markdown

CubeGeometryTest

This repository contains the code and figures for two related but distinct pieces of work:

1. cube_geometry/ – Self-organized criticality (SOC) experiments on a cube with WWWWHW geometry.
2. inquiry_studio/ – The cubic interrogative instrument used in the paper, which measures interrogative entropy and answer-side behavior for local LLM runs.

---

Repository Layout

• cube_geometry/

  • scripts/ – Sandpile and WWWWHW cube simulations:
    • cube_sandpile.py, cube_sandpile_micro*.py
    • cube_WWWWHW_optionA_flux*.py
    • cube_WWWWHW_optionB_micro_fit.py
    • cube_WWWWHW_optionC_micro_fit.py
    • sandpile_2d*.py
  • figures/ – Example avalanche plots produced by the scripts.

• inquiry_studio/

  • src/ – The interrogative measurement system:
    • cubic_dynamics.py, inquiry_state.py, loop.py, session.py, metrics.py, studio_config.py
    • adapters/ – LLM adapter implementations (llm_dummy.py, llm_local.py)
    • backstop/ – Safety / logging utilities and rules
  • README.md – How to run the Type A / Type B / Type C sessions.
  • requirements.txt – Minimal dependencies for the instrument.
  • test_questions.txt – Example question sets.

• requirements.txt (root) – Shared minimal scientific stack (e.g., numpy, matplotlib, scipy).

• LICENSE.md – License for this repository.

• .gitignore – Keeps virtual environments, logs, and compiled Python files out of version control.

---

cube_geometry: Running the SOC Cube Experiments

From the repo root:

cd cube_geometry
python -m pip install -r ../requirements.txt

Example runs (from within cube_geometry):

# Baseline sandpile (no WWWWHW geometry)
python scripts/cube_sandpile_micro_fit.py

# WWWWHW option A – flux variants
python scripts/cube_WWWWHW_optionA_flux_true.py
python scripts/cube_WWWWHW_optionA_flux_safe.py

# WWWWHW option B / C – external forcing variants
python scripts/cube_WWWWHW_optionB_micro_fit.py
python scripts/cube_WWWWHW_optionC_micro_fit.py

Each script generates avalanche statistics and, in some cases, saves plots into cube_geometry/figures/.

---

inquiry_studio: Interrogative Instrument Used in the Paper

The inquiry_studio directory contains the code for the cubic interrogative instrument used in the manuscript. It:

• Represents questions in a WWWWHW field (who/what/when/where/why/how).
• Computes interrogative entropy (H_I) over that field.
• Logs answer-side behavior (H_A and related metrics) for each step.
• Runs reproducible Type A / Type B / Type C sessions against a local LLM or dummy adapter.

Basic usage (from the repo root):

cd inquiry_studio
python -m pip install -r requirements.txt

# Example: run the configured sessions
python -m src.loop

The exact session definitions and question sets used in the paper are encoded in src/studio_config.py and the associated question files. To reproduce the manuscript, use this repository as-is and do not alter those definitions.

---

Reproducibility Notes

• cube_geometry/ provides SOC experiments for geometric and physical intuition (supporting work).
• inquiry_studio/ is the instrument actually used to generate the interrogative entropy results in the paper.

Together, they form the computational backbone for the work described in the Zenodo-linked manuscript.

::contentReference[oaicite:0]{index=0}

Inspirations

The initial cube and sandpile experiments were inspired by the sandpile simulations demonstrated in Veritasium’s videos and interactive examples:

• Sandpile Avalanche Simulation https://www.veritasium.com/simulation3
• 2D Ising Model: Criticality https://www.veritasium.com/simulation4
• Forest Fire Model (Drossel–Schwabl) https://www.veritasium.com/simulation5

The code in this repository is an independent implementation.

Main Python deps:

• numpy
• matplotlib
• scipy (for fitting, if used)

Simulations