Steepest descent, non-analytic roots

  • Context: Graduate 
  • Thread starter Thread starter CompuChip
  • Start date Start date
  • Tags Tags
    Roots
Join the discussion
Registration is free. Ask a follow-up in this thread, or start your own.
1 reply · 3K views
Messages
4,305
Reaction score
49
Hi,

For a physics class, I am supposed to evaluate the following integral
[tex]I_n(a) = \int_{-\infty}^\infty \mathrm dx \, e^{-n x^2/2 + n a x} \cosh^n(x)[/tex]
as a function of the real non-zero parameter a, in the limit as [itex]n \to \infty[/itex] using the method of steepest descent. The question adds: "Express the ersult in parametric form as a function of the saddle point position."

The problem I ran into, is that the extremum position [itex]x_0[/itex] satisfies the equation
[tex]x_0 - a - \tanh(x_0) = 0[/tex]
which cannot be solved analytically. So I'm getting a bit confused, whether I should just leave the result
[tex]I_n(a) \simeq e^{- n x_0^2 + n a x_0} \sqrt{\frac{2\pi}{f''(x_0)}}[/tex]
where f''(x_0) = tanh(x_0) = x_0 - a, or whether I am missing something here.

I'd appreciate your input.
 
Physics news on Phys.org
I've tried some more approaches, but in all of them, the problem is in the log(cosh(x)) that's in the exponent. There is no way it can be rewritten without having [itex]\log\cosh(x_0))[/itex], with x0 the position of the minimum of the exponent, in the final expression.

I will try to see the TA tomorrow, but your suggestions are still welcome.
(The exercise was given Thursday afternoon, due on Tuesday).