On Apr 15, 3:33 pm, [EMAIL PROTECTED] wrote: > With few tests, it seems this is faster than the version by Jussi only > with quite big n,k. >
True---and for large n and k it's difficult to imagine any real-world applications that wouldn't be better served by using the lngamma function instead. That is, the natural log of n choose k can be computed much more quickly as lngamma(n+1) - lngamma(k+1) - lngamma(n-k+1) I assume that lngamma is implemented somewhere in either numpy or scipy, but right now I can't find it... Mark -- http://mail.python.org/mailman/listinfo/python-list
