Microtubule Resonance Simulator
🌲 Part of the Brokenbranch Lab — one human and a cluster of AI agents shipping strange software in public. This is one experiment among many; the front door lists them all.
An interactive computational exploration of fractal electromagnetic resonance in biological nanostructures, based on findings by Bandyopadhyay et al. (2020, 2022).
This is NOT peer-reviewed research. It is an exploratory tool for sharpening questions and testing mathematical plausibility.
Status: 🟡 Complete (exploratory tool — not formally validated)
This experiment is finished, not abandoned. It shipped its full result set in one publication and reached its intended scope: 7 hypotheses tested honestly, with negative and undecided results reported as such. A 2026-10-05 math audit corrected several engines: H2 (chirality) moved from falsified to inconclusive, and H6 (stochastic resonance) from plausible to inconclusive, now tested with positive and negative controls. H7 (Schumann alignment, p=0.183) stays not significant. There is no open backlog. If you find it sitting still, that stillness is done, not neglect.
It remains 🟡 (an exploratory toy, not a validated result) by design. It has not formally passed the Brokenbranch Lab validation gate (the "Four Tests" — reproduced, externally checked, pre-registered, and survives adversarial review). Promoting it to 🟢 would require, at minimum:
- an independent reproduction of the simulator's key results from the published parameters,
- external review of the physics engines (especially the H5 pitch-angle robustness and H6 stochastic-resonance parameterization),
- a pre-registered hypothesis set rather than the post-hoc meta-analysis presented here, and
- adversarial review of the model-tests-model concerns flagged in H3 (scale invariance is tautological as built).
None of those have been done, so the honest label stays 🟡. The value here is the honesty of the negative results, not a validated claim.
Reproducibility
Monte Carlo runs use a fixed seed (20260906, mulberry32, DEFAULT_SEED in physics.js); node scripts/recompute-figures.mjs reprints every figure the pages quote, and docs/figures-20260906.txt is the committed receipt.
Quick Start
- Clone the repository
- Open
index.html in a browser (or deploy to any static host)
- No build step, no dependencies, no server required
Structure
| File | Purpose |
|---|
index.html | Landing page with context, results summary, and explanations |
simulator.html | The interactive simulator (5 panels + hypothesis lab + meta-analysis) |
whitepaper.html | Detailed technical methodology and results |
sim.js | Visualization and UI logic (~85 KB) |
physics.js | 10 computational engines (~72 KB) |
style.css | Simulator design system |
landing.css | Landing page styles |
whitepaper.css | Whitepaper reading styles |
assets/og.png | Open Graph / Twitter card image (1200×630) |
scripts/recompute-figures.mjs | Node harness: recomputes every quoted figure from the fixed seed |
docs/figures-20260906.txt | Receipt: the harness output the pages' numbers come from |
docs/SYNC-DESIGN.md | How brokenbranch.dev mirrors this repo (daily pull; verbatim except the site chrome on two pages) |
Tech Stack
- Pure HTML / CSS / JavaScript (no frameworks, no build step)
- Canvas-based visualizations with
requestAnimationFrame
- Physics engines: RK4 integration, Monte Carlo sampling, Berry phase computation, stochastic resonance analysis, energy budget calculation
Key Results
| Hypothesis | Verdict | Notes |
|---|
| H1: Fractal Coherent Amplification | Plausible | Regular lattice beats fractal |
| H2: Chirality Creates Triplets | Inconclusive | With all modes counted, the helix's triplet count sits inside a matched random-splitting null (was "Falsified" before the 2026-10-05 audit) |
| H3: Scale Invariance | Consistent | Tautological (model tests model) |
| H4: Temporal Cascade | Consistent | Expected from oscillator structure |
| H5: Pitch Angle Optimality | Inconclusive | Parameter-dependent (55% robust) |
| H6: Noise-Fueled Resonance | Inconclusive (not demonstrated at the model's drive) | Fair test: driven slowly (f = 0.01), the same double well peaks at noise D = 0.40, close to the D* = 0.45 where the Kramers escape rate matches the drive; but at the model's own drive (0.7 to 1.4), across noise D = 0.2 to 7.5, no frequency met the criterion (an interior peak at least 3x a run with no signal; each peak came in at about 1x), so stochastic resonance is not demonstrated there. |
| H7: Schumann Alignment | Unvalidated | p=0.183, not significant |
Meta-analysis: 75.7% overall robustness (recomputed after the audit and the fair H6 test). Regular lattice beats fractal for amplification. Active oscillation energetically implausible (~40x neuron budget).
2026-10-05 math audit: the Flywheel math gate found errors in several engines and hypothesis tests; two verdicts changed (H2, H6). Every fix and its check are listed in AUDIT-2026-10-05.md.
References
- Bandyopadhyay, A. et al. "Fractal, Scale Free Electromagnetic Resonance of a Single Brain Extracted Microtubule Nanowire, a Single Tubulin Protein, and a Single Neuron." Fractal and Fractional, 2020.
- Bandyopadhyay, A. "A century-old picture of the nerve impulse is wrong." Communicative & Integrative Biology, 2022.
- Berry, M.V. "Quantal Phase Factors Accompanying Adiabatic Changes." Proc. R. Soc. A, 1984.
- Gammaitoni, L. et al. "Stochastic Resonance." Reviews of Modern Physics, 1998.
- Naaman, R. & Waldeck, D.H. "Chiral-Induced Spin Selectivity Effect." J. Phys. Chem. Lett., 2012.
Built With
Built with Claude agents (Anthropic), as a collaboration between human scientific curiosity and AI computational capability. This repo is the source of truth; https://www.brokenbranch.dev/labs/microtubule/ pull-mirrors it daily (see docs/SYNC-DESIGN.md).
License
MIT License. See LICENSE.
Contributing
See CONTRIBUTING.md for guidelines.
