[BradyPlanden]

Description

Several solver inputs could be wrong without anyone noticing:

• model.convert_to_format was a plain attribute, so a typo was accepted and surfaced at solve time, if at all.
• JSON save/load dropped convert_to_format, so a model saved as "python" came back as "casadi".
• IDAKLUSolver and NonlinearSolver merged unknown options keys through, so {"num_thread": 4} silently kept the defaults.
• Each solver carried its own copy of the input-stacking and flattening code.

Consolidating that code turned up several bugs. Restarting a sensitivity solve from an output_variables solve seeded the state sensitivities from the output sensitivities: repeated step calls failed with an unrelated shape error or returned wrong gradients, and an SPM discharge-then-rest Experiment returned a voltage sensitivity 19% below finite differences. Several input sets with nproc=1 went through a one-process multiprocessing pool, which ran the same solves in the same order as a loop but spawned an interpreter and pickled the model and every solution. For a 32-set DFN with calc_ic=False, finding consistent initial conditions took 1.8 s instead of 3 ms, and on macOS and Windows an unguarded script hung. There were also three smaller bugs: a per-state atol array crashed the root-solver setup, AlgebraicSolver's "minimize" gradient was wrong for non-CasADi formats, and the initial event check raised UnboundLocalError for convert_to_format=None.

Models and simulations pickled by earlier versions still load, and JSON files written by earlier versions load with the model's default format.

• BaseModel.convert_to_format becomes a validated property that raises OptionError, with a __setstate__ migration for old pickles and a uses_stacked_inputs flag that names the input convention rather than the format.
• The JSON serialisers write and read convert_to_format.
• Shared stack_inputs/flatten_inputs helpers in base_solver.py replace the per-solver copies.
• BaseSolver._overlay_options rejects unknown option keys, and IDAKLUSolver and NonlinearSolver use it.
• BaseSolver._integrate loops in the calling process when nproc == 1.
• step() rejects a sensitivity restart from an outputs-only solution with a SolverError. In an Experiment this ends the experiment after its first step, as other step failures do.
• Fixes for the root-solver atol, the "minimize" gradient and the None-format event check.

This is one of the backend-neutral pre-flight PRs split out of the Rust execution engine (#5732). It has no Rust dependency.

Type of change

Add an entry under # [Unreleased] in CHANGELOG.md, in one of ## Breaking changes, ## Deprecated, ## Features, or ## Bug fixes (include PR number; breaking and deprecation entries need a one-line migration note). Internal-only PRs — refactor, docs, CI, tests — can skip this. See RELEASE.md for the full policy.

Important checks:

Please confirm the following before marking the PR as ready for review:

• No style issues: nox -s pre-commit
• All tests pass: nox -s tests
• The documentation builds: nox -s doctests
• Code is commented for hard-to-understand areas
• Tests added that prove fix is effective or that feature works

[codecov]

Codecov Report

❌ Patch coverage is 98.75000% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 98.00%. Comparing base (2677897) to head (d41088c).

Files with missing lines	Patch %	Lines
...ackages/pybamm/src/pybamm/solvers/idaklu_solver.py	85.71%	1 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #5784   +/-   ##
=======================================
  Coverage   98.00%   98.00%           
=======================================
  Files         346      346           
  Lines       35259    35300   +41     
=======================================
+ Hits        34554    34595   +41     
  Misses        705      705           

☔ 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.

[martinjrobins]

looks great, but I don't think convert_to_format = None is valid? this is just the same as , say, convert_to_format = 'asdf', which you are erroring on.

[github-actions]

Bencher Report

Project	PyBaMM
Branch	a3/solver-inputs-options
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