Abstract This work presents Neural Equivariant Interatomic Potentials (NequIP), an E(3)-equivariant neural network approach for learning interatomic potentials from ab-initio calculations molecular dynamics simulations. While most contemporary symmetry-aware models use invariant convolutions and only act on scalars, NequIP employs interactions of geometric tensors, resulting in a more information-rich faithful representation atomic environments. The method achieves state-of-the-art accuracy...

Abstract Novel functional materials enable fundamental breakthroughs across technological applications from clean energy to information processing 1–11 . From microchips batteries and photovoltaics, discovery of inorganic crystals has been bottlenecked by expensive trial-and-error approaches. Concurrently, deep-learning models for language, vision biology have showcased emergent predictive capabilities with increasing data computation 12–14 Here we show that graph networks trained at scale...

Abstract Machine learned force fields typically require manual construction of training sets consisting thousands first principles calculations, which can result in low efficiency and unpredictable errors when applied to structures not represented the set model. This severely limits practical application these models systems with dynamics governed by important rare events, such as chemical reactions diffusion. We present an adaptive Bayesian inference method for automating interpretable,...

A simultaneously accurate and computationally efficient parametrization of the potential energy surface molecules materials is a long-standing goal in natural sciences. While atom-centered message passing neural networks (MPNNs) have shown remarkable accuracy, their information propagation has limited accessible length-scales. Local methods, conversely, scale to large simulations but suffered from inferior accuracy. This work introduces Allegro, strictly local equivariant deep network...

Abstract This work examines challenges associated with the accuracy of machine-learned force fields (MLFFs) for bulk solid and liquid phases d -block elements. In exhaustive detail, we contrast performance force, energy, stress predictions across transition metals two leading MLFF models: a kernel-based atomic cluster expansion method implemented using sparse Gaussian processes (FLARE), an equivariant message-passing neural network (NequIP). Early present higher relative errors are more...

Abstract Molecular dynamics simulation is an important tool in computational materials science and chemistry, the past decade it has been revolutionized by machine learning. This rapid progress learning interatomic potentials produced a number of new architectures just few years. Particularly notable among these are atomic cluster expansion, which unified many earlier ideas around atom-density-based descriptors, Neural Equivariant Interatomic Potentials (NequIP), message-passing neural...

The rapid progress of machine learning interatomic potentials over the past couple years produced a number new architectures. Particularly notable among these are Atomic Cluster Expansion (ACE), which unified many earlier ideas around atom density-based descriptors, and Neural Equivariant Interatomic Potentials (NequIP), message passing neural network with equivariant features that showed state art accuracy. In this work, we construct mathematical framework unifies models: ACE is generalised...

Computing accurate reaction rates is a central challenge in computational chemistry and biology because of the high cost free energy estimation with unbiased molecular dynamics. In this work, data-driven machine learning algorithm devised to learn collective variables multitask neural network, where common upstream part reduces dimensionality atomic configurations low dimensional latent space separate downstream parts map predictions basin class labels potential energies. The resulting shown...

Deep learning has emerged as a promising paradigm to give access highly accurate predictions of molecular and material properties. A common short-coming shared by current approaches, however, is that neural networks only point estimates their do not come with predictive uncertainties associated these estimates. Existing uncertainty quantification efforts have primarily leveraged the standard deviation across an ensemble independently trained networks. This incurs large computational overhead...

Ionic liquids (ILs) are an exciting class of electrolytes finding applications in many areas from energy storage to solvents, where they have been touted as "designer solvents" can be mixed precisely tailor the physiochemical properties. As using machine learning interatomic potentials (MLIPs) simulate ILs is still relatively unexplored, several questions need answered see if MLIPs transformative for ILs. Since often not pure, but either together or contain additives, we first demonstrate...

Abstract This work presents Neural Equivariant Interatomic Potentials (NequIP), a SE(3)-equivariant neural network approach for learning interatomic potentials from ab-initio calculations molecular dynamics simulations. While most contemporary symmetry-aware models use invariant convolutions and only act on scalars, NequIP employs interactions of geometric tensors, resulting in more information-rich faithful representation atomic environments. The method achieves state-of-the-art accuracy...

The properties of lithium metal are key parameters in the design lithium-ion and lithium-metal batteries. They difficult to probe experimentally due high reactivity low melting point as well microscopic scales at which exists batteries where it is found have enhanced strength, with implications for dendrite suppression strategies. Computationally, there a lack empirical potentials that consistently quantitatively accurate across all properties, ab initio calculations too costly. In this...

This work brings the leading accuracy, sample efficiency, and robustness of deep equivariant neural networks to extreme computational scale. is achieved through a combination innovative model architecture, massive parallelization, models implementations optimized for efficient GPU utilization. The resulting Allegro architecture bridges accuracy-speed tradeoff atomistic simulations enables description dynamics in structures unprecedented complexity at quantum fidelity. To illustrate...

Abstract Crystallization of amorphous precursors into metastable crystals plays a fundamental role in the formation new matter, from geological to biological processes nature synthesis and development materials laboratory. Reliably predicting outcome such process would enable research directions these areas, but has remained beyond reach molecular modeling or ab initio methods. Here we show that candidates for crystallization products can be predicted many inorganic systems by sampling local...

This work brings the leading accuracy, sample efficiency, and robustness of deep equivariant neural networks to extreme computational scale. is achieved through a combination innovative model architecture, massive parallelization, models implementations optimized for efficient GPU utilization. The resulting Allegro architecture bridges accuracy-speed tradeoff atomistic simulations enables description dynamics in structures unprecedented complexity at quantum fidelity. To illustrate...

A simultaneously accurate and computationally efficient parametrization of the energy atomic forces molecules materials is a long-standing goal in natural sciences. In pursuit this goal, neural message passing has lead to paradigm shift by describing many-body correlations atoms through iteratively messages along an atomistic graph. This propagation information, however, makes parallel computation difficult limits length scales that can be studied. Strictly local descriptor-based methods, on...

Generative models trained at scale can now produce text, video, and more recently, scientific data such as crystal structures. In applications of generative approaches to materials science, in particular structures, the guidance from domain expert form high-level instructions be essential for an automated system output candidate crystals that are viable downstream research. this work, we formulate end-to-end language-to-structure generation a multi-objective optimization problem, propose...

Crystallization of the amorphous phases into metastable crystals plays a fundamental role in formation new matter, from geological to biological processes nature synthesis and development materials laboratory. Predicting outcome such phase transitions reliably would enable research directions these areas, but has remained beyond reach with molecular modeling or ab-initio methods. Here, we show that crystallization products can be predicted any inorganic chemistry by sampling pathways their...

Machine learned force fields typically require manual construction of training sets consisting thousands first principles calculations, which can result in low efficiency and unpredictable errors when applied to structures not represented the set model. This severely limits practical application these models systems with dynamics governed by important rare events, such as chemical reactions diffusion. We present an adaptive Bayesian inference method for automating interpretable,...

The properties of lithium metal are key parameters in the design ion and batteries. They difficult to probe experimentally due high reactivity low melting point as well microscopic scales at which exists batteries where it is found have enhanced strength, with implications for dendrite suppression strategies. Computationally, there a lack empirical potentials that consistently quantitatively accurate across all ab-initio calculations too costly. In this work, we train Machine Learning...