orateurs invités

Orateurs invités:

 

- Nicolas Ollinger (LIFO, Orléans)

   "Qubits et portes quantiques : premiers pas en informatique quantique"

 

 

- David Richard (Computational Soft Matter Lab, University of Amsterdam)

  "Sampling methods to study rare events in Molecular Dynamics"

Rare events are ubiquitous in nature, from phase transitions to folding and self-assembly in biological systems. From a perspective of the transition state theory, these events fall in the same description where two basins (reactants and product states) are separated by a large free energy barrier. Numerically, the sampling of transitions between these two states is very difficult and often prohibited due to the limited time scale available by standard Molecular Dynamics (MD). For example, modern supercomputers reach at most microseconds to tens of milliseconds, while the biological time scale (conformational changes, binding, and unbinding) can easily span seconds. In the last two decades, a series of advanced sampling methods have been developed to bypass brute force MD and enable the study of complex phenomena from the nucleation of water in our atmosphere to drug binding processes. Here, we will give an overview of numerical methods to compute free energy barriers, transition rates, and unveil microscopic pathways. In particular, we will shortly discuss standard techniques such as umbrella sampling and metadynamics and focus on more novel path sampling algorithms such as transition path sampling and interface path sampling (Transition Interface Sampling and Forward Flux Sampling). A discussion of pitfalls and some of the remaining challenges will be given. In particular, the problem of finding good collective variables to describe the process and studying complex energy landscapes with multiple metastable states. In that context, we will briefly discuss how rare event sampling methods can be combined with Markov State Modelling to study more complex landscapes.

 

 

 

 

 

Personnes connectées : 8 Vie privée
Chargement...