Integration by parts is a technique for performing indefinite integration or definite integration
by expanding the differential of a product of functions
and expressing the original integral in terms of a known integral
. A single integration by parts starts with
|
(1)
|
and integrates both sides,
|
(2)
|
Rearranging gives
|
(3)
|
For example, consider the integral and let
|
(4)
| |
|
(5)
|
so integration by parts gives
|
(6)
| |||
|
(7)
|
where
is a constant of integration.
The procedure does not always succeed, since some choices of may lead to more complicated integrals than the original.
For example, consider again the integral
and let
|
(8)
|
giving
|
(9)
| |||
|
(10)
|
which is more difficult than the original (Apostol 1967, pp. 218-219).
Integration by parts may also fail because it leads back to the original integral. For example, consider and let
|
(11)
|
then
|
(12)
|
which is same integral as the original (Apostol 1967, p. 219).
The analogous procedure works for definite integration by parts, so
|
(13)
|
where .
Integration by parts can also be applied times to
:
|
(14)
|
Therefore,
|
(15)
|
But
|
(16)
|
|
(17)
|
so
|
(18)
|
Now consider this in the slightly different form . Integrate by parts a first time
|
(19)
|
so
|
(20)
|
Now integrate by parts a second time,
|
(21)
|
so
|
(22)
|
Repeating a third time,
|
(23)
|
Therefore, after
applications,
|
(24)
|
If
(e.g., for an
th
degree polynomial), the last term is 0, so the sum
terminates after
terms and
|
(25)
|