b***k 发帖数: 2673 | 1 ☆─────────────────────────────────────☆
pere (pere) 于 (Sat Jan 16 14:34:12 2010, 美东) 提到:
As an application of change of measure to Monte Carlo, we look at importance
sampling, whose goal is to use change of measure to reduce the variance of
a sample. The idea is that some of the paths generated have little impact on
the final result (usually an expectation), so time spent generating these
paths is essentially wasted. Using importance sampling we make paths which
have more impact more likely |
|