Testing suite

This page describes the MORS test suite and how to run it. The suite is organised into tiers so that fast, mocked tests run on every pull request while the slower tests that run the real evolutionary-track model run nightly. Every source file has a companion test file: the spectral synthesis pipeline, the Spectrum class, the rotation, activity, and high-energy emission model, the rotation-evolution solver, the Cluster class, the Baraffe and Spada track interpolation, the coupled Star model, and the command-line, logging, and data-download utilities.

Test tiers

Every test file declares a tier marker that decides where and how often it runs:

The integration tests download and run the real tracks, so they need FWL_DATA set and take minutes; keeping them in the nightly tier keeps the pull-request feedback fast.

Running the tests

Prerequisites

Install MORS with the development dependencies:

pip install -e .[develop]

Download the stellar evolution data (required for the integration tier):

export FWL_DATA=/path/to/your/data
mors download all

Run the pull-request tiers

pytest -m "(unit or smoke) and not skip"

Run the full suite (including the real-track tiers)

pytest -m "(unit or smoke or integration or slow) and not skip"

Run a specific file or tier

pytest tests/test_spectrum.py
pytest -m unit
pytest -m integration

View coverage

coverage run -m pytest -m "(unit or smoke or integration or slow) and not skip"
coverage report

Test-quality tooling

Two checks run on every pull request alongside the tests:

# Confirm every test carries a tier marker
bash tools/validate_test_structure.sh

# Enforce the test-quality rules against a locked baseline
python tools/check_test_quality.py --check

The quality check rejects new happy-path tests: each test needs a docstring, an edge case, more than one meaningful assertion, and, on a physics source, a physical invariant. Two advisory reports list follow-up work:

python tools/check_test_quality.py --reference-pinned-status
python tools/check_test_quality.py --physics-invariant-status

Current test files

The suite runs about 270 tests and covers roughly 99% of the source. The mocked unit tier runs on every pull request; the real-model integration tier runs nightly.

Unit tier

• test_spectrum.py: the Spectrum class and band helpers. Wavelength-to-band lookup, surface / 1 AU flux scaling, Planck surface flux, spectrum sanitisation, band integration, spectral extensions, and TSV round-trip.
• test_synthesis.py: spectral synthesis. Flux-budget closure, band-scale factors, per-band rescaling, and the modern-property fit, with the stellar model mocked.
• test_physicalmodel.py: rotation, activity, and high-energy emission. The X-ray, EUV, and Lyman-alpha luminosities, the Rossby number and convective turnover time, the core and envelope torques, the mass-loss and escape relations, and the habitable-zone boundaries, with the structure model mocked.
• test_rotevo.py: rotation-evolution integration. The forward-Euler, Runge-Kutta, Runge-Kutta-Fehlberg, and Rosenbrock steppers and the spin-down invariants, with the rate law and structure mocked.
• test_cluster.py: the Cluster orchestration and percentile statistics, with the member stars mocked.
• test_miscellaneous.py: shared helpers. Index lookups, array loading, the activity lifetime, and the integrated-emission fluence.
• test_data.py: evolutionary-track download, with the network and filesystem mocked.
• test_cli.py: the command-line interface, with the downloads mocked.
• test_logs.py: the logger setup and the exception hook.

Integration tier

• test_baraffe.py: the Baraffe et al. (2015) track interpolation. Pins luminosity, radius, and insolation at two masses and asserts luminosity rises with mass. Baraffe tracks use time in years, not Myr.
• test_stellarevo.py: the Spada et al. (2013) structure-track interpolation. Pins the solar calibration and asserts the main-sequence mass-luminosity monotonicity.
• test_star.py: the coupled Star model. Pins the solar luminosity in cgs for a solar-mass star at the solar age and the model output for both time-integration methods, and exercises the full quantity-getter surface.

Validation anchors

Each physics source has a validation page naming the published benchmark or analytical limit that anchors its reference-pinned test: Baraffe tracks, stellar evolution tracks, Star model, rotation evolution, activity and habitable zone, cluster percentiles, Spectrum, and spectral synthesis.

Continuous integration

Coverage gates move one way: a pull-request check blocks any change that lowers them, and the thresholds are raised toward the 90% ecosystem target as coverage improves. The pull-request tier covers the mocked unit surface; the nightly tier adds the real-track coverage.