Clp.old
Clp.old copied to clipboard
[TRAC] Incorrect solution to quadratic programme
Moved from TRAC as part of COIN fORgery: https://projects.coin-or.org/Clp/ticket/23
The attached MPS file contains a small problem that has the ability to confuse CLP, whereas I would expect the solution to be around 1.713e15 I have observed values in the range of 2.484e15 without presolve and 5.076e15 with. Barrier will yield results of the order of 1023.
Attachment: MPS file describing the programme https://projects.coin-or.org/Clp/attachment/ticket/23/degenerate.mps Comments: Might also add that the objective is positive semidefinite
Owner changed from somebody to forrest
Status changed from new to assigned
Matrix should only have one copy of off diagonal elements. Still gives problems but I can get 1.71323e15