Timeline for Faster Convergence in CLT for sums and convolutions of Gaussians?
Current License: CC BY-SA 4.0
5 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jan 27 at 3:14 | history | edited | Mark Schultz-Wu | CC BY-SA 4.0 |
added 183 characters in body
|
| Jan 27 at 1:54 | comment | added | Michael Hardy | I think you ought to explain that in the question. | |
| Jan 26 at 22:59 | comment | added | Mark Schultz-Wu | @MichaelHardy The random vectors $\vec X, \vec X'$ are viewed as polynomials $\vec X \sim \sum_{i} \vec X_i y^i$, and then multiplied in $\mathbb{R}[y]/(y^n-1)$. As an example, the first coordinate of $\vec X\ast \vec X'$ is $\sum_{i = 0}^n \vec X_i \vec X'_{n-i}$ (up to a potential off-by-one error in the indices). | |
| Jan 26 at 22:56 | comment | added | Michael Hardy | Where you speak of "convolutions of random variables,", I wonder whether the things whose convolutions you speak of are actually the density functions rather than the random variables themselves? | |
| Jan 26 at 22:25 | history | asked | Mark Schultz-Wu | CC BY-SA 4.0 |