numerical evaluation of quotients of gamma functions

>>>>> "Barton" == Barton Willis <willisb at unk.edu> writes:

    Barton> Yes that's also a good idea. But I think I'd need to write
    Barton> my own log-base-two-gamma-function? Or am I missing
    Barton> something?

Oh, you wanted something that was already built in? Then I don't
really know, other than doing log(gamma(x)), which might be good
enough for now if gamma(x) doesn't overflow and you don't care about
the precision loss in log(gamma(x)).  I guess that's at most 10 bits of
precision lost.

But we might consider providing a loggamma function since it's so easy
for gamma to overflow a double float. This doesn't help you solve your
problem though.

Ray