Re: looking for a suitable numericla method


smilee wrote:
Hi. I am looking for a way of integration whose integrand is like
"exp(-x^2) tanh(a+b x)" and the integral range is from -infinity to
+infinity. Does Gaussian-Hermite quadrature method give a correct
answer for this case? Or is there any more suitable way to perform
this?
Thanks.

If it is numerical, ordinary Gauss Legendre does very well (I just
checked it out, got 16 places with no sweat) since the integrand,
if real, is smooth and decays rapidly. Gauss Hermite
could be a refinement for the two seminfinite intervals.

.

Quantcast