In article <[EMAIL PROTECTED]>, Gary <[EMAIL PROTECTED]> wrote:
>I really need some help for a problem I encountered in coding a matlab >program. I would like to normalize a series of random variables by a >STD, sigma; saying that the process is divided by the sigma. The STD >is drawed from a Inverse Gamma Distribution with alpha = beta = 0.001. > >But the Inverse Gamma Distribution with parameters above will always >produce some zero valued sigma. We cannot divide a number by zero-so >is there anybody can give some interesting hints? The numbers aren't really zero, of course - just so close that they underflow to zero on the computer. The hint these results are trying to tell you is that you don't actually want to draw your values from this distribution. You don't say what led you to think that you want to do this, but I might speculate that you've encountered the idea, distressingly common in Bayesian papers, that the gamma(0.001,0.001) distribution is "vague", and suitable as a prior distribution in situations with little knowledge. As an examination of a sample from this distribution will show, this idea is utterly wrong. Radford Neal ---------------------------------------------------------------------------- Radford M. Neal [EMAIL PROTECTED] Dept. of Statistics and Dept. of Computer Science [EMAIL PROTECTED] University of Toronto http://www.cs.utoronto.ca/~radford ---------------------------------------------------------------------------- . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
