8 commits

Author	SHA1	Message	Date
David Simcha	8aede6061f	Attempt number 2: Work around Bug 4298 in gammafunction and errorfunction code.	2010-11-29 13:57:37 +00:00
Don Clugston	6d5104ceb5	Relaxed tolerance on betaIncompleteInv unit test. The fact that a 0.8-bit error in log() caused a 38 bit change in the result, is a nice demonstration of how difficult this function is to calculate. BTW Mathematica crashes on this calculation.	2010-11-29 08:31:51 +00:00
Don Clugston	5db6dad8c1	Increase tolerance of this test, now that we know that failure is caused by a small imprecision in AMD's microcode.	2010-11-28 20:11:36 +00:00
Don Clugston	d0c8affabc	Some tougher, more accurate unit tests for betaIncomplete. Possibly these will shed more light on the AMD/Intel difference.	2010-11-23 21:50:05 +00:00
David Simcha	172ec1dc1a	Revert the last changeset.	2010-11-20 22:48:31 +00:00
David Simcha	66bdf06479	Work around Bug 4298 in gammafunction and errorfunction code.	2010-11-20 22:02:35 +00:00
Don Clugston	09667828b6	Enable unittests which used to fail, but are working now.	2010-11-19 22:25:28 +00:00
Don Clugston	876bf9789d	Added std.mathspecial. Only 14 functions are included at this stage.	2010-11-18 19:55:57 +00:00