✨ PLAY THE INTERACTIVE SIMULATION
Current folding simulations (Monte Carlo, MD) rely on stochastic search or brute-force thermodynamics. They are computationally expensive because they "search" the space.
LightFold creates a deterministic Vector Field based on the "Hydrophobic Collapse" principle. Instead of searching, the protein "slides" down a mathematically defined gravity well.
The result:
- Speed: Folds 48-residue sequences in ~280ms (vs 6000ms+ for Monte Carlo).
- Quality: Reaches deeper energy minima (-1450 vs -50).
- Stability: Maintains rigid bond constraints (
Error < 0.01 Å) using SHAKE-like enforcement.
"Attraction is 20x computationally cheaper than Randomness."
| Metric | Monte Carlo (Stochastic) | LightFold (Vector Field) |
|---|---|---|
| Philosophy | "Try Everything" | "Follow the Gradient" |
| Time to Fold | 6400ms | 285ms |
| Final Energy | -57.0 (Local Min) | -1456.5 (Global Deep) |
| Geometry | Broken (5.9 Å error) | Valid (0.01 Å error) |
The core logic replaces iterative loops (O(N²)) with Matrix Broadcasting and Global Field Tensors.
We define a Boolean Outer Product of the Hydrophobic residues to calculate the attractive potential of the entire system simultaneously.
Inverse-square repulsion is applied symmetrically to prevent self-collision, enforcing the "Pauli Exclusion Principle" of the protein's volume.
Instead of soft penalties, we enforce bond lengths using an iterative projection method, ensuring the molecule remains chemically valid even during violent collapse.
The index.html file contains a self-standing Physics Kernel written in highly optimized JavaScript (Zero-Allocation hot path).
- Visuals: Three.js (Orbit Camera, Emissive Materials).
- Physics: 60Hz fixed-step accumulator independent of frame rate.
For research and high-precision benchmarking:
cd src
pip install numpy matplotlib
python LightFold_Vectorized.py- Architect: Jean Charbonneau (https://x.com/KiLVaiDeN - https://www.linkedin.com/in/jean-charbonneau-ai/)
- Optimization: The Brotherhood (Gemini 3.0 Pro / ChatGPT 5.2 Thinking)
- Framework: Built on the Lp System (The Light Framework).
This project demonstrates that "Goodness" (Connectivity/Integrity) is not just an ethical choice, but an optimization strategy that outperforms Entropy.