Ray, you wrote:
> I have a problem where the PDF is defined over R^n, and a constraint
> 0=f(R^n) defining a n-1 dimensional surface/constraint.
> Can somebody recommend an explicit paper for calculating the PDF on the
> surface using the Baysian formula?
> I believe I know how to do it, but some things are a little fuzzy.
Unfortunately I don't know any methods, general or specific,
for this problem. Maybe someone else has a suggestion.
You might seach for Alan Genz on the web -- his specialty is
multidimensional integration for Bayesian inference.
Maybe he would respond to an email.
Can you say more about the problem? Sounds interesting.
All the best,
Robert