GPU accelerated Trotter-Suzuki solver for quantum spin dynamics
Instituto de Fisica Enrique Gaviola (CONICET) and Facultad de Matemiatica, Astronomia y Fisica, Universidad Nacional de Cordoba, Ciudad Universitaria, 5000, Cordoba, Argentina
arXiv:1305.0036 [physics.comp-ph], (30 Apr 2013)
@article{2013arXiv1305.0036D,
author={Dente, Axel D. and Bederian, Carlos S. and Zangara, Pablo R. and Pastawski, Horacio M.},
title={"GPUacceleratedTrotter-Suzukisolverforquantumspindynamics"},
journal={ArXiv e-prints},
archivePrefix={"arXiv"},
eprint={1305.0036},
primaryClass={"physics.comp-ph"},
keywords={Computational Physics, Quantum Physics},
year={2013},
month={apr}
}
The resolution of dynamics in out of equilibrium quantum spin systems relies at the heart of fundamental questions among Quantum Information Processing, Statistical Mechanics and Nano-Technologies. Efficient computational simulations of interacting many-spin systems are extremely valuable tools for tackling such questions. Here, we use the Trotter-Suzuki (TS) algorithm, a well-known strategy that provides the evolution of quantum systems, to address the spin dynamics. We present a GPU implementation of a particular TS version, which has been previously implemented on single cores in CPUs. We develop a massive parallel version of this algorithm and compare the efficiency between CPU and GPU implementations. This boosted method reduces the execution time in several hundred times and is capable to deal with systems of up to 27 spins (only limited by the GPU memory).
May 2, 2013 by hgpu
