Let’s assume I have a nonlinear function g(x,y) where x and y are two input random variables of joint PDF f(x,y) possibly correlated and encoded into an input UQLab object. How can I sample the conditional distribution f(y|x)? My goal is to approximate the following integral via MC,
G(x) = \int g(x,y) f(y|x) dy
As an alternative, is it possible to evaluate the joint PDF?