Skip to content

R: integration problem

3 messages · Clark Allan, Uwe Ligges, Ravi Varadhan

#
hi all 

an integration problem. i would like an exact or good approximation for
the following, but i do not want to use a computer. any suggestions:


integral of exp(b*x)/sqrt(1-x^2)

where "b" is a constant greater than or equal to 0
and
the integral runs from 0 to 1


any help would be apreciated

/
allan
#
Clark Allan wrote:

            
Sounds like the problem of integrating the Gaussian density...

Uwe Ligges
#
Hi,

If your limits were to be from -1 to +1 (instead of lower limit being 0),
your integral is:

pi * I_0(b)

Where I_0 is the modified Bessel's function of the zeroth order.  

If it is from 0 to 1, then there is no closed form (the integrand is not
symmetric about 0). You must evaluate the integral with exp(a*cos(t)) as the
integrand from 0 to pi/2.

Hope this is helpful,
Ravi.