Problem libraries

Problem libraries#

OptiProfiler separates the benchmarking engine from the collections of optimization problems used by an experiment. The Python and MATLAB distributions include S2MPJ as the default library. Other providers can be installed independently without copying their source into the core package.

For a typical experiment, the workflow is:

  1. install OptiProfiler and any optional provider;

  2. inspect or select providers by their stable public names;

  3. pass those names to Python plibs or MATLAB options.plibs; and

  4. update, detach, uninstall, or remove files according to each provider’s ownership boundary.

The provider pages distinguish the OptiProfiler engine, the adapter or distribution, an upstream runtime or data collection, the MATLAB registry, and generated caches or experiment output. Uninstall and detach operations affect only their documented ownership boundary; they do not silently delete independently managed source, runtimes, or user data.