Parallel Algorithms for Molecular-Dynamics Simulations of Coulombic Systems

1992 ◽  
Vol 291 ◽  
Author(s):  
Wei Li ◽  
Rajiv K. Kalia ◽  
Simon De Leeuw ◽  
Aiichiro Nakano ◽  
Donald Greenwell ◽  
...  
Keyword(s):  
Molecular Dynamics ◽  
Computational Complexity ◽  
Parallel Algorithms ◽  
Molecular Dynamics Simulations ◽  
Execution Time ◽  
Divide And Conquer ◽  
Three Dimension ◽  
Interprocessor Communication ◽  
Time Step ◽  
Dynamics Simulations

ABSTRACTIn molecular-dynamics simulations for the long-range Coulomb interaction, a great deal of effort is devoted to reducing the computational complexity of the usual N2operations in the direct calculation. For bulk systems, we have designed a parallel algorithm based on the domain-decomposition strategy for the Ewald summation. The performance of the algorithm is evaluated on the in-house iPSC/860 system. We find that this algorithm reduces the computational complexity to O(N). For a 64,000-particle plasma in three dimension, the execution time on an 8-node system is 27.4 sec per MD time step. The interprocessor communication is a small fraction of the total execution time. We find linear speedups and a parallel efficiency of 0.85. For comparison, parallel algorithms are also designed for the Fast Multipole Method (FMM) - a divide and conquer scheme in which the system is divided into cubic subdomains and interactions between distant charged regions are calculated with a truncated multipole expansion. The performance of the FMM on Touchstone Delta machine is discussed.

PARALLEL MOLECULAR DYNAMICS SIMULATIONS OF ORGANIC MATERIALS

1994 ◽  
Vol 05 (02) ◽  
pp. 295-298 ◽  
Author(s):  
STEVE PLIMPTON ◽  
BRUCE HENDRICKSON
Keyword(s):  
Molecular Dynamics ◽  
Parallel Algorithms ◽  
Molecular Dynamics Simulations ◽  
Parallel Algorithm ◽  
Communication Cost ◽  
Dynamics Model ◽  
Organic Systems ◽  
Intel Paragon ◽  
Parallel Molecular Dynamics ◽  
Dynamics Simulations

A new parallel algorithm suitable for molecular dynamics simulations of organic systems is presented. It reduces the communication cost and memory requirements of other commonly-used parallel algorithms by a factor of [Formula: see text] where P is the number of processors. The algorithm has been implemented in a CHARMM-like molecular dynamics model and its performance on 1024-processor nCUBE 2 and Intel Paragon machines is discussed.

scholarly journals Variable time domain discretization methodology for molecular dynamics simulation of metallic compounds

2021 ◽  
Author(s):  
Jamal Zeinalov
Keyword(s):  
Molecular Dynamics ◽  
Molecular Dynamics Simulation ◽  
Potential Energy ◽  
Molecular Dynamics Simulations ◽  
Dynamics Simulation ◽  
Time Step ◽  
Variable Time ◽  
Metallic Compounds ◽  
Dynamics Simulations ◽  
Computational Resources

The present work proposes a methodology to improve the computational requirements of molecular dynamics simulations while maintaining or improving the fidelity of the obtained results. The most common method of molecular dynamics simulation at present is the multi-force, constant time-step, explicit computation, which advances a single time step at a time to determine the next state of the system. The present work proposes a variable time-step strategy, where a single large simulation is subdivided into multiple time domains which redistribute computational resources where they are needed the most: in areas of higher than average potential or kinetic energy or highly dynamic areas around impurity clusters, void formations and crack propagations. The research focuses on the simulation of metallic compounds, as these form the basis of most common molecular dynamics simulations, and have been very thoroughly investigated over the years, thus providing a very extensive body of work for the purpose of comparison and validation of the proposed methodology. The novel methodology presented in this work allows to alleviate some of the limitations associated with the molecular dynamics methodologies and go beyond traditional scales of simulation. The proposed method has been observed to deliver 5 to 20 percent increase in simulation size domain while maintaining or improving the accuracy and computational cycle time. The benefits were observed to be greater for large simulations with one or more areas of higher than average kinetic or potential energy levels, such as those found during crack initiation and propagation, coating-substrate interface, localized pressure application or large thermal gradient. The large difference allows for very clear prioritization of computational resources for high energy areas and as a result provides for faster and more accurate simulation even with increased domain size. Conversely, this method has been observed to provide little to no benefit when simulating stable systems that are undergoing very slow change, such as (relatively) slow change in ambient temperature or pressure, or otherwise homogeneous internal and external boundary conditions. However, for the majority of applications described above, including coating deposition and additive manufacturing, the proposed methodology will yield substantial increase in both simulation size and accuracy, since in the aforementioned processes kinetic and potential energy gradients across the simulation are typically very significant

Sign in / Sign up

Export Citation Format

Share Document