-
Notifications
You must be signed in to change notification settings - Fork 87
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[Bridges] fix adjoint functions in SetDotInverseScalingBridge #2264
Conversation
Mosek still broken, but because you explicitly coded the bridge type into a test: |
This is weird. The adjoint is the transpose and the transpose if a Diagonal should be the matrix itself |
We should add a test that sets and gets ConstraintDualStart in the bridge test, it should clarify |
There is a test. But it just tests that the round trip preserves. So it doesn't help if it gets the wrong value... 😆 |
Isn't the MathOptInterface.jl/src/Bridges/Constraint/bridges/set_dot_scaling.jl Lines 90 to 108 in fdf2559
|
It's good that I left a comment to explain this to my future self 😆 . So this fix is indeed correct because the adjoint is not the transpose. Can you add a test that don't just check the round trip ? |
Potential fix for #2262 (comment).
I don't really understand what's going here, but it makes sense that the adjoint should be the opposite of the forward map? That's what it is for
SetDotScalingBridge
, so I think this is just a typo.Seemed to fix SDPA, so I'll re-run solver-tests on this branch: https://github.com/jump-dev/MathOptInterface.jl/actions/runs/6103555292