We describe collective-move Monte Carlo algorithms designed to approximate the overdamped dynamics of self-assembling nanoscale components equipped with strong, short-ranged and anisotropic interactions. Conventional Monte Carlo simulations comprise sequential moves of single particles, proposed and accepted so as to satisfy detailed balance. Under certain circumstances such simulations provide an approximation of overdamped dynamics, but the accuracy of this approximation can be poor if e.g. particle-particle interactions vary strongly with distance or angle. The twin requirements of simulation efficiency (trial moves of appreciable scale are needed to ensure reasonable sampling) and dynamical fidelity (true in the limit of vanishingly small trial moves) then become irreconcilable. As a result, single-particle moves can underrepresent important collective modes of relaxation, such as self-diffusion of particle clusters. However, one way of using Monte Carlo simulation to mimic real collective modes of motion, retaining the ability to make trial moves of reasonable scale, is to make explicit moves of collections of particles. We will outline ways of doing so by iteratively linking particles to their environment. Linking criteria can be static, conditioned upon properties of the current state of a system, or dynamic, conditioned upon energy changes resulting from trial virtual moves of particles. We argue that the latter protocol is better-suited to approximating real dynamics.
Comment on: Thermostatistics of Overdamped Motion of Interacting Particles [arXiv:1008.1421]
在最近的一篇论文中,Phys. Rev. Lett. 105 260601(2010)[arXiv:1008.1421],Andrade等人,他认为,在T=0时被限制在抛物阱中的经典粒子按照Tsallis统计分布。为了证明他们的观点,作者进行了分子动力学模拟。在这里,我们证明Andrade等人的模型可以精确求解。T=0时粒子的分布与Tsallis熵无关,而仅仅由力平衡决定。