Document and Test Dual Sign Convention#3528
Conversation
|
I believe this affects MindtPy and GDPOpt as in the implementation of outer-approximation algorithms; we are making assumptions on the directions or the multipliers |
|
Are MindtPy and/or GDPOpt using the new solver interfaces in |
Not that I know of, but it might be worth revisiting this |
There's nothing to stop someone from using them as subsolvers--we have those as config arguments. And we aren't testing with them currently, so I think it is possible we could be running afoul of the dual sign convention? But of course, it was before too since we let anyone stick any solver in there... |
|
For what it is worth, I did not have to change any code for this PR. I just added documentation and tests. |
|
I still need to address maximization problems as mentioned in #3519. I'll convert this to a draft for now. |
Robbybp
left a comment
Robbybp
left a comment
There was a problem hiding this comment.
Can you document the following:
- What happens for maximization problems?
- How does this formulation map onto "range inequalities"? (I believe the dual's sign implies one side and the other side is implicitly zero.)
|
Great suggestions. Will do. |
|
Okay, this should be ready. @Robbybp - I added documentation and tests for maximization problems and range constraints. |
7 participants
Fixes #3519.
Summary/Motivation:
This PR adds documentation and tests for the dual sign convention adopted in the new solver interfaces.
Legal Acknowledgement
By contributing to this software project, I have read the contribution guide and agree to the following terms and conditions for my contribution: