Timeline for Asymptotic behavior of gradient descent on a smooth, convex, non-negative function with no finite minimum
Current License: CC BY-SA 3.0
14 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Oct 7, 2017 at 13:58 | vote | accept | Daniel Soudry | ||
| Oct 6, 2017 at 17:03 | answer | added | Pietro Majer | timeline score: 3 | |
| Oct 6, 2017 at 8:31 | comment | added | Pietro Majer | Btw, the two variable function $f(x,y)=e^x + x^2y^2$ is not convex. | |
| Oct 6, 2017 at 6:46 | history | edited | Daniel Soudry | CC BY-SA 3.0 |
edited tags
|
| Oct 6, 2017 at 4:18 | history | edited | Daniel Soudry | CC BY-SA 3.0 |
edited body
|
| Oct 5, 2017 at 16:40 | history | edited | Daniel Soudry | CC BY-SA 3.0 |
added 36 characters in body
|
| Oct 5, 2017 at 13:33 | comment | added | Daniel Soudry | $x_n / || x_n || $ converges to $(-1,0)$ . | |
| Oct 5, 2017 at 13:23 | comment | added | user100927 | Can you check $f(x)=e^{x(1)}+x(1)^2x(2)^2$? | |
| Oct 5, 2017 at 13:00 | comment | added | Daniel Soudry | This function can be negative, and also this is not a counter-example since $ x_n / || x_n ||$ converges to $(0,1)$ or $(0,-1)$. | |
| Oct 5, 2017 at 12:07 | comment | added | user100927 | Ok then take $f(x)=ax(1)^2-x(2)$ with $a>\eta^{-1}$ instead. | |
| Oct 5, 2017 at 11:51 | comment | added | Daniel Soudry | Thanks for the answer, but this function appears to be non-convex. | |
| Oct 5, 2017 at 11:38 | comment | added | user100927 | The function $f(x)=atan2(x(1),x(2))$ is a counterexample. You can also find a continuous counterexample by changing this function locally at the half line of discontinuity. | |
| Oct 5, 2017 at 9:22 | history | edited | Daniel Soudry | CC BY-SA 3.0 |
added 56 characters in body
|
| Oct 3, 2017 at 11:26 | history | asked | Daniel Soudry | CC BY-SA 3.0 |