[DNM] Adds Plasmo.jl interface in contrib folder for benders algorithms on linopy models

[coroa]

Changes proposed in this Pull Request

Adds a new sub-packge linopy.contrib.plasmo to run Benders or nested benders decompositions on linopy models.

A partitioning algebra that is available through linopy.contrib.plasmo.Partition and a few small helpers allows splitting up a linopy model into named nodes based on its dimensions. Each nodes is subset of variables and constraints. Variable sets are typically overlapping, while constraint sets are disjunct. Within Python CSR matrices and variable bounds for each node are constructed, passed to helpers in a small julia module and constructed as individual subgraphs in a Plasmo.jl OptiGraph. A Benders or NestedBendersAlgorithm can then be used to solve the OptiGraph in Julia/JuMP.

The result is retrieved as a Result object that can be assigned to the original Linopy model.

The contrib.plasmo code probably will eventually not live in linopy, but instead a small adjacent package. The experiment folder is a small example which should move to the docs (but without the dependencies).

Look at the experiment/run_benders.py file or the commented experiment/plasmo-benders-decomposition.nb notebook, both run benders decomposition on the small testProblem that was provided by @leonardgoeke through https://github.com/leonardgoeke/Linopy2Plasmo.jl , see also discussion in PyPSA/PyPSA#1712 .

Dependencies

The interface depends on juliacall and juliapkg to set up a julia installation and an environment with the required Julia packages. They are set up automatically within the pixi environment in the experiment folder.

Open issues

• Lacks implementation for non-continuous variables and special constraint types.
• Experiment should not be part of the repository
• Unit tests

Checklist

• AI-generated content is marked (see AGENTS.md).
• Code changes are sufficiently documented; i.e. new functions contain docstrings and further explanations may be given in doc.
• Unit tests for new features were added (if applicable).
• A note for the release notes doc/release_notes.rst of the upcoming release is included.
• I consent to the release of this PR's code under the MIT license.

[coroa]

@odow Promised ping, please have a look.

[codspeed]

Merging this PR will degrade performance by 14.68%

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 1 improved benchmark
❌ 3 regressed benchmarks
✅ 177 untouched benchmarks
⏩ 181 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
❌	test_to_lp[qp-n=1000]	1.9 MB	2.6 MB	-24.96%
❌	test_to_lp[nodal_balance_sparse-severity=50]	2.8 MB	3.7 MB	-24.08%
❌	test_to_lp[merge_balance-severity=0]	2.7 MB	3.3 MB	-18.24%
⚡	test_to_lp[nodal_balance-severity=0]	3.7 MB	3.3 MB	+13.77%

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing feat/plasmo-jl-benders (8ccbdbd) with master (718c0c1)}

Footnotes

1. 181 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[odow]

Yeah we should chat about this.