Computational Modelling Group

Seminar  10th May 2011 2 p.m.  27/2001

Squeezing the Matrix: polynomially scaling spin dynamics simulation algorithms

Dr Ilya Kuprov
Physical and Theoretical Chemistry Laboratory, University of Oxford

Categories
Bioinformatics, Biomathematics, C, C++, Complex Systems, Computer Science, Electromagnetism, Energy, Finite differences, Finite elements, Fortran, HPC, Molecular Dynamics, Optimisation, Scientific Computing
Submitter
Chris-Kriton Skylaris

Dr Ilya Kuprov

Abstract

We present a family of polynomially scaling algorithms for time domain spin dynamics simulations in Liouville space, collectively called "State Space Restriction" methods. The exact or approximate reduction in matrix dimension is achieved using conservation law analysis, symmetry factorization, system trajectory parsing and the analysis of state space connectivity graphs. Liouville space simulations (including symmetry, relaxation and chemical kinetics) of most liquid state NMR experiments on 40+ spin systems can now be performed without effort on a desktop workstation. Much progress has also been made with improving the efficiency of ESR, solid state NMR and Spin Chemistry simulations. Brute force search for long-lived states using complete Redfield relaxation superoperators can be performed for systems with 10+ spins