Trying to work out Gaussian integral using contour integral and Stokes' Theorem

71 Views Asked by At

I'm not sure what I'm getting wrong here. Based on this question, I'm trying to compute the double Gaussian integral using Stokes' theorem with complex forms.

I'm writing: $$ \int_R dx\wedge dy \;e^{-x^2 - y^2} = \int_R \frac{d\bar{z}\wedge dz}{2i}\; e^{-z\bar{z}}$$ for a circular region $R$ of radius $r$ centred at the origin. Then Stokes' theorem says $\int_R d\omega = \int_{\partial R} \omega$ for a differential form $\omega$ and $d = dz\frac{\partial}{\partial z} + d\bar{z}\frac{\partial}{\partial \bar{z}}$. In this case: $$\omega = -\frac{1}{4iz} \; e^{-z\bar{z}} \, dz + \frac{1}{4i\bar{z}} e^{-z\bar{z}} \, d\bar{z} \implies d\omega = \frac{1}{2i} e^{-z\bar{z}} d\bar{z}\wedge dz$$ So writing $z = r e^{i\theta}$, the integral is $$\int_{\partial R} -\frac{1}{4iz} e^{-z\bar{z}} dz + \frac{1}{4i\bar{z}} e^{-z\bar{z}} d\bar{z} = \int_{0}^{2\pi} -\frac{1}{2} e^{-r^2} d\theta = -\pi\, e^{-r^2}$$

But the answer should be $\pi$ if $r\rightarrow \infty$. What am I doing wrong or not understanding here?

1

There are 1 best solutions below

0
On

In its current form, the theorem does not apply because there is a singularity at the origin in the middle of the domain. Cut it out properly by considering an annulus $r\leq |z| \leq R$ instead which gives

$$I = \pi\left(e^{-r^2}-e^{-R^2}\right)$$

with each boundary having reversed orientations. Taking the proper limits $r\to0^+$ and $R\to\infty$ now gives the correct answer.