Differentiable Resist Modeling via PINN
A Physics-Informed Neural Network that learns the resist development PDE end-to-end, replacing the threshold model in the OpenLithoHub forward chain. Open research target — your implementation lands here.
Open research problems and merged community algorithms that build on OpenLithoHub. If you're a PhD student or ML researcher looking for an EDA / AI venue paper that ships with reproducible code, you're in the right place.
on GitHub
research-topic label.subclass LithographyModel, return PredictionResult) — EPE, PV-band, MRC, GDS export are handled by us.A Physics-Informed Neural Network that learns the resist development PDE end-to-end, replacing the threshold model in the OpenLithoHub forward chain. Open research target — your implementation lands here.
Curvilinear sub-resolution assist features generated by a diffusion model or RL agent, MRC-aware by construction. Direct foundation for an ICCAD / DAC paper.
Per-pixel failure-probability heatmap from a Bayesian U-Net (MC-Dropout, ensembles, or variational). The single most-watched topic at SPIE Advanced Lithography 2024–2026.
Differentiable MRC penalty (min CD / spacing / curvature) reusing OpenLithoHub's morphology utilities. A clean good-first-issue with broad downstream impact.
Open an issue on the OpenLithoHub repo with a one-paragraph proposal. If it's a good fit, we'll label it research-topic and feature it here too.