Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

How to calculate the principal part of this improper integral via contour integration?

\begin{equation} P\int_{0}^{+\infty}\frac{dx}{x^2+x-2} \end{equation}

I have seen some examples where you integrate along a semicircle and then take the limit $R\to\infty$, where $R$ is the radius. But here the integrand is not a even function and in addition there are poles on the real line (that's why is divergent)... Any help is appreciated.

share|improve this question
    
Does it have to be contour integration? –  Fabian Feb 18 '12 at 13:10
    
Of course not! what matters is the final answer. But it should be possible via contour integration... –  quark1245 Feb 18 '12 at 14:04
1  
In my opinion, it's not the final answer that matters, it's the process of reasoning that leads to the final answer that matters. –  David Mitra Feb 18 '12 at 14:33
1  
I simply meant that a problem can be solved in different ways, following different lines of thought and you don't have to be single-minded about a particular way to tackle it if it doesn't work... anyway, I agree that in an exercise like this nobody cares if the final digit is wrong or not... –  quark1245 Feb 18 '12 at 14:46

1 Answer 1

up vote 3 down vote accepted

Let's factorize the denominator : $x^2+x-2=(x+2)(x-1)$

so that $\displaystyle \frac1{x^2+x-2}=\frac13\left(\frac1{x-1}-\frac1{x+2}\right)$

Let's try a direct proof without complex integration. $$P.V. \int_0^{\infty} \frac 1{x^2+x-2}=\frac13 P.V.\int_0^{\infty} \frac1{x-1}-\frac1{x+2} dx=$$ $$ =\frac13 \lim_{\epsilon\to 0}\left[\int_0^{1-\epsilon} \frac1{x-1}-\frac1{x+2} dx+\int_{1+\epsilon}^{\infty} \frac1{x-1}-\frac1{x+2} dx\right] $$

$$ =\frac13 \lim_{\epsilon\to 0}\left[ \left[\log(1-x)-\log(x+2)\right]_0^{1-\epsilon}+ \left[\log(x-1)-\log(x+2)\right]_{1+\epsilon}^{\infty}\right] $$ $$ =\frac13 \lim_{\epsilon\to 0}\left[\log(\epsilon)-\log(3-\epsilon)+\log(2)-\log(\epsilon)+\log(3+\epsilon)\right] $$ (using $\lim_{R\to \infty}\log\left(\frac{R-1}{R+2}\right)=0$) $$ =\frac{\log(2)}3 $$

share|improve this answer
    
Thanks for your help! I believe it's all right but, in general, do you know about some special technique to handle contour integrals with poles on the real axis? –  quark1245 Feb 18 '12 at 14:34
1  
@quark1245: when simple poles are on the integration line we just have to divide the residue by $2$. –  Raymond Manzoni Feb 18 '12 at 14:40
    
Can I ask you why? –  quark1245 Feb 18 '12 at 14:49
    
@quark1245: This makes sense only when considering Principal values! Let's consider a disk centered around a simple pole then the integral around the disk is $2\pi i\mathrm{residu}\ $. Now cut this disk in two equal parts with a line (crossing the pole of course!). The integral around the disk will be the sum of the two integrals around the half-disks (the line in the middle will be crossed in both directions). Each integral around the half disk will add a contribution $\pi i $ (the important thing is the angle at the pole $\pi\ $ here, with other angles the contributions will be unequal!). –  Raymond Manzoni Feb 18 '12 at 15:51
    
Ok, now I see.. thank you again –  quark1245 Feb 18 '12 at 16:40

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.