Skip to content

roadmap.md

Latest commit

 

History

History
87 lines (69 loc) · 4.55 KB

roadmap.md

File metadata and controls

87 lines (69 loc) · 4.55 KB
Raw

Roadmap — beyond HST/ACS phase 1

Skeleton only: each section becomes its own design pass (and its own PyAutoMind prompt) when its predecessor nears shipping. Nothing here is committed design; it exists so phase-1 code doesn't paint us into a corner.

Instrument-adapter abstraction (the phase-1 obligation)

The one thing phase 1 must get right for everything below: pipeline stages (acquire/align/drizzle/noise/psf/package) speak to instruments only through an adapter in autoreduce.instruments that owns detector geometry, calibrated-product naming (_flc vs _flt vs _cal), units and gain, recommended combine parameters, and the PSF-model source. ACS/WFC is adapter #1; nothing outside instruments/ may mention a detector by name.

HST/WFC3 (IR + UVIS) — in progress (phase 2, PyAutoReduce#4)

  • Design deltas live in wfc3.md; adapters wfc3_uvis / wfc3_ir implemented. UVIS is ACS-like (CTE-corrected _flc); IR differs (_flt, no CTE correction, up-the-ramp CR rejection, 0.128″ native → recommended 0.065″ output).
  • Other ACS/WFC3 filters (F435W, F606W…) are config, not code: the adapter already parameterizes the filter-dependent pieces.

JWST (NIRCam first) — in progress (phase 3, PyAutoReduce#6)

  • Design deltas live in jwst.md; adapters nircam_sw/nircam_lw
    • the combine-backend dispatch (astrodrizzle | jwst_image3) implemented; noise = read propagated ERR × R; validated on the COSMOS-Web ring, four bands, against the autolens_assistant demo dataset.
  • PSF: STPSF tier-2 back-end still open (tier-1 ePSF carries over).

Keck NIRC2 AO — in progress (phase 4, PyAutoReduce#11)

  • Design deltas live in keck_ao.md; adapters nirc2_narrow / nirc2_wide, the acquire-backend seam (archive: koa | mast, PyKOA), the ground pre-combine stages (calibrate, sky), the nirc2_native combine backend (distortion-as-drizzle-pixmap), and the provisional-PSF contract (tier-A PSF-star candidates) implemented. SHARP-grounded (Lagattuta 2012; Chen 2016/2019); validation anchor B1938+666.
  • Open: wide-camera distortion, subarrays, MULTISAM read noise, north-up resampling (see keck_ao.md open items).

ALMA interferometry — in progress (phase 5, PyAutoReduce#14)

  • Design lives in alma.md; the first visibility-domain product family: calibrated measurement sets → the al.Interferometer.from_fits triplet (visibilities / uv_wavelengths / noise_map, (Nvis, 2)), via the visibility branch acquire → split → extract → assemble → package.
  • Headless modular CASA (casatools/casatasks) replaces the interactive casa shell; grounded in an active ALMA modeler's continuum recipe; validation anchor 2016.1.00282.S / G09v1.40.
  • Open: scriptForPI restore automation, emission-line/cube extraction (reduction side of the shipped alma-datacube modelling work).

Per-exposure frame products (_flt/_flc with cosmic rays)

Shipped (HST) as the opt-in TargetSpec.frame_products packaging mode — per-exposure cutouts, ERR-based native-pixel noise maps, DQ + deepCR cosmic-ray masks, the WCS manifest with measured relative registration (issue #19), and tier-1 native ePSFs per frame (issue #21); design in hst_acs_pipeline.md ("Per-exposure frame products"). Open items:

  • Tier-2 model PSF, HST side (TinyTim / focus-diverse ePSF grid) — the fallback for HST frames whose star field cannot support a tier-1 ePSF (recorded per frame in the manifest today); shared with the mosaic path's psf/fallback.model_psf stub. The JWST side is live (issue #29, STPSF tier-2b — jwst.md PSF tiering).
  • JWST analogue — shipped (issue #27): frames/registration/PSF chain extended to NIRCam per jwst.md §"Per-exposure frame products — feasibility" (_crf inputs, native MJy/sr, DO_NOT_USE-only masking).
  • Per-adapter dq_bad_bits — refinement of the any-nonzero-bit masking policy, if a consumer ever needs to keep e.g. warm-pixel bits.

Other instruments / surveys

  • Ground-based (e.g. archival CFHT/Subaru for environment studies) and other space missions as need arises — adapters again.
  • Euclid VIS is explicitly out of scope: Euclid reduction and lens-modeling glue live in euclid_strong_lens_modeling_pipeline.

Quality-of-life (unscheduled)

  • Sample-level driver: table in → datasets out, streaming acquire→evict.
  • PSF-model library caching (focus-diverse ePSFs, STPSF grids) in the size- capped cache.
  • verify_install-style smoke path once the package is released.