Skip to content

Inject controlled PSF-model error in make_data (testing)#828

Open
cailmdaley wants to merge 1 commit into
developfrom
pr/sim-psf-model-error
Open

Inject controlled PSF-model error in make_data (testing)#828
cailmdaley wants to merge 1 commit into
developfrom
pr/sim-psf-model-error

Conversation

@cailmdaley

Copy link
Copy Markdown
Contributor

make_data renders the galaxy with the true PSF and, by default, hands ngmix that same PSF as the model stamp — zero deconvolution error. Two optional knobs let the model stamp differ from the truth: psf_model_fwhm_ratio (size error) and psf_model_shear (shape error), so the digital twin can probe the one shape systematic metacal's response doesn't divide out.

Defaults reproduce the legacy output byte-for-byte; injected error perturbs only the PSF model stamp, never the galaxy image. Both pinned by tests.

🤖 Generated with Claude Code

— Claude (Opus) on behalf of Cail

make_data renders the galaxy with the true PSF and, by default, hands ngmix
that same PSF as the model stamp — zero deconvolution error, the one shape
systematic the digital twin could not previously probe. Two optional knobs let
the model stamp differ from the truth: psf_model_fwhm_ratio (size error) and
psf_model_shear (shape error). Defaults reproduce the legacy output byte-for-
byte; injected error perturbs only the PSF model stamp, never the galaxy image.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01C9zx7LC72tPMhNGtKnADpp
@martinkilbinger

Copy link
Copy Markdown
Contributor

@claude not sure I understand: The test asserts that shear and PSF are equal with and without perturbed PSF size and shape. Why whould these tests not fail? Would PSF size and shape residual (or is this not it?) not introduce additive and multiplicative shear biases?

@cailmdaley

Copy link
Copy Markdown
Contributor Author

yes exactly, this is less of a test-of-correctness than adding a knob to allow us to characterize PSF model error (e.g. what happens to the response when the PSF is mis-modelled in a controlled way).

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants