Thursday, June 14, 2012

Lumps 'n' Bumps: Discrete probability integral transform for large N

You may or may not have heard of the probability integral transform, but it’s got to be up there as one of my favourite integral transforms (yeah – I have favourites). The basic idea is that you want a sample of random variables from some none-standard probability distribution, but all you have is a basic random number generator that spits out samples from the uniform distribution. With a bit of magic involving the conditional distribution function (cdf) of your target distribution...

Read more at Lumps 'n' Bumps