An important variance reduction technique in Monte Carlo simulation is importance sampling. The idea behind this concept is to approximate an expectation by a weighted average value

where the normalized weights are proportional to the so-called likelihood ratio

and the sampling is conducted with respect to the importance density g(x), not the original density f(x). Show that the expectation of h?_?is still unbiased.

Transcribed Image Text:

h-Σ νω), Wnh n=1 Ax") N. g(x") and wn = 1, n=1