These posts are a collection of concepts I encountered throughout my research—ideas that were hard to understand at first, but brought real intellectual joy once they clicked. I wrote these posts to share that joy with more people! All posts are fully written with AI (Claude) through interactive editing and direction by the author, as an ongoing experiment in AI-assisted technical writing and figure generation.
-
From Jarzynski's Equality to Diffusion Models
From Jarzynski's equality to diffusion models — path measures unify free energy estimation, AIS, diffusion models, and GFlowNets as instances of the same mathematics.
-
Generative Flow Networks
An introduction to GFlowNets from the perspective of probabilistic ML — sampling proportionally to rewards, training objectives, and connections to MaxEnt RL, variational inference, and diffusion models.
-
Ensembles, Thermostats, and Barostats for ML Researchers
Statistical mechanics for ML researchers — from Newton's equations to ensembles, thermostats, barostats, Monte Carlo, and connections to generative modeling.
-
Protein Design for ML Researchers
An introduction to protein structure, function, and computational design — from amino acids to the RFDiffusion/ProteinMPNN pipeline.
-
Heterogeneous Electrocatalysis for ML Researchers
Heterogeneous electrocatalysis for ML researchers — the energy storage problem, why oxides matter, the solid-liquid interface, and the complexities of real catalyst design.
-
Quantum Chemistry and DFT for ML Researchers
Quantum chemistry and density functional theory for ML researchers — from the Schrödinger equation to Kohn-Sham DFT and modern deep learning approaches.
-
The Fokker-Planck Equation
Three routes to the Fokker-Planck equation — intuition, heuristic discretization, and rigorous Itô calculus — building from physical pictures to mathematical proof.
-
Spherical Equivariant Layers for 3D Atomic Systems
Understanding the spherical equivariant layers that power modern molecular neural networks, from group theory foundations to Clebsch-Gordan tensor products.