Experiments
Interactive versions of lectures I have given: the maths behind my work, running live in your browser. No libraries and no video. Every solver on these pages is implemented from scratch in a few hundred lines of TypeScript.
Live demos
- 01 · Probabilistic ML
Gaussian process regression
Click to place observations and watch a GP posterior re-fit live: mean, uncertainty band and sampled functions. Then let Bayesian optimisation tell you where to measure next, and read how the same technique scales to years of real plant data.
- 02 · Numerical methods
Navier–Stokes, live
A real-time incompressible Navier–Stokes solver: semi-Lagrangian advection with a pressure projection, running a continuous jet of ink. Stir it with your pointer and change the viscosity mid-flow. Grown out of a teaching lecture on modelling formulated products.