Hi, as rightly pointed out by Christian, the bug reported in MATH-718 is in fact a result of the current implementation of the incomplete beta function I(x, a, b), which is inaccurate when a and/or b are large-ish. I think this is a very interesting issue, but also one which will not be solved in one or two days' work. I've skimmed through slatec [1], GSL, Boost [2] as well as you-know-who. Neither uses the same method to compute this function. I would like to review in detail those methods before coming up with a fix for this issue. So I propose that we mark MATH-718 as won't fix, and open a new JIRA issue to affect 3.1 or 4.0. Working on special functions (which is a weak side of CM anyway) is something I would be very much interested in.
Do you agree with this proposal? Sébastien PS: my lab does not subscribe to ACM. Does anyone have access to this paper TRANSACTIONS ON MATHEMATICAL SOFTWARE, VOL. 18, NO. 3, SEPTEMBER, 1992, PP. 360-373z. The FORTRAN code is freely available [3]. [1] http://www.netlib.org/slatec/fnlib/betai.f [2] http://www.boost.org/doc/libs/1_38_0/libs/math/doc/sf_and_dist/html/math_toolkit/special/sf_beta/ibeta_function.html [3] http://www.netlib.org/toms/708 --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
