Test Methodology

Status: CANONICAL

Validation Philosophy

SSZ validation follows a forward-prediction pipeline:

Mass M → r_s → Ξ(r) → D(r) → Observable → Compare to data

The chain is one-directional. Reverse inference (observable → Ξ → M) is used only for consistency checks, never for calibration.

Anti-Circularity Principle

SSZ formulas must never be calibrated against the data they predict:

Ξ formulas are derived from φ-geometry, not data
Regime/formula-domain boundaries are fixed canonical rules; r* values are D_SSZ = D_GR comparisons for declared Xi forms, not fitted thresholds
Δ(M) is from φ-geometry, not object-specific tuning
No per-object parameter adjustment is allowed

Test: test_anti_circularity.py in segmented-calculation-suite

Test Categories

1. Weak-Field Validation (SSZ ≡ GR)

SSZ must reproduce all confirmed GR results to measurement precision.

Test	Measurement	SSZ Prediction	Status
GPS time correction	45.85 μs/day	45.85 μs/day	✅ PASS
Pound-Rebka	Δν/ν = 2.46×10⁻¹⁵	2.46×10⁻¹⁵	✅ PASS
Mercury perihelion	42.98"/century	42.98"/century	✅ PASS
Eddington lensing	1.75"	1.75" (PPN)	✅ PASS
Cassini Shapiro	γ = 1.000021±0.000023	γ = 1 (exact)	✅ PASS
S2 star redshift	z = 0.00018	0.00018	✅ PASS

2. Strong-Field Predictions (SSZ ≠ GR)

Prediction	GR Value	SSZ Value	Testable?
NS redshift excess	z_GR	z_GR + 13%	NICER (2025-2027)
Shapiro strong-field	Δt_GR	Δt_GR + 12%	SKA (2028+)
BH shadow diameter	d_GR	d_GR - 1.3%	ngEHT (2028-2030)
D(r_s)	0	0.555	Indirect
Pulsar timing	Δt_GR	Δt_GR + 30%	NANOGrav (2025-2028)

3. Internal Consistency

Check	What it verifies	Status
C² continuity at blend	Ξ, dΞ/dr, d²Ξ/dr²	✅
r* comparisons	`D_SSZ(r) = D_GR(r)` for the declared Xi form	✅
Asymptotic flatness	D → 1 as r → ∞	✅
PPN recovery	β = γ = 1 in weak field	✅
Energy conditions	WEC, DEC, SEC for r ≥ 5r_s	✅

Test Infrastructure

Canonical All-Tests Snapshot

The current validation baseline is the all-tests run from 2026-05-07:

Metric	Value
Total expected tests/checks	1296
Total passed	1296
Total failed	0
Pass rate	100.0%
Source output	`E:\clone\ssz-all-tests\really-full-output.md`

See all_tests_2026-05-07.md. Older partial counts in this documentation set are superseded by this snapshot.

Repositories with Tests

Repository	# Tests	Focus
ssz-qubits	184	GPS, Pound-Rebka, S2, qubits
ssz-metric-pure	36	4D tensors, Einstein equations
segmented-calculation-suite	158	Core Ξ/D/regime
ssz-schumann	178	Schumann resonance
ssz-lensing	279	Gravitational lensing
Unified-Results	147	Unified validation framework
ssz-trajectories	63	Geodesic trajectory integration
g79-cygnus-tests	5	G79 LBV nebula
ssz-lagrange	54	Lagrange, Kerr, quantum corrections
segmented-energy	7	Stellar energy and astronomical objects
frequency-curvature-validation	82	PPN, Shapiro, lensing
chord-partition (local)	103	Eigenmodes and phi resonance
TOTAL	1296

Unified Validation

The all-tests runner executes native test runners per repository and captures verbose output in really-full-output.md. Unified-Results remains one component of the full validation suite, not the complete validation source by itself.

Reproducibility

All repos are public on GitHub (github.com/error-wtf)
Python 3.10+ with numpy, scipy, matplotlib
Automated test runner: run_consistency_suite.ps1
Commit hashes documented for exact reproduction

Method Assignment in Tests

CRITICAL: Tests must use the correct method for each observable:

Observable	Method	Test file
Time dilation	Ξ-based	test_time_dilation.py
Redshift	Ξ-based	test_redshift.py
Lensing	PPN (1+γ)	test_ppn_lensing.py
Shapiro	PPN (1+γ)	test_shapiro_delay.py
Perihelion	PPN (β,γ)	test_perihelion.py

Using Ξ-only for null observables gives 50% of GR — this is NOT a bug, it's the g_tt-only contribution.

FilesExpand file tree

test_methodology.md

Latest commit

History