DiracDelta



> Something with an exp(-x^2) in it just seems perversely hard to integrate!

The popularity of the gaussian in this context almost surely derives from the heat kernel.

--bw