Talk:Projection on Polyhedral Cone
From Wikimization
(New page: The definition of projection should be made clear. If, by projection you mean the nearest point to the cone, then this results in a quadratic programming problem, i.e. given the point <ma...) |
|||
| Line 7: | Line 7: | ||
No explicit formula for a solution of a quadratic program exists (or for the solution of a linear program). It is doubtful that such a formula will be found due to the combinatorial nature of the problem. | No explicit formula for a solution of a quadratic program exists (or for the solution of a linear program). It is doubtful that such a formula will be found due to the combinatorial nature of the problem. | ||
| + | |||
| + | ==Reply== | ||
| + | Yes, the nearest point belonging to the cone. | ||
| + | |||
| + | I have been asked this question several times; ''i.e''., is there a '''formula''' for projecting on a polyhedral cone. | ||
| + | We know some formulas; ''e.g''., projection on the positive semidefinite cone, or on the various orthants. | ||
| + | People are surprised when they discover there are no formulae for projection on simple polyhedral cones, | ||
| + | because it seems like there ought to be. | ||
| + | |||
| + | I think it remains a fascinating question. Perhaps we could collect here what we do know about this problem. | ||
| + | --[[User:Dattorro|Dattorro]] 13:49, 9 June 2008 (PDT) | ||
Revision as of 13:49, 9 June 2008
The definition of projection should be made clear.
If, by projection you mean the nearest point to the cone, then this results in a quadratic programming problem, i.e. given the point and the cone
, then the quadratic program is
No explicit formula for a solution of a quadratic program exists (or for the solution of a linear program). It is doubtful that such a formula will be found due to the combinatorial nature of the problem.
Reply
Yes, the nearest point belonging to the cone.
I have been asked this question several times; i.e., is there a formula for projecting on a polyhedral cone. We know some formulas; e.g., projection on the positive semidefinite cone, or on the various orthants. People are surprised when they discover there are no formulae for projection on simple polyhedral cones, because it seems like there ought to be.
I think it remains a fascinating question. Perhaps we could collect here what we do know about this problem. --Dattorro 13:49, 9 June 2008 (PDT)