Writing

Case studies benchmarking the hydration-site detection I wrote for Rowan as a part of their MD post-processing. I look at three (very studied) protein-ligand systems and two (less studied) molecular glue systems.

Large-deviation theory can be used to turn a rare-fluctuation problem into an equation of state (EOS) problem. In an interacting 1D chain, a low-order Padé approximation to that EOS reproduces every limit used to construct it, but becomes nonmonotone at stronger coupling. The resulting rate function is nonconvex, which is thermodynamically forbidden.

TLDR; pruning radius is a first-order parameter in protein-ligand simulations due to their effects on solvation statistics.

A super short introduction to the two main types of molecular states seen in polaritonic chemistry. Motivated by the increasing discourse regarding the role of "dark states" in optical materials.

Final project for my Numerical Solution of Partial Differential Equations (APMA 2550 at Brown) class that I thought I did a good job on.

Observations from using contact maps as kinetic descriptors for protein-ligand MD simulations.

Reflection on attempts at 'mid-tier' binding affinity prediction from my time at Rowan.

Neural network potential protein-ligand interaction energy benchmarks from my time at Rowan.