ترغب بنشر مسار تعليمي؟ اضغط هنا

Efficient swap algorithms for molecular dynamics simulations of equilibrium supercooled liquids

126   0   0.0 ( 0 )
 نشر من قبل Camille Scalliet
 تاريخ النشر 2018
  مجال البحث فيزياء
والبحث باللغة English




اسأل ChatGPT حول البحث

It was recently demonstrated that a simple Monte Carlo (MC) algorithm involving the swap of particle pairs dramatically accelerates the equilibrium sampling of simulated supercooled liquids. We propose two numerical schemes integrating the efficiency of particle swaps into equilibrium molecular dynamics (MD) simulations. We first develop a hybrid MD/MC scheme combining molecular dynamics with the original swap Monte Carlo. We implement this hybrid method in LAMMPS, a software package employed by a large community of users. Secondly, we define a continuous time version of the swap algorithm where both the positions and diameters of the particles evolve via Hamiltons equations of motion. For both algorithms, we discuss in detail various technical issues as well as the optimisation of simulation parameters. We compare the numerical efficiency of all available swap algorithms and discuss their relative merits.



قيم البحث

اقرأ أيضاً

Glasses are solid materials whose constituent atoms are arranged in a disordered manner. The transition from a liquid to a glass remains one of the most poorly understood phenomena in condensed matter physics, and still no fully microscopic theory ex ists that can describe the dynamics of supercooled liquids in a quantitative manner over all relevant time scales. Here we present such a theoretical framework that yields near-quantitative accuracy for the time-dependent correlation functions of a supercooled system over a broad density range. Our approach requires only simple static structural information as input and is based entirely based on first principles. Owing to this first-principles nature, the framework offers a unique platform to study the relation between structure and dynamics in glass-forming matter, and paves the way towards a systematically correctable and ultimately fully quantitative theory of microscopic glassy dynamics.
238 - H. Watanabe , M. Suzuki , 2010
Efficient implementations of the classical molecular dynamics (MD) method for Lennard-Jones particle systems are considered. Not only general algorithms but also techniques that are efficient for some specific CPU architectures are also explained. A simple spatial-decomposition-based strategy is adopted for parallelization. By utilizing the developed code, benchmark simulations are performed on a HITACHI SR16000/J2 system consisting of IBM POWER6 processors which are 4.7 GHz at the National Institute for Fusion Science (NIFS) and an SGI Altix ICE 8400EX system consisting of Intel Xeon processors which are 2.93 GHz at the Institute for Solid State Physics (ISSP), the University of Tokyo. The parallelization efficiency of the largest run, consisting of 4.1 billion particles with 8192 MPI processes, is about 73% relative to that of the smallest run with 128 MPI processes at NIFS, and it is about 66% relative to that of the smallest run with 4 MPI processes at ISSP. The factors causing the parallel overhead are investigated. It is found that fluctuations of the execution time of each process degrade the parallel efficiency. These fluctuations may be due to the interference of the operating system, which is known as OS Jitter.
In the replica-exchange molecular dynamics method, where constant-temperature molecular dynamics simulations are performed in each replica, one usually rescales the momentum of each particle after replica exchange. This rescaling method had previousl y been worked out only for the Gaussian constraint method. In this letter, we present momentum rescaling formulae for four other commonly used constant-temperature algorithms, namely, Langevin dynamics, Andersen algorithm, Nos{e}-Hoover thermostat, and Nos{e}-Poincar{e} thermostat. The effectiveness of these rescaling methods is tested with a small biomolecular system, and it is shown that proper momentum rescaling is necessary to obtain correct results in the canonical ensemble.
We introduce a scheme for deriving an optimally-parametrised Langevin dynamics of few collective variables from data generated in molecular dynamics simulations. The drift and the position-dependent diffusion profiles governing the Langevin dynamics are expressed as explicit averages over the input trajectories. The proposed strategy is applicable to cases when the input trajectories are generated by subjecting the system to a external time-dependent force (as opposed to canonically-equilibrated trajectories). Secondly, it provides an explicit control on the statistical uncertainty of the drift and diffusion profiles. These features lend to the possibility of designing the external force driving the system so to maximize the accuracy of the drift and diffusions profile throughout the phase space of interest. Quantitative criteria are also provided to assess a posteriori the satisfiability of the requisites for applying the method, namely the Markovian character of the stochastic dynamics of the collective variables.
We show that spherical truncations of the 1/r interactions in models for water and acetonitrile yield very accurate results in bulk simulations for all site-site pair correlation functions as well as dipole-dipole correlation functions. This good per formance in bulk simulations contrasts with the generally poor results found with the use of such truncations in nonuniform molecular systems. We argue that Local Molecular Field (LMF) theory provides a general theoretical framework that gives the necessary corrections to simple truncations in most nonuniform environments and explains the accuracy of spherical truncations in uniform environments by showing that these corrections are very small. LMF theory is derived from the exact Yvon-Born-Green (YBG) hierarchy by making physically-motivated and well-founded approximations. New and technically interesting derivations of both the YBG hierarchy and LMF theory for a variety of site-site molecular models are presented in appendices. The main paper focuses on understanding the accuracy of these spherical truncations in uniform systems both phenomenologically and quantitatively using LMF theory.
التعليقات
جاري جلب التعليقات جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
mircosoft-partner

هل ترغب بارسال اشعارات عن اخر التحديثات في شمرا-اكاديميا