From: SourceForge.net <no...@so...> - 2012-11-11 17:44:03
|
Bugs item #3586225, was opened at 2012-11-11 09:44 Message generated for change (Tracker Item Submitted) made by rtoy You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=104933&aid=3586225&group_id=4933 Please note that this message will contain a full copy of the comment thread, including the initial issue submission, for this request, not just the latest update. Category: Lisp Core - Floating point Group: None Status: Open Resolution: None Priority: 5 Private: No Submitted By: Raymond Toy (rtoy) Assigned to: Nobody/Anonymous (nobody) Summary: inverse_erf(1d-10) fails to converge Initial Comment: inverse_erf(1d-10) doesn't converge. But there shouldn't really be a problem since the taylor series converges nicely there. ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=104933&aid=3586225&group_id=4933 |
From: SourceForge.net <no...@so...> - 2012-11-14 16:13:24
|
Bugs item #3586225, was opened at 2012-11-11 09:44 Message generated for change (Comment added) made by rtoy You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=104933&aid=3586225&group_id=4933 Please note that this message will contain a full copy of the comment thread, including the initial issue submission, for this request, not just the latest update. Category: Lisp Core - Floating point Group: None Status: Open Resolution: None Priority: 5 Private: No Submitted By: Raymond Toy (rtoy) Assigned to: Nobody/Anonymous (nobody) Summary: inverse_erf(1d-10) fails to converge Initial Comment: inverse_erf(1d-10) doesn't converge. But there shouldn't really be a problem since the taylor series converges nicely there. ---------------------------------------------------------------------- >Comment By: Raymond Toy (rtoy) Date: 2012-11-14 08:13 Message: Interesting. This happens because there's a bug in erf. In particular, slatec::derf is wrong. When x < sqeps (approx 1.49e-8), derf uses erf = 2*x*x/sqrt(%pi). But the taylor series for erf is 2*x/sqrt(%pi). When this is fixed, inverse_erf(1d-10) converges very quickly. ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=104933&aid=3586225&group_id=4933 |
From: SourceForge.net <no...@so...> - 2012-11-14 18:09:22
|
Bugs item #3586225, was opened at 2012-11-11 09:44 Message generated for change (Comment added) made by rtoy You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=104933&aid=3586225&group_id=4933 Please note that this message will contain a full copy of the comment thread, including the initial issue submission, for this request, not just the latest update. Category: Lisp Core - Floating point Group: None >Status: Closed >Resolution: Invalid Priority: 5 Private: No Submitted By: Raymond Toy (rtoy) Assigned to: Nobody/Anonymous (nobody) Summary: inverse_erf(1d-10) fails to converge Initial Comment: inverse_erf(1d-10) doesn't converge. But there shouldn't really be a problem since the taylor series converges nicely there. ---------------------------------------------------------------------- >Comment By: Raymond Toy (rtoy) Date: 2012-11-14 10:09 Message: Closing. ---------------------------------------------------------------------- Comment By: Raymond Toy (rtoy) Date: 2012-11-14 08:13 Message: Interesting. This happens because there's a bug in erf. In particular, slatec::derf is wrong. When x < sqeps (approx 1.49e-8), derf uses erf = 2*x*x/sqrt(%pi). But the taylor series for erf is 2*x/sqrt(%pi). When this is fixed, inverse_erf(1d-10) converges very quickly. ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=104933&aid=3586225&group_id=4933 |