Replacing Lab Experiments with Computation

An alchemical free energy method that achieves sub-chemical accuracy for the hydration free energies of organic molecules, combining machine learning force fields with rigorous statistical mechanics.

A method for training generative models on Boltzmann distributions using only molecular energy functions, without requiring samples from the target distribution. Enables accurate sampling of complex molecular systems.

An equivariant generative model for molecular conformations that respects the physical symmetries of 3D space. Achieves sampling more than an order of magnitude faster than traditional molecular dynamics.