Short technical notes, derivations, and learning summaries — things that don’t need a full blog post but I’d like to keep accessible.
Why ML's backprop and physics' adjoint sensitivity analysis are the same algorithm.
A minimal, self-contained derivation of finite-difference time-domain.
Score matching, Langevin dynamics, and the Fokker-Planck connection.
This page will grow as I write things down. If you’re looking for longer essays, check the Blog.
Powered by Jekyll and Minimal Light theme.