Billion atom molecular dynamics simulations of carbon at extreme conditions and experimental time and length scales

Billion atom molecular dynamics (MD) using quantum-accurate machine-learning Spectral Neighbor Analysis Potential (SNAP) observed long-sought high pressure BC8 phase of carbon at extreme pressure (12 Mbar) and temperature (5,000 K). 24-hour, 4650 node production simulation on OLCF Summit demonstrated an unprecedented scaling and unmatched real-world performance of SNAP MD while sampling 1 nanosecond of physical time. Efficient implementation of SNAP force kernel in LAMMPS using the Kokkos CUDA backend on NVIDIA GPUs combined with excellent strong scaling (better than 97% parallel efficiency) enabled a peak computing rate of 50.0 PFLOPs (24.9% of theoretical peak) for a 20 billion atom MD simulation on the full Summit machine (27,900 GPUs). The peak MD performance of 6.21 Matom-steps/node-s is 22.9 times greater than a previous record for quantum-accurate MD. Near perfect weak scaling of SNAP MD highlights its excellent potential to advance the frontier of quantum-accurate MD to trillion atom simulations on upcoming exascale platforms.

[1]  Volker L. Deringer,et al.  Origins of structural and electronic transitions in disordered silicon , 2021, Nature.

[2]  Gilbert W. Collins,et al.  X-ray diffraction at the National Ignition Facility. , 2020, Review of Scientific Instruments.

[3]  D. Queloz.  Nobel Lecture: 51 Pegasi b and the exoplanet revolution , 2020 .

[4]  A. Thompson,et al.  Computational aspects of many-body potentials , 2012 .

[5]  J. Behler.  Perspective: Machine learning potentials for atomistic simulations. , 2016, The Journal of chemical physics.

[6]  A. Thompson,et al.  Beryllium-driven structural evolution at the divertor surface , 2021 .

[7]  E Weinan,et al.  Deep Potential Molecular Dynamics: a scalable model with the accuracy of quantum mechanics , 2017, Physical review letters.

[8]  A. Thompson,et al.  Explicit Multi-element Extension of the Spectral Neighbor Analysis Potential for Chemically Complex Systems. , 2020, The journal of physical chemistry. A.

[9]  B. Militzer,et al.  A multiphase equation of state for carbon addressing high pressures and temperatures , 2013, 1311.4577.

[10]  Gilbert W. Collins,et al.  Melting temperature of diamond at ultrahigh pressure , 2010 .

[11]  Volker L. Deringer,et al.  Machine Learning Interatomic Potentials as Emerging Tools for Materials Science , 2019, Advanced materials.

[12]  Aidan P Thompson,et al.  Extending the accuracy of the SNAP interatomic potential form. , 2017, The Journal of chemical physics.

[13]  Paolo Bientinesi,et al.  The Vectorization of the Tersoff Multi-body Potential: An Exercise in Performance Portability , 2016, SC16: International Conference for High Performance Computing, Networking, Storage and Analysis.

[14]  Adrian E. Roitberg,et al.  Less is more: sampling chemical space with active learning , 2018, The Journal of chemical physics.

[15]  H. L. Hanshaw,et al.  Review of pulsed power-driven high energy density physics research on Z at Sandia , 2020, Physics of Plasmas.

[16]  Gilbert W. Collins,et al.  Metastability of diamond ramp-compressed to 2 terapascals , 2021, Nature.

[17]  Christian Trott,et al.  Spectral neighbor analysis method for automated generation of quantum-accurate interatomic potentials , 2014, J. Comput. Phys..

[18]  R. F. Smith,et al.  Ramp compression of diamond to five terapascals , 2014, Nature.

[19]  Volker L. Deringer,et al.  Growth Mechanism and Origin of High sp^{3} Content in Tetrahedral Amorphous Carbon. , 2018, Physical review letters.

[20]  Cas van der Oord,et al.  Performant implementation of the atomic cluster expansion (PACE) and application to copper and silicon , 2021, npj Computational Materials.

[21]  M. Mayor.  Nobel Lecture: Plurality of worlds in the cosmos: A dream of antiquity, a modern reality of astrophysics , 2020, Reviews of Modern Physics.

[22]  J. Behler,et al.  A Performance and Cost Assessment of Machine Learning Interatomic Potentials. , 2019, The journal of physical chemistry. A.

[23]  E Weinan,et al.  Pushing the Limit of Molecular Dynamics with Ab Initio Accuracy to 100 Million Atoms with Machine Learning , 2020, SC20: International Conference for High Performance Computing, Networking, Storage and Analysis.

[24]  Raymond F. Smith,et al.  Ultra-High Pressure Dynamic Compression of Geological Materials , 2019, Front. Earth Sci..

[25]  M. McMahon.  Synchrotron and FEL Studies of Matter at High Pressures , 2018, Synchrotron Light Sources and Free-Electron Lasers.

[26]  Steve Plimpton,et al.  Fast parallel algorithms for short-range molecular dynamics , 1993 .

[27]  T. Mattsson,et al.  Thermodynamically constrained correction to ab initio equations of state , 2014 .

[28]  Ralf Drautz.  Atomic cluster expansion of scalar, vectorial, and tensorial properties including magnetism and charge transfer , 2020 .

[29]  A. Thompson,et al.  Quantum accurate SNAP carbon potential for MD shock simulations , 2020, SHOCK COMPRESSION OF CONDENSED MATTER - 2019: Proceedings of the Conference of the American Physical Society Topical Group on Shock Compression of Condensed Matter.

[30]  Nicholas Lubbers,et al.  Automated discovery of a robust interatomic potential for aluminum , 2020, Nature Communications.

[31]  R. Kondor,et al.  Gaussian approximation potentials: the accuracy of quantum mechanics, without the electrons. , 2009, Physical review letters.

[32]  Brian D. Wirth,et al.  Data-driven material models for atomistic simulation , 2019, Physical Review B.

[33]  Daniel Sunderland,et al.  Kokkos: Enabling manycore performance portability through polymorphic memory access patterns , 2014, J. Parallel Distributed Comput..

[34]  L. Fried,et al.  ChIMES: A Force Matched Potential with Explicit Three-Body Interactions for Molten Carbon. , 2017, Journal of chemical theory and computation.

[35]  Simon D. Hammond,et al.  SNAP: Strong Scaling High Fidelity Molecular Dynamics Simulations on Leadership-Class Computing Platforms , 2014, ISC.

[36]  Duncan Poole,et al.  Routine Microsecond Molecular Dynamics Simulations with AMBER on GPUs. 1. Generalized Born , 2012, Journal of chemical theory and computation.

[37]  Justin S. Smith,et al.  Hierarchical modeling of molecular energies using a deep neural network. , 2017, The Journal of chemical physics.

[38]  C. Dominik,et al.  Mineralogy, Structure, and Habitability of Carbon-Enriched Rocky Exoplanets: A Laboratory Approach. , 2018, Astrobiology.

[39]  J. Behler.  Atom-centered symmetry functions for constructing high-dimensional neural network potentials. , 2011, The Journal of chemical physics.

[40]  E Weinan,et al.  86 PFLOPS Deep Potential Molecular Dynamics simulation of 100 million atoms with ab initio accuracy , 2020, Comput. Phys. Commun..

[41]  Alexander V. Shapeev,et al.  Moment Tensor Potentials: A Class of Systematically Improvable Interatomic Potentials , 2015, Multiscale Model. Simul..

[42]  Conrad W. Rosenbrock,et al.  Machine-learned interatomic potentials for alloys and alloy phase diagrams , 2019, npj Computational Materials.

[43]  Jonathan Vandermause,et al.  On-the-fly active learning of interpretable Bayesian force fields for atomistic rare events , 2020, npj Computational Materials.