Duality gap as a benchmark metric #9
Replies: 3 comments 2 replies
-
A consequence of this fact is that the solutions returned by some solvers can be severely sub-optimal. The benchmark shouldn't trust them to be optimal, and perform its own independent check. Idea 1: the benchmark checks the duality gap directly
|
Beta Was this translation helpful? Give feedback.
-
Idea 2: check the cost errorTake the primal solution returned by a solver, compute the corresponding cost, compare it to the known optimum.
|
Beta Was this translation helpful? Give feedback.
-
There is some follow-up to this discussion in the ProxQP documentation:
|
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
OSQP does not constrain the duality gap which can make it artificially faster, but in reality it's returning incorrect solutions, e.g. if the problem is strictly convex and thus a zero duality gap is part of optimality conditions.
@bodono discusses this in a paper:
Original context: bodono/scs-python#63 (comment)
Beta Was this translation helpful? Give feedback.
All reactions