Long time evolution of complex materials : algorithmic developments and applications to metals, semiconductors and proteins

点击翻译按钮获取中文摘要

In spite of the continual increase in computer power over many decades, it is still a challenge, today, to reproduce on computer the long-time atomistic dynamics that is responsible for the evolution of materials and complex molecules such as proteins, for example. Yet, many macroscopic properties of interest to us, such as strength and conductivity are determined by specific atomic position. As even a tiny fraction of atoms hops around, these properties can be modified in depth, affecting the use of the objects upon which our civilization relies. Over the past years, I have been developing various algorithms and methods to try to access the right time scale or ensure a sufficient thermodynamical sampling in order to study understand the kinetics of these complex materials : glasses, proteins, defects in metals and semiconductors, etc. The program presented here first aims at expanding this development in order to address a wider range of problems - the question is complex and much remains to be done in order to cross both time and length scales. On the condensed matter and materials science side, I also plan to focus on specific scientific questions with the goal to develop a better understanding of the kinetics of point defects, both self-defects and impurities, near extended defects such as grain boundaries and dislocations. On the biophysics side, my focus will be two-fold: first, finish the development of a simplified all-atom forcefield based on a coarse-grained potential that we have used for many years and, second, apply this potential to study the kinetics of amyloid aggregation and the interaction between potential inhibitors, coming from natural molecules, and amyloid proteins. In parallel with this work, I will continue to collaborate with groups from around the world who are using the computer codes I have developed that implement methods I have created over the years : the activation-relaxation technique (ART nouveau), a very efficient open-ended saddle-point search method, the kinetic activation-relaxation technique (kinetic ART), a unique off-lattice kinetic Monte Carlo algorithm with an on-the-fly cataloguing capability and, soon, the new all-atom forcefield for protein we are developing. These codes are used today by more than 20 groups from around the world and their use continues to increase, providing considerable leverage to the development that will take place over the coming years. Overall, therefore, this research program will advance significantly our understanding of the kinetics of complex materials from the atomic level both with the tools that I will continue to develop and distribute and their applications that I plan to do over the next few years, particularly regarding defect kinetics in metals and semiconductors and protein aggregation.