[BradyPlanden]

Description

Models default to convert_to_format = "rust" instead of "casadi" (BaseModel._DEFAULT_CONVERT_TO_FORMAT). Every model is lowered to the Rust core unless it opts out with model.convert_to_format = "casadi". With the rest of the stack:

• the default IDAKLUSolver solves and reads rust models in Rust;

• ScipySolver, AlgebraicSolver and EIS evaluate them through compiled functions;

• CasadiSolver and CasadiAlgebraicSolver convert them to CasADi, logging the conversion at INFO for a model left at the default format and at WARNING for a format chosen explicitly, so the deprecated solvers' DeprecationWarning stays the one warning a default model sees.

• Serialisation: saving falls back to the default format, and a file saved without a convert_to_format key loads as "rust".

• Tests:

  • Tests that exercise CasADi on purpose pin "casadi": the CasADi solver suites (through autouse fixtures), CasADi leaves and pickles, AOT compilation, the lead-acid comparisons, the IDAKLU sparse-output sensitivity guard, and single tests in ten more files, among them the bitwise CasADi parity of input-dependent discontinuity times. Among those, test_multiple_inputs keeps covering CasADi with fewer solvers than threads.
  • standard_output_tests raises its absolute tolerance from 2e-14 to 5e-14, since round-off differs by backend.

• Docs:

  • A new user guide page, user_guide/rust_backend.rst, covers choosing the backend, its differences from CasADi, DiffsolSolver's output times and Hermite reads, and adding a new expression type.
  • The multithreading notebook gains a prose section on num_threads across the backends.
  • rust_idaklu_build.md and the pybamm-rust README point to the guide.

What raises on the Rust backend:

• IDAKLUSolver raises SolverError for sensitivities with respect to an input of width above 1, and solves with num_solvers equal to num_threads, as the guide says.
• return_jacp_stacked and TensorField output variables also raise SolverError.
• A symbol that cannot be lowered raises TypeError or NotImplementedError, and an interpolant table Rust refuses raises ModelError.
• There is no automatic fallback to CasADi.

Performance

A release needs pybammsolvers 0.12.0 on PyPI (#5961) and the platform wheels from #5958.

Type of change

Breaking change, under ## Breaking changes in CHANGELOG.md, with the migration note "Set model.convert_to_format = "casadi" to restore the previous backend."

Key checklist

• No style issues: pre-commit run
• Unit tests pass (N), as do the integration tests (N)
• The documentation builds: nox -s doctests
• Tests: the existing suites run on the Rust default, and the tests listed above pin "casadi". The CasADi solvers' conversion test, from feat(rust): evaluate rust models through compiled evaluators #5960, now checks the log level for a default and a chosen format

Rust execution engine stack: proposal #5667, showcase #5732

Each PR is one commit on the one above it.

#	PR	Adds
0	#5954	Function.create_copy keeps the function's class
0b	#5955	CasADi's RegPower and uniform-grid interpolant lookups match evaluate
1	#5956	pybamm-core: expression graph, typed IR, interpreters, forward-mode AD
2	#5957	pybamm-core: CompiledModel, coloured Jacobians, the C ABI
3	#5958	pybamm.rust bindings, built into the pybamm wheel
4	#5959	Symbol.to_rust
5	#5960	convert_to_format = "rust" for ScipySolver, AlgebraicSolver and EIS
6	#5961	pybammsolvers: IDAKLU over the C ABI
7	#5962	IDAKLUSolver on rust models
8	#5963	variables and sensitivities read in Rust
9	#5964	pybamm-core: diffsol DAE solver
10	#5965	pybamm.DiffsolSolver
11	#5966	"rust" as the default convert_to_format ← this PR

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 98.14%. Comparing base (b9bb622) to head (fd167d2).

Additional details and impacted files

@@                     Coverage Diff                     @@
##           stack/10-diffsol-solver    #5966      +/-   ##
===========================================================
- Coverage                    98.15%   98.14%   -0.02%     
===========================================================
  Files                          354      354              
  Lines                        37278    37279       +1     
===========================================================
- Hits                         36592    36589       -3     
- Misses                         686      690       +4     

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[BradyPlanden]

I'm marking this ready-for-review to get benchers results with rust as the default backend. It should not be merged until the rust backend is verified as stable.

[github-actions]

Bencher Report

Project	PyBaMM
Branch	stack/11-default-rust
Testbed	bare-metal

⚠️ WARNING: Truncated view!

The full continuous benchmarking report exceeds the maximum length allowed on this platform.

🚨 1 Alert

🐰 View full continuous benchmarking report in Bencher