Scribd
Upload
55 views

A Generalization of Integrals

This document presents a method for deriving an infinite series representation for the indefinite integral of a function using repeated applications of integration by parts. Starting with an initial integral, each application of integration by parts generates a new integral involving higher-order derivatives of the function with polynomial weights. The author shows that this recursion yields an infinite series with terms involving successive derivatives divided by factorials. Exceptions are noted for functions whose derivatives cancel the polynomial weights like 1/x. The method is demonstrated on several examples.

Uploaded by

liza
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
55 views

A Generalization of Integrals

This document presents a method for deriving an infinite series representation for the indefinite integral of a function using repeated applications of integration by parts. Starting with an initial integral, each application of integration by parts generates a new integral involving higher-order derivatives of the function with polynomial weights. The author shows that this recursion yields an infinite series with terms involving successive derivatives divided by factorials. Exceptions are noted for functions whose derivatives cancel the polynomial weights like 1/x. The method is demonstrated on several examples.

Uploaded by

liza
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 6

Universidad Interamericana de Puerto Rico - Recinto de Ponce

A Generalization of Integrals by the Formula of Integration by


Parts
By: Jorge E. Mendoza-Guzmn .
Instituto de Educacin y Pedagoga, Universidad del Valle, Cali- Colombia
Abstract
In this short paper I show an unusual route to deduce an infinite series, using the
formula of integration by parts, starting a recurrence and finding a pattern that can be rewritten
as an infinite series.
Keywords Integral Calculus, Integration methods, Infinite Series.
Introduction
In our courses of integral calculus, we learn to calculate the primitive of a function

in

terms of the usual techniques of integration, such as simple substitution, partial fractions,
quadratic factors, trigonometric methods, and integration by parts. Thus, if you show me an
integral, I try to guess the most convenient of those methods to calculate it; but if I consider the
you show me, can I solve it by always using the same method?
As we shall see, the answer for this question is Yes, I do: the solution of

is

almost always possible using the technique provided by the usual formula of integration by
parts1 iterated indefinitely. Assume we start by doing a first round of integration by parts.
Where can we go from there and what should we do to keep iterating indefinitely? What
happens in the long run?
What I did was precisely to repeat the application of the parts formula until I found a
recognizable pattern. Then, the surprise came: I could now calculate a series expansion for any
integral! (Wait, for almost any integral. The innocent-looking exceptions will be dealt with
later).

Procedure
Let be
a function which can be derived repeatedly, i.e., a C-function. Assume we wish to
obtain the indefinite integral of f(x):
(1)
To calculate this integral, let us apply the parts formula:
(2)
Then, as usual, the first step is to define what is to be taken as ,

in the following order:

The parts formula or integration by parts formula can be obtained from Leibnitz formula for the derivation of
a product:
. By inspection, it is clear that the primitive of
or
is
. Using Leibnitz differential notation and
integrating,
, which is often written as
.

Revista 360/ No.8/ 2013

Universidad Interamericana de Puerto Rico - Recinto de Ponce

Then the integral (1), by the parts formula (2), is simply:


(3)
Notice that in the right-hand side of (3) there appears now a new integral, indicated in boldface, which
involves a first derivative with a linear weight x:

Again, let us proceed to calculate this new integral by using the parts formula:

(4)

Again, I noticed that a second new integral appears on the right-hand side of (4), which involves a
second derivative with a quadratic weight:

Thus, I simply repeated the same procedure:

Revista 360/ No.8/ 2013

Universidad Interamericana de Puerto Rico - Recinto de Ponce

3
(5)

When I repeated the same procedure for the integral in boldface in (5), the numbers 2 and 6 gave
me a hint: they are just the well-known factorials 2! and 3!

(6)
It took me only the next step to realize what the recurrent pattern was. Try it! Then, I inserted the
calculated results in equation (3). There, the second term of the right-hand side was calculated in (4),
so I inserted (4) in (3) to get (7), etc.

(7)

(8)

Following the pattern, the integral of

can be written as follows:


(9)

Another way to write the nth term of this series is to consider the integral

The usual notation for

, in general

Revista 360/ No.8/ 2013

Universidad Interamericana de Puerto Rico - Recinto de Ponce

We can consider it as an error term. Note that if this integral is repeatedly performed, (9) is reduced
to:

Where c is the usual constant of integration.


This result might cause some problems in the moment of plugging in specific functions, for
example,
, because it can lead to contradictions when used without care. Indeed, if I am not
careful, I might get 0

, so it is better to consider the equivalent case of the definite integral

between 0 and . So, redefining the last equation we get

This way we no longer need a constant of integration, and we can set

. Expanding,

It is puzzling to see how easily I deduced an infinite series using only this method, which thus
promises a way to find the primitive of any indefinitely differentiable function expressed as an
alternating series of its derivatives with coefficients the powers of x divided by the respective
factorials. Is it true that I can always work this way with any C-function
?
Let us consider the function

; then,

. When we plug this

result in the infinite series, the denominators will cancel with the power weights:

This is the divergent harmonic series, which excludes from our treatment the function
with its powers and its linear variants, since they are the functions that force all variable terms to
become constant in the power series (9), because of the cancellation of the powers
with the
derivatives in the term

In general, all the inverse powers suffer the same problem

because the derivatives are all unbounded near zero.


So for

Revista 360/ No.8/ 2013

Universidad Interamericana de Puerto Rico - Recinto de Ponce

On the other hand, it is easy to work when the functions are polynomials or exponentials; for
example if
, then, if
,

In this case I easily got the primitive for

using only the derivatives, but in those cases

involving square roots or higher roots, the calculation of the derivatives becomes more and more
complicated when we attempt to directly calculate the expression for the primitive. Part of the fun is
to work out the limitations of the universe of functions that can be safely subjected to this treatment.
As a final particular case, consider the following well-known function

, whose nth

derivative is
.

If we rewrite

as

, we can directly check that the last equation is correct, because we are

simply killing the first term (+1) and shifting indices in the usual expansion for
case

. For the special

, this result is the usual expansion for the number e.

One could, of course, assume that the Taylor series expansion exists, and integrate term by
term to compare with the above formula, but the surprising detour through the parts formula yields
new insights into the relationship between derivatives and integrals and allows a new hold of the error
terms.

Revista 360/ No.8/ 2013

Universidad Interamericana de Puerto Rico - Recinto de Ponce

References
[1] Stewart, J. (2007). Calculus (6th edition). St. Paul, MN: Brooks Cole Publ. Co.

Jorge E. Mendoza-Guzman [email protected] Instituto de Educacin y


Pedagoga, Universidad del Valle, Cali- Colombia.

Revista 360/ No.8/ 2013

You might also like