Products and ratios Bessel functions -> any known approximations?

[nronald]

Hi,

I work in a computational neuroscience lab, where we study human perception using Bayesian models. In our models we often have to compute products and ratios of Bessel functions (specifically, zeroth-order modified Bessel functions of the first kind).

Our computations could speedup considerably if we would replace these Bessel evaluations by approximations. Does anyone know if approximations exist for such products and ratios?

Any other information/references on how to compute efficiently with Bessel functions would also be highly appreciated.

Thanks!

[Feldoh]

You could try doing a stationary phase approximation using Bessel's integral.