Timeline for Are there examples of functions with Nesterov's convergence bound between convex quadratic and strongly convex cases?
Current License: CC BY-SA 3.0
7 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Aug 22, 2017 at 19:27 | comment | added | littleO | Can you elaborate on how Su et al expands the influence of Nesterov's method in the stats community? | |
| May 8, 2017 at 17:52 | history | edited | Henry.L | CC BY-SA 3.0 |
deleted 68 characters in body
|
| May 8, 2017 at 17:51 | comment | added | Henry.L | @Dirk yes, corrected. | |
| May 8, 2017 at 17:29 | comment | added | Dirk | Actually, you need that $\nabla f$ exists and is Lipschitz with constant $L$ and not that all subgradients are bounded by $L$. If you merely have subgradients available, the optimal rate is $O(1/\sqrt{t})$ and the subgradient achieves that (Theorem 3.2.1 and 3.2.2 in Nesterov). | |
| May 8, 2017 at 16:33 | comment | added | Dirk | I think, the question is about the linearly convergenz method for strongly convex functions. There results there are of flavor, though. | |
| May 8, 2017 at 14:24 | history | edited | Henry.L | CC BY-SA 3.0 |
added 9 characters in body
|
| May 8, 2017 at 14:18 | history | answered | Henry.L | CC BY-SA 3.0 |