MULTIDYN - Non-Adiabatic Molecular Dynamics With Explicitly Treated Electronic Degrees of Freedom

Principal Investigator:
Rene Kalus

Technical University of Ostrava (Czech Republic)

Local Project ID:

HPC Platform used:
Hermit of HLRS

Date published:

Molecular dynamics methods represent most powerful theoretical tools for getting insight in how processes run on the atomic scale. Everything seems to be fairly simple and maybe routine if classical Newtonian mechanics can be used and if (most importantly!) the electronic state of the studied system does not change during the time evolution. However, if the latter is not the case (we speak about non-adiabatic processes in this case), the situation is much more involved and very much more computationally demanding.

Simply because the forces needed in the Newtonian equations of motion cannot be calculated from more or less (but always explicit) formulas and have to be evaluated via highly demanding quantum chemistry approaches. This was the main reason why, before the advent of the most powerful (petascale) computers, only small systems (consisting of a few atoms) and a couple of their electronic states could be considered. The main challenge of the present project was to further develop and optimize the methods for non-adiabatic molecular dynamics simulations and, finally, use them in a series of calculations on fairly large systems (tens of atoms) and with very many (tens to hundreds) electronic states involved.

Why supercomputing power?

The calculations proposed for the project consisted in numerical integrations of large sets (hundreds) of coupled non-linear differential equations (equations of motion) for sufficiently long time periods (hundreds of picoseconds) and all this had to be repeated for many times (typically about 10 000 trajectories were integrated within each simulation run of this project) to get well converged data. In addition, since transitions had to be considered in our calculations among tens to hundreds of active electronic states, the right-hand sides of the equations of motion (forces) had to be calculated using quantum mechanics. The simulation project was made possible through the European HPC initiative PRACE (Partnership for Advanced Computing in Europe), using supercomputer Hermit of HLRS Stuttgart.

The primary goal of the project was to get an insight in how ionic complexes of rare-gas atoms (clusters) disintegrate after they were electronically excited (either by a collision or via absorbing a photon). Besides the fundamental science, a strong motivation for such a research came from the field of applied plasma physics. Namely, it has been clearly demonstrated during the last decade that cold rare-gas plasmas have a big potential for broad biomedical applications. For example, they can be used in an almost perfect surface sterilization, wound healing, or even for malignant cells inactivation. However, still a lot of fundamental research is needed on the processes beyond these marvelous properties in order to optimize plasma generators and develop possible biomedical applications. This was one of the most important benefits followed by the project.

The project team consisted of three researchers of the IT4Innovations National Supercomputing Center & Department of Applied Mathematics, VSB - Technical University of Ostrava, Martin STACHON, Ales VITEK and Rene KALUS, and one scientist of the Institute of Geonics, Academy of Sciences of the Czech Republic, Ostrava, Ivan JANECEK. In addition, three students of the University of Ostrava (another university located in Ostrava) were also involved through their diploma theses, Tomas JANCA, Pavel NAAR and Jakub ZAVACKY, as well as a secondary school student, Jan PREMUS, via his internship at the Institute of Geonics.


Photofragmentation of the Ar19+ Cluster (Video)

Scientific Contact:

Rene Kalus
IT4Innovations National Supercomputing Center and Department of Applied Mathematics
VSB - Technical University of Ostrava
17. listopadu 15/2172, CZ-708 33 Ostrava, Czech Republic

Tags: Materials Science HLRS