Ompr v1

[dirkschumacher]

Towards a new MIPModel. It is currently named MIPModel2.

[sbmack]

Towards a new MIPModel. It is currently named MIPModel2.

Dirk, I am sort of confused. So MIPModel2 is a new branch from MIPModel with MILPModel being set aside? Is there a link with MIPModel2 formulation information and examples available for MIPModel2? Thanks, SteveM

[dirkschumacher]

Dirk, I am sort of confused. So MIPModel2 is a new branch from MIPModel with MILPModel being set aside? Is there a link with MIPModel2 formulation information and examples available for MIPModel2? Thanks, SteveM

Well, if everything works, there will be a new MIPModel which is near 100% backwards compatible. It won't affect MILPModel. As of right know, I would like to make that new version of MIPModel good enough also for larger models, such that we can deprecate MILModel. I believe the vectorized semantics just cause too many problems. But that is another discussion and not part of that PR.

[sbmack]

Well, if everything works, there will be a new MIPModel which is near 100% backwards compatible. It won't affect MILPModel. As of right know, I would like to make that new version of MIPModel good enough also for larger models, such that we can deprecate MILModel. I believe the vectorized semantics just cause too many problems. But that is another discussion and not part of that PR.

Thanks for the update. Agree about the problematic vectorized semantics in MILPModel. Ideally, an updated constraint function in MIPModel2 would accommodate sparse formulations, (row, column, value). I am not a developer, but it seems like the dplyr pivot_longer function could pre-process the sparse constraint set in rcv format and that could be passed to MIPModel2.

Again, appreciate your efforts. SteveM

[dirkschumacher]

Ideally, an updated constraint function in MIPModel2 would accommodate sparse formulations, (row, column, value). I am not a developer, but it seems like the dplyr pivot_longer function could pre-process the sparse constraint set in rcv format and that could be passed to MIPModel2.

@sbmack can you elaborate a bit what you mean exactly? Ideally post that in the Ideas category of the discussions. Appreciate your input!

[sbmack]

@sbmack can you elaborate a bit what you mean exactly? Ideally post that in the Ideas category of the discussions. Appreciate your input!

Thanks for the reply. Idea posted in the Ideas category of the Discussions section. SteveM