“Know how to solve every problem that has been solved.” “What I cannot create, I do not understand.” — Richard Feynman

Projects

Built, validated, demonstrable work. The flagships are the pieces to show first — each carries a verified line stating the check that makes it trustworthy. Below them, working prototypes that do one canonical thing well, each paired with a ladder of graduated extensions. Every project also has a home in the subject notes, where the derivations live.

Flagships

Section overview
SCF in the Browser (Knot → C → wasm)
Caffarel ZVMC Reproduction
Spectral Functions from Krylov Moments
Brute-Force Hartree-Fock (C++)
Learning to Sample (SLMC on PySCF)

Formula Discovery (Knot → wasm)

Symbolic Regression in the Browser
NN-Guided Symmetry Discovery
Formula Discovery Without Search (Meijer G)
The Knot Playground

Statistical Mechanics & MD

2D Ising Model (Metropolis MC)
GPU-Accelerated Ising (PyTorch/MPS)
Critical Exponents from Finite-Size Scaling
Lennard-Jones MD (Velocity Verlet)
Wang-Landau on a PySCF Surface
Self-Learning MC on a PySCF Surface

Statistics & ML

Transformer From Scratch (GPT)
Reverse-Mode Autodiff Engine
MCMC From Scratch
Probabilistic Learning on Manifolds

Biomedical Physics

Gamma Knife Dose Simulation