Gökhan wrote in message <[EMAIL PROTECTED]>... > >Hi! >I wonder how the public is evaluating the normal distribution function >in realworld applications. I am implementing some methods where i have >to calculate different times probability functions relying on normal >distribution functions with steadily changing covariance matrix and mean > >values... >I know that my methods work in matlab but implementing them on c++ makes > >one think about speed ;-) > >Can anyone give me a link or a hint how to evaluate the determinant btw >the inverse >of the SYMETRIC covariance matrix and so the normal distribition >function rapidly? >Thanks in advance. >gökhan > >-- > >==================================== >Gökhan BakIr >Insitute of Robotics and Mechatronics >German National Research Institute for Aero and Space >82234 Oberpfaffenhofen >Tel: + 49-8153 - 28 2440 >ICQ : 82040497 >www.fastray.de > > I don't understand this either. Mention of the covariance matrix suggests that it is the multivariate normal which is wanted, not the simple univariate normal. For the univariate normal, see the Applied Statistics algorithm AS66, either from statlib: http://lib.stat.cmu.edu or from my ozemail web site; The multivariate normal integral is much more difficult. There is some code at my ozemail website. You could also try Alan Genz's web site at Washington State Univ. All of the above code is in some version of Fortran, some ancient, some modern. -- Alan Miller, Retired Scientist (Statistician) CSIRO Mathematical & Information Sciences Alan.Miller -at- vic.cmis.csiro.au http://www.ozemail.com.au/~milleraj http://users.bigpond.net.au/amiller/ ================================================================= Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =================================================================
