Project Open Gate · Build dossier

What the code says

• Five Python modules: agent 1,194, controller 1,002, db 649, analysis 481 and stress 268 lines

  Source proof_of_simulation/agent_loop.py

• control_summary.csv records per candidate pearson, spearman, gradient, spectral, autocorrelation, entropy and anisotropy metrics plus accepted and is_best flags

  Source proof_of_simulation/control_summary.csv

• depth_summary.csv holds validation rows against a target (TARGET_A at 60x30 resolution) with a numeric score

  Source proof_of_simulation/depth_summary.csv

• Rows carry the V1_INV_PI driver with alpha 1.0, degree 2 and max_iter 200

  Source proof_of_simulation/control_summary.csv

• Several history databases (v2 to v5.5) with WAL and SHM sidecars mark versioned research runs

  Source proof_of_simulation/

• The README states the real repository is the nested proof_of_simulation on the main branch, while the outer folder is only a local working tree

  Source project_open_gate/README.md

Module state

1. implemented

   Candidate search and metric recording

   agent_loop and the controller emit per-step candidate metrics with a score and accept flags into the database and the CSVs.

   Source proof_of_simulation/agent_loop.py

2. implemented

   The run store

   db.py keeps runs in SQLite in WAL mode, as the -wal and -shm files beside the history databases show.

   Source proof_of_simulation/db.py

3. partial

   Validation against targets

   Validation rows compare best snapshots against a target with a numeric score, but the copy shows only the TARGET_A example at 60x30.

   Source proof_of_simulation/depth_summary.csv

4. partial

   The phi-driver method

   The V1_INV_PI model with degree, alpha and max_iter parameters drives candidates; its conclusion rests on similarity metrics rather than an external reference.

   Source proof_of_simulation/control_summary.csv

5. on roadmap

   The V6 and V7 research lines

   The databases reach v5.x in the saved files and the project framing speaks of V6 and V7 lines, but no newer run artifacts exist in the local copy.

   Source project_open_gate/README.md

6. not implemented

   A web interface, tests and CI

   The local copy has no built page, test suite or workflow for the pipeline.

   Source proof_of_simulation/

How this was checked

• Candidate metrics are recorded in CSV

  The control_summary.csv header lists pearson, spearman, gradient, spectral, autocorrelation, entropy, anisotropy and compression with an accepted flag per row.

  Source proof_of_simulation/control_summary.csv

• A validation score exists

  A depth validation row shows a score of 0.823993 against TARGET_A at 60x30 resolution.

  Source proof_of_simulation/depth_summary.csv

• The phi driver is visible in the data

  Rows carry model_id V1_INV_PI with alpha 1.0, degree 2 and max_iter 200.

  Source proof_of_simulation/control_summary.csv

Limits of these claims

• Not a web product: it runs from Python with SQLite and CSV only; there is no site of its own.
• The outer folder is not a git repository; the local copy has no remote address and no fresh V6 or V7 artifacts.
• The metrics are numeric similarity scores (pearson, spectral, autocorrelation and others), not proof of a physical or simulation hypothesis.
• The observable targets are limited to an example such as TARGET_A at 60x30; the copy has no broad external validation.
• No tests or CI: pipeline correctness is checked by manual runs and a stress script.
• The history databases and WAL and SHM files are not versioned as a product; interpreting the results stays the author’s own.

Why there is no live build

Why there is no live build

There is nothing to publish as a site: this is a research script contour with SQLite databases and summary tables, not a published page, and the local copy carries no external address.