202406121510
Status: #idea
Tags: Probability Theory, Statistics
State: #nascient

Inverse Transformations

It's a really cool result that shows that as long as some cdf is continuous, it is possible to sample from it through a uniform distribution.

Let F(x) be the cdf such that it is injective, find F1(x) and then plug a uniform distribution from [0,1] in there and boom, even though you sampled from a uniform distribution, through that process you can mathemagically sample from any continuous distribution trivially.