant of the integration,
zn -n
r-1 log 2r -1
n
(z) = + (-1) Σ(-1)
n
n =0, 1, 2, (4.1h)
n! r=2 2r -1
z
Applying Euler transformation to this second term, we obtain (4.1h') . By this transformation, (4.1h) is
analytically continued from Rez > 0 to the whole complex plane.
- 14 -
Note
If the left side of (4.1h) is represented by integral symbols, it is as follows.
z z
zn log -n2r -1
(z)dz = + (-1) nΣ(-1) r-1
an a1 n! r=2 2r -1
z
zp k (-1) r-1 k log -p2r -1
(z) = + e p iΣΣ
<p>
(4.1s')
(1+p) k =2 r=2 2k+1 r 2r -1
z
Proof
n! with 1+n , replacing (-1) -n with e -n i
In Formula 26.4.1h , replacing and replacing the natural
number n with a complex number p , we obtain the desired expressions.
- 15 -
26.4.2 Higher and Super Derivative of Dirichlet Beta Function
(4.2h')
Proof
As seen in the proof of Formula 26.4.1h ,
1 1 1
(z) = 1 + Σ(-1)r-1e -z log2r-1 = 1 - + - +-
r=2 3z 5z 7z
Hereafter, in a way similar to the proof of Formula 26.3.2h , (4.2h) is obtained.
And applying Euler transformation to this second term, we obtain (4.2h') . By this transformation, (4.2h) is
analytically continued from Rez > 0 to the whole complex plane.
- 16 -
(2) The following expression holds on whole complex plane.
r-1
z -p k (-1) k log p (2r -1)
p
(z) = + e -p iΣΣ k+1 (4.2s')
(1-p) k =2 r=2 2 r (2r -1)z
Proof
In Formula 26.4.2h , replacing (-1) -n with e -n i and replacing the natural number n with a complex
Note
If the sign of p is inverted in (4.2s), it becomes as follows.
zp p i
r-1 log
-p
(2r -1)
-p
(z) = + e Σ(-1)
(1+p) r=2 (2r -1)z
This results in Formula 26.4.1s . That is, the lineal super calculus of Dirichlet beta function is seamless.
2018.01.13
Kano Kono
Alien's Mathematics
- 17 -
27 Zeros of Super Derivative of Riemann Zeta
constant, the1st order derivative (1)z of Rieamnn zeta fanction is expanded to Laurent series around 1
as follows
1 (z -1)r-1
(1)
(z) = - + Σ(-1) r r
(1.0)
(z -1)2 r=0 (1+r -1)
As is clear from this formula and the figure, (1)z has a pole of order 2 at z =1 .
In this section, we actually ask for the zeros of (1)z bearing these things in mind.
At the zeros of (1)z , both the real part Re (1)z and the imaginary part Im (1)z have
to be 0. That is, solutions of the following simultaneous equations must be obtained.
Since it is impossible to solve this, we are obliged to use the numerical solution method like the Newton-Raphson
method. Since the function FindRoot [] for it is implemented in formula manipulation software Mathematica,
we use this. However, for that purpose, approximate position of the zeros must be known. In order to know this,
- 1-
we draw contour plots of (1.1r) and (1.1i) and find the intersection. Since the function ContourPlot [] for it
is implemented in formula manipulation software Mathematica, we use this. At this time, x can be up to 3
at most from the above (ii) .
(1) 0 y 15
The contour plot in this interval is drawn as follows. Although (1 ,0) looks like the intersection of the real part
and the imaginary part, this is not a zero but a singular point (pole) . Therefore, there is no zero of (1)z
in this interval.
(2) 15 y 25
Zero is found near (2.5 ,23) . The result of the numerical calculation is written on the right. This is the first
zero of (1)z .
- 2-
(3) 25 y 35
Zero is found near (1 ,32) . However, an exact figure cannot be drawn by the series of (1.0) and the upper
one becomes a ghost. Of course, accurate numerical calculation is also impossible. So, it was calculated by
the function Zeta'[] of Mathematica . It is written on the right.
(4) 35 y 50
For y 35 , we draw and calculate by the left side ( Zeta'[] of Mathematica ) rather than the right side
of (1.0) . Then, the contour plot is drawn as follows. 3 zeros are seen. The results of the numerical calculation
are written on the right.
As mentioned above, 5 zeros were obtained in the Interval 0 y 50 . If these are recalculated with
significant 16 digits, it is as follows.
- 3-
For trial, if the first zero is substituted for Zeta'[z ] , it is as follows.
If these are plotted on the complex plane with the zeros of z , it is as follows. Blue is a zero of z
and Red is a zero of (1)z . The latter number is reduced by half from the former. Further, each of the latter
exists on the right side ( y > 1/2 ) of the former.
I have not seen trivial zeoros of (1)z . So, first, let us draw the 2D diagram and the contour plot. Then,
- 4-
Observing these diagrams, we can see that trivial zeros of (1)z are distributed in the negative region of
the x -axis. If the first few are calculated with significant 16 digits, it is as follows.
- 5-
27.2 Zeros of
(1/2)
z
the p (p)z of Rieamnn zeta fanction is expanded to Laurent series around 1 as follows
th order derivative
Though the super differentiation for arbitrary p >0 is possible using (2.0) , we deal with the super derivative
between (0)z and (1)z in this chapter. And, in order to overlook the zeros of the super derivative
in this interval, it seems good to observe the zeros of (1/2)(z) which is in the middle.
From (2.0) , the function used for this is as follows.
logz -1--1/2 - 0 z -1
r-1/2
(1/2)
(z)= z -1
-3/2
+Σ(-1) r r
(2.1)
-1/2 r=0 r +1/2
(1) 0 y 15
The contour plot in this interval is drawn as follows. The intersection of the real part and the imaginary part
( zero ) is found in two places. The results of the numerical calculation are written on the right. The bottom
intersection is not a zero but a singularity point.
- 6-
(2) 15 y 25
2 zeros are seen also in this interval. The results of the numerical calculation are written on the right. However,
The first zero is an estimate.
(3) 25 y 35
The contour plot is drawn as follows. Most are ghosts and the location of the intersection can not be seen.
This figure shows that it is impossible to obtain zeros in wider interval, by the above formula (2.1) and my
personal computer & the software.
- 7-
As mentioned above, 4 zeros were obtained in the Interval 0 y 25 . If the first 3 are recalculated with
significant 16 digits, it is as follows.
- 8-
If these are calculated with significant 16 digits, it is as follows.
- 9-
27.3 Transition of Non-Trivial Zeros associated with the Super Dderivative.
In this section, we investigate the transition of non-trivial zeros associated with the Super Derivative using
formula (2.0) .
The zero of (p)z appears near p =3/10 , moves to the right at p =4/10 , 5/10 , 6/10 , and
disappears at p =7/10 . This is a zero peculiar to super derivative that appears only on non-integer order.
- 10 -
The blue line is the real part and the orange line is the imaginary part. The red dot is the intersection of these
namely the zero. The red dot on the left is the first zero 1/2+ i 14.1347 of (z) . This zero moves to
the right at p =1/10 9/10 and exists as long as p <1 . And the zero disappears at the moment of p =1 .
Because, at this time, the real part and the imaginary part become parallel and do not cross.
Next, we investigate the transition of the zeros of (0)z (1)z in 18 y 22.5 . The interval
of x is set to 0 x 7.5 . if these contour lines and intersections are drawn on 1 figure, it is as follow.
The blue line is the real part and the orange line is the imaginary part. The red dot is the intersection of these
namely the zero. The red dot on the left is the 2nd zero 1/2+ i 21.0220 of (z) . This zero moves to
the right at p =1/10 9/10 and exists as long as p <1 . And the zero disappears at the moment of p =1 .
Because, at this time, the real part and the imaginary part become parallel and do not cross.
Last, we investigate the transition of the zeros of (0)z (1)z in 22.5 y 26 . The interval
of x is set to 0 x 3. if these contour lines and intersections are drawn on 1 figure, it is as follow.
The blue line is the real part and the orange line is the imaginary part. The red dot is the intersection of these
namely zero.
- 11 -
The red dot on the left is the 3rd zero 1/2+ i 25.0108 of (z) . This zero moves to the right at
p =1/10 9/10 and finally reaches the zero 2.4631+ i 23.2983 of (1)(z) .
Summary
Investigating the transition of the non-triviall zeros of (p)(z) p = 0 1 in interval 0 y 26 ,
we obtained the following results.
(1) There exist non-trivial zeros peculiar to super derivative that appear only on non-integer order 0 < p < 1 .
(2) The first two zeros 1/2+ i 14.1347 and 1/2+ i 21.0220 of (z) move to the lower right with
increasing p , and disappear suddenly at (1)(z) . This is consistent with the fact that the number of
zeros of (1)z is less than the number of zeros of z ( 27.1.1 (iii) ) .
(3) The 3rd zero 1/2+ i 25.0108 of (z) moves to the lower right with increasing p , and becomes
the 1st zero 2.4631+ i 23.2983 of (1)(z) . This assists the assertion that the zeros of
- 12 -
27.4 Transition of Trivial Zeros associated with the Super Derivative.
In this section, we investigate the transition of trivial zeros associated with the Super Derivative using formula (2.0) .
(z) 's zero -2 draws a ballistic trajectory on a complex plane with increasing p , and falls near 1.77 .
0 0.67 show the transition. In 0.68 , this zero is separated into two zeros on the x -axis. In 0.78 , two
zeros move to the right and the left respectively. In 0.82 , the left zero is swallowed in the singular point 1 and
side. This looks like the transmigration from x =1 . Additionally, the zeros existing on the half line x >1, y = 0
at 0.670597 p < 1 are trivial zeros peculiar to the non-integer order derivative.
- 13 -
27.4.2 Transition of the trivial zeros z = -4 , -6 at z
(0) (1)
z
z
(0) (1)
Here, we investigate the transition of the trivial zeros z = -4 , -6 at z . The interval of x
is set to -7.5 x -3 . When the zeros are traced with nine contour plots, it is as follows.
The blue line is the real part, the orange line is the imaginary part, and the red dot is the zero.
In 0 , there are two zeros -4 , -6 of (z) . Zero -4 disappears at the moment p increases. Zero -6 draws
a lofted trajectory on a complex plane with increasing p , and falls to (1)(z) ' s zero -4.93676 .
Simultaneously, another (1)(z) ' s zero -7.07459 appears on the left side. This looks like the trans-
migration from -6 .
Here, we investigate the transition of the trivial zeros z = -8 , -10 at (0)z (1)z . The interval of x
is set to -11.5 x -7.5 . When the zeros are traced with nine contour plots, it is as follows.
The blue line is the real part, the orange line is the imaginary part, and the red dot is a zero .
- 14 -
In 0 , there are two zeros -8 , -10 of (z) . Zero -8 disappears at the moment p increases. Zero -10
draws a lofted trajectory on a complex plane with increasing p , and falls to (z) ' s zero -9.17049 .
(1)
Summary
Investigating the transitions of (z) 's trivial zeros -2 , -4 , , - 10 associated (p)(z) p = 0 1 ,
we obtained the following results.
(1) There exist trivial zeros peculiar to super derivative that appear only on non-integer order 0.670597 p < 1 .
And, these are real numbers greater than 1.
(2) Half of (z) 's zero -2 transmigrates from the singular point 1 to (1)(z) 's zero -2.71726 , and.
the other half disappears at the moment of p =1 . The above (1) occurs in this process.
(3) (z) 's zero -6 (1)(z) 's zero -4.93676 , and transmigrates to (1)(z) 's zero
moves to
- 15 -
(4) (z) 's zero -10 (1)(z) 's zero -9.17049 , and transmigrates to (1)(z) 's zero
moves to
2018.12.07
2018.12.20 Renewed
Kano Kono
Alien's Mathematics
- 16 -
28 Zeros of Super Integral of Riemann Zeta
28.1 Zeros of
<1>
z
constant, the1st order lineal primitive <1>z of Rieamnn zeta fanction is expessed as follows
r+1
z -1
<1>(z) = logz -1 + Σ(-1) r r (1.0)
r=0 r +1!
The 3D figures of the real part and the imaginary part are as follows. The left is the real part and the right is
the imaginary part. As is clear from both figures, there is no zero on the half line x <0 , y =0 .
28.1.1 Zeros of
<1>
z
At the zeros of <1>z , both the real part Re <1>z and the imaginary part Im <1>z have
to be 0. That is, solutions of the following simultaneous equations must be obtained.
Re (x +iy) = 0
<1>
(1.1r)
Im (x +iy) = 0
<1> (1.1i)
Since it is impossible to solve this, we are obliged to use the numerical solution method like the Newton-Raphson
method. Since the function FindRoot [] for it is implemented in formula manipulation software Mathematica,
we use this. However, for that purpose, approximate position of the zeros must be known. In order to know this,
we draw contour plots of (1.1r) and (1.1i) and find the intersection. Since the function ContourPlot [] for it
is implemented in Mathematica, we use this.
- 1-
Fig.1 is -43 x -20 . There is no zero here. Fig.2 is -20 x 3 . There are nine zeros here.
Fig.3 is 3 x -26 . There is no zero here.
When these upper ( y 32 ) contour plots were drawn, the three plots became ghosts. It is difficult to draw
any more by the formula (1.0) and my personal computer & the software. However, the followings may be said.
(i) There seems to be no zero in the upper part of Fig.1 and Fig.3.
(iI) There seems to be innumerous zeros in the upper part of Fig.2.
- 2-
- 3-
28.2 Transition of Non-Trivial Zeros associated with the Super Integral.
In this section, we investigate the transition of non-trivial zeros associated with the Super Integral.
<0> <1>
28.2.1 Non-Trivial Zeros of z z 12.5 y 32
z
<0> <1>
Here, we investigate the transition of the zeros of z in 12.5 y 32 . The interval
of x is set to -3 x 2 . When the zeros are traced with 5 contour plots, it is as follows. The blue line is
the real part, the orange line is the imaginary part, and the red dot is a zero.
In Fig.1, the non-trivial zeros of <0>z are drawn. In Fig.2, the zeros of <0.2>z are drawn. Looking at
these figures, we can see that the zeros are moving to the left with increasing p . And finally they reach Fig.5.
If Fig.5 and Fig.1 are shown numerically, it is as follows.
- 4-
-2.3007 + i 16.0854 1/2 + i 14.1347
-2.2574 + i 22.0090 1/2 + i 21.0220
-1.9080 + i 26.4561 1/2 + i 25.0108
-2.0862 + i 30.8578 1/2 + i 30.4248
Summary
The non-trivial zeros of z move to the upper left on the complex plane with increasing p .
- 5-
28.3 Transition of Trivial Zeros associated with the Super Integral.
In this section, we investigate the transition of trivial zeros associated with the Super Integral using formula (2.0) .
z
<0> <1>
Here, we investigate the transition of the trivial zero z = -2 at z . The interval of x is set
to -2.25 x 2.25 . When the zero is traced with nine contour plots, it is as follows. The blue line is the real
part, the orange line is the imaginary part, and the red dot is a zero.
-6-
After this, the zero point moves to the right with increasing p , and reaches 1.6690082124782 at 1.
-7-
In 0 , (z) 's trivial 8 zeros are drawn. -4 , -8 , -12 , -16 are drawn with a red dot and -6 , -10 , -14 , -18
are drawn with a purple dot.
Although all zeros disappear at the moment of p > 0 , among them, -4 , -8 , -12 , -16 revive in the order near
the origin with increasing p , and rises on the complex plane. On the contrary, -6 , -10 , -14 , -18 do not revive.
0.01 , 0.2 , 0.445 show the transitions. After
0.5 , these zeros rise further in the complex plane.
Although partner changes of the contour plots are also observed In 0.445 0.972 , the four zeros finally
reach 1 . The results of these transitions are numerically shown as follows.
-4 -5.2076 + i 6.7222
-8 -10.4916 + i 4.6111
-12 -14.1483 + i 2.6351
-16 -17.1410 + i 0.7711
Summary
(z) 's trivial zeros -2 , -4 , -6 , p(z) p = 0 1 ,
< >
Investigating the transitions of associated
we obtained the following results.
(1) (z) 's zero -2 transmigrates to the immediate right of singular point 1 at the moment of p >0 ,
and moves to the right on the x axis with increasing p
(2) Although (z) 's trivial zeros -4 , -6 , -8 , disappear at the moment of p >0 , among them, zeros
-4 , -8 , -12 , -16 revive in the order near the origin with increasing p , and move to the upper left on the
complex plane. In addition, zeros -20 , -24 , -28 , can not revive at 0<p 1 .
2018.12.28
Kano Kono
Alien's Mathematics
-8-