Timewarp: Transferable Acceleration of Molecular Dynamics by Time-Coarsened Dynamics

Explore a comprehensive conference talk on Timewarp, an innovative method for accelerating molecular dynamics simulations. Delve into the challenges of conventional molecular dynamics and discover how Timewarp utilizes normalizing flows as proposal distributions in Markov chain Monte Carlo methods to target the Boltzmann distribution. Learn about the transferability of Timewarp between molecular systems and its potential to generalize to unseen small peptides. Gain insights into the background, datasets, model details, sampling techniques, and experimental results. Understand the related work and future directions in this field. Engage with the speakers' expertise during the Q&A session to deepen your understanding of this cutting-edge approach to enhancing molecular dynamics simulations.

- Intro
- Background
- Timewarp
- Datasets
- Model details
- Sampling
- Experiments
- Related work and future work
- Q+A