Документ взят из кэша поисковой машины. Адрес оригинального документа : http://hea-www.harvard.edu/AstroStat/slog/groundtruth.info/AstroStat/slog/2009/use-and-misuse-of-chi-square/feed/index.html Дата изменения: Unknown Дата индексирования: Mon Feb 4 06:34:20 2013 Кодировка: Поисковые слова: п п п п п п п п п п п

http://hea-www.harvard.edu/AstroStat/slog/2009/use-and-misuse-of-chi-square/ Weaving together Astronomy+Statistics+Computer Science+Engineering+Intrumentation, far beyond the growing borders Fri, 01 Jun 2012 18:47:52 +0000 hourly 1 http://wordpress.org/?v=3.4 http://hea-www.harvard.edu/AstroStat/slog/2009/use-and-misuse-of-chi-square/comment-page-1/#comment-871 vlk Wed, 01 Apr 2009 19:37:14 +0000 http://hea-www.harvard.edu/AstroStat/slog/?p=1862#comment-871 XSPEC has a function called "renorm" which simply rescales the unfrozen normalizations of all the model components such that the predicted counts equal the observed counts. (As also PINTofALE's FITLINES.) Sherpa figures out the default values based on the data, so the initial guess is usually in the ballpark. This is all done before the actual fit. The result of the fit, naturally, is not expected to produce counts identically equal to the observed counts, but will differ as appropriate to the assumed error model. i.e., the difference between the summed predicted and observed counts will be consistent with the error bar on the normalizations (when there are no odd correlations to mess it up). XSPEC has a function called “renorm” which simply rescales the unfrozen normalizations of all the model components such that the predicted counts equal the observed counts. (As also PINTofALE’s FITLINES.) Sherpa figures out the default values based on the data, so the initial guess is usually in the ballpark. This is all done before the actual fit. The result of the fit, naturally, is not expected to produce counts identically equal to the observed counts, but will differ as appropriate to the assumed error model. i.e., the difference between the summed predicted and observed counts will be consistent with the error bar on the normalizations (when there are no odd correlations to mess it up).

]]>