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

Cong, Kien Nguyen; Willman, Jonathan; Moore, Stan Gerald; Belonoshko, Anatoly B.; Gayatri, Rahulkumar; Weinberg, Evan; Wood, Mitchell; Thompson, Aidan P.; Oleynik, Ivan

doi:10.1145/3458817.3487400

Cited by 25 publications

(23 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…MLIPs using strictly local descriptors such as Behler-Parrinello neural networks [5], GAP [6], SNAP [7], DeepMD [20], Moment Tensor Potentials [8], or ACE [12] do not suffer from this obstacle due to their strict locality. As a result they can easily be parallelized across devices and have successfully been scaled to extremely large system sizes [28][29][30][31]. Approaches based on local atom-density based descriptors, however, have so far fallen behind in accuracy compared to state-of-the-art equivariant message passing interatomic potentials [15].…”

Section: Introductionmentioning

confidence: 99%

Learning Local Equivariant Representations for Large-Scale Atomistic Dynamics

Musaelian¹,

Batzner²,

Johansson³

et al. 2022

Preprint

View full text Add to dashboard Cite

A simultaneously accurate and computationally efficient parametrization of the energy and atomic forces of molecules and materials is a long-standing goal in the natural sciences. In pursuit of this goal, neural message passing has lead to a paradigm shift by describing many-body correlations of atoms through iteratively passing messages along an atomistic graph. This propagation of information, however, makes parallel computation difficult and limits the length scales that can be studied. Strictly local descriptor-based methods, on the other hand, can scale to large systems but do not currently match the high accuracy observed with message passing approaches. This work introduces Allegro, a strictly local equivariant deep learning interatomic potential that simultaneously exhibits excellent accuracy and scalability of parallel computation. Allegro learns many-body functions of atomic coordinates using a series of tensor products of learned equivariant representations, but without relying on message passing. Allegro obtains improvements over stateof-the-art methods on the QM9 and revised MD-17 data sets. A single tensor product layer is shown to outperform existing deep message passing neural networks and transformers on the QM9 benchmark. Furthermore, Allegro displays remarkable generalization to out-of-distribution data. Molecular dynamics simulations based on Allegro recover structural and kinetic properties of an amorphous phosphate electrolyte in excellent agreement with first principles calculations. Finally, we demonstrate the parallel scaling of Allegro with a dynamics simulation of 100 million atoms.

show abstract

Section: Introductionmentioning

confidence: 99%

Learning Local Equivariant Representations for Large-Scale Atomistic Dynamics

Musaelian¹,

Batzner²,

Johansson³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…1) The speed of our model exceeds both the non-reactive DeePMD benchmark on simple thermal dynamics in homogeneous systems by 30× for bulk water [26] and 4× for bulk copper [27], the SNAP benchmark by 1.7× for bulk carbon [25], and the reactive force field ReaxFF by at least 5×. ReaxFF benchmark was on 1M atoms on one V100 GPU node, and does not scale linearly with the system size.…”

Section: Innovations Realizedmentioning

confidence: 86%

“…Importantly, none of the existing models describe reactive dynamics, in either hetero-geneous or even homogeneous systems. The SNAP model was employed on bulk carbon system with peak performance 6.21 M atoms•steps/s/node [25]. Likewise, DeePMD achieved 0.3 M atoms•steps/s/node for homogeneous bulk systems -water and copper -on the OLCF Summit machine [26], with the performance recently improved to 2.0 M atoms•steps/s/node [27].…”

Section: A Scalability and Speedmentioning

confidence: 99%

“…Time-to-solution for MLFF has previously been defined as the production simulation speed of the finally trained model, i.e. the M atom-steps/s/node [26], [25]. However, the cost of generating relevant and sufficiently diverse first-principles reference data and training the force fields can take months and requires domain knowledge and arduous human trial and error.…”

Section: Time-to-solution From First Principlesmentioning

confidence: 99%

“…The strong scaling results are shown in Figure 7. In the top panel, the speed of FLARE is compared to the peak performance of SNAP [25] and DeePMD [27]. FLARE outperforms SNAP and DeePMD by factors of 1.7 and 5.3, respectively, when in the regime of linear scaling, achieving a peak performance of 10.5 M atom-steps/s/node.…”

Section: A Scalabilitymentioning

confidence: 99%

See 2 more Smart Citations

Micron-scale heterogeneous catalysis with Bayesian force fields from first principles and active learning

Johansson¹,

Xie²,

Owen³

et al. 2022

Preprint

View full text Add to dashboard Cite

Quantum-mechanically accurate reactive molecular dynamics (MD) at the scale of billions of atoms has been achieved for the heterogeneous catalytic system of H 2 /Pt(111) using the FLARE Bayesian force field. This achievement provides accelerated time-to-solution from first principles, with Bayesian active learning enabling efficient and autonomous training of the machine learning model. The resulting model is then deployed in LAMMPS on GPUs using the Kokkos performance portability library. The Bayesian force field provides quantitative uncertainty of predictions on every atomic environment, critical for detecting configurations in large reactive simulations that are outside of the training set. Scaling benchmarks were performed using realapplication MD of the H 2 /Pt(111) heterogeneous catalysis on the Summit supercomputer, with simulations reaching 0.5 trillion atoms on 4556 GPU nodes.

show abstract

Construction of High Accuracy Machine Learning Interatomic Potential for Surface/Interface of Nanomaterials—A Review

Wan,

He,

Shi

2023

Advanced Materials

View full text Add to dashboard Cite

The inherent discontinuity and unique dimensional attributes of nanomaterial surfaces and interfaces bestow them with various exceptional properties. These properties, however, also introduce difficulties for both experimental and computational studies. The advent of machine learning interatomic potential (MLIP) addresses some of the limitations associated with empirical force fields, presenting a valuable avenue for accurate simulations of these surfaces/interfaces of nanomaterials. Central to this approach is the idea of capturing the relationship between system configuration and potential energy, leveraging the proficiency of machine learning (ML) to precisely approximate high‐dimensional functions. This review offers an in‐depth examination of MLIP principles and their execution, and elaborates on their applications in the realm of nanomaterial surface and interface systems. We also discuss the prevailing challenges faced by this potent methodology.This article is protected by copyright. All rights reserved

show abstract

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

Cited by 25 publications

References 37 publications

Learning Local Equivariant Representations for Large-Scale Atomistic Dynamics

Learning Local Equivariant Representations for Large-Scale Atomistic Dynamics

Micron-scale heterogeneous catalysis with Bayesian force fields from first principles and active learning

Construction of High Accuracy Machine Learning Interatomic Potential for Surface/Interface of Nanomaterials—A Review

Contact Info

Product

Resources

About