S2MPJ Python Subset

This repository provides a specialized Python-only subset of the S2MPJ collection.

Contents

This repository preserves only the files relevant to Python users from the original source. These files are located in the src/ directory:

• src/python_problems/: Directory containing the optimization problems converted to Python.
• src/list_of_python_problems: A listing of all available problems.
• src/s2mpjlib.py: Supporting library script.

OptiProfiler Lifecycle

S2MPJ is the bundled default Python problem library in OptiProfiler. Ordinary users install it with the core optiprofiler distribution and should not install this repository as a separate Python package. It is discoverable under the public name s2mpj:

from optiprofiler import list_problem_libraries

assert "s2mpj" in list_problem_libraries()

Select it with benchmark(..., plibs=["s2mpj"]); no custom filesystem path is needed. Removing optiprofiler also removes its bundled S2MPJ files, but it does not remove benchmark output or other user data.

This repository keeps a reviewed S2MPJ snapshot for OptiProfiler maintenance. An automated workflow checks upstream and reports differences, but it never changes src/, metadata, or the OptiProfiler lock. Users receive a new S2MPJ snapshot only after maintainers review the candidate, commit it explicitly, update the locked gitlink, and publish or install a matching core revision.

Configuration

Selection limits use finite integer lower bounds (mindim >= 1, other min* >= 0). Each of maxdim, maxb, maxlcon, maxnlcon and maxcon accepts an integer at least the corresponding lower bound or float('inf') for no upper cutoff. NaN and negative infinity are not valid limits. benchmark validates these inputs; direct s2mpj_select callers should provide criteria satisfying the same contract.

The file config.txt in this directory controls how s2mpj_select filters problems (e.g., variable_size and test_feasibility_problems). See the comments in config.txt for a full description of each option.

This repository keeps the legacy s2mpj_load / s2mpj_select interface while also exposing the same API-v1 adapter callbacks used by separately installed problem-library plugins.

For a reproducible OptiProfiler experiment, pass the options explicitly for this run:

from optiprofiler import benchmark

benchmark(
    solvers,
    plibs=['s2mpj'],
    plib_options={
        's2mpj': {
            'variable_size': 'all',
            'test_feasibility_problems': 2,
        },
    },
)

OptiProfiler stores the validated effective mapping with the experiment. For a process-level default shared by subsequent calls, the compatibility API remains available:

from optiprofiler import set_plib_config, get_plib_config

# View the current effective configuration
print(get_plib_config('s2mpj'))

# Override subsequent calls in the current Python process
set_plib_config('s2mpj', variable_size='all', test_feasibility_problems=2)

The precedence is per-run plib_options, process-level set_plib_config, environment variables, config.txt, then built-in defaults. You can also set S2MPJ_VARIABLE_SIZE and S2MPJ_TEST_FEASIBILITY_PROBLEMS directly. The adapter merges these layers first and validates the final mapping once, so an explicit valid per-run value can replace an invalid lower-priority value.

Testing

The CI workflow runs daily and on pushes. It checks the OptiProfiler adapter layer by:

• selecting a small set of representative u, b, l, and n problems;
• loading each selected problem through s2mpj_load;
• evaluating fun, cub, and ceq at the initial point;
• checking variable_size and test_feasibility_problems environment overrides;
• checking the OptiProfiler API-v1 adapter callbacks used by the core loader;
• checking nonlinear lower/upper/two-sided constraint Hessians against Jacobian differences, including stacking and multiplier order;
• sampling a few additional small problems each day with at most two numerical-library threads.

Locally, from this repository:

python -m unittest discover -s tests -p 'test_*.py'

Maintenance

Check S2MPJ Upstream compares the managed Python subset with the latest GrattonToint/S2MPJ revision every day. A difference creates or updates an upstream-update issue and uploads a report. The workflow has no permission to push source changes. Collect Info is manual and uploads candidate metadata as an artifact; adopting either source or metadata requires a reviewed commit.

Provenance and Citation

The files under src/ originate from S2MPJ by Serge Gratton and Philippe L. Toint. Their BSD-3-Clause license is preserved verbatim in LICENCE.txt. The third-party notice records the exact license revision, source comparison and attribution, and distinguishes the independently added OptiProfiler adapter and metadata. The upstream copyright notice does not assign ownership of those independent additions.

Please cite S. Gratton and Ph. L. Toint, S2MPJ and CUTEst optimization problems for Matlab, Python and Julia, Optimization Methods and Software 40(4), 871-903 (2025), doi:10.1080/10556788.2025.2490640.

Distributions must carry LICENCE.txt and THIRD_PARTY_NOTICES.md together with the source. CI checks their exact contents in source archives and in the core wheel/sdist built with this bundled provider. To check a built archive:

python3 tests/check_distribution.py /path/to/archive.whl

For the full collection or other languages, please visit the original repository.