Integrate Gaussian$\times$Log

114 Views Asked by At

Show that

$$\int_1^\infty e^{-\frac{x t^2}{2}}\log(t) dt \sim \frac{e^{-\frac{x}{2}}}{x^2}$$

for large positive $x$.

I tried Taylor expanding the log. Each of the resulting terms can be integrated separately and yields a hypergeometric function upon integration. However, even after using the asymptotic form of these hypergeometric functions, the sum can't be done in closed form.

1

There are 1 best solutions below

1
On BEST ANSWER

Recognizing that this is dominated by what happens near the lower limit and Taylor expanding $\log(t)\approx t-1$, this is asymptotic to $$ \int_1^\infty (t-1)e^{-\frac{1}{2}xt^2}dt = \int_0^\infty te^{-\frac{1}{2}x(t+1)^2}dt \\= \frac{e^{-x/2}}{x^2}\int_0^\infty te^{-t-\frac{1}{2}t^2/x}dt \\\sim \frac{e^{-x/2}}{x^2}\int_0^\infty te^{-t}dt \\= \frac{e^{-x/2}}{x^2}.$$