Closed-form expression for the Gaussians on infinite lattices $\sum_{n=-\infty}^\infty e^{-\frac{(x-Ln)^2}{2\sigma^2}}$

104 Views Asked by At

Does this sum have a closed-form expression?

$$\sum_{n=-\infty}^\infty e^{-\frac{(x-Ln)^2}{2\sigma^2}}$$

I'm trying to find the closed-form expression for the function composed of the Gaussians on infinite lattices of which interval is $L$.

Any suggestions or advice is greatly appreciated. Thanks in advance!

1

There are 1 best solutions below

0
On BEST ANSWER

Expanding upon Graham Hesketh's comment, we can solve this in terms of the Jacobi theta function.

The original expression is: $$\sum_{n=-\infty}^{\infty} e^{-\frac{(x-Ln)^2}{2\sigma^2}}$$

The Jacobi theta function can be defined by: $$\vartheta(z;\tau) = \sum_{n=-\infty}^{\infty}e^{\pi in^2\tau + 2\pi inz }$$

First we'll expand the original expression a bit. $$\sum_{n=-\infty}^{\infty} e^{-\frac{(x-Ln)^2}{2\sigma^2}} = \sum_{n=-\infty}^{\infty} e^{-\frac{x^2 - 2Lnx + L^2n^2}{2\sigma^2}}$$

$$= \sum_{n=-\infty}^{\infty} e^{-\frac{x^2}{2\sigma^2}} e^{-\frac{L^2n^2}{2\sigma^2} + \frac{Lnx}{\sigma^2}}$$

$$= e^{-\frac{x^2}{2\sigma^2}} \sum_{n=-\infty}^{\infty} e^{-\frac{L^2n^2}{2\sigma^2} + \frac{Lnx}{\sigma^2}}$$

Now equating terms, first for $\tau$: $$\pi i n^2\tau = -\frac{L^2n^2}{2\sigma^2}$$

$$\tau = -\frac{L^2}{2\pi i \sigma^2}$$

And for $z$: $$2\pi i nz = \frac{Lnx}{\sigma^2}$$

$$z = \frac{Lx}{2\pi i \sigma^2}$$

Combining all of the parts:

$$\sum_{n=-\infty}^{\infty} e^{-\frac{(x-Ln)^2}{2\sigma^2}} = e^{-\frac{x^2}{2\sigma^2}}\vartheta\left(\frac{Lx}{2\pi i \sigma^2};\frac{L^2}{2\pi i \sigma^2}\right)$$