feat: first-class lensing latent variable API in PyAutoLens - #534
Merged
Conversation
Add autolens/analysis/latent.py with a registry of five tracer-derived latent variables — total_lens_flux_mujy, total_lensed_source_flux_mujy, total_source_flux_mujy, magnification, effective_einstein_radius — plus the latent_keys_enabled() config reader. autolens/config/latent.yaml provides the user-toggle layer with all keys default `false` (same regression reasoning as PyAutoGalaxy #441: compute_latent_samples runs on every fit, and magzero-requiring latents would crash existing users if enabled by default). AnalysisImaging gains LATENT_KEYS @Property + compute_latent_variables method, dispatching through the shared registry. NotImplementedError short-circuit when no latents are enabled keeps autofit's existing handler path clean. Helpers ab_mag_via_flux_from / flux_mujy_via_ab_mag_from are imported from PyAutoGalaxy (shipped in #441) — no duplication. The latent functions take a generic `fit` argument and use fit.tracer / fit.galaxy_image_dict / fit.dataset.grids.lp — APIs shared between FitImaging and FitInterferometer — so the registry can be wired into AnalysisInterferometer in a follow-up without code duplication. Refs: #533
This was referenced May 23, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Adds the lensing latent-variable catalogue to PyAutoLens, building on the infrastructure PyAutoGalaxy #441 put in place. Five tracer-derived latents now live in
autolens/analysis/latent.py:total_lens_flux_mujy— integrated flux of the lens galaxytotal_lensed_source_flux_mujy— image-plane flux of the source after lensingtotal_source_flux_mujy— source-plane intrinsic fluxmagnification— image-plane / source-plane flux ratioeffective_einstein_radius— viaLensCalc.einstein_radius_jit_fromAnalysisImagingoverrides PyAutoFit's stubbed hooks (LATENT_KEYS@property+compute_latent_variablesmethod) to dispatch through the registry. All five latents shipfalseinautolens/config/latent.yamlso the change is opt-in — existing fits remain unchanged. Users enable them in their workspace'sconfig/latent.yamland passmagzerovia Analysis kwargs.Closes #533. After this lands, sub-prompt #3 of the latent_refactor epic (euclid pipeline migration) can drop the bespoke
LATENT_KEYS+compute_latent_variablesineuclid_strong_lens_modeling_pipeline/util.py:306-490and inherit the library catalogue (aperture-flux latents remain Euclid-specific by user direction).API Changes
New public module
autolens.analysis.latentexposing the five latent functions +LATENT_FUNCTIONSregistry +latent_keys_enabled()reader.AnalysisImaginggainsLATENT_KEYS@propertyandcompute_latent_variables(parameters, model). Newautolens/config/latent.yaml(all keys defaultfalse). Helpersab_mag_via_flux_from/flux_mujy_via_ab_mag_fromare imported fromautogalaxy.imaging.model.latent(shipped in PyAutoGalaxy #441). No PyAutoFit changes. Latents take a genericfitargument and use APIs shared betweenFitImagingandFitInterferometer, so a futureAnalysisInterferometerwiring can reuse the registry without duplication.See full details below.
Test Plan
pytest test_autolens/analysis/test_latent.py -x -v— 17/17 passpytest test_autolens/— 311/311 pass (no regression)autolens_workspace/scripts/imaging/start_here.py, inspectlatent_summary.jsonFull API Changes (for automation & release notes)
Added
autolens.analysis.latent— new moduleautolens.analysis.latent.total_lens_flux_mujy(fit, magzero, xp=np)— total integrated lens-galaxy flux in muJyautolens.analysis.latent.total_lensed_source_flux_mujy(fit, magzero, xp=np)— image-plane source flux after lensingautolens.analysis.latent.total_source_flux_mujy(fit, magzero, xp=np)— source-plane intrinsic source fluxautolens.analysis.latent.magnification(fit, magzero, xp=np)— dimensionless flux ratioautolens.analysis.latent.effective_einstein_radius(fit, magzero, xp=np)— via LensCalcautolens.analysis.latent.LATENT_FUNCTIONS: Dict[str, Callable]— flat registry of name → functionautolens.analysis.latent.latent_keys_enabled(yaml_config=None) -> List[str]— config readerautolens.AnalysisImaging.LATENT_KEYS— new `@property` overriding the inherited class-level `[]`autolens.AnalysisImaging.compute_latent_variables(parameters, model)— overrides PyAutoFit's stub; returns tuple aligned to keys; raisesNotImplementedErrorwhenLATENT_KEYSis emptyautolens/config/latent.yaml— new config file (5 keys, allfalse)Changed Behaviour
false, soLATENT_KEYSis[],compute_latent_variablesraisesNotImplementedError, and autofit's existingexcept NotImplementedError: return Nonecleanly skips the latent pipeline. Matches today's behaviour.magzero(all four flux-derived latents), the latent computation raisesValueErrorloudly ifmagzerois not passed viaAnalysisImaging(..., magzero=<value>).effective_einstein_radiusandmagnificationdo not requiremagzero.Migration
None required. Existing code continues to work unchanged. To opt into a latent: set its key to
truein<workspace>/config/latent.yamland (for flux latents) passmagzerowhen constructingal.AnalysisImaging.🤖 Generated with Claude Code