CuOpt Solver Integration

[Iroy30]

No description provided.

[CLAassistant]

All committers have signed the CLA.

[Iroy30]

Incorporated review comments from #837

[pchtsp]

Thanks. I'm not sure why you closed the previous PR. As long as you push the code to the branch, you can leave the PR open and it will update with the new changes. This makes easier to follow the changes/ suggestions and comments.

For example, there seems to be an error with the types when checking them with mypy, as I mentioned in the previous PR. I've added instructions to run mypy locally in the contributing site: https://coin-or.github.io/pulp/develop/contribute.html#checking-types-with-mypy

Also, you seem to not have signed the CLA in this PR.

[bbozkaya]

Hi, we have re-submitted this PR via the NVidia fork of the Pulp repo using NV org account, as discussed with COIN-OR and agreed upon. All the reviews requested for the original PR are addressed.
#839

[pchtsp]

as I mentioned, running mypy ./ is still failing. I've added the lines that need to be changed in the code.

[pchtsp]

change for:

            from cuopt.linear_programming import data_model  # type: ignore[import-not-found, import-untyped, unused-ignore]

[pchtsp]

change for:

import cuopt  # type: ignore[import-not-found, import-untyped, unused-ignore]

[Iroy30]

Updated, thanks!

[pchtsp]

I'd do the changes myself but I cannot push to nvidia/pulp/master it seems. You also need to merge the current master into the branch

[Iroy30]

@pchtsp Currently I am running all the pulp tests in test_pulp.py with CUOPT as the solver instance (Updated the test in the PR). I can paste the results as a comment here since the CI cannot run cuopt. Is there anything else we need to verify wrt testing?

[Iroy30]

Also mypy has started failing again. I can fix the type issue in test_pulp.py , however, I cannot reproduce the numpy import issue in apis/cuopt_api.py
Could you confirm it would be the right place to do a numpy import ?

[pchtsp]

hello, thanks for the reminder to look into this. Two things:

1. I fixed the typing issue in the master branch, you should just need to pull the master again.
2. for numpy I'd recommend just doing the same as for the other libraries: import numpy as np # type: ignore[import-not-found, import-untyped, unused-ignore] and hope it doesn't complain.

thanks and let's hope this is the last change : )

[Iroy30]

@pchtsp The final question :)
#839 (comment)

[pchtsp]

Sorry I forgot to answer, yes: please paste here all the logs from the tests so we can take a look. Ideally for the next release, we should try to set up the github actions. But with additional solvers what we mostly care is that we're not breaking other solvers, which doesn't seem the case. I will merge as soon as all the tests pass