# Simpler way to solve this integral?

#### nacho

##### Active member
This problem came up in worksheet for a real analysis class, so I can only assume they want us to use a method from real analysis.

It would be pretty rigorous to solve otherwise, could anyone point me in the correct direction?

Or do they simply want us to test our differentiating skills?

#### Attachments

• 8.1 KB Views: 36

#### chisigma

##### Well-known member
This problem came up in worksheet for a real analysis class, so I can only assume they want us to use a method from real analysis.

It would be pretty rigorous to solve otherwise, could anyone point me in the correct direction?

Or do they simply want us to test our differentiating skills?
Take into account that...

$\displaystyle \frac{x^{2}}{(x^{2}+9)\ (x^{2}+4)^{2}} = \frac{\frac{9}{25}}{x^{2}+ 9} - \frac{\frac{9}{25}}{x^{2}+4} + \frac{\frac{4}{5}}{(x^{2} + 4)^{2}}$

Kind regards

$\chi$ $\sigma$

#### Ackbach

##### Indicium Physicus
Staff member
This problem came up in worksheet for a real analysis class, so I can only assume they want us to use a method from real analysis.

It would be pretty rigorous to solve otherwise, could anyone point me in the correct direction?

Or do they simply want us to test our differentiating skills?
Do you need to prove that the integral converges?

Take into account that...

$\displaystyle \frac{x^{2}}{(x^{2}+9)\ (x^{2}+4)^{2}} = \frac{\frac{9}{25}}{x^{2}+ 9} - \frac{\frac{9}{25}}{x^{2}+4} + \frac{\frac{4}{5}}{(x^{2} + 4)^{2}}$

Kind regards

$\chi$ $\sigma$
Excellent! That's where my mind went, too.

#### nacho

##### Active member
Do you need to prove that the integral converges?

Excellent! That's where my mind went, too.
Thanks for the response sigma,

Uh, as far as I am sure, no we do not. Although, briefly, what would one need to do in order to find convergence for an integral.

Thanks again!

#### Ackbach

##### Indicium Physicus
Staff member
Thanks for the response sigma,

Uh, as far as I am sure, no we do not. Although, briefly, what would one need to do in order to find convergence for an integral.

Thanks again!
Well, with this integral, you could probably get away with noting two things: 1. The denominator is always strictly positive, and bounded away from zero. Hence, the lower limit of the region of integration will give you no problems. 2. Furthermore, the integrand behaves like $1/x^4$ for large $x$, and hence goes to zero faster than $1/x^2$. Therefore, the upper limit of the region of integration will give you no problems. This is a hand-waving style of argument, but you could probably use it to give you an outline of the proof.

#### Prove It

##### Well-known member
MHB Math Helper
Thanks for the response sigma,

Uh, as far as I am sure, no we do not. Although, briefly, what would one need to do in order to find convergence for an integral.

Thanks again!
If you are unsure if an improper integral converges, you CAN always just evaluate the integral and see if you get a numerical answer. In this case you will

#### nacho

##### Active member
Haha, well there's that.

Ok so after simplifying the integral, I would have to apply some use Cauchy Residue theorem?

#### Prove It

##### Well-known member
MHB Math Helper
Haha, well there's that.

Ok so after simplifying the integral, I would have to apply some use Cauchy Residue theorem?
No, an indefinite integral can be found. Substitute \displaystyle \begin{align*} x = 3\tan{(\theta)} \end{align*} into the first term, and \displaystyle \begin{align*} x = 2\tan{(\theta)} \end{align*} into the others.

#### Bacterius

##### Well-known member
MHB Math Helper
If you are unsure if an improper integral converges, you CAN always just evaluate the integral and see if you get a numerical answer. In this case you will
[JUSTIFY]Just as a quick off-topic question, does this always work? Or are there integrals which do not converge yet give bogus numerical results if you go ahead and still try to numerically evaluate them? I'm asking because I know that messing with infinite series without knowing if they converge is recipe for disaster, not sure if it's the same for integrals..

Might be a stupid question but I've learned to never take anything for granted [/JUSTIFY]

Last edited:

#### Prove It

##### Well-known member
MHB Math Helper
[JUSTIFY]Just as a quick off-topic question, does this always work? Or are there integrals which do not converge yet give bogus numerical results if you go ahead and still try to numerically evaluate them? I'm asking because I know that messing with infinite series without knowing if they converge is recipe for disaster, not sure if it's the same for integrals..[/JUSTIFY]
I don't see any reason why this method shouldn't work. If the integral doesn't converge, then why would it give a number for its value? I could be wrong though...

Just for clarification as I don't know if I was clear enough, when I said see if you get a numerical answer, I meant see if you get a number, not if you can integrate it using a numerical method...

#### ZaidAlyafey

##### Well-known member
MHB Math Helper
[JUSTIFY]Just as a quick off-topic question, does this always work? Or are there integrals which do not converge yet give bogus numerical results if you go ahead and still try to numerically evaluate them? I'm asking because I know that messing with infinite series without knowing if they converge is recipe for disaster, not sure if it's the same for integrals..

Might be a stupid question but I've learned to never take anything for granted [/JUSTIFY]
Generally if you follow the correct steps that will work even for series but most of the time you need to test the uniform convergence in an intermediate step . For series I think you are talking about absolute convergence and rearrangement theorem . Assuming an integral or series converge only tells us that the problem has an analytic value .Solving a divergent integral or series should not result in an analytic value unless we are making a mistake somewhere . Of course we first need to define what we mean by divergent and in what sense .

#### Bacterius

##### Well-known member
MHB Math Helper
I don't see any reason why this method shouldn't work. If the integral doesn't converge, then why would it give a number for its value? I could be wrong though...

Just for clarification as I don't know if I was clear enough, when I said see if you get a numerical answer, I meant see if you get a number, not if you can integrate it using a numerical method...
No sorry this is indeed what I meant, e.g. integrate say an improper integral such as $\int_{-1}^1 \frac{1}{x} ~ \mathrm{d}{x}$ and get, say, $2.5$ as a result even though the integral does not converge. I thought maybe there were exotic integrals which could give misleading results like these if one was not careful enough about establishing convergence.

Generally if you follow the correct steps that will work even for series but most of the time you need to test the uniform convergence in an intermediate step . For series I think you are talking about absolute convergence and rearrangement theorem . Assuming an integral or series converge only tells us that the problem has an analytic value .Solving a divergent integral or series should not result in an analytic value unless we are making a mistake somewhere . Of course we first need to define what we mean by divergent and in what sense .
Ah, that makes sense. Yes, I was thinking about the rearrangement theorem but there are probably many such examples.

Last edited: