Killer Validation Index

Purpose

This page is the compact index for Moira's adversarial validation layer.

It does not replace the detailed result tables. It tells you which proof covers which failure mode, where the measured outputs live, and which suites to run when someone sends a new "killer test."

Validation layers

Layer	What it proves	Primary artifacts
Segment continuity	No public jump at real DE441 split boundaries	tests/unit/test_de441_segment_boundaries.py, wiki/03_release/KILLER_TEST_RESULTS.md
Delta-T robustness	Smooth, symmetric response to Delta T perturbations across coverage	tests/unit/test_ephemeris_stress_proofs.py, wiki/03_release/DELTA_T_500_YEAR_CHECKPOINTS.md
Coverage-edge truth	Finite results inside coverage and explicit failure outside it	tests/unit/test_ephemeris_stress_proofs.py
Conjunction solver rigor	Near-zero residual roots and time-reversal sanity at conjunctions	tests/unit/test_ephemeris_stress_proofs.py
Fixed-case routing coherence	Direct, chained, and routed topocentric paths agree at one TT-pinned epoch	tests/unit/test_topocentric_multi_path_consistency.py, tests/integration/test_topocentric_multi_path_horizons_anchor.py
Breadth gauntlet	Cross-epoch, cross-observer, cross-body route, round-trip, smoothness, and external-anchor stability	tests/unit/test_ephemeris_breadth_gauntlet.py, tests/integration/test_ephemeris_breadth_horizons_gauntlet.py
Polar high-latitude oracle	Supported polar-capable house systems match the cached Swiss oracle above critical latitude	tests/integration/test_houses_polar_external_reference.py, wiki/03_validation/POLAR_HOUSE_EXTERNAL_REFERENCE.md
Polar house gauntlet	Dynamic critical-latitude doctrine, strict-policy honesty, and fallback placement parity	tests/unit/test_polar_house_breadth_gauntlet.py
Polar chart public-path gauntlet	Chart vessel, house fallback, body placement, angularity, and lots remain coherent on the public API path	tests/unit/test_polar_chart_public_gauntlet.py
Public doctrine surface audit	Public doctrine and policy vessels remain reachable, frozen, default-constructible, and correctly classified	tests/unit/test_public_doctrine_surfaces.py, wiki/03_release/KILLER_TEST_RESULTS.md
Adversarial singularity suite	Coordinate transforms, planetary geometry, time system, house cusps, and boundary ownership under degenerate inputs; reveals silent failures, not correctness	tests/unit/test_adversarial_singularities.py, tests/unit/test_adversarial_house_singularities.py, wiki/03_validation/ADVERSARIAL_SINGULARITY_DEFECT_LEDGER.md
Compound adversarial singularity suite	Dual and triple seam composition across segment boundaries, TT/UT conversion, public chart fallback, deep-time chart products, and exact house ownership	tests/unit/test_adversarial_dual_singularities.py, tests/unit/test_adversarial_triple_singularities.py, wiki/03_validation/ADVERSARIAL_SINGULARITY_DEFECT_LEDGER.md
Hard-mode compound adversarial suite	Higher-pressure seam composition across public topocentric continuity, chart admission at coverage edges, downstream fallback products, subnormal pole-adjacent vectors, and station-neighborhood truth	tests/unit/test_adversarial_hardmode_singularities.py, wiki/03_validation/ADVERSARIAL_SINGULARITY_DEFECT_LEDGER.md
Quad-axis adversarial suite	Four-boundary public-path composition across apparent-route coherence, coverage admission, fallback doctrine, exact cusp ownership, and retrograde semantics	tests/unit/test_adversarial_quad_singularities.py, wiki/03_validation/ADVERSARIAL_SINGULARITY_DEFECT_LEDGER.md
Oracle-hostile semantic audit, wave 1	Public houses, Gauquelin sectors, and rise/set routes remain semantically aligned with cached Swiss and Horizons authorities, not merely stable	tests/integration/test_oracle_hostile_semantic_audit.py, wiki/03_validation/ADVERSARIAL_SINGULARITY_DEFECT_LEDGER.md
Oracle-hostile semantic audit, wave 2	Public fixed-star and twilight routes remain aligned with cached Swiss, ERFA, and USNO authorities across centuries and published boundary tables	tests/integration/test_oracle_hostile_semantic_audit_wave2.py, wiki/03_validation/ADVERSARIAL_SINGULARITY_DEFECT_LEDGER.md

How to read the stack

• The stress suite attacks local numerical pathologies: boundaries, Delta T, root-finding, and coverage edges.
• The breadth suite attacks routing drift: different observers, epochs, bodies, coordinate conversions, and compact external anchors.
• The polar high-latitude oracle attacks supported-system truth against Swiss at the actual polar cap.
• The polar-house suite attacks fallback doctrine: whether high-latitude house policy stays explicit and downstream-consistent.
• The polar chart suite attacks public-path coherence: whether fallback truth survives all the way up to chart assembly, placement, angularity, and lots.
• The doctrine audit attacks API drift at the policy layer: whether exposed doctrine vessels still exist, stay frozen, and preserve stable default construction semantics.
• The adversarial singularity suite attacks isolated seams under degenerate inputs.
• The compound adversarial suite attacks boundary marriages: whether two or three individually lawful seams remain truthful when they compose.
• The hard-mode compound suite attacks the same territory under higher pressure and with more downstream public-product assertions.
• The quad-axis suite attacks small hostile public pipelines: four lawful boundaries composed end-to-end.
• The oracle-hostile waves attack plausible lies: whether public routes remain semantically correct against stronger cached authority even when they already look stable internally.

Detailed reports

• wiki/03_release/KILLER_TEST_RESULTS.md
• wiki/03_release/DELTA_T_500_YEAR_CHECKPOINTS.md
• wiki/03_validation/POLAR_HOUSE_EXTERNAL_REFERENCE.md
• wiki/03_validation/ADVERSARIAL_SINGULARITY_DEFECT_LEDGER.md

Recommended run order

1. python -m pytest tests/unit/test_de441_segment_boundaries.py -q
2. python -m pytest tests/unit/test_ephemeris_stress_proofs.py -q
3. python -m pytest tests/unit/test_topocentric_multi_path_consistency.py -q
4. python -m pytest tests/unit/test_ephemeris_breadth_gauntlet.py -q
5. python -m pytest tests/unit/test_polar_house_breadth_gauntlet.py -q
6. python -m pytest tests/unit/test_polar_chart_public_gauntlet.py -q
7. python -m pytest tests/unit/test_public_doctrine_surfaces.py -q
8. python -m pytest tests/unit/test_adversarial_singularities.py tests/unit/test_adversarial_house_singularities.py tests/unit/test_adversarial_dual_singularities.py tests/unit/test_adversarial_triple_singularities.py -q
9. python -m pytest tests/unit/test_adversarial_hardmode_singularities.py -q
10. python -m pytest tests/unit/test_adversarial_quad_singularities.py -q
11. python -m pytest tests/integration/test_oracle_hostile_semantic_audit.py -q
12. python -m pytest tests/integration/test_oracle_hostile_semantic_audit_wave2.py -q
13. python -m pytest tests/integration/test_topocentric_multi_path_horizons_anchor.py -q
14. python -m pytest tests/integration/test_ephemeris_breadth_horizons_gauntlet.py -q
15. python -m pytest tests/integration/test_houses_polar_external_reference.py -q

Decision rule

• If a new challenge is about astronomical continuity, time scale behavior, or root precision, start in the stress suite.
• If it is about observer routing, coordinate transforms, or external anchors, start in the breadth suite.
• If it is about supported house systems at the polar cap, start in the polar external oracle.
• If it is about fallback doctrine for unsupported semi-arc systems, start in the polar-house suite.
• If it is about whether that fallback doctrine survives chart assembly or lot computation, start in the polar chart suite.
• If it is about whether a policy or doctrine surface is still publicly exposed and structurally stable, start in the doctrine audit.
• If it is about two or three seams interacting, start in the compound adversarial suite.
• If it still survives that first compound layer, escalate to the hard-mode compound suite.
• If it survives hard mode, escalate to the quad-axis suite.
• If it still survives while looking finite and stable, escalate to the oracle-hostile waves.
• If it survives all of those layers, it is probably testing a genuinely new semantic surface rather than an already-covered edge.