4 May 2022 in Nature Communications
This work presents Neural Equivariant Interatomic Potentials (NequIP), an E(3)-equivariant neural network approach for learning interatomic potentials from ab-initio calculations for molecular dynamics simulations. While most contemporary symmetry-aware models use invariant convolutions and only act on scalars, NequIP employs E(3)-equivariant convolutions for interactions of geometric tensors, resulting in a more information-rich and faithful representation of atomic environments. The method achieves state-of-the-art accuracy on a challenging and diverse set of molecules and materials while exhibiting remarkable data efficiency. NequIP outperforms existing models with up to three orders of magnitude fewer training data, challenging the widely held belief that deep neural networks require massive training sets. The high data efficiency of the method allows for the construction of accurate potentials using high-order quantum chemical level of theory as reference and enables high-fidelity molecular dynamics simulations over long time scales.
3 February 2023 in Nature Communications
A simultaneously accurate and computationally efficient parametrization of the potential energy surface of molecules and materials is a long-standing goal in the natural sciences. While atom-centered message passing neural networks (MPNNs) have shown remarkable accuracy, their information propagation has limited the accessible length-scales. Local methods, conversely, scale to large simulations but have suffered from inferior accuracy. This work introduces Allegro, a strictly local equivariant deep neural network interatomic potential architecture that simultaneously exhibits excellent accuracy and scalability. Allegro represents a many-body potential using iterated tensor products of learned equivariant representations without atom-centered message passing. Allegro obtains improvements over state-of-the-art methods on QM9 and revMD17. A single tensor product layer outperforms existing deep MPNNs and transformers on QM9. Furthermore, Allegro displays remarkable generalization to out-of-distribution data. Molecular simulations using Allegro recover structural and kinetic properties of an amorphous electrolyte in excellent agreement with ab-initio simulations. Finally, we demonstrate parallelization with a simulation of 100 million atoms.
1 January 2011 in Journal of the Electrochemical Society
RSRoel S. Sanchez-Carrera
+2
Lithium/air batteries, based on their high theoretical specific energy, are an extremely attractive technology for electrical energy storage that could make long-range electric vehicles widely affordable. However, the impact of this technology has so far fallen short of its potential due to several daunting challenges. In nonaqueous Li/air cells, reversible chemistry with a high current efficiency over several cycles has not yet been established, and the deposition of an electrically resistive discharge product appears to limit the capacity. Aqueous cells require water-stable lithium-protection membranes that tend to be thick, heavy, and highly resistive. Both types of cell suffer from poor oxygen redox kinetics at the positive electrode and deleterious volume and morphology changes at the negative electrode. Closed Li/air systems that include oxygen storage are much larger and heavier than open systems, but so far oxygen- and OH−-selective membranes are not effective in preventing contamination of cells. In this review we discuss the most critical challenges to developing robust, high-energy Li/air batteries and suggest future research directions to understand and overcome these challenges. We predict that Li/air batteries will primarily remain a research topic for the next several years. However, if the fundamental challenges can be met, the Li/air battery has the potential to significantly surpass the energy storage capability of today's Li-ion batteries.
1 January 2016 in Computational Materials Science
18 March 2020 in npj Computational Materials
Machine learned force fields typically require manual construction of training sets consisting of thousands of first principles calculations, which can result in low training efficiency and unpredictable errors when applied to structures not represented in the training set of the model. This severely limits the practical application of these models in systems with dynamics governed by important rare events, such as chemical reactions and diffusion. We present an adaptive Bayesian inference method for automating the training of interpretable, low-dimensional, and multi-element interatomic force fields using structures drawn on the fly from molecular dynamics simulations. Within an active learning framework, the internal uncertainty of a Gaussian process regression model is used to decide whether to accept the model prediction or to perform a first principles calculation to augment the training set of the model. The method is applied to a range of single- and multi-element systems and shown to achieve a favorable balance of accuracy and computational efficiency, while requiring a minimal amount of ab initio training data. We provide a fully open-source implementation of our method, as well as a procedure to map trained models to computationally efficient tabulated force fields.
24 January 2011 in Physical Review Letters
The thermal conductivity of disordered silicon-germanium alloys is computed from density-functional perturbation theory and with relaxation times that include both harmonic and anharmonic scattering terms. We show that this approach yields an excellent agreement at all compositions with experimental results and provides clear design rules for the engineering of nanostructured thermoelectrics. For SixGe1-x, more than 50% of the heat is carried at room temperature by phonons of mean free path greater than 1\,\hskip0.056em\hskip0.056em$\mu${}m, and an addition of as little as 12% Ge is sufficient to reduce the thermal conductivity to the minimum value achievable through alloying. Intriguingly, mass disorder is found to increase the anharmonic scattering of phonons through a modification of their vibration eigenmodes, resulting in an increase of 15% in thermal resistivity.