Beruflich Dokumente
Kultur Dokumente
Mat Hunt
October 10, 2011
Abstract
This set of notes provides an introduction to integration, what it is
and some of the techniques that are used. The numerical evaluation
of the integral via the trapezium rule. The level of material will cover
up to A-level.
1
1.1
Suppose we have a curve y = f (x) and we want to find the area under the
graph from a point x = a to a point x = b. One way that we can possibly
can be calculated. The thinner the strip then the more accurate the answer
for the area. Say we split up the interval [a, b] into n pieces each of length
(b a)/n, the points of the interval are
ba
xi = a +
i
(1)
n
The set of points a = x0 < x1 < x2 < . . . < xn1 < xn = b is called a
partition of the interval [a, b]. To compute the area choose strips that lie
completely within the area: Mathematically we are choosing the minimum
n1
baX
min(f (xi ), f (xi+1 ))
n
i=0
i=0
(2)
This will be called the lower sum for the area. Note that the term xi+1 xi is
used for the interval (b a)/n because in general the partition can be uneven
in the widths of the strips xi+1 xi , we have made then even for convenience
only. Likewise we can overestimate the area under the curve by choosing the
maximum of f (xi ) and f (xi+1 ) for a strip starting at xi The area under the
AL (n) =
n1
X
baX
max(f (xi ), f (xi+1 ))
AU (n) =
max(f (xi ), f (xi+1 ))(xi+1 xi ) =
n
i=0
i=0
(3)
This will be called the upper sum for the area. The approximations will be
valid if we can find an n > N for which |AU (n) AL (n)| < , this is called
the Riemann criterion. The Riemann criterion basically states that if the
difference of the upper and lower limits is vanishingly small as n gets larger
and larger, then the approximations we made are valid for computing the
area under the curve. Once we know that the Riemann criterion holds then
we can take the limit as n in either the upper or lower limits to obtain
the area.
Example. Take the equation y = x and we will compute the area under
the curve from the point x = 0 to x = a. The interval [0, a] would be split
up into n different points, each of length a/n, the points in the partition are
ai/n. Then computing the lower limit of the area shows that:
n1
n1
a X ai
a2 X
AL (n) =
i
= 2
n i=0 n
n i=0
Now:
n1
X
i=0
i=
n(n 1)
2
(4)
Then:
a2
AL (n) =
2
1
1
n
(5)
(6)
Now geometrically AU (n) > AL (n) and so |AU (n)AL (n)| = AU (n)AL (n),
so computing this condition shows that:
1
a2
1
a2
a2
1+
1
=
AU (n) AL (n) =
2
n
2
n
n
So taking N > a2 / will ensure that the the Riemann criterion will hold and
the approximations for the area under the curve is valid and we may take the
limit as n to obtain the exact area under the curve to obtain the area
to be a2 /2 which is geometrically intuitive because we computed the area of
a right angled triangle which is just half the area of a square with length of
sides a.
Example. Take the function y = x2 and as before we will compute the area
under the graph from x = 0 to x = a. The partition of the interval [0, a] will
be divided equally into n pieces, each of length a/n and the points of the
partition will be ai/n. Computing the lower sum shows that:
n1
aX
AL (n) =
n i=0
ai
n
2
n1
a3 X 2
i
= 3
n i=0
(7)
i2 =
n(n 1)(2n 1)
6
shows that:
a3 n(n 1)(2n 1)
a3
AL (n) = 3
=
n
6
6
3
1
2 + 2
n n
(8)
(9)
(10)
So this can be made less that any given by taking N > a3 /. So the
Riemann criterion is satisfied and we may take the limit as n to obtain
the area as a3 /3.
Definition The upper integral from a point x = a and x = b of a function
y = f (x) is denoted as:
Z b
f (x)dx
(11)
a
(12)
i=0
(14)
i=0
If the Riemann criterion holds then the upper integral is the same as lower
integral and we simply call it the integral and we write it as:
Z b
f (x)dx
(15)
a
Geometrically the integral represents the area under the curve from the points
x = a to the point x = b. The point a is called the lower limit and the point
b is called the upper limit. Functions which can be integrated are called
integrable. The function f (x) inside the integral is called the integrand. It
should be noted that x in the integral is what is known as a dummy variable
and any variable can be used. It is good practice to use a different letter to
represent the dummy variable than what appears in the limits.
Integration is the act of performing an integral and along with differentiation
makes up a subject called calculus. Integration is sometimes called integral
calculus
5
1.2
Basic Properties
(16)
(17)
(18)
It is easy to see why the first of these properties holds. The geometric
interpretation of an integral is the area under the graph, so the first of the
properties ask about the area under the graph from x = a to x = a and
this is 0. For the second statement, known as the domain splitting property,
consider the following illustration: From the above diagram it can be seen
and so:
b
a
f (x)dx =
f (x)dx
b
Although the above properties may at first sight seem somewhat useless, they
are however a very useful set of properties.
2
2.1
F (x) =
f (u)dx
(19)
The geometric picture for the fundamental theorem of calculus is given by:
The quotient for the derivative is given by:
F (y) F (x)
=
yx
=
=
f (u)du
f (u)du
yx
Z
f (u)du +
1
yx
7
yx
f (u)du
f (u)du
Now the task is to compute the integral. The way to compute the integral
So the function will either be increasing or decreasing on the interval [x, y].
The lower sum is simply given by:
AL = f (x)(y x)
1
f (x) 6
yx
f (u)du 6 f (y).
x
(20)
F (y) F (x)
6 f (y).
yx
(21)
(22)
(23)
(25)
so h(x) and f (x) will differ by a constant (as both h(x) and f (x) are antiderivatives for f (x)), h(x) f (x) = C, evaluating this at x = a shows that
9
b
a
(26)
The square brackets is just notation for taking the difference. So what we
have shown here is that differentiation and integration are truly inverse processes of each other, this very important result is known as the fundamental
theorem of calculus. The fundamental theorem of calculus allows is to
immediately write down a whole load of integrals without having to resort to
computing the upper and lower sums. It is common practice to call integrals
of the form:
Z b
F (x) =
f (x)dx
(27)
a
(28)
where C is an arbitrary constant (remember antiderivatives of the same function differ by a constant). As an example examine the function:
f (x) =
1
xn+1
n+1
2.2
(29)
Given two functions f (x) and g(x) with primitives F (x) and G(x) respectively and let and be two numbers, then the following result holds for
the integral:
Z
f (x) + g(x)dx =
f (x) +
a
g(x)dx
(30)
This very important property is called linearity. to show this define H(x) =
F (x) + G(x), differentiate this to obtain:
H (x) = f (x) + g(x)
10
(31)
= H(b) H(a)
= F (b) + G(b) F (a) G(a)
= (F (b) F (b)) + (G(b) G(a))
Z b
Z b
=
f (x)dx +
g(x)dx
a
Combining these results give the result sought after. Although this doesnt
seem all that useful at first sight, it is indeed one of the most important
properties that the integral has, for example we can evaluate the following
integral:
Z
Z
Z
x3
2
3
2
x + 4x dx = x dx + 4 x3 dx =
+ x4 + C
3
There are numerous examples of when linearity is useful in calculus.
2.3
Negative Area
=
4 a
4
4
4
a
This gives a negative result, which is odd as previously we have given the
geometrical interpretation as the area below the graph and the y = 0 axis
which is positive. However when we pot the graph of y = x3 we notice that
the the values of f (x) = x3 are negative. As f (x) = x3 is an odd function, it
is possible to find the area under the graph between x = 0 and x = a as:
4 a
Z a
a4 04
a4
x
3
x dx =
=
=
4 0
4
4
4
0
So from this calculation we can see that this represents the area under the
graph as it gives a positive result. So the geometrical interpretation of the
11
f (x)dx = 0
(32)
f (x)dx = 2
3.1
f (x)dx
(33)
This section we show how the standard results for integration may be obtained, one was covered earlier:
Z
xn dx =
xn+1
+ C,
n+1
3.2
Trigonometric Functions
The integral of trigonometric functions are found this way, note that:
d 1
1
d
cos ax = sin ax
sin ax = cos ax
dx a
dx
a
and so we can say:
Z
1
sin axdx = cos ax + C
a
cos axdx =
1
sin ax + C.
a
(34)
(35)
d
sinn+1 x = (n + 1) sinn x cos x,
dx
and therefore:
Z
Z
1
sinn+1 x + C ,
n+1
1
cosn+1 x + C
n+1
The above results are useful for computing the integrals of odd powers of sin x
and cos x. As an example of the method lets integrate f (x) = sin3 x. now
sin3 x = sin2 x sin x and using the trigonometric identity sin2 x + cos2 x 1
in the form of sin2 x = 1 cos2 x shows that sin3 x = (1 cos2 x) sin x and
so:
Z
Z
3
sin xdx =
(1 cos2 x) sin xdx
Z
Z
=
sin xdx cos2 x sin xdx
= cos x +
1
cos3 x + C
3
The same idea can be used for integrating odd powers of cos x.
13
3.2.2
There is no general formula for this but there is a general method, the method
us based upon the expression for cos 2x, as:
cos 2x = 2 cos2 x 1
So, integrate f (x) = cos4 x. Note cos4 x = (cos2 x)2 = ((1 + cos 2x)/2)2 ,
expanding this out completely shows that:
cos4 x =
1 + 2 cos 2x + cos2 2x
4
1 cos 2x 1 + cos 4x
+
+
4
2
8
The result can now be integrated using the standard rules given before.
For even powers of sin x, turn sin2 x into 1 cos2 x any apply the technique
above.
3.2.3
Integration of tann x
1
tann+1 x + C.
n+1
Use tann x = tan2 x tann2 x = tann2 x(sec2 x 1) and we are lead to the
relation:
Z
Z
1
n1
n
tan
x tann2 xdx
tan xdx =
(36)
n1
Equations of type (36) are called recurrence relations. When n = 2 we have:
Z
Z
2
tan xdx = sec2 x 1dx = tan x x + C
This completes all the integration for tann x.
14
3.3
d
dx
so:
Z
Likewise:
1 ax
e
a
= eax
1
eax dx = eax + C.
a
d
dx
and so:
1 x
a
ln a
ax dx =
(37)
= ax
1 x
a +C
ln a
(38)
15
1
dx
a x
should have a value as the area above the curve us well defined geometrically,
to get a way around this multiple both denominator and numerator by 1
to get:
Z b
Z b
1
1
dx =
d(x) = [ln(x)]ba ,
a x
a x
for values of x < 0. So we can see that in general that:
Z
1
dx = ln |x| + C
(40)
x
Z
Generalising this result slightly, note that if f (x) = ln(f (x)), then:
f (x) =
f (x)
,
f (x)
(41)
and so:
Z
f (x)
dx = ln |f (x)| + C.
f (x)
(42)
dx
1+x 1+x
Z
1
dx
=
1
1+x
= x ln |1 + x| + C.
16
The trick here was to add and subtract 1 in the numerator. Another application of this in integrating tan x, note that cos x = sin x and so:
Z
tan xdx = ln | cos x| + C.
Integration by Parts
(43)
d
(uv)dx =
dx
b
a
du
v dx +
dx
dv
dx
dx
(44)
Using the fundamental theorem of calculus on the LHS of (44) shows that:
Z
d
(uv)dx = [u(x)v(x)]ba
dx
(45)
du
v dx +
dx
u
a
dv
dx.
dx
(46)
Taking the first term in the RHS over to the LHS gives the equation:
Z
dv
u dx = [u(x)v(x)]ba
dx
v
a
du
dx
dx
(47)
This is the expression for integration by parts. The indefinite integral version
of this expression is:
Z
dv
u dx = u(x)v(x)
dx
du
dx
dx
(48)
v = ex
2
2
which is a more complicated than the original integral.
Example. Find the indefinite integral of f (x) = ex sin x, so we compute:
Z
ex sin xdx
For this let u(x) = ex and dv/dx = sin x, then:
du
= ex ,
dx
v(x) = cos x
(49)
ex cos xdx
v = sin x
(50)
Upon re-arranging
Z
ex sin xdx =
ex
(sin x cos x) + C
2
du
1
= ,
dx
x
v=x
Integration by Substitution
5.1
Basic Idea
Define:
F (x) =
f (u)du.
The next step is to examine (F (g(x))) which by the chain rule is:
(F (g(x))) = F (g(x))g (x) = f (g(x))g (x)
(51)
(52)
Now:
F (g(b)) F (g(a)) =
g(b)
c
f (u)du
g(a)
f (u)du =
g(b)
f (u)du
f (g(x))g (x)dx =
f (u)du
a
(53)
g(a)
(54)
g(a)
The way that integration by substitution usually works is that we are given
an integral usually1 in the form of the RHS of (54) and our task to to come
up with a substitution u = g(x) that will simplify the integration.
1
19
5.2
Examples
x
0
du
=
1 + u2
tan1 x
0
sec2 v
dv =
1 + tan2 v
Example. Evaluate:
tan1 x
1dv = [v]0tan
= tan1 x
2x
dx
1 + x2
Use g(x)
= 1 + x , then g (x) = 2x, the limits transform as x = 0 g = 0
and x = 3 g = 4, then inserting these values in (54) shows that:
Z
2x
dx =
1 + x2
4
dg
dg = [2 g]0 = 2 4 2 0 = 4
g
Example. Evaluate:
du
b2 u2
du
=
2
b u2
sin1 (a/b)
0
b cos x
dx =
b cos x
Example. Evaluate:
Z
dx
1+ x
20
sin1 (a/b)
dx = sin1 (a/b)
Let x = u2 then
x (u) = 2u and the limits become x = 0 u = 0 and
x = a u = a, then inserting the values into the LHS of (54) shows that:
Z a
Z a
u
dx
= 2
du
1+u
x
0
0 1+
Z a
1+u1
du
= 2
1+u
0
Z a
Z a
du
du
= 2
1+u
0
0
= [u ln |1 + u|]0 a
a ln(1 + a)
=
Example. Evaluate:
Z
x2 x3 + 1dx
Let g(x) = x + 1 then g (x) = 3x2 , with the limits g(0) = 1 and g(1) = 2,
The integral becomes transformed as:
Z
Z 1
1 2
2
2
3
udu = (2 2 1)
x x + 1dx =
3 1
9
0
This section is all about the one off techniques which can be useful in certain
situations
6.1
dx
a (x + )(x + )
The idea is to use partial fractions to decompose into a simpler integrand.
So using partial fractions the integral becomes:
1
1
1
1
=
(x + )(x + )
x+ x+
Z
dx =
a x+ x+
x+ a
a (x + )(x + )
21
6.2
can be easily integrated by the use of a specific substitution. The trick here
is to use the substitution:
x
t = tan
2
or in the notation that we are used to:
x = 2 tan1 t
Using the double angle formula for tan(x/2) we see that:
tan x =
2 tan(x/2)
2t
=
.
2
1 tan (x/2)
1 t2
Upon using the identities 1 + tan2 z = sec2 x and 1 + cot2 = cosec2 x we can
write:
1 t2
2t
cos x =
, sin x =
2
1+t
1 + t2
Differentiating the original substitution shows that:
x (t) =
2
1 + t2
Example. Evaluate:
Z
dx
2 + sin x
22
dx
=
2 + sin x
=
=
=
=
=
=
2
1 + t2
dt
2
2
0 2(t + t + 1) 1 + t
Z 1
dt
2
0 t +t+1
Z 1
dt
1 2
0
t + 2 + 43
1
2t + 1
2
1
tan
3
3
0
2
1
1
1
(tan
3 tan
3
3
2
6
3 3
3 3
Z
Trapezium Rule
There is a refinement on how the area under the graph or area above the
graph, that is computing the integral:
Z b
f (x)dx.
a
23
f (x)dx
n
X
Ai
i=0
1X
(f (xi ) + f (xi+1 ))(xi+1 xi )
2 i=0
n
hX
(f (xi ) + f (xi+1 ))
2 i=0
h
(f (x0 ) + f (x1 ) + f (x1 ) + f (x2 ) + + f (xn2 ) +
2
+ f (xn1 ) + f (xn1 ) + f (xn )
!
n1
X
h
f (a) + f (b) + 2
f (xi )
2
i=1
This is the basis of the approximation for the integral. As an example take:
Z 1
x2 dx.
0
The value of this integral is 1/3. Let us take 10 strips in our numerical
scheme, so the partition will be 0 < 0.1 < 0.2 < < 0.9 < 1.0 with
24
b(x)
K(x, t)u(t)dt
(55)
a(x)
The function K(x, t) is called the kernel of the integral equation. In general
integral equations are rather difficult to solve but there are certain cases
where solutions can be found. We take the case where a(x) and b(x) are
constant, g(x) = 1 and K(x, t) = (x)(t). With this (55) becomes:
u(x) = f (x) + (x)
(t)u(t)dt
(56)
(57)
(58)
So this is the solution, all that needs to be done how is to compute the
constant , This is done by inserting (58) into (56) and obtaining an equation
for which can be solved. Doing this shows that:
Z b
u(x) = f (x) + (x)
(t)(f (t) + (t))dt
(59)
a
25
We recognise that f (x) = cos x + 2x, (x) = x and (t) = t and that is
given by:
Z
tu(t)dt
u(x) = cos x + (2 + )x
So calculating shows that:
Z
=
t(cos t + (2 + )t)dt
0
Z
Z
=
t cos tdt + (2 + )
t2 dt
0
0
Z
1
= [t sin t]0
sin tdt + (2 + ) t3
3 0
0
3
= 2 + (2 + )
3
Re-arranging shows that = 2 and the solution is u(x) = cos x.
26