Access Type

Open Access Thesis

Date of Award

January 2021

Degree Type

Thesis

Degree Name

M.S.

Department

Computer Science

First Advisor

Loren Schwiebert

Abstract

In a collaboration between researchers in the physics and computer science departments at Wayne State University, we have developed and implemented a GPU-accelerated application to study artificial spin systems emulating the Random Bond Ising model. To emulate the Random Bond Ising Model, we generate quadrupolar lattices with both ferroquadrupolar and antiferroquadrupolar ordered phases. We also introduce structural disorder into the system by randomly removing a percentage of individual spins. The presence of structural disorder gives rise to phase transitions that can be observed and mapped using our application. Our algorithm implements a Monte Carlo simulation based on the Metropolis model and allows multiple lattices to execute in parallel. By executing multiple lattices in parallel, we can derive correlation values to compare and track the evolution of multi-lattice systems. This correlation among the several lattices can be used to analyze disorder-driven phase transitions that can be useful in attempting to observe a spin glass phase. To accommodate large lattices and large systems comprised of many lattices, our application, written in the C++ language and parallelized using the CUDA Application Programming Interface (API), calculates the energy of independent spins as well as the correlation among the multiple lattices on the GPU.

Share

COinS