Interfaces to the SAS solvers

[phchri]

Fixes

Provides an interface to the SAS LP/MILP optimization solvers.

Summary/Motivation:

Provides an interface to the SAS LP/MILP optimization solvers on both the legacy 9.4 platform (via the saspy package) and on the SAS Viya cloud platform (CAS, via the swat package).

I'm repurposing the solver_io option to choose between the two interfaces, if there is a better way to do this, please advise. The idea is that the interface chooses automatically depending on which package is available.

Changes proposed in this PR:

• Add the solver interface classes
• Add 52 SAS specific unit tests for the two interfaces, two are currently skipped because the status between LP and MILP is inconsistent (still investigating what's going on there, probably just a status reporting inconsistency on the solver side).

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:

1. I agree my contributions are submitted under the BSD license.
2. I represent I am authorized to make the contributions and grant the license. If my employer has rights to intellectual property that includes these contributions, I represent that I have received permission to make contributions and grant the required license on behalf of that employer.

[mrmundt]

Thanks for this PR! Overall, it looks good; however, we do not (as far as I am aware) have SAS in our testing environment. I will approve GitHub Actions to run for this PR to confirm.

Please look at the two test_*.yml files in https://github.com/Pyomo/pyomo/tree/main/.github/workflows

[mrmundt]

Confirmed my previous comment: https://github.com/Pyomo/pyomo/actions/runs/5374476640/jobs/9760403183?pr=2886#step:24:8439 <- SAS tests are skipped.

[phchri]

Confirmed my previous comment: https://github.com/Pyomo/pyomo/actions/runs/5374476640/jobs/9760403183?pr=2886#step:24:8439 <- SAS tests are skipped.

Too bad, I was hoping you might have an academic version of at least SAS 9.4 available somewhere.

I ran the unit tests with various SAS versions in my environment, is this good enough for you?

[mrmundt]

Too bad, I was hoping you might have an academic version of at least SAS 9.4 available somewhere.

I ran the unit tests with various SAS versions in my environment, is this good enough for you?

Unfortunately, no. Because this PR proposes adding the solver into the core offering of Pyomo, we need to ensure its capabilities on our various Python and OS versions before it can be merged.

If you have information on how we might go about obtaining an academic license, we are willing to pursue that.

[phchri]

I will investigate how to most easily give you a license or access to machines that run SAS. I'll get back to you as soon as I hear something.

Which institution would a license be for?

[mrmundt]

I will investigate how to most easily give you a license or access to machines that run SAS. I'll get back to you as soon as I hear something.

Which institution would a license be for?

Thank you! The primary developers are at Sandia National Laboratories, so likely there. Please feel free to email wg-pyomo@sandia.gov so we can discuss further.

[codecov]

Codecov Report

Patch coverage: 24.25% and project coverage change: -0.24% ⚠️

Comparison is base (176015b) 87.83% compared to head (ac2dfa3) 87.59%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2886      +/-   ##
==========================================
- Coverage   87.83%   87.59%   -0.24%     
==========================================
  Files         770      771       +1     
  Lines       89642    89976     +334     
==========================================
+ Hits        78733    78814      +81     
- Misses      10909    11162     +253     

Flag	Coverage Δ
linux	84.64% <24.25%> (-0.23%)	⬇️
osx	74.40% <24.25%> (-0.19%)	⬇️
other	84.81% <24.25%> (-0.23%)	⬇️
win	81.90% <24.25%> (-0.22%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files Changed	Coverage Δ
pyomo/solvers/plugins/solvers/SAS.py	24.02% <24.02%> (ø)
pyomo/solvers/plugins/solvers/__init__.py	100.00% <100.00%> (ø)

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[mrmundt]

Good news, @phchri ! We have SAS approval for our test machines.

[phchri]

This is great @mrmundt ! I will look into resolving your comments by the end of the week so that we can move ahead with this.

[phchri]

I did the requested changes. It's now possible to add options for saspy.SASsession or swat.CAS when creating the solver factory. All options are just passed through. For example:

SAS94: opt = pyo.SolverFactory('sas', solver_io="_sas94", cfgfile="sasconf.py")
SAS Viya: opt = pyo.SolverFactory('sas', solver_io="_sascas", hostname="HOSTNAME", port="4711", passsword="YOURPASSWORD")

I also fixed a few issues with non-optimal return codes and the version checking.

[phchri]

For the unit tests, the test classes have session_options that are passed to the SolverFactory. For SASCAS (SAS Viya) the session options are currently set up to come from environment variables, but this could be adjusted. For SAS 9.4, the default is to use sascfg_personal.py, so currently no session_options are set, but that could be added to point to file or to specify the session options explicitly.

The documentation for the two types of sessions are here:
https://sassoftware.github.io/saspy/api.html
and here:
https://developer.sas.com/apis/swat/python/v1.4.0/generated/swat.cas.connection.CAS.html