[R-SIG-Finance] bessel functions

[stefano iacus]

Hi,

does anyone know about a reliable  algorithm to evaluate the  
exponentially rescaled bessel function of first find, any order, real  
argument?

what I need is the equivalent of R function

besselI(x, nu, expon.scaled = TRUE)

for large real x (say 5000) and large nu (say 10 to 30)

The R implementation does not satisfy my needs because, for example,


 > besselI(5539.947191,11,exp=TRUE)
[1] NaN

but if you ask Mathematica you get this number: 0.00530180603357

To be more specific I need this to evaluate the likelihood of the CIR  
model. I know there are many different estimators for the CIR but I  
need to evaluate the likelihood.

Googling around did not help me.

thanks in advance.

stefano