fix: rename composite ESOH phase-capacity inputs off the MSMR name pattern

[rtimms]

Fixes #5737.

Description

electrode_soh_composite named its phase-capacity InputParameters Q_n_1 / Q_p_1 / Q_n_2 / Q_p_2. Those match the deprecated-MSMR name pattern in msmr.py:

^(X|Q|w|U0|a|j0_ref)_(n|p)(_[ld])?_[0-9]+$

so when the inputs reach ParameterValues.check_parameter_values via BaseSolver._set_up_model_inputs, the deprecation branch fires. Two effects, on every composite / multi-phase initial-state solve:

1. A DeprecationWarning telling the user to rename a parameter they never set and cannot change — it names PyBaMM's own internal input.
2. A phantom key added to the returned dict: values[new_param] = values.get(param) adds e.g. Negative electrode host site occupancy capacity (1) [A.h] alongside Q_n_1. On this path it goes no further — _set_up_model_inputs keeps only the names the model declares, so the phantom entry is discarded before the solver sees it. The warning is the whole user-visible symptom; I overstated this in the issue and have corrected it there.

The collision is accidental: the composite Q_n_1 is a phase capacity (primary/secondary particle), the MSMR Q_n_1 is a host site capacity.

Fix

Rename to Q_n_prim / Q_n_sec / Q_p_prim / Q_p_sec, which matches the prim/sec vocabulary already used for the phases (param.n.prim, param.n.sec) and the existing symbol-suffix convention elsewhere (R_n_prim in standard_spatial_vars, the var_pts keys).

On whether this is breaking. No documented API takes these names: solve_split, solve_full, and get_initial_stoichiometries_composite all build them internally from parameter_values, their inputs kwarg is a different namespace (it is forwarded to parameter_values.evaluate), and the returned dicts hold only stoichiometries. They do not appear in summary_variables (all 102 use display names) or in the user model's solution.all_inputs ([{}]).

They are visible in one place, so I would not call them unreachable: the ESOH sub-simulation's solution, which a caller can hold via the public esoh_sim kwarg on solve_full —

ElectrodeSOHComposite.solve_full(0.5, pvals, options=options, esoh_sim=esoh_sim)
esoh_sim.solution.all_inputs
# [{'Q_Li': [...], 'Q_n_1': [...], 'Q_n_2': [...], 'Q_p_1': [...], 'SOC_init': [...]}]

Reading those keys is introspection rather than a documented contract, and code that does it fails with a loud KeyError rather than silently wrong numbers, so I have treated this as non-breaking and spelled the old -> new mapping out in the CHANGELOG instead. Happy to be overruled.

I also considered aligning with the "Primary: Negative electrode capacity [A.h]" convention and did not: PHASE_NAMES = ["Primary: ", "Secondary: "] (parameters/bpx.py:95) builds ParameterValues keys, whereas these are InputParameter names sitting in all_inputs next to Q_Li, V_init, SOC_init, and z_1. A long display-style string there would be the only one of its kind and would read as a parameter-set key.

Note the names are also built dynamically at _get_electrode_capacity_equation (f"Q_{e}_1" / f"Q_{e}_2"), which a grep for the literal strings misses; both forms are updated.

Narrowing _VALID_NAME_RE instead would be wrong: Q_n_1 genuinely is a legacy MSMR name, so the check is correct on the name it is handed. The problem was that the composite path chose that name for something else.

Verification

Before, on main:

import warnings, pybamm

model = pybamm.lithium_ion.SPMe({"particle phases": ("2", "1")})
sim = pybamm.Simulation(model, parameter_values=pybamm.ParameterValues("Chen2020_composite"))
with warnings.catch_warnings(record=True) as caught:
    warnings.simplefilter("always")
    sim.solve([0, 10], initial_soc="2.8 V")
print(len([w for w in caught if issubclass(w.category, DeprecationWarning)]))
# 3   ->   0 with this branch

Added TestElectrodeSOHComposite::test_phase_capacity_inputs_are_not_deprecated_msmr_names, which fails on main and passes here.

Local runs (tests/unit/test_models, test_parameters, test_simulation.py, test_base_simulation.py, test_experiments, test_serialisation, test_batch_study.py): 2770 passed, 13 skipped. test_solvers/test_idaklu_jax.py fails identically on unmodified main in my environment, so it is untouched by this change and left to CI.

Type of change

• Bug fix (non-breaking change which fixes an issue)

Checklist:

• No style issues: nox -s pre-commit
• All tests pass: nox -s tests (see note above on the environmental idaklu_jax failures)
• The documentation builds: nox -s doctests — not run; no docs touched
• Code is commented for hard-to-understand areas
• Tests added that prove the fix is effective

[codecov]

Codecov Report

❌ Patch coverage is 93.54839% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 98.20%. Comparing base (78fe10a) to head (7c62e60).

Files with missing lines	Patch %	Lines
...tery_models/lithium_ion/electrode_soh_composite.py	93.54%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #5738      +/-   ##
==========================================
+ Coverage   98.12%   98.20%   +0.08%     
==========================================
  Files         340      340              
  Lines       32743    32743              
==========================================
+ Hits        32128    32156      +28     
+ Misses        615      587      -28     

☔ 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.