SEXA-Public-Validation

SEXA Master Public Audit

This repository is the audience-facing reproducibility package for the SEXA framework.

One-command run

Windows

Double-click:

RUN_WINDOWS.bat

macOS / Linux

Run:

bash RUN_MAC_LINUX.sh

Any Python 3 installation

python RUN_SEXA_MASTER_AUDIT.py

No third-party Python packages are required.

What this package does

The runner audits the supplied SEXA stack at several distinct levels:

1. Core dimensional and sexagesimal arithmetic.
2. Sigmatics cascade arithmetic, ratios, information-retention values, and 5-bit reconstruction.
3. Γ six-stage kill-switch semantics.
4. Horizon/discrete-geometry convergence stress tests.
5. Standard GR/QFT reference reproductions used by the SEXA compatibility papers.
6. Activation-threshold consistency.
7. Deterministic recursive-computation stress tests.
8. Source provenance, SHA-256 hashes, exact duplicate detection, and scan-only source detection.
9. Explicit classification of claims that cannot be independently regenerated from the supplied PDFs.
10. Explicit separation of computational survivability from experimental confirmation.

Result classes

• PASS — the coded target reproduced.
• FAIL — the stated numerical target failed its declared tolerance.
• REFERENCE_REPRODUCTION — a reported benchmark is retained as a reference because the source does not provide enough case-specific input data for an independent derivation.
• NOT_INDEPENDENTLY_EXECUTABLE — the paper reports an aggregate/result but the complete raw dataset or graph needed to regenerate it was not supplied.
• PENDING_EXPERIMENT — a physical/observational claim remains experimentally untested in the supplied materials.
• PENDING_IMPLEMENTATION — hardware or engineering realization is described as future work.
• UNRESOLVED — source material could not be made machine-executable without inventing missing information.

Ruthless-audit rule

This repository does not convert a mathematical or computational PASS into a claim of experimental proof. It also does not hide discrepancies. For example, the suite directly recomputes 60^18.86 and separately tests the source-stated 3.47×10^33 figure.

Repository files

• RUN_SEXA_MASTER_AUDIT.py — master public runner
• data/source_manifest.json — source hashes and extraction inventory
• data/duplicate_source_groups.json — exact duplicate-source groups
• data/evidence_classifications.json — non-executable and pending claims
• data/all_numeric_mentions.csv — every machine-extracted numerical mention with source/line context across the uploaded stack
• data/equation_like_lines.csv — machine-extracted equation/relationship lines
• evidence/extracted_text/ — text extraction for each supplied PDF, bound to the source manifest hashes
• reports/MASTER_AUDIT_REPORT.md — generated human-readable report
• reports/master_results.csv — generated machine-readable table
• reports/master_results.json — generated full result objects
• docs/AUDIENCE_HANDOFF.md — short instructions for the Australia audience

Source integrity

The source manifest binds the audit to the exact uploaded PDF bytes using SHA-256 hashes. Exact duplicate documents are identified so they are not treated as independent evidence.

Important scope limitation

The supplied Sigmatics PDF reports aggregate orbit statistics and reachability data, but the complete machine-readable 96-class transition graph / excitation vector is not embedded in the supplied PDF stack. Those claims are therefore not labeled as independently regenerated by this package.

Likewise, the conference hardware paper describes FPGA/ASIC/AI-accelerator realization as future work. This package tests deterministic computational properties but does not relabel planned hardware as completed hardware validation.