Beruflich Dokumente
Kultur Dokumente
∞
∞ X
X 1 1 1 1
= +
K=1 L=1
x − Kπ x − Lπ x − Kπ x + Lπ
1 1 1 1
+ +
x + Kπ x − Lπ x + Kπ x + Lπ
(Now, separate out the terms where K = L.)
∞
X 1 2 1
= 2
+ + 2
k=1
(x − kπ) (x − kπ)(x + kπ) (x + kπ)
∞ X ∞
X 1 1 1 1
+ +
K=1 L=1
x − Kπ x − Lπ x − Kπ x + Lπ
K6=L 1 1 1 1
+ +
x + Kπ x − Lπ x + Kπ x + Lπ
∞ ∞
2
X 1 X 4+2
f (x) − −
k=−∞
(x − kπ)2 k=1
x 2 − k2π 2
M M !
X X 1 1 4K
= lim +
M →∞
K=1 L=1
K−L K+L x2 − K 2 π 2
L6=K
M
X 1 1
Note specifically that the sum + equals
L=1,L6=K
K−L K+L
M M M +K M −K
X 1 X 1 X 1 X 1
− = −
L=1
L+K L=1
L−K L=K+1
L L=1−K
L
L6=K L6=K L6=2K L6=0
2K−1 M +K −1 M −K
X 1 X 1 X 1 X 1
= + − −
L=K+1
L L=2K+1
L L=1−K
L L=1
L
2K−1 M +K K−1 M −K
X 1 X 1 X 1 X 1 P
= + + − L −L in the third
L=K+1
L L=2K+1
L L=1
L L=1
L
2K M +K K M −K
X 1 X 1 X 1 X 1 1 1
= + + − − −
L=K+1
L L=2K+1
L L=1
L L=1
L 2K K
M +K M −K M +K
X 1 X 1 3 X 1 3
= − − = − .
L=1
L L=1
L 2K L=M −K+1
L 2K
∞
2 0
X 6
f (x) − f (x) − 2 − k2π 2
k=1
x
M M !
X X 1 1 4K
= lim +
M →∞
K=1 L=1
K−L K+L x2 − K 2 π 2
L6=K
M M +K !
X X 1 3 4K
= lim −
M →∞
K=1 L=M −K+1
L 2K x2 − K 2 π 2
M M +K !
X X 1 4K
= lim
M →∞
K=1 L=M −K+1
L x2 − K 2 π 2
M !
X 3 4K
+ lim −
M →∞
K=1
2K x2 − K 2 π 2
∞
2 0
X 6
f (x) − f (x) − 2 − k2π 2
k=1
x
M M +K
X X 1 4K 4K 4K
= lim − + +
M →∞
K=1 L=M −K+1
L K 2π 2 K 2π 2 x2 − K 2 π 2
M
X −6
+ lim 2 − K 2π 2
M →∞
K=1
x
M M +K 2
X X 1 4K 4Kx
= lim − +
M →∞
K=1 L=M −K+1
L K 2π 2 K 2π 2(x2 − K 2π 2)
∞
X 6
− (which cancels above)
K=1
x2 − K 2π 2
We have, after cancellation,
f 2(x) − f 0(x) !
M M +K
X X 1 4K 4Kx2
= lim − 2 2+ 2 2 2 2π 2)
M →∞
K=1 L=M −K+1
L K π K π (x − K
Why?
C equals 1.
tan (x − D) → ±∞ when, and only when,
x approaches the numbers D ± π2 , D ± 3π
2
, D± 5π
2
,...
∞
1 X 1 1
f (x) = + + →∞
x k=1
x − kπ x + kπ
when, and only when,
x approaches the numbers 0, ±π, ±2π, ±3π, . . .
∞
X 1 (−1)n−122n−1π 2n
We have the identities = B2n
k2n (2n)!
for n = 1, 2, . . . k=1
∞
X 1 (−1)n−122n−1π 2n
= B2n
k=1
k2n (2n)!
∞
X 1 (−1)1−122−1π 2 (−1)021π 2 1 π2
= B2 = =
k=1
k2 (2)! 2 6 6
∞
X 1 (−1)2−124−1π 4 (−1)23π 4 −1 π4
= B4 = =
k=1
k4 4! 24 30 90
∞
X 1 (−1)3−126−1π 6 (−1)225π 6 1 π6
= B6 = =
k=1
k6 6! 720 42 945
∞
X 1 (−1)4−128−1π 8 (−1)27π 8 −1 π8
= B8 = =
k=1
k8 8! 40320 30 9450
∞
X 1 (−1)5−129π 10 (−1)429π 10 5 π 10
= B10 = =
k=1
k10 10! 362880 66 93555
∞
X 1 1 1 1 1 1 1
1< =1+ + + + + + + ...
k=1
k2n 22n 32n 42n 52n 62n 72n
1 1 1 1 1 1
<1+ + + + + + + ...
22n 22n 42n 42n 42n 42n
2 4 8
=1+ + + + ...
22n 42n 82n
1 1 1 1
=1+ + + + ... =
22n−1 42n−1 82n−1 1
1−
22n−1
∞
X 1 1
yields both the inequalities 1 < < ≤ 2,
k2n 1
k=1 1−
∞
22n−1
X 1
and the limit lim = 1.
n→∞
k=1
k2n
For the Bernoulli numbers the corresponding inequalities are
(−1)n−122n−1π 2n 1
1< B2n < ≤2
(2n)! 1
1−
22n−1
(2n)! (2n)!
< B2n| = (−1)n−1B2n < (sharper)
22n−1π 2n π 2n 22n−1 − 1
(2n)! 4(2n)!
< B2n| = (−1)n−1B2n < (simpler)
22n−1π 2n 22nπ 2n
n−1 (2n)!
with the asymptotic formulas B2n ∼ (−1)
22n−1π 2n
2n+ 12
n−1 4n
and B2n ∼ (−1) 1
e2nπ 2n− 2
∞
X Bn(x)
Now we shall consider the function g(x, t) = tn .
n=0
n!
∞ ∞ ∞
d d X Bn(x) X d Bn(x) X B 0 (x)
n
g(x, t) = tn = tn = tn
dx dx n=0 n! n=0
dx n! n=0
n!
∞ ∞
X nBn−1(x) n
X nBn−1(x)
= t = tn
n=0
n! n=1
n(n − 1)!
∞ ∞ ∞
X Bn−1(x) X Bk (x) X Bn(x)
= tn = tk+1 = tn+1
n=1
(n − 1)! k=0
k! n=0
n!
∞
X Bn(x)
=t tn = tg(x, t), for each value of t.
n=0
n!
d
For each constant value of t, we have g(x, t) = tg(x, t)
dx
d −xt d −xt d −xt
and g(x, t)e = g(x, t) e + g(x, t) e
dx dx dx
= tg(x, t) e−xt + g(x, t)te−xt
= 0.
As a function of x, g(x, t)e−xt must be a constant,
which depends on the value of t:
g(x, t)e−xt equals C(t) for some function C.
∞
X Bn(x)
We now have tn = g(x, t) equal to C(t)ext,
n=0
n!
for all x and t satisfying |t| < 2π.
∞
X Bn(x)
Starting from C(t)ext = tn ,
n=0
n!
we keep t constant and integrate these functions of x:
Z 1 Z 1 X ∞
xt Bn(x) n
C(t)e dx = t dx,
x=0 x=0 n=0 n!
Z 1 ∞ R1
x=0 Bn(x) dx n
X
xt
C(t) e dx = t ,
x=0 n=0
n!
xt 1
R1
e x=0 B0(x) dx 0
C(t) = t ,
t x=0 0!
et − e 0
C(t) = 1,
t
t
C(t) = , for 0 < |t| < 2π.
et −1
A Generating Function for Bernoulli Polynomials
∞
X Bn(x) text
For 0 < |t| < 2π, tn =
n=0
n! et − 1
∞ ∞
X B̂n(x) X Bn({x}) te{x}t
For 0 < |t| < 2π, tn = tn =
n=0
n! n=0
n! et − 1
∞
X Bn n t
For 0 < |t| < 2π, t =
n=0
n! et − 1
∞ ∞
d X Bn(x) X d Bn(x)
To justify the tn = tn step above,
dx n=0 n! n=0
dx n!
check the pattern
g(u) = g(x) + g 0(x)(u − x) + o(u − x):
Bn(u) Bn(x) Bn0 (x) Bn(2)(cn)(u − x)2
= − (u − x) + ,
n! n! n! 2n!
Bn(u) Bn(x) Bn−1(x) Bn−2(cn)(u − x)2
= − (u−x) + ,
n! n! (n − 1)! 2(n − 2)!
where each cn lies between u and x.
∞ ∞ ∞
X Bn(u) n X Bn(x) n X Bn−1(x)
t = t − (u − x)tn
n=0
n! n=0
n! n=0
(n − 1)!
∞
1 X Bn−2(cn) n
+ t (u − x)2.
2 n=0 (n − 2)!
∞
1 X Bn−2(cn)
tn(u − x)2 will be o(u − x)
2 n=0 (n − 2)!
∞
X Bn−2(cn) n
if t converges,
n=0
(n − 2)!
and it does, for 0 ≤ x ≤ 1 and 0 ≤ u ≤ 1,
since it converges absolutely there:
∞
X Bn−2(cn) n
t converges by the root test:
(n − 2)!
n=0 s
n
Bn−2(cn) n |t|
for even n, t ∼ < 1.
(n − 2)! 2π