Void Dynamics Model - Declassified Public Overview

Current Status:

The WORKING directory for this project is located on my personal account and can be found here: justinlietz93/Prometheus_VDM
Applying physics discoveries to the VDM intelligence model

Status Last Updated: Nov 4, 2025

Author: Justin K. Lietz

Contact: [email protected]

Created: August 9, 2025
Updated: September 29, 2025

This research is protected under a dual-license to foster open academic research while ensuring commercial applications are aligned with the project's ethical principles. Commercial use requires written permission from the author..

See LICENSE file or click the LICENSE badge above for full terms.

🔥News:

• October 31, 2025
  • Added a T8 grade proposal for a new Axiom candidate as A8 - Lietz Infinity Conjecture which provides an elegant hypothesis for how our universe evolved to be how it is from a beginning state to now. It includes the derived rules for how our universe would be structured at hierarchical scales to dissipate away the instabilities of void fluctuations.
• October 30, 2025
  • Added a historical/ folder including early original work like:
    • A self healing knowledge graph using Topological Data Analysis
    • As well as a Self Improvement Engine that integrates multiple reward components like novelty, self benefit, habituation, and TD error into a single "total reward" signal used to modulate its own neural plasticity for stable self-improvement that avoids weight saturation.
  • Validated the Counterfactual Echo Gain hypothesis by proving the trustworthiness and accuracy of the instrument, and showing that echo assist does modulate and improve the performance of a self aware system with 0 difference in cost compared to baseline.
• October 28, 2025
  • Published an article on Medium about upcoming work.
    • Published a "The Physics of Choice" video, along with a couple other videos on YouTube.
• October 23, 2025:
  • Created a sparsely populated CANON_PROGRESS.md document to post updates on private work to prevent this public repo from going stale.
• September 29, 2025:
  • First public code release + creation of private Void Dynamics package which can now be imported and run in this repository using workflows and repository secrets.
• September 28, 2025:
  • Posted two pre-prints to Zenodo. If you've published similar or relevant work on Reaction-Diffusion in the past 3 years on arXiv and want to support this work by endorsing me in a related category, submit an issue, post in the discussion board, or send me an email with Subject: RD Endorsement to get my attention. It would be much appreciated!
• August 21, 2025: Launched public repo
• ...
• March, 2025
  • Released first falsifiable, reproducible simulations that validated initial claims.
• February, 2025
  • SIE and Emergent KG + Real-Time TDA show very strong statistically significant findings, making feasible the entire model
• October, 2024
  • Initial realization of the idea.

DOIs:

---

❗NOTE

This organization is currently managed and operated by me (J. Lietz) alone as a solo developer / researcher. I may not respond by email right away. If you want to get my attention post in the discussion board briefly about what you'd like to talk about and let me know you sent me an email.

📜Background Context

You may see reference to FUM (Fully Unified Model or FUVDM (Fully Unified Void Dynamics Model) within the repository, this is referencing the VDM (Void Dynamics Model). These were earlier names given to the model for specific reasons, and it's been my decision to simplify and specify the name to improve clarity. Fully Unified Model was originally named so because of it's unique architecture.

I effectively broke down every type of machine learning model to it's fundamental principles and strategies. Then I began unifying them into a single architecture. This created my first variant of the model I called AMN (Adaptive Modular Network). AMN was able to learn to solve quadratic equations with ~85% accuracy in 65 seconds of training and only a handful of examples.

While doing this I quickly realized I could get the same behavior or better while removing a lot of the heavy solutions that LLMs, GNNs, SNNs, CNNs, and other ANNs used just by allowing a principle in physics called the path of least action to occur in the space between interactions. Eventually I realized I accidentally connected a lifelong hypothesis of mine ("intelligence emerges from the void space within the interactions of entities") to the model. Thus, the Void Dynamics Model earned it's name.

💬Discussion Board:

• https://github.com/orgs/Neuroca-Inc/discussions

---

This repository includes provides a public, declassified view of the Void Dynamics Model. It includes the theory, write-ups, code, notebooks, figures, logs, and validations for review by physicists, applied mathematicians, and scientifically minded engineers. Reproducible code released for the public is now available. Proprietary code is not available to view directly, but can be run via a private package with this repository. Remaining proprietary work must be requested directly.

Classified dependency notice

Certain executable simulations load void-dynamics and memory-steering modules from a private Neuroca, Inc package repository. If those files are missing, the code exits with:

Attempted to import classified code, ask the author for access to run this simulation. Otherwise you can still view figures and logs.

Affected entry points include:

• code/fluid_dynamics/fluids/lbm2d.py (and benchmarks such as taylor_green_benchmark.py and lid_cavity_benchmark.py).
• code/memory_steering/memory_steering_experiments.py.

Contact the maintainer for access. After gaining access, the simulations will run normally.

🧭What this is

• A set of derivation papers that establish a clean baseline physics slice using reaction-diffusion (RD).
• Additional documents that explore a future, quarantined effective field theory (EFT) branch, clearly labeled as future work.
• Each paper separates what is proven from what is plausible or speculative and, where applicable, includes acceptance criteria for simple numerical checks.

🤖Why it relates to AI (brief)

• The project studies how simple, local rules can yield stable, interpretable global behavior under resource constraints.
• That design philosophy is relevant to AI systems that favor locality, event-driven updates, and transparent evaluation instead of opaque heuristics.
• “Memory steering” (covered separately) frames slow routing bias and retention/decay as structured influences over faster dynamics, an analogy for directing computation without black-box shortcuts.

☄️What’s inside (papers)

• Program overview and banner:
  • VDM_Overview.md
• Papers and pre-prints:
  • papers/
• Foundations:
  • discrete_to_continuum.md
  • continuum_stack.md
• Reaction-Diffusion (canonical baseline):
  • rd_front_speed_validation.md
  • rd_dispersion_validation.md
  • rd_validation_plan.md
• Memory Steering (slow routing bias; separate layer):
  • memory_steering_acceptance_verification.md
• Finite-domain EFT modes (quarantined future work):
  • finite_tube_mode_analysis.md
• Change log / scoping notes:
  • CORRECTIONS.md

🚫What’s not included

• Source code, executables, or private runtime harnesses.
• Logs, figures, or any artifacts that would be sufficient to easily reconstruct proprietary implementations. Notebooks with clear falsifiable code will be provided soon.
• Drafts outside the derivation index above.

📖How to read these papers

• Each file follows a consistent structure: Purpose · Assumptions/Parameters · Discrete law · Continuum limit · PDE/Action/Potential · Fixed points & stability · Dispersion · Conservation/Lyapunov · Numerical plan + acceptance · Results · Open questions.
• Claim labels:
  • [PROVEN] = sign/dimension/limit checks plus a minimal numerical test that passes stated tolerances.
  • [PLAUSIBLE] = future work with rationale; quarantined until derivation + checks are complete.
  • [SPECULATIVE] = exploratory; not used for baseline claims.

⚖️Licensing and scope

• These materials are shared for academic review and discussion. Commercial use requires prior written permission. See the project’s license notice in the distribution or parent repository materials.
• The scope stays within theoretical physics and simulation. Broad cosmological claims are withheld or clearly labeled until backed by derivation + numeric checks.

🔖Citations

• When referencing specific results, cite the overview and the relevant validation paper, for example:
  • VDM_Overview.md
  • rd_front_speed_validation.md
  • rd_dispersion_validation.md

📫Contact

• For scope questions or clarifications about acceptance criteria, refer to the headers in the overview and topic files listed above. If you are reading this as part of a paper-only bundle, the maintainer’s contact is provided alongside the distribution materials.