Re: random variable
- To: mathgroup at smc.vnet.net
- Subject: [mg103302] Re: random variable
- From: fd <fdimer at gmail.com>
- Date: Mon, 14 Sep 2009 07:11:11 -0400 (EDT)
- References: <200908280943.FAA11851@smc.vnet.net> <28A3460B9CB94660BE9B0EF0E222E8E2@laertes>
On Sep 11, 7:29 pm, omar bdair <bdai... at yahoo.com> wrote: > Dear all, > > Generating random variables using the rejection method is the best > method that I want to use. But the file sent by Harker only discussed > the case of one variable pdf. The question is: Could we use the > rejection method to generate a sequence of random variables when the > pdf is of multi-variables? > > Thank you > > Omar Bdair I'd suggest you to use a Markov chain monte carlo method. You can find an excellent exposition in Mackay's book (chapter 29) http://www.inference.phy.cam.ac.uk/mackay/itila/book.html The author explains why rejection isn't the best for multi-variable sampling.