High-performance computing of phase transitions using GPU

2016 Fiscal Year Final Research Report

• Project/Area Number: 25400406
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Mathematical physics/Fundamental condensed matter physics
• Research Institution: Tokyo Metropolitan University
• Principal Investigator: OKABE Yutaka 首都大学東京, 理工学研究科, 客員教授 (60125515)
• Project Period (FY): 2013-04-01 – 2017-03-31
• Keywords: モンテカルロ法 / GPU / クラスターフリップ法 / フラストレーション / イジングモデル / 準結晶

Outline of Final Research Achievements

I developed parallel calculation of cluster-flip Monte Carlo algorithm using the Graphical Processing Unit (GPU). I enrolled the program of our group in the program library of Comp. Phys. Commun. for the convenience of other researchers.
I applied the GPU algorithm of the cluster-flip Monte Carlo simulation to the problem of the phase transition of the spin models on the quasi-lattices, that is, the Penrose lattice and its dual lattice. I also applied the GPU algorithm to the problem of ferromagnetic spin models on the pyrochlore lattice.
I also studied the frustrated antiferromagnetic systems on the pyrochlore lattice. Using the Wang-Landau method, which calculates the energy density of states directly, I studied the residual entropy of the diluted spin-ice models. Using the replica-exchange method for both the temperature and magnetic field, I investigated the magnetization curve of the diluted spin-ice problem.

Free Research Field

数物系科学