High-performance scientific computing

In this research area we develop algorithms and methods for solving complex physics problems using state-of-the-art scientific computing resources.

Scientific high performance computing

A large and common thread for many of the activities within the Theoretical Physics Group is the use of high-performance scientific computing. This includes the development of novel algorithms and computational methods to solve complex physics problems allowing phenomena to be revealed that are inaccessible to conventional techniques. A considerable part of this work involves implementing sophisticated optimisations and parallelisation schemes as well as exploiting the latest developments in multicore and GPU hardware. As such we make extensive use of the Advanced Computing Research Centre’s high-performance supercomputing facility BlueCrystal  located within the H.H. Wills Physics Laboratory.
 
One of the most powerful tools employed by our Theoretical Physics researchers are Monte Carlo simulations. This is a computational approach that relies on repeated random sampling to stochastically estimate quantities that are intractable to compute in a direct deterministic way. Within soft matter research this involves tracking the statistics of motion of many interacting classical particles that model complex colloids and other unusual phases of liquids. Much of this work is devoted to developing new methods to overcome problems such as large free energy barriers, critical slowing down and handling rare events. Within quantum many-body physics Monte Carlo is used to variationally optimise complex Jastrow and neural-network based wavefunctions that can capture the ground state properties of interacting quantum particles.
 
We also tackle strongly correlated quantum systems using a variety of tensor-network based methods, such as density matrix renormalisation group and time-evolving block decimation, as well as specialised adaptations for impurity problems for dynamical mean-field theory and the study of open quantum systems. Other key methodologies widely applied by the group include classical and quantum density functional theory, crucial for understanding liquids at interfaces and for predicting the electronic structure of novel materials, respectively.
 
Edit this page