Name: Stewart Reed
Institution: University of Leeds
Research: Quantum Mechanics modelling using the NGS
Often the behaviour of small molecules of fewer than 10 atoms cannot be described by the Newtonian mechanics that rules the behaviour of objects around us. Instead quantum mechanics have to be used. Unlike conventional Newtonian mechanics, quantum mechanics is probabilistic in nature and allows things to happen in the macroscopic world that should not. For example, particles described by quantum mechanics can `tunnel' through barriers that would stop particles that are governed by Newtonian mechanics.
Stewart Reed from the University of Leeds, is aiming to develop new methods of performing accurate computer simulations of these “tunnelling” molecules; tunnelling is very important for a large range of atomic scale processes. His research has provided a new methodology for studying the quantum behaviour of molecules. To date he has used his methodology to study the dissociation of small atomic clusters containing a bromine molecule with one or more neon atoms loosely bound to it. In the future, he will use it to study quantum phenomena, such as tunnelling, in larger molecules.
As Stewart’s research involves developing methodologies, he runs his own specially made simulation programs on the NGS. To do this, he makes use of the high quality compilers and libraries that are provided as part of the NGS. In particular he uses the Intel Fortran compiler, Lapack, ScaLapack and ideally MPI2 libraries. Stewart has written his code so that it can be compiled into a serial code or a parallel code.
The rate limiting step of the CCS (Coupled Coherent States) method is solving a system of linear equations using either Lapack or ScaLapack but unfortunately Stewart’s code does not scale particularly with the number of processors. This is not a significant problem as he needs to repeat each calculation several times with different initial conditions in order to get good statistics. A recent extension of technique can require hundreds of even thousands of simulations in order to get a more accurate description of the dissociation of the molecular clusters. He therefore typically finds it more efficient overall to run more simulations on fewer processors (i.e. 1) than to use lots of processors for each individual simulation.
Quantum mechanical simulations are inherently expensive in terms of the computer resources required to perform them. Stewart explains that “the NGS provides excellent computing resources with which to perform these calculations. The computational capacity available from the NGS allows larger systems to be studied more accurately than are possible with standard workstations”.
There is still much work to be done as Stewart outlines future plans “Although there is much to be gained by studying small molecules, quantum mechanics also plays an important role in larger systems, for example in the biologically important reactions of DNA and proteins. The NGS being a network of dedicated high performance computing resources with the requisite support infrastructure ensures that more time can be spent doing research”.
Project funding - EPSRC
Grant no. - EP/E009824/1
Project PI - Dr Dmitry Shalashilin
Download a summary slide of this case study.
