Validation: src/proteus/outgas/binodal.py
This page tracks the @pytest.mark.reference_pinned tests that anchor the
behaviour of proteus.outgas.binodal against the Rogers, Young &
Schlichting (2025) H2-MgSiO3 miscibility model.
| Test id | Reference | Scope |
|---|---|---|
tests/outgas/test_binodal.py::test_h2_mole_total_uses_h2_molecular_mass_per_rogers2025 |
Rogers, Young & Schlichting (2025), MNRAS, 544, 3496 (doi:10.1093/mnras/staf1940) | Pins the molar-mass conversion the binodal partition applies to every H2 reservoir against an independent computation using scipy.constants.Avogadro and M(H2) = 2.016e-3 kg/mol. The kg-closure atm + liquid + solid = total is structurally trivial for the sigma * T + (1 - sigma) * T + 0 assignment, so the discriminating assertion is the molar-mass pin (rel=1e-12): a regression mistaking M(H2) for the atomic mass M(H) = 1.008e-3 kg/mol would inflate the mole total by a factor of two and be caught. |
Re-derivation note
The Rogers et al. (2025) binodal partitions H2 between the atmosphere and the dissolved (liquid mantle) reservoir at the mass mixing-ratio set by thermodynamic miscibility, not Henry's law:
sigma = rogers2025_suppression_weight(P_surf, T_magma, w_H2, w_sil)
H2_kg_liquid = sigma * H2_kg_total
H2_kg_atm = (1 - sigma) * H2_kg_total
H2_kg_solid = 0 # H2 does not partition into solid silicate
By construction the three reservoirs sum to H2_kg_total for any sigma
in [0, 1], so the kg-closure is structurally trivial. The discriminating
check is instead the molar-mass conversion: the reference-pinned test patches
rogers2025_suppression_weight to a specific value (0.4), then verifies that
H2_mol_total = H2_kg_total / M(H2) / N_av uses the molecular mass
M(H2) = 2.016e-3 kg/mol rather than the atomic mass M(H).
Pending verification
The cross-implementation comparison against atmodeller's binodal handling is a separate work item, tracked under the CALLIOPE-vs-atmodeller science verification handover.