Scientific computing library for optics, computer graphics and visual perception.

Differentiable wave optics simulation library built on PyTorch

Integrated Machine and Deep Learning for Optical Design

A Python-based tool for modeling optical photolithography

[Optica, 2025] Accurate and efficient diffraction modeling between arbitrary planes

Large-Area Fabrication-Aware Computational Diffractive Optics (SIGGRAPH Asia & TOG 2025)

[Nature Communications, 2025] In situ characterization of tightly focused fields

Optical neural computation on a commodity smartphone: OLED screen + mirror + front camera as an analog matrix engine. 101 experiments, bilingual docs, and 6 Zenodo papers.

Scripts developed for Computational Optics course

Physics-Informed Neural Networks for Optimal Beam Shaping in Flat Optics

repo for the website of neural litho project

[Photonics Research, 2026] Accurate wave propagation modeling with reduced precision

A growable codex of research on neural holography displays.

FFT-DI is a compact Python implementation of the FFT-based direct integration method for the Rayleigh-Sommerfeld diffraction integral. The project is aimed at researchers and engineers who need a readable, modifiable baseline for scalar diffraction propagation rather than a closed optical-design package.

Convert optical simulation requests into validated, solver-ready spec JSON.

...

A growable open-source computer-generated holography framework built on top of PADO, a PyTorch differentiable optics library.

Sovereign pure-Rust (zero-dep) differentiable wave-optics simulation + diffractive-optical-network design toolkit. Verified decoder, optical security seal, multi-focal AR combiner, all-optical D2NN, optical matmul, photonic attention, fabricable mask export. cargo test = the oracle.

Welcome to the era of Artificial Cambrian Explosion.

UV-Vis extinction → gold-nanoparticle size + speciation. Monomer/dimer/trimer plasmonic modeling (Mie, coupled-dipole, T-matrix) with temperature-dependent equilibria and inverse fitting.