General Purpose Molecular Dynamics Research Articles

An exa-scale high-performance molecular dynamics simulation program: MODYLAS.

A new version of the highly parallelized general-purpose molecular dynamics (MD) simulation program MODYLAS with high performance on the Fugaku computer was developed. A benchmark test using Fugaku indicated highly efficient communication, single instruction, multiple data (SIMD) processing, and on-cache arithmetic operations. The system's performance deteriorated only slightly, even under high parallelization. In particular, a newly developed minimum transferred data method, requiring a significantly lower amount of data transfer compared to conventional communications, showed significantly high performance. The coordinates and forces of 101 810 176 atoms and the multipole coefficients of the subcells could be distributed to the 32 768 nodes (1 572 864 cores) in 2.3ms during one MD step calculation. The SIMD effective instruction rates for floating-point arithmetic operations in direct force and fast multipole method (FMM) calculations measured on Fugaku were 78.7% and 31.5%, respectively. The development of a data reuse algorithm enhanced the on-cache processing; the cache miss rate for direct force and FMM calculations was only 2.74% and 1.43%, respectively, on the L1 cache and 0.08% and 0.60%, respectively, on the L2 cache. The modified MODYLAS could complete one MD single time-step calculation within 8.5ms for the aforementioned large system. Additionally, the program contains numerous functions for material research that enable free energy calculations, along with the generation of various ensembles and molecular constraints.

RUMD: A general purpose molecular dynamics package optimized to utilize GPU hardware down to a few thousand particles

RUMD is a general purpose, high-performance molecular dynamics (MD) simulation package running on graphical processing units (GPU’s). RUMD addresses the challenge of utilizing the many-core nature of modern GPU hardware when simulating small to medium system sizes (roughly from a few thousand up to hundred thousand particles). It has a performance that is comparable to other GPU-MD codes at large system sizes and substantially better at smaller sizes. RUMD is open-source and consists of a library written in C++ and the CUDA extension to C, an easy-to-use Python interface, and a set of tools for set-up and post-simulation data analysis. The paper describes RUMD’s main features, optimizations and performance benchmarks.

Electronic coarse graining enhances the predictive power of molecular simulation allowing challenges in water physics to be addressed

One key factor that limits the predictive power of molecular dynamics simulations is the accuracy and transferability of the input force field. Force fields are challenged by heterogeneous environments, where electronic responses give rise to biologically important forces such as many-body polarisation and dispersion. The importance of polarisation in the condensed phase was recognised early on, as described by Cochran in 1959 [Philosophical Magazine 4 (1959) 1082–1086] [32]. Currently in molecular simulation, dispersion forces are treated at the two-body level and in the dipole limit, although the importance of three-body terms in the condensed phase was demonstrated by Barker in the 1980s [Phys. Rev. Lett. 57 (1986) 230–233] [72]. One approach for treating both polarisation and dispersion on an equal basis is to coarse grain the electrons surrounding a molecular moiety to a single quantum harmonic oscillator (cf. Hirschfelder, Curtiss and Bird 1954 [The Molecular Theory of Gases and Liquids (1954)] [37]). The approach, when solved in strong coupling beyond the dipole limit, gives a description of long-range forces that includes two- and many-body terms to all orders. In the last decade, the tools necessary to implement the strong coupling limit have been developed, culminating in a transferable model of water with excellent predictive power across the phase diagram. Transferability arises since the environment automatically identifies the important long range interactions, rather than the modeller through a limited set of expressions. Here, we discuss the role of electronic coarse-graining in predictive multiscale materials modelling and describe the first implementation of the method in a general purpose molecular dynamics software: QDO_MD.

Optimization of Analytical Potentials for Coarse-Grained Biopolymer Models.

The increasing trend in the recent literature on coarse grained (CG) models testifies their impact in the study of complex systems. However, the CG model landscape is variegated: even considering a given resolution level, the force fields are very heterogeneous and optimized with very different parametrization procedures. Along the road for standardization of CG models for biopolymers, here we describe a strategy to aid building and optimization of statistics based analytical force fields and its implementation in the software package AsParaGS (Assisted Parameterization platform for coarse Grained modelS). Our method is based on the use and optimization of analytical potentials, optimized by targeting internal variables statistical distributions by means of the combination of different algorithms (i.e., relative entropy driven stochastic exploration of the parameter space and iterative Boltzmann inversion). This allows designing a custom model that endows the force field terms with a physically sound meaning. Furthermore, the level of transferability and accuracy can be tuned through the choice of statistical data set composition. The method-illustrated by means of applications to helical polypeptides-also involves the analysis of two and three variable distributions, and allows handling issues related to the FF term correlations. AsParaGS is interfaced with general-purpose molecular dynamics codes and currently implements the "minimalist" subclass of CG models (i.e., one bead per amino acid, Cα based). Extensions to nucleic acids and different levels of coarse graining are in the course.

Rapid calculation of hydrodynamic and transport properties in concentrated solutions of colloidal particles and macromolecules

A new method for calculating the resistance tensors of arbitrarily shaped particles and the translational and rotational self-diffusivity in suspensions of such particles is developed. This approach can be harnessed to efficiently and accurately predict the hydrodynamic and transport properties of large macromolecules such as antibodies in solution. Particles are modeled as a rigid composite of spherical beads, and the continuum equations for low Reynolds number fluid mechanics are used to calculate the drag on the composite or its diffusivity in a solution of other composites. The hydrodynamic calculations are driven by a graphics processing unit (GPU) implementation of the particle-mesh-Ewald technique which offers log-linear scaling with respect to the complexity of the composite-bead particles modeled as well as high speed execution leveraging the hyper-parallelization of the GPU. Matrix-free expressions for the hydrodynamic resistance and translational and rotational diffusivity of composite bead particles are developed, which exhibit substantial improvements in computational complexity over existing approaches. The effectiveness of these methods is demonstrated through a series of calculations for composite-bead particles having a spherical geometry, and the results are compared to exact solutions for spheres. Included in the supplementary material is an implementation of the proposed algorithm which functions as a plug-in for the GPU molecular dynamics suite HOOMD-blue (http://codeblue.umich.edu/hoomd-blue) [J. A. Anderson, C. D. Lorenz, and A. Travesset, “General purpose molecular dynamics simulations fully implemented on graphics processing units,” J. Comput. Phys. 227(10), 5342–5359 (2008) and Glaser et al., “Strong scaling of general-purpose molecular dynamics simulations on GPUs,” Comput. Phys. Commun. 192, 97–107 (2015)].

Strong scaling of general-purpose molecular dynamics simulations on GPUs

We describe a highly optimized implementation of MPI domain decomposition in a GPU-enabled, general-purpose molecular dynamics code, HOOMD-blue (Anderson and Glotzer, 2013). Our approach is inspired by a traditional CPU-based code, LAMMPS (Plimpton, 1995), but is implemented within a code that was designed for execution on GPUs from the start (Anderson et al., 2008). The software supports short-ranged pair force and bond force fields and achieves optimal GPU performance using an autotuning algorithm. We are able to demonstrate equivalent or superior scaling on up to 3375 GPUs in Lennard-Jones and dissipative particle dynamics (DPD) simulations of up to 108 million particles. GPUDirect RDMA capabilities in recent GPU generations provide better performance in full double precision calculations. For a representative polymer physics application, HOOMD-blue 1.0 provides an effective GPU vs. CPU node speed-up of 12.5×.

MODYLAS: A Highly Parallelized General-Purpose Molecular Dynamics Simulation Program for Large-Scale Systems with Long-Range Forces Calculated by Fast Multipole Method (FMM) and Highly Scalable Fine-Grained New Parallel Processing Algorithms.

Our new molecular dynamics (MD) simulation program, MODYLAS, is a general-purpose program appropriate for very large physical, chemical, and biological systems. It is equipped with most standard MD techniques. Long-range forces are evaluated rigorously by the fast multipole method (FMM) without using the fast Fourier transform (FFT). Several new methods have also been developed for extremely fine-grained parallelism of the MD calculation. The virtually buffering-free methods for communications and arithmetic operations, the minimal communication latency algorithm, and the parallel bucket-relay communication algorithm for the upper-level multipole moments in the FMM realize excellent scalability. The methods for blockwise arithmetic operations avoid data reload, attaining very small cache miss rates. Benchmark tests for MODYLAS using 65 536 nodes of the K-computer showed that the overall calculation time per MD step including communications is as short as about 5 ms for a 10 million-atom system; that is, 35 ns of simulation time can be computed per day. The program enables investigations of large-scale real systems such as viruses, liposomes, assemblies of proteins and micelles, and polymers.

Accelerating Molecular Dynamics Simulation Using Graphics Processing Unit

We have developed CUDA-enabled version of a general purpose molecular dynamics simulation code for GPU. Implementation details including parallelization scheme and performance optimization are described. Here we have focused on the non-bonded force calculation because it is most time consuming part in molecular dynamics simulation. Timing results using CUDA-enabled and CPU versions were obtained and compared for a biomolecular system containing 23558 atoms. CUDA-enabled versions were found to be faster than CPU version. This suggests that GPU could be a useful hardware for molecular dynamics simulation.

GUI-BioPASED: A program for molecular dynamics simulations of biopolymers with a graphical user interface

There are several software packages that allow the user to model main types of biopolymers with molecular dynamics methods. However, the work with most of these packages is quite complicated, especially for non-experts in computational structural biology. We have created GUI-BioPASED, a Web-based GUI for molecular dynamics software BioPASED, to partially automate the tasks usually responsible for the majority of errors. GUI-BioPASED creates a task as a standalone executable file, which can be exported to the client side, and ensures its execution. A user-friendly cross-platform interface is created that performs data check, reports errors and hints at possible ways to correct them. BioPASED, a general purpose molecular dynamics program, and GUI-BioPASED along with their typical example of usage are described.

JINR CICC in computational chemistry and nanotechnology problems: DL_POLY performance for different communication architectures

This report compares the performance of the DL POLY general-purpose molecular dynamics simulation package on the LIT JINR computing cluster CICC with various communication systems. The comparison involved two cluster architectures: Gigabit Ethernet and InniBand technologies, respectively. The code performance tests include some comparison of the CICC cluster with the special-purpose computer MDGRAPE-3 developed at RIKEN for a high-speed acceleration of the MD (molecular dynamics) without axed cutoff. The DL POLY benchmark covers a set of typical MD system simulations detailed below.

General purpose molecular dynamics simulations fully implemented on graphics processing units

Graphics processing units (GPUs), originally developed for rendering real-time effects in computer games, now provide unprecedented computational power for scientific applications. In this paper, we develop a general purpose molecular dynamics code that runs entirely on a single GPU. It is shown that our GPU implementation provides a performance equivalent to that of fast 30 processor core distributed memory cluster. Our results show that GPUs already provide an inexpensive alternative to such clusters and discuss implications for the future.

Use of massively parallel molecular dynamics simulations for radiation damage in pyrochlores

DL_POLY_3 is a general purpose molecular dynamics (MD) simulation package designed to simulate systems of the order of tens of millions of particles and beyond by efficiently harnessing the power of modern computer clusters. Here we discuss the package design, functionality and report on performance and capability limits. We then report the application of DL_POLY_3 to study radiation cascades in Gd2Ti2O7 and Gd2Zr2O7, potential materials for high-level radioactive waste storage and discuss problems associated with the analysis of the cascades. We see little direct amorphisation but rather the start of a transition to the fluorite structure which is more pronounced for the Zr than the Ti compound.

A short description of DL_POLY

DL_POLY is a general purpose molecular dynamics simulation package with in-built parallel algorithms. It may be run on a wide selection of distributed memory parallel computers, from national supercomputers with thousands of processors, to single processor workstations and can simulate small systems with order 100 atoms, to systems with millions of atoms. This introduction provides an outline of the features of the package and the underlying methodology.

DL_POLY_3: the CCP5 national UK code for molecular-dynamics simulations.

DL_POLY_3 is a general-purpose molecular-dynamics simulation package embedding a highly efficient domain decomposition (DD) parallelization strategy. It was developed at Daresbury Laboratory under the auspices of the Engineering and Physical Sciences Research Council. Written to support academic research, it has a wide range of applications and will run on a wide range of computers; from single-processor workstations to multi-processor computers, with accent on the efficient use of multi-processor power. A new DD adaptation of the smoothed particle mesh Ewald method for calculating long-range forces in molecular simulations, incorporating a novel three-dimensional fast Fourier transform (the Daresbury Advanced Fourier Transform), makes it possible to simulate systems of the order of one million particles and beyond. DL_POLY_3 structure, functionality, performance and availability are described in this feature paper.

DL_POLY: Application to molecular simulation

DL_POLY is a general-purpose molecular dynamics simulation package, which was developed by Daresbury Laboratory in the mid-1990s for the molecular simulation community in the United Kingdom. The package now has a world-wide user base and applications in many areas of molecular simulation. In this article we briefly review the history and design of the package and highlight some recent applications in the areas of; liquids and solutions; spectroscopy; ionic solids; molecular crystals; polymers; glasses; membranes; proteins; solid and liquid interfaces; catalysis; liquid crystals; intercalation and clathrates; and novel systems. The strengths and weaknesses of the code and its future in the near term are also discussed.

Development of General Purpose Molecular Dynamics Simulator for Large Scaled System of Complex Materials

Development of General Purpose Molecular Dynamics Simulator for Large Scaled System of Complex Materials

A domain decomposition parallel processing algorithm for molecular dynamics simulations of systems of arbitrary connectivity

We describe in this paper methods for applying domain decomposition to a general purpose molecular dynamics program. The algorithm is suitable for either distributed memory parallel computers or shared memory machines with message passing libraries. A method is discussed in detail which allows molecules of arbitrary connectivity to be simulated within the domain decomposition approach. The algorithm also contains techniques to handle both rigid bond constraints and special CH 2 constraints which involve five atoms. Examples are given for molecular systems containing both types of constraints plus two, three- and four-body intramolecular potentials as well as short-range and long-range non-bonded potentials. The algorithm has been implemented on a Silicon Graphics Power Challenge machine using the MPI message passing library.

Parallelism in computational chemistry

An account is given of experience gained in implementing computational chemistry application software, including quantum chemistry and macromolecular refinement codes, on distributed memory parallel processors. In quantum chemistry we consider the coarse-grained implementation of Gaussian integral and derivative integral evaluation, the direct-SCF computation of an uncorrelated wavefunction, the 4-index transformation of two-electron integrals and the direct-CI calculation of correlated wavefunctions. In the refinement of macromolecular conformations, we describe domain decomposition techniques used in implementing general purpose molecular mechanics, molecular dynamics and free energy perturbation calculations. Attention is focused on performance figures obtained on the Intel iPSC/2 and iPSC/860 hypercubes, which are compared with those obtained on a Cray Y-MP/464 and Convex C-220 minisupercomputer. From this data we deduce the cost effectiveness of parallel processors in the field of computational chemistry.