SFT — Structural Field Theory (Docs + Runners)

This repository is a verification-first release of SFT: a theory/framework plus a set of runners designed for external audit.

Start here: read Doc 0 — Abstract & Orientation for the ontological stance, scope, current limitations, and falsifiable predictions.

The key idea is simple:

• The “vacuum” is treated as a structural medium/field.
• “Particles” are stable field configurations (discrete in manifestation, continuous in availability).
• The project is organized so third parties can verify pipelines via PASS/FAIL gates, hash manifests, and schema-validated artifacts.

Important: many included datasets are synthetic (C) because this release is built to be verified on CPU-only.
Physical end-to-end confirmation requires REAL (P) solver outputs. CI only validates the verification pipeline (CPU-only), not physical validation.

• REAL-ready: verifiers + artifact contract are in place; can consume third-party real-solver outputs.
• REAL-verified: a real-solver run has been published with artifacts + hashes and passes the declared gates.

---

Quickstart (3 minutes)

1) Setup

git clone https://github.com/Xaquer69/sft-theory-and-runners
cd sft-theory-and-runners

python -m venv .venv
source .venv/bin/activate   # Windows: .venv\Scripts\activate

pip install -r requirements.txt

2) Run one demo runner

Pick one of the “packs” below (recommended: Spin (Doc.5) or DoubleSlit (Doc.7)) and run its verify script.

Example pattern (adjust the path to the pack you choose):

cd <PATH_TO_A_PACK>
sha256sum -c checksums_SHA256.txt
python3 verify_*.py

If the verifier ends with GLOBAL_PASS = true, you successfully validated that pipeline.

Recommended verification path (in order)

1. CI / convergence sanity (C)
   Run the EOC/CI (Doc.6) synthetic check to verify your environment and regression gates.

2. One physics-style demo with PASS/FAIL (C)
   Run either: Double-slit (Doc.7) or Spin-½ (Doc.5).

3. External Review Pack (C) (Doc.9)
   Validate SCAN → REGION → REPORT artifacts, JSON schema compliance, and manifests/hashes for audit readiness.

4. (Optional) Doc 10.1 — Alpha-Elasticity CI Suite (C) A CI/synthetic protocol for an alpha-like estimator (anti-leakage discipline, preregistered decision rules + hashing, null/adversarial controls). Not physical validation of α; included as a verification methodology appendix.

5. (Optional) Doc 8 - Limitations and Validation Plan includes External Break Test #01 (field rescaling) as a Doc9/Doc10.1-style anti-post-hoc audit protocol.

Repository structure

Doc_01/ ... Doc_14/ The full documentation set (each doc in its own folder).

*_CLEANED.zip and runner bundles When a document has an executable verification component, the corresponding zipped runner bundle is included.

Schemas/ (if present) JSON schemas for standardized artifacts (reports, scans, manifests).

Artifact contract (what “verification” means here)

Most runners produce a standard set of artifacts:

COMPATIBILITY_SCAN_*.json A parameter sweep (grid/scan) with per-point metrics.

EXISTENCE_REGION_*.json The PASS region extracted from the scan (mask/region geometry + thresholds).

EXISTENCE_REPORT_*.json A single verdict + provenance (version, seed, gates, hashes).

Typical verdict labels:

natural — stable without “maintenance”

maintained — stable only under explicit constraints/controls

All packs include:

SHA-256 manifests (checksums_SHA256.txt, etc.)

PASS/FAIL gates with explicit thresholds

(often) JSON schema validation

What is (C) vs (P)?

This repo uses a strict separation:

(C) Convenience / CI / Synthetic

Runs on CPU.

Validates pipelines, regression checks, schemas, and audit trails.

Does not claim physical validation by itself.

(P) Physical / REAL

Requires the real solver outputs (often GPU-scale).

Intended to be run by third parties, then checked with the same verification pipelines.

Minimal requirements

Python 3.10+

CPU-only is sufficient for the (C) packs.

OS: Linux/macOS/Windows should work if Python dependencies install cleanly.

Where to start (most user-friendly packs)

If you only run one thing, run one of these:

DoubleSlit_PACKAGE — verifies phase slope linearity and visibility monotonicity (PASS/FAIL). (Doc.7).

Spin / Appendix S pack — verifies FR 2π sign flip and rotor spectrum fit (PASS/FAIL). (Doc.5).

External Review Pack — verifies SCAN→REGION→REPORT with schema checks and manifests. (Doc.9).

(Each pack contains a README/HOW-TO-RUN plus verify_*.py scripts.)

Notes for reviewers

This repository is intentionally organized around external verifiability:

deterministic seeds,

explicit thresholds,

hashes/manifests,

and schema-validated outputs.

If you want to contribute REAL (P) runs, please follow this minimal contract:

• Export SCAN / REGION / REPORT artifacts using the same filename patterns used by the pack.
• Include full provenance in SCAN metadata: solver version/hash, grid(s), Δt/CFL, BCs, seed(s), platform.
• Provide ≥2 resolutions for any physical claim (or an explicit pre-asymptotic analysis).
• Include a SHA-256 manifest covering all exported files, and reference the exact gate_id used.
• Gates must be pre-registered (locked before the solver run); post-hoc loosening is not allowed.

Distributed under the MIT License. See the LICENSE file.

DOI: https://doi.org/10.5281/zenodo.18615390

Author & Contact

Francisco Queral Rallo Murcia, España ✉ xaviqueral@gmail.com