In article <[EMAIL PROTECTED]>, Radford Neal <[EMAIL PROTECTED]> wrote: >In article <[EMAIL PROTECTED]>, >nicolexz <[EMAIL PROTECTED]> wrote: >>Thank you very much for the information. I will do further research >>on it. BTW, how would you do if you happen to run into a similiar >>problem under the setting of univariate t-distributiion?
>The procedure I posted depends only on the distribution being unimodal, >so it would work fine for a t distribution. I hadn't noticed that your >previous question concerned only normal distributions, for which someone >must have worked out a nice special-purpose procedure. The t distribution is not log-concave. It would probably be better to transform it to the symmetric beta, which is. But it is faster to use a normal for the acceptance-rejection for a symmetric beta than a general log-concave. -- This address is for information only. I do not claim that these views are those of the Statistics Department or of Purdue University. Herman Rubin, Department of Statistics, Purdue University [EMAIL PROTECTED] Phone: (765)494-6054 FAX: (765)494-0558 . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
