Finalise the star shear-response test definition (star cutout vs input PSF) #772
Description
The star shear-response standing test (Tier-1 guardrail #763) needs its exact definition pinned. The star-response investigation showed the metacal response is ≈0 when the exact PSF (the star's own cutout) is used, and climbs at the bright end with the interpolated PSFEx model — so an "exact-cutout" test definition returns ≈0 by construction (trivially satisfiable), while an "input simulation PSF" definition measures something real but conflates PSF-model fidelity.
The open question (2026-06-23 tomography call): a definition that is both physically meaningful and not trivially satisfiable. The meeting proposed redefining the test to use the star cutout directly, to be settled offline.
Meeting record assigns this to Cail + Fabian. Feeds the Tier-1 guardrail #763 and the verification-suite epic #765.
— Claude on behalf of Cail
Activity
cailmdaley commentedon Jul 16, 2026
let's use the interpolated PSF model as fiducial, and ablate if we need to understand the problem
cailmdaley commentedon Sep 6, 2026
Decision (2026-07-16): the star shear-response test uses the interpolated PSF model as its fiducial input; the exact-cutout variant is kept as an ablation for when the bright-end response needs diagnosing. Closing; the test definition lives in #765.