Timeline for Minimize the variance of a Boltzmann distribution
Current License: CC BY-SA 4.0
4 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| May 29, 2018 at 7:34 | comment | added | dohmatob | Not quite, as it doesn't use any information given in the problem. My best guess might be to resort to "control variates" cooked using the transformed variables $f_i(x)$). BTW, thanks to one of pdfs your referred has helped me identify that he problem is linked to something financial economists refer to as "logarithmic returns". So basically my problem is to produce a low variance estimate of the derivative of logarithmic returns. Rather unexpected connection! | |
| May 28, 2018 at 18:36 | comment | added | Carlo Beenakker | Well, you could use the "antithetic" variance reduction technique described in the linked references, which combines points $x$ where the integrand is small with those where it is large to achieve a variance reduction; this exploits the structure of the problem, doesn't it? | |
| May 28, 2018 at 18:29 | comment | added | dohmatob | I should precise in the question that i don't want Monte Carlo (or other black-box simulation technique). I need something more principled which exploits the structure of the problem... | |
| May 28, 2018 at 18:19 | history | answered | Carlo Beenakker | CC BY-SA 4.0 |