Timeline for Sampling according to a product of a known density and a probability function
Current License: CC BY-SA 4.0
6 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jun 22, 2022 at 9:20 | comment | added | Xi'an | "Probability function" has no particular meaning in this setting. For rare events, cross-entropy is a way to build a better proposal than $p(\cdot)$ | |
| Jun 22, 2022 at 0:48 | comment | added | Glen_b | There are many potential algorithms that might be used. If you can find a suitable envelope for $q$ there's accept-reject (and a variety of potential ways to obtain some). If you have a suitable approximation to use in a proposal distribution and don't mind dependence in your generated values you may be able to use Metropolis-Hastings. etc etc | |
| Jun 21, 2022 at 5:28 | comment | added | kykim | @jbowman For the specific problem I'm looking at, it's more a scoring function whose output is in [0, 1] e.g., $f(x)$ being how likely $x$ to meet some criteria. Perhaps I shouldn't have called $f$ a probability function. | |
| Jun 21, 2022 at 1:27 | comment | added | jbowman | Is $f$ a probability density or a cumulative density? | |
| S Jun 21, 2022 at 0:20 | review | First questions | |||
| Jun 21, 2022 at 1:19 | |||||
| S Jun 21, 2022 at 0:20 | history | asked | kykim | CC BY-SA 4.0 |