Timeline for Independent Decomposition of a Random Vector
Current License: CC BY-SA 3.0
5 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Aug 19, 2016 at 16:23 | vote | accept | EmmGee | ||
| Aug 17, 2016 at 18:32 | comment | added | Christian Remling | @EmmGee: Your edited version runs into the exact same problem: we can take an $F$ for which $F(g_j)$ is approximately the Bernoulli RV from my example. While admittedly I haven't thought about my remark from the second paragraph much, my feeling is that it does describe what this problem is about. | |
| Aug 17, 2016 at 7:21 | comment | added | EmmGee | But I wonder "how crucial" the extra property of rotation invariance is. Is this really necessary or can this be relaxed? I slightly reformulated the problem issue in this direction. | |
| Aug 17, 2016 at 6:16 | comment | added | EmmGee | Thank you! This counterexample is already helpful for me because the above question is actually part of a more general statement, which is not true for $\pm 1$-values entries. | |
| Aug 16, 2016 at 16:10 | history | answered | Christian Remling | CC BY-SA 3.0 |