Yamaguchi Chiaki, Kawashima Naoki
Department of Physics, Tokyo Metropolitan University, Hachioji, Tokyo 192-0397, Japan.
Phys Rev E Stat Nonlin Soft Matter Phys. 2002 May;65(5 Pt 2):056710. doi: 10.1103/PhysRevE.65.056710. Epub 2002 May 22.
We propose a method for Monte Carlo simulation of statistical physical models with discretized energy. The method is based on several ideas including the cluster algorithm, the multicanonical Monte Carlo method and its acceleration proposed recently by Wang and Landau. As in the multibondic ensemble method proposed by Janke and Kappler, the present algorithm performs a random walk in the space of the bond population to yield the state density as a function of the bond number. A test on the Ising model shows that the number of Monte Carlo sweeps required of the present method for obtaining the density of state with a given accuracy is proportional to the system size, whereas it is proportional to the system size squared for other conventional methods. In addition, the method shows a better performance than the original Wang-Landau method in measurement of physical quantities.
我们提出了一种用于对具有离散能量的统计物理模型进行蒙特卡罗模拟的方法。该方法基于多种思想，包括团簇算法、多正则蒙特卡罗方法及其最近由王和兰道提出的加速方法。如同扬克和卡普勒提出的多键系综方法一样，当前算法在键数分布空间中进行随机游走，以得到作为键数函数的态密度。对伊辛模型的测试表明，对于本方法，以给定精度获得态密度所需的蒙特卡罗扫描次数与系统大小成正比，而对于其他传统方法，该次数与系统大小的平方成正比。此外，在物理量的测量方面，该方法比原始的王 - 兰道方法表现更好。
