Tighten analytic validation before visual upgrades

[Biswajit1999]

Why this pass

This upgrade follows the research-led sequence: scientific claims and equations first, then reproducibility and validation, before any new interface animation or rendering.

Repository worked on

Biswajit1999/Cosmic-Lens-Lab

Branch

science-validation-tightening-20260706

Changed files

• python/tests/test_lensing_invariants.py
• docs/validation.md
• docs/science-audit-2026-07-06.md
• README.md

Scientific checks added

• Point-mass scalar image-position invariants:
  • theta_plus + theta_minus = beta
  • theta_plus * theta_minus = -theta_E^2
  • each image maps back through beta = theta - theta_E^2 / theta
• Point-mass signed magnification parity and total absolute magnification consistency.
• SIS image-count boundary at |beta| = theta_E.
• SIS opposite-sign image locations inside the Einstein radius.
• SIS antisymmetry under beta -> -beta.

Documentation tightened

• Added a science audit that states the current thin-lens, single-plane, geometric-optics scope.
• Added a visual-truth policy: critical curves, caustics, magnification maps, and animations must trace back to equations or declared scene parameters.
• Updated validation docs with the new analytic invariant tier.
• Linked the audit from the README.

Validation status

Not merged. This is a draft PR for review. The expected local validation command is:

PYTHONPATH=python pytest python/tests -q

I could inspect and update the repository through GitHub, but I did not run the test suite in a local checkout during this automation pass.

Next scientific gap

Add browser/Python parity fixtures: export one point-mass scene from the web app, recompute the same ray mapping in Python, and assert agreement away from the singular origin and critical curve.