Additional on Microlensing observables + PDF vetting report + Harris+2026 citation

[eagnespuerto]

Finally adds the frequently used TIC + sector direct-fetch mode (parallels the transit tab's MAST input) and refactors coord- and TIC-keyed autoload to share the sector-walk

• parse + pack pipeline.

Observable parameters (from PSPL fit alone, no external priors):

• t0 in BTJD and BJD
• Einstein timescale tE with 1σ error
• Impact parameter u0 with 1σ error
• Intrinsic + blended peak magnification
• Peak brightening in magnitudes
• Einstein-crossing duration (2·tE·sqrt(1-u0²), 0 if u0>1)
• Magnification FWHM (numerical inversion of A(u) = (A_peak+1)/2)
• Source and blend flux fractions
• μ_rel under fiducial θ_E = 0.5 mas (bulge typical)

Predicted planet parameters (fiducial-lens bulge priors — M_L = 0.3 M☉, D_L = 6 kpc, D_S = 8 kpc):

• θ_E from sqrt(κ·M_L·π_rel) with κ = 8.144 mas/M_sun
• Physical Einstein radius r_E (AU)
• Transverse relative velocity v_rel (km/s)
• Closest-approach distance u0·r_E (AU)
• Planet-detection floor q_min ≈ (cadence/2·tE)² and corresponding planet mass in Earth/Jupiter units for the fiducial lens

Both blocks are surfaced in the classifier's results panel and included in the PDF vetting report.

PDF vetting report:

• New endpoint POST /api/microlensing/report (JSON in, PDF blob out) — accepts a fit result + metadata + optional base64 plot PNG, renders a reportlab-styled report matching the transit-pipeline PDF conventions (running header/footer, unified tables, verdict banner, Helvetica).
• Frontend: Download PDF report button in the export toolbar. Auto- rasterises the SVG plot and attaches it if not already cached.
• New file: backend/app/microlensing_report.py.

TIC + sector fetch (parallels transit MAST mode):

• New endpoint POST /api/microlensing/lightcurve_by_tic
• Shared _fetch_and_pack_lightcurve helper de-duplicates the sector-walk
  • parse + quality-mask + response-pack code used by both autoload paths.
• DataLoader UI adds TIC/sector inputs with List sectors + Fetch buttons, matching the transit tab's ergonomics.

Bugfix: microlensing_report style-name collision. reportlab's getSampleStyleSheet already registers h1-h5 as aliases; renaming to ml_h2 / ml_small / ml_verdict.

Credit for Harris, Dragomir, Bachelet, Fausnaugh & Johnson (2026), "TESS's First Bound Microlensing Planet — A Binary Microlensing Event Revealing a Planetary Companion toward the Galactic Plane", ApJL 1005 L33 (DOI 10.3847/2041-8213/ae7a50):

• New SCIENCE.md §17 (observable parameters math + planet-detection floor derivation) and §18 (TESS-specific caveats informed by the paper: short single-sector baseline, 21″ pixel blending, single-band achromaticity limitation, and the standard best-practice of joint TESS + Gaia fitting via pyLIMA plus difference-image photometry).
• Footer credit line in MicrolensingPanel with DOI link.
• Sensitivity-floor note in compute_planet_predictions references the Harris result and TESS-only detection limits.

Tests: 26 pass (13 microlensing + 13 coverage), including 3 new observables tests (Paczyński A_max match, wide-u0 edge case, response integration).

Done with the aid of Claude Code. Ideas are purely human but implementation is machine-assisted.