Multi-Scale Molecular Dynamics
Biomolecular functions emerge from complex molecular motions that span a vast range of spatial and temporal scales. However, conventional atomistic molecular dynamics (MD) simulations are fundamentally limited in the scales they can access. Our research aims to overcome these limitations by developing new computational methodologies that enable the exploration of biomolecular dynamics far beyond the conventional boundaries of MD simulations. Our approach integrates three key elements: advances in computational science for large-scale acceleration, efficient algorithms grounded in statistical mechanics, and multiscale molecular modeling. These methods are implemented in the molecular simulation software GENESIS ( https://mdgenesis.org/ ), whose development is led by our group. The software is openly distributed to the scientific community under the LGPLv3 license.
We will continue to develop multiscale MD methods and GENESIS. In particular, we aim to accelerate MD simulations by a factor of ten to a hundred by fully leveraging GPUs. We also emphasize research in the field of “AI for Science”, which involves innovating MD by integrating machine learning.
Methodological developments in Multi-Scale MD