This notebook implements a denoising diffusion probabilistic model (DDPM) for generating abstract art paintings.

Predicting dark matter initial conditions. Python, JAX, Equinox.

Evolutionary Multi Agent MLP Training in Rust.

Personal static webpage. Made using svelte, ts and svelte-router. 2D particles using twgl.js

Official codebase of the paper "Graph Spring Neural ODEs for Link Sign Prediction".

Fast and differentiable tools for analysis and optimization on structured and unstructured data in cosmology. Powered by JAX.

Implementations of differentiable ODE and PDE solvers for gradient based initial state optimiziation in JAX.

three.js drag & drop asset browser proof of concept.

Virtual 3D village using three.js and svelte.

Solution for the course: "Computational Methods for Radiative Transfer" taught at UZH in 2024.

Quick & dirty implementation of a generative adversarial network with jax, optax and equinox.

Quick dirty implementation of a convolutional image autoencoder with JAX, Flax and optax.

Short implementation of the Neural ODE, including the adjoint based version in pure JAX and OPTAX

Differentiable particle simulation for drawing images

Notes and exercise solutions for the course advanced quantum algorithms taught @UZH 2023.

Webgpu 3D noodle extrusion.

GPU accelerated ODE solver, 3D extrustion and rendering in WebGPU.

Minimal WebGPU Compute Shader Particle System

Some creative coding sketches using nannou (rust)

Advanced High Performance Computing Course

Projects for the course Advanced Simulation in Natural Science. Includes an FFT implemtation, a pytorch based raytracer and a multrigrid PDE solver.

WebGL shader playground in Typescript

Neural Cellular Automata (NCA) to learn Physics simulations