To comment on the following update, log in, then open the issue: http://www.openoffice.org/issues/show_bug.cgi?id=43040
------- Additional comments from [email protected] Tue Jul 28 18:55:55 +0000 2009 ------- The asymptotic approximation is unsuitable. Even with x-values above 20000 (which are far to large for double) you get only a relative accuracy of four decimals. On the other hand the iterative solution is well conditioned and stable. Therefore I drop the asymptotic approximation. The iteration converges fast (<100) for "normal" cases. In cases when x is large, where the iteration would take long, the iteration will break before ending with overflow error, because Besseli is growing like exp(x). I think that no "NoConvergenceException" is needed. In addition to the errors for x>30, the version in OOo3.1 is wrong too for small x and larger N, like N=31 in my examples. This is caused by the absolute comparison to epsilon instead of a relative comparison. For small x the iteration stops too early, although only a few additional steps are needed to get an accurate result. I'll attach a document with test cases. Open it in OOo3.1 to see the problems. --------------------------------------------------------------------- Please do not reply to this automatically generated notification from Issue Tracker. Please log onto the website and enter your comments. http://qa.openoffice.org/issue_handling/project_issues.html#notification --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
