Wednesday, 13 February 2019

integration - Calculating a Complicated Integral of Two Variables



I have encountered the following integral
I=2π02π0c1+c2cosθ2a+cosθ1+cosθ2cos(mθ1)cos(nθ2)dθ1dθ2

where a>2 and I am having a hard time moving forward. I would be happy with an analytic solution, converting this to some non-elementary but known special function, or finding the asymptotic behavior of the integral as a2 from the positive reals.



I was able to perform the θ1 integration by looking in Gradshteyn and Ryzhik and this yields something proportional to
2π0c1+c2cosθ2a22acos(θ2)sin(θ2)2cos[nθq]×(a22acos(θ2)sin(θ2)2a+cos(θ2))mdθ2.
Trying to find this form in a table was hopeless, so I changed variables to x=cosθ2. Having to expand cos(nθ2) introduced a sum, and the resulting integral is proportional to
11dx1x2c1+c1x(a+x)21×((a+x)21a+x))mr=02rn(1)r(n2r)xn2r(1x2)r.
I am pretty stumped at this point.




I tried doing some further expansions, but I got infinite sums of infinite sums, so this didn't seem like progress (the hypergeometric 3F2 and the regularized 2˜F1 appeared in this approach).



There are reduction formulas for integrals, but many of them don't seem to apply, as I have two distinct square roots (1x2 and (a+x)21). However, my integrand is a rational function of these square roots and x.



There is a small parameter in the problem, namely ϵ>0 in a=2+ϵ. The integral diverges for a=2, so this is a singular perturbation problem (if this approach is even helpful).



I would appreciate any suggestions or advice on this problem!


Answer



Extracting the leading order is not hard. Indeed, let f:R2C be a continuous function such that f(k1,k2)=f(k1+2π,k2)=f(k1,k2+2π) for all k1,k2R. Then with ,




\int_{0}^{2\pi}\int_{0}^{2\pi} \frac{f(k)}{a-\cos k_1 - \cos k_2} \, dk_1 dk_2 = \int_{B(0,\pi)} \frac{f(k)}{a - 2 + \frac{1}{2}|k|^2} \, dk + \mathcal{O}(\|f\|).



Now applying the polar coordinates change followed by the substitution r=\sqrt{2(a-2)u},



\begin{align*} \int_{B(0,\pi)} \frac{f(k)}{a - 2 + \frac{1}{2}|k|^2} \, dk & = \int_{0}^{\pi} \frac{r}{a-2+\frac{1}{2}r^2} \left( \int_{S^1} f(r\omega) \, d\omega \right) \, dr \\ &= \int_{0}^{\frac{\pi^2}{2(a-2)}} \left( \int_{S^1} f(\omega\sqrt{2(a-2)u}) \, d\omega \right) \, \frac{du}{u+1} \\ &\sim 2\pi f(0) \log \left( \frac{\pi^2}{2(a-2)} \right) \\ &\sim -2\pi f(0) \log(a-2). \end{align*}



As an alternative direction, I guess that a probabilistic interpretation may perhaps help analyze the behavior of I. Indeed,



G^p(x,y) = \frac{1}{2\pi^2} \int_{0}^{2\pi}\int_{0}^{2\pi} \frac{e^{-ik\cdot y}}{\frac{2}{1-p} - \cos(k_1) - \cos(k_2)} \, dk



is the Green function of the 2-D simple random walk on \mathbb{Z}^2 killed at rate p.


No comments:

Post a Comment

real analysis - How to find lim_{hrightarrow 0}frac{sin(ha)}{h}

How to find \lim_{h\rightarrow 0}\frac{\sin(ha)}{h} without lhopital rule? I know when I use lhopital I easy get $$ \lim_{h\rightarrow 0}...