Validation: src/zephyrus/boiloff.py
This page tracks the @pytest.mark.reference_pinned tests that anchor the boil-off branch of zephyrus.boiloff.
| Test id | Reference | Scope |
|---|---|---|
tests/test_boiloff.py::test_parker_mach_sonic_limit_and_shutoff |
Owen & Wu (2016), ApJ 817, 107 (isothermal transonic Parker wind, Lambert-W form) | The photospheric Mach number is 1 identically with the launch level at the Bondi radius (the analytical sonic-point limit), falls monotonically as the level retreats inward, and has collapsed by more than six decades at their published shutoff R_p/R_B = 0.1. |
tests/test_boiloff.py::test_lambda_equals_two_bondi_radii_over_rp |
Fossati et al. (2017), A&A 598, A90 (restricted Jeans parameter); Owen & Wu (2016), ApJ 817, 107 | The identity Lambda = 2 R_B / R_p holds for every mean molecular mass, which is what makes the Owen & Wu shutoff equal to Lambda = 20 for every composition. The literature band 15 to 35 brackets the default threshold. |
tests/test_boiloff.py::test_bondi_cap_is_misener_eq10_at_the_wind_temperature |
Misener et al. (2025), ApJ 980, 152, Eq. (10); Gupta & Schlichting (2020), MNRAS 493, 792, Eq. (10) | The Bondi cap equals the closed form at the wind temperature T_eq / 2^(1/4) on a hydrostatic launch level, differs from the same form at T_eq by 2^(-3/8) exp(Lambda_launch (2^(1/4) - 1)), and stands to the Parker rate as tau_launch (T_wind / T_launch) exp(1/2 - Mach^2 / 2), so it binds on an optically thin launch level and not on a thick one. |
tests/test_boiloff.py::test_luminosity_cap_carries_the_tidal_barrier_reduction |
Erkaev et al. (2007), A&A 472, 329, Eq. (17) and Table 1 | The luminosity cap divides by the tidally reduced escape barrier, so it rises by the enhancement factor 1 / K, whose value at xi = 3 is the 1.92 printed in their Table 1. K = 1 reproduces the untidal cap identically. |
Notes
The Bondi cap follows Misener et al. (2025, ApJ 980, 152, Eq. (10)) at their recommended wind temperature, which carries a factor e^2 the Gupta & Schlichting (2020, MNRAS 493, 792, Eq. (10)) original lacks, and the luminosity cap Gupta & Schlichting (2019, MNRAS 487, 24, Eq. (8)), whose barrier carries the Erkaev tidal factor at xi = R_Hill / R_p so that an admitted residual and the energy-limited rate it competes against measure the same barrier. The cap ordering and the gate semantics are asserted as physics invariants in the same file. The Tang et al. (2024, ApJ 976, 221, Eq. (8)) timescale comparison runs as a diagnostic only.
Anchor type
Analytical limit plus published benchmark (the printed shutoff).
Date of last comparison against the sources: 2026-09-28.