To comment on the following update, log in, then open the issue: http://www.openoffice.org/issues/show_bug.cgi?id=20819
------- Additional comments from [email protected] Thu Sep 23 20:59:21 +0000 2010 ------- @jumbo444 a) You need upwards of 9 SF in many of the coefficients to achieve a decent fit. Realistically, we need 11 digits for this case, which is a lot when inverting a poorly-conditioned matrix using brute-force methods. b) The fact that even the algorithm used at XUXU* works reasonably** if you transpose X to be symmetrical about zero (i.e. -5, -3, -1, 1, 3, 5) may be indicative of other possible useful techniques. *I notice that for moderately large zero order coefficients, XUXU truncates zero-order coefficient at the decimal point even when the resolution is much lower than for the other coefficients - is this indicative of other strange behaviour? **But even here adding third-order fits to successive recorded errors provides successively better fits - I assume that this indicates significant residual issues with the algorithms? --------------------------------------------------------------------- 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]
