Timeline for convex representation of a combinatorial constraint
Current License: CC BY-SA 3.0
4 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jul 5, 2016 at 6:37 | comment | added | Christian Clason | You write $X=t A + (1-t)B$, where $A$ and $B$ are the two matrices you list. What you wish to do is to do minimize over $t\in \{0,1\}$, which is non-convex. Convex relaxation would be to minimize over $t\in[0,1]$ instead and hope that the optimal value is either $0$ or $1$ (which it often is, but this depends on the rest of the problem). | |
| Jul 5, 2016 at 1:07 | comment | added | Michael Fan Zhang | what convex relaxation technique can i use? Any references? Thanks | |
| Jul 4, 2016 at 14:32 | comment | added | Christian Clason | Such a constraint cannot be convex (otherwise you'd have to admit "two-and-a-half zeros"). You could introduce a convex relaxation, which often (but not always) leads to the desired property of the solution. | |
| Jul 4, 2016 at 13:52 | history | asked | Michael Fan Zhang | CC BY-SA 3.0 |