Validation: src/proteus/outgas/compaction.py¶
This page tracks the @pytest.mark.reference_pinned tests that anchor
proteus.outgas.compaction, the melt drainage of the front scheme.
| Test id | Reference | Scope |
|---|---|---|
tests/outgas/test_compaction.py::test_mobility_reproduces_the_interior_solver_in_every_regime |
Cross-implementation check against the interior solver: aragog.eos.entropy_phase.EntropyPhaseEvaluator.relative_velocity (fwl-aragog), the three-regime permeability of Bower et al. (2018), section 2.1 |
Calls the solver's own separation velocity on a stand-in mush and backs the mobility out of it, for three grain sizes and porosities spanning all three regimes and both blends. Agreement is rtol=2e-4 from a porosity of 0.05 up and 1.5e-3 below, where the solver's soft porosity clip (width 1e-3) lifts the porosity by about 1e-6 / (4 phi). Also pins the drainage velocity as the solver's velocity times (1 - phi), the buoyancy taken against the mixture (McKenzie 1984, Eq. 3.7) rather than the solid (Abe 1995, Eq. 40). The mobility is the solver's own aragog.eos.mobility_function, so this checks that the drainage uses the law the separation velocity uses. |
tests/outgas/test_compaction.py::test_the_lever_rule_porosity_is_the_solvers_where_the_melt_is_lighter |
Cross-implementation check against the interior solver's porosity, aragog.eos.entropy_phase.EntropyPhaseEvaluator._porosity_from (fwl-aragog 26.10.03), which it writes as porosity_b |
On a two-phase mush with the melt lighter, the lever rule the trapping step applies to the solver's basic-node densities equals the solver's porosity to atol=2e-5, the soft clip of 2.5e-7 / phi at each edge. Where the melt is the denser phase the solver's porosity reads below 1e-3, as its docstring warns, while the lever rule returns the melt volume, and a fully molten node takes its melt fraction, 1, where the lever rule on its single-phase density gives 0.95. |
tests/outgas/test_compaction.py::test_drainage_matches_an_independent_quadrature_over_five_martian_fronts |
Analytic limits and a reference calculation: the closed-form timescales tau_D = L eta_m / ((1 - phi) drho g F(phi)) and tau_s = mu_s / (drho g L); the separable form of the drainage equation, t_res = int_F^phi_c max(tau_D, tau_s) dphi / phi, evaluated by quadrature; and its series solution in the Rumpf-Gupte regime, H(phi) = sum_k phi^(k - 4.5) / (k - 4.5) falling by C t_res / L |
A 60 km martian front entered at phi_c = 0.3 (drho = 330 kg/m3, g = 3.711 m/s2), crossed at 70 and 5 cm/yr, for five sets of grain size, melt viscosity and mush viscosity. The shipped model, (1 - phi) included, matches the quadrature to rel=1e-6 in all ten trapped fractions and the series to 1e-5 in the three Rumpf-Gupte cases; the fractions are pinned at rel=1e-4 (0.27165 and 0.18134 down to 0.0094342 and 0.0025379) and the timescale ratios likewise (tau_D / t_res from 8.1653 to 8.1653e-5, tau_s / t_res 0.50314, 50.314 and 0.0050314). Dropping (1 - phi) from w_D shifts the percolation-limited fractions by more than 0.007. |
Analytic limit checked alongside¶
test_drainage_is_limited_by_the_slower_of_the_two_processes pins the
matrix-limited case to its closed form phi_c exp(-t_res / tau_s) at
rel=1e-6.