[claude]

Tests-only change to tests/test_magnitude_scaling.py, covering source_modelling/magnitude_scaling.py.

What the new tests check

• Leonard rake boundaries: leonard_area_to_magnitude uses the strike-slip constant (3.99) at exactly ±45° and ±135°, and the dip-slip constant (4.0) at ±46° and ±134°.

• Leonard random coefficients: with a fixed seed, 2000 samples have the mean and standard deviation from Table 4 of the paper. That is 3.99 ± 0.26 for strike-slip and 4.03 ± 0.3 for dip-slip. The existing test only checked that the samples were normally distributed.

• Contreras interface sign convention: the normal sample is mocked to a fixed deviation. contreras_interface_area_to_magnitude must then invert contreras_interface_magnitude_to_area exactly, and the area must equal exp(a1 + a2·M + δ).

• Range warnings and boundaries:

  • Contreras interface area: no warning at 137.76 or 138; an exact message match below that.
  • Contreras interface magnitude, in both the area and aspect-ratio functions: warns at 5.9, not at 6.0.
  • Strasser slab area: no warning at 130 or 5212; warns at 129.9 and 5212.5.
  • Strasser slab magnitude: warns at 5.8 and 7.9.
  • Contreras slab aspect ratio: warns below magnitude 5.
  • Leonard width: warns outside the strike-slip range (3.4 to 19 km) and at or below the dip-slip limit of 5.4 km.

  Each of these tests also checks the value returned.

• Aspect ratio values: the ratio is 1 below the hinge magnitude and exp(a3·(M − m1)) above it, for both the interface and slab relations.

• Missing rake: magnitude_to_length_width, magnitude_to_area and area_to_magnitude raise ValueError for LEONARD2014 when no rake is given.

Coverage of magnitude_scaling.py

Before: 94%. Lines 250, 256, 400, 437, 522, 572, 613, 690, 735 and 773 were not covered.
After: 100%.

Surviving mutants

Killed (23):

• leonard_area_to_magnitude mutants 6–11 and 13: the rake boundary comparisons and constants.
• Mutants 20–23 and 32–35: loc or scale dropped or changed by 1. Caught by the sample mean and standard deviation test.
• contreras_interface_area_to_magnitude mutants 2 and 3: the threshold changed to <= or 138.76.
• Mutants 4–7: the warning message changed. Caught by the exact message match.
• Mutant 28: - sigma_a changed to + sigma_a. Caught by the inversion test with a fixed deviation.
• strasser_slab_area_to_magnitude mutants 3 and 5: the area bounds changed to 131 and 5213.

Equivalent (1):

• contreras_interface_area_to_magnitude mutant 13 removes loc=0 from sp.stats.norm(loc=0, ...). The default loc is already 0, so the behaviour cannot change.

No issues filed.

🤖 Generated with Claude Code