Machine learning predicts atomization energies accurately from low-fidelity calculations.
problem Predicting accurate atomization energies of organic molecules efficiently.
method Machine learning models trained on low-fidelity B3LYP energies to predict high-fidelity G4MP2 energies.
result Predicted G4MP2 atomization energies within 0.012 eV for molecules with 10-14 heavy atoms.
We introduce a machine learning model to predict atomization energies of a diverse set of organic molecules, based on nuclear charges and atomic positions only. The problem of solving the molecular Schrödinger equation is mapped onto a non-linear statistical regression problem of reduced complexity. Regression models a…
GRAPE uses graph kernels to predict molecular energies efficiently.
problem Efficiently predicting molecular energies with physical constraints.
method GRAPE approach based on graph theory incorporating symmetries.
result GRAPE predicts atomization energies accurately on organic molecules.
Graph neural network predicts protonation energies of oxygen atoms in bio-oil molecules.
problem Predicting protonation energies of oxygen atoms in bio-oil molecules for chemical upgrading.
method Site-specific graph neural network approach using iterative local nonlinear embedding.
result Effective prediction of protonation energies of individual oxygen atoms in bio-oil molecules.
EBM predicts protein conformations at atomic scale using crystallized data.
problem Predicting the conformation of a side chain from its context within a protein structure.
method Energy-based model trained on crystallized protein data, evaluating performance on rotamer recovery task.
result EBM achieves performance close to state-of-the-art methods, including Rosetta energy function.
Gaussian process regression cuts energy evaluations for atomic rearrangement paths.
problem Reducing computational effort for minimum energy paths in complex systems.
method Gaussian process regression to approximate energy surfaces and converge to minimum energy paths.
result Significant reduction in energy evaluations (less than a fifth for a test problem).
Study compares atom representations in graph neural networks for molecular properties.
problem Incorrect attribution of results in molecular property prediction due to varying atom features.
method Evaluated multiple atom representations on free energy, solubility, and metabolic stability predictions.
result Different atom representations can lead to varying predictive performance in graph neural networks.
LC-GAP uses localized Coulomb descriptors for accurate molecular potential predictions.
problem Creating accurate molecular potentials for large molecules.
method Combining localized Coulomb matrix representations with Gaussian approximation potential.
result LC-GAP generates accurate potentials for molecules larger than training data with chemical accuracy.
Normalizing flows model atomic solids without needing ground-truth samples.
problem Modeling atomic solids without ground-truth samples.
method Normalizing flows to transform a base distribution into the target solid.
result Excellent agreement between model estimates and literature values for Helmholtz free energy.
Model predicts DFT formation energies without atomic positions.
problem Fast prediction of material properties without atomic positions.
method Symmetry-labeled graphs and message passing neural network.
result Mean absolute error below 0.1 eV for selenides.
The study shows that certain graphs are regular at boundary points.
problem Boundary regularity of anisotropic minimal Lipschitz graphs.
method Proves regularity for graphs with bounded anisotropic mean curvature and atomic energy condition.
result Regularity at boundary points with density bounded above by 1/2 + σ.
A new method predicts electron density accurately from atom-centered models.
problem Predicting electron density accurately from atom-centered models.
method Gradient-based approach to minimize loss function in an optimized sparse feature space.
result Extremely accurate predictions of electron density and total energies.
Proposes linking energy and force uncertainty in deep learning potentials.
problem Uncertainty in predicted energies and forces in machine learning models.
method Introduces a spatially correlated noise process to link energy and force uncertainty.
result Demonstrates the approach on molecular datasets, linking energy and force uncertainties.
Predicts atomization energy with high accuracy using graph kernels and active learning.
problem Predicting molecular atomization energy with high accuracy.
method Gaussian process regression with marginalized graph kernel, active learning.
result Achieves mean absolute error of 0.62 +- 0.01 kcal/mol with 2000 training samples.
New algorithm efficiently trains machine learning models to atomic forces data.
problem Efficiently training machine learning models to large amounts of force data.
method Developed an efficient algorithm for training machine learning models to all available force data.
result Training to all available force data is only a few times more expensive than training to energies alone.
Novel ML model predicts solvation free energies from atom interactions.
problem Predicting solvation free energies from atomistic interactions.
method Two encoding functions extract atomic feature vectors, interactions calculated by inner product.
result Outstanding performance and transferability on 6,493 experimental measurements.
New model predicts protein-ligand binding affinity from atomic coordinates.
problem Predicting protein-ligand binding affinity using empirical scoring functions.
method Developed atomic convolutional neural network to learn chemical interactions directly from atomic coordinates.
result Atomic convolutional networks outperform or compete with cheminformatics methods in predicting binding free energy.
ShotgunCSP predicts crystal structures using machine learning, achieving high accuracy with minimal computation.
problem Predicting stable or metastable crystal structures of large systems.
method Noniterative screening using transfer learning and generative models.
result ShotgunCSP achieves 93.3% accuracy in benchmark tests with 90 different crystal structures.
New method uses geometric moments for accurate machine learning potentials.
problem Creating high-dimensional potential energy surfaces efficiently.
method Feed-forward neural networks with invariant local molecular descriptors based on geometric moments.
result Accuracy comparable to established models, high efficiency.
Graphs with bounded anisotropic mean curvature are regular almost everywhere.
problem Understanding the regularity of graphs with anisotropic mean curvature.
method Proving regularity for m-dimensional Lipschitz graphs with anisotropic mean curvature bounded in Lp. result Graphs with bounded anisotropic mean curvature are regular almost everywhere.
New method trains deep neural networks for non-interacting kinetic-energy functionals in DFT.
problem Lack of exact relationship between electron density and non-interacting kinetic energy.
method Variational principle to regularize machine-learned density functionals.
result Excellent results on kinetic-energy functionals for various systems.
A2I Transformer predicts atom energies from coordinates, avoiding heavy featurization.
problem Efficiently predicting atom energies from molecular coordinates with minimal featurization.
method End-to-end model using self-attention, permutation-equivariant.
result Stable predictions with significantly smaller errors than molecular dynamics simulations.
Improved molecular property prediction using updated neural message passing.
problem Predicting properties of molecules and materials accurately.
method Extended neural message passing model with edge update network.
result Superior prediction of formation energies and other properties on multiple datasets.
Hydrogen atom confined in an inverted-Gaussian potential, with detailed numerical methods and results.
problem Studying hydrogen atom in a specific potential.
method Three numerical methods: Lagrange-mesh, fourth order finite differences, and finite element method.
result Accurate numerical results for hydrogen atom energies and eigenfunctions, improving previous literature.
Large GNNs trained with Graph Parallelism improve atomic simulation accuracy.
problem Training memory-intensive GNNs for high-order atomic interactions.
method Graph Parallelism to distribute graphs across GPUs.
result Graph-parallelized models achieve state-of-the-art results on catalyst datasets.
New bin-wise scaling methods improve prediction uncertainty calibration for machine learning.
problem Improving prediction uncertainty calibration for machine learning regression.
method Adaptations of Binwise Variance Scaling (BVS) with alternative loss functions and feature-based binning.
result Improved adaptivity and consistency in prediction uncertainty calibration.
Gaussian process regression speeds up nudged elastic band calculations for transitions.
problem Reducing computational effort for calculating minimum energy paths in thermalized systems.
method Approximate energy surface generation and refinement using Gaussian process regression.
result The number of energy and force evaluations can be reduced by an order of magnitude.
Autoencoders discover and accelerate molecular dynamics simulations.
problem Efficient sampling of macromolecular folding landscapes with high free energy barriers.
method Employing auto-associative artificial neural networks to learn nonlinear collective variables (CVs) that are explicit and differentiable functions of atomic coordinates.
result Substantial speedups in exploration of configurational space and discovery of data-driven CVs.
Deep neural network predicts molecular wave functions in minimal basis.
problem Improving accuracy and efficiency in quantum chemistry calculations.
method Adapted SchNet for Orbitals (SchNOrb) model in quasi-atomic minimal basis.
result Model accurately predicts molecular orbital energies and wavefunctions for large molecules.
Wavelet scattering predicts material properties beyond training data.
problem Predict material properties beyond training data.
method Atomic orbital wavelet scattering transform.
result Extrapolation of material properties achieved.
GDML learns effective CG models from all-atom data.
problem Learning effective coarse-grained force fields efficiently.
method Ensemble learning with stratified sampling and GDML.
result GDML yields smaller free energy error than neural networks.
New neural networks explain quantum chemistry predictions atomically.
problem Need for interpretable quantum chemical models.
method Atomistic neural networks and aggregation of atom-wise contributions.
result Atom-wise explanations reveal chemical insights.
Cormorant learns molecular properties via rotationally covariant neural networks.
problem Learning molecular potential energy surfaces and properties.
method Rotationally covariant neural network architecture with tensor products and Clebsch-Gordan decomposition.
result Significantly outperforms competing algorithms in learning molecular Potential Energy Surfaces.
Spherical CNNs tackle 3D data analysis, especially spherical images.
problem Learning problems involving spherical images, like omnidirectional vision and molecular regression.
method Defined spherical cross-correlation, developed a generalized FFT for efficient computation.
result Demonstrated spherical CNNs' effectiveness in 3D model recognition and atomization energy regression.
We introduce multiscale invariant dictionaries to estimate quantum chemical energies of organic molecules, from training databases. Molecular energies are invariant to isometric atomic displacements, and are Lipschitz continuous to molecular deformations. Similarly to density functional theory (DFT), the molecule is re…
Machine learning predicts electronic density of states for condensed matter.
problem Predicting the electronic density of states (DOS) in complex condensed matter systems.
method Developed a machine learning framework to predict DOS from density functional theory data, considering geometric configurations of atoms.
result Demonstrated the model's effectiveness in predicting DOS and its components for various silicon configurations.
TACE unifies scalar and tensorial modeling in Cartesian space for accurate, stable, and efficient atomistic predictions.
problem Complexity and challenges in equivariant atomistic machine learning models.
method Tensor Atomic Cluster Expansion (TACE) in Cartesian space, decomposing local environments into irreducible Cartesian tensors (ICT).
result Universal invariant and equivariant embeddings, enabling explicit control at inference.
A multi-scale model predicts atomic-scale properties using both local and long-range information.
problem Inability of machine-learning schemes to capture long-range physical effects.
method Combines local and non-local information in a multipole expansion framework.
result Demonstrates the ability to model electrostatics, polarization, and dispersion.
This article addresses the issue of representing electroencephalographic (EEG) signals in an efficient way. While classical approaches use a fixed Gabor dictionary to analyze EEG signals, this article proposes a data-driven method to obtain an adapted dictionary. To reach an efficient dictionary learning, appropriate s…
Study calculates curvature functionals in space forms, proving sphere stability.
problem Understanding curvature functionals in space forms.
method Computed first and second variations of a general functional, applying stability criteria.
result Proved spheres are stable in a generalized Willmore functional.
3D scattering model predicts lithium-silicon formation energies.
problem Predicting formation energies of amorphous Li-Si materials.
method Steerable wavelet scattering for 3D signals, invariant to translations and rotations.
result State-of-the-art results compared to other machine learning methods.
We consider few-body bound state systems and provide precise definitions of Borromean and Brunnian systems. The initial concepts are more than a hundred years old and originated in mathematical knot-theory as purely geometric considerations. About thirty years ago they were generalized and applied to the binding of sys…
Unified theory linking atom-centered and message-passing models for molecular properties.
problem Combining atom-centered and message-passing models for accurate molecular property prediction.
method Generalizing ACDC framework to include multi-centered information, providing a complete linear basis for regression.
result Unified understanding of atom-centered and message-passing models, providing a coherent foundation.
Proposes atomic swaptions for trustless cryptocurrency derivatives.
problem Lack of trustless derivatives for cryptocurrency exchanges.
method Extends atomic swap protocol to include derivatives without oracles.
result Atomic swaptions enable trustless exchange of derivative assets.
wACSFs improve machine learning potentials for molecular structures.
problem Improving accuracy and efficiency of machine learning potentials for molecular structures.
method Introducing wACSFs based on ACSFs, optimizing parameters with a genetic algorithm.
result wACSFs lead to better generalization performance and require fewer descriptors.
DenSNet learns electron densities for molecular dynamics, enabling accurate spectroscopic predictions.
problem Lack of accurate electronic observables in MLIPs for molecular dynamics.
method DenSNet uses SE(3)-equivariant neural networks to predict electron densities and total energy.
result DenSNet predicts infrared spectra with excellent agreement to experimental data.
TeaNet uses GCNs to model complex atomic interactions inspired by electronic relaxation.
problem Creating a universal interatomic potential for all elements.
method Tensor-embedded atom network (TeaNet) using graph convolutional neural networks (GCNs).
result TeaNet achieves good performance (19 meV/atom) for structures and reactions involving elements from H to Ar.
Researchers use manifold learning to analyze 4D-STEM data of graphene, revealing atomic structure details.
problem Challenges in processing and interpreting large 4D-STEM datasets, especially for light materials.
method Data-driven manifold learning approaches for visualization and exploration of 4D-STEM datasets.
result Extracted patterns relate to individual atom sites and sublattice structures, effectively discriminating single dopant anomalies.