Timeline for Symmetry group in posterior distribution/inference
Current License: CC BY-SA 4.0
8 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Feb 28, 2019 at 18:37 | comment | added | whuber♦ | Since you are using Bayesian methodology, I don't see why you even need to compute expectations: aren't you interested in the posterior distribution? Use properties of that distribution that are suitable for your analysis, such as modes and measures of spread around those modes. | |
| Feb 28, 2019 at 17:39 | comment | added | Justin Solomon | Sure, this amounts to reasoning about the quotient space directly. But notions like expectation are weird in this quotient. For example, using [0,2pi) to represent the unit circle neglects the fact that 2pi=0 (mod 2pi), which gets in the way of computing expectations and other quantities. | |
| Feb 28, 2019 at 15:08 | comment | added | whuber♦ | en.wikipedia.org/wiki/Fundamental_domain | |
| Feb 28, 2019 at 13:00 | vote | accept | Justin Solomon | ||
| Feb 28, 2019 at 5:32 | answer | added | Xi'an | timeline score: 3 | |
| Feb 28, 2019 at 0:35 | comment | added | Justin Solomon | Afraid I don’t know what that means! Might you have a pointer? Do statistical theorems carryover to this case? | |
| Feb 28, 2019 at 0:08 | comment | added | whuber♦ | Usually one just chooses a fundamental domain for the group action. | |
| Feb 27, 2019 at 23:54 | history | asked | Justin Solomon | CC BY-SA 4.0 |