3D neural network models atomistic potentials in complex alloys.
problem Designing robust atomistic potentials for complex alloys is computationally expensive and time-consuming.
method Voxelized atomic configurations and 3D convolutional neural networks to learn atomic interactions.
result 3D convolutional neural networks effectively model atomistic potentials in complex alloys.
Improved atomistic model predicts molecular properties using weighted skip-connections.
problem Understanding the relative importance of interactions in molecular property prediction.
method Extended SchNet architecture with weighted skip-connections to analyze molecule properties.
result Relative weighting of interaction blocks depends on molecule's chemical composition and configurational degrees of freedom.
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.
Deep neural network generates molecular structures close to equilibrium.
problem Discovery of atomistic systems with desirable properties.
method Autoregressive, convolutional deep neural network architecture.
result Model generates molecules close to equilibrium for C7O2H10 isomers.
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.
ASLA learns atomic structures using neural networks and reinforcement learning.
problem Designing materials and drugs with desired properties.
method Atomistic structure learning algorithm (ASLA) using a convolutional neural network and reinforcement learning.
result ASLA can predict optimal structural arrangements of atoms for various target properties.
New framework embeds physics in coarse-grained models without big data.
problem Lack of big data and computational demand in data-driven coarse-graining.
method Proposes a novel objective based on reverse Kullback-Leibler divergence that incorporates physics in the form of force fields.
result Generative coarse-grained model predicts atomistic configurations and reveals physicochemical CVs.
SchNetPack 2.0 enhances atomistic machine learning with improved neural networks.
problem Improving atomistic machine learning methods and applications.
method Improved data pipeline, equivariant neural networks, PyTorch implementation, PyTorch Lightning, Hydra configuration framework.
result Easy extension and complex training tasks support.
Bayesian models discover CVs for complex systems, enhancing sampling methods.
problem Limitations in modeling complex systems in biochemistry and materials science.
method Formulated CV discovery as a Bayesian inference problem, using deep learning and variational inference.
result Discovered CVs improve predictive ability for alanine dipeptide and ALA-15 peptides.
Machine learning improves atomic property predictions by optimizing structure representations.
problem Improving accuracy of machine learning models for atomic-scale properties.
method Generalized SOAP kernel with distance-dependent weights and chemical species correlations.
result Optimized representation improves model performance and reveals chemical insights.
New method uses normalizing flows to improve force fields for coarse-grained molecular dynamics.
problem Lack of reference atomistic forces makes force matching infeasible for MLCG force fields.
method Introduces noise-based kernels adapted to low-data regimes using normalizing flows.
result Flow-based kernels reduce local distortions while preserving global accuracy.
Neural networks learn molecule and material representations.
problem Learning efficient representations for molecules and materials.
method Continuous-filter convolutional network SchNet.
result SchNet accurately predicts chemical properties across various datasets.
Study evaluates uncertainty quantification for atomistic neural networks, revealing complex relationships between error and uncertainty.
problem Uncertainty quantification for predictions of atomistic neural networks.
method Modified PhysNet NN architecture, evaluated with various metrics, analyzed QM9 and tautomerization reaction databases.
result Error and uncertainty are not linearly related; redundancy and noise complicate predictions, especially for small changes.
We present a hybrid continuum-atomistic scheme which combines molecular dynamics (MD) simulations with on-the-fly machine learning techniques for the accurate and efficient prediction of multiscale fluidic systems. By using a Gaussian process as a surrogate model for the computationally expensive MD simulations, we use…
A framework to compare atomistic descriptors and their transformations.
problem Comparing and understanding different atomistic descriptors and their transformations.
method Introducing a framework to compare different sets of descriptors and their transformations by metrics and kernels.
result Diagnostic tools to determine equivalent information and distorted common information between feature spaces.
Optimizes atomic descriptors to reduce redundancy and improve machine learning models.
problem Redundant descriptors in atomistic machine learning models increase computational burden and limit model expressivity.
method Employing techniques from pattern recognition, we refine and augment existing atomistic representations to produce optimal sets of descriptors.
result New architectures recognize up to 5-body patterns with low computational cost and high accuracy.
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.
New model predicts molecular wavefunctions and densities with unprecedented accuracy.
problem Challenging task of predicting wavefunctions due to molecular rotations.
method Introduces SE(3)-equivariant operations for deep learning.
result Achieves speedups and error reductions over ab initio methods.
Introduce a thermodynamically informed, temperature-transferable MLCG framework for proteins.
problem Temperature transferability of MLCG models for proteins.
method Explicit decomposition of CG potential into energetic and entropic components.
result Reproduces temperature-dependent quantities like heat capacity.
The CAPM's market returns are endogenously determined, affecting all assets' expected returns.
problem The standard CAPM's market return assumption is not endogenously consistent.
method Demonstrates the impact of endogenously determined market returns on asset returns and the range of feasible market returns.
result Expected returns are influenced by all assets' risks, and market returns are limited by asset distribution.
Unified machine learning predicts molecular wavefunctions efficiently.
problem Lack of explicit electronic structure in machine learning models for chemistry.
method Deep neural network for quantum mechanical wavefunction prediction.
result Efficient prediction of molecular wavefunctions with full electronic structure access.
We seek to deepen understanding of the micro-foundations of institutionalization while contributing to a sociological theory of markets by investigating the puzzle of price bubbles in financial markets. We find that such markets, despite textbook conditions of high efficiency -- perfect information, atomistic agents, n…
Generative models accelerate molecular dynamics by four orders of magnitude.
problem Femtosecond time steps limit access to slow molecular processes.
method Deep generative modeling framework that accelerates sampling.
result Quantitative characterization of equilibrium ensembles and dynamical relaxation processes.
Ancient grain boundaries resemble atoms in their formation and properties.
problem Understanding the formation and properties of ancient grain boundaries.
method Analyzing ancient grain boundaries as analogous to atoms and using geometric flow techniques.
result New examples of convex ancient and translating solutions to mean curvature flow.
Auto-encoders learn atomistic to coarse-grained mappings for molecular dynamics.
problem Simulating large systems in molecular dynamics is computationally expensive.
method Auto-encoders learn both atomistic to coarse-grained mappings and the coarse-grained potential energy function.
result Auto-encoders enable efficient simulation of larger systems in molecular dynamics.
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.
Improved neural network models predict molecular and material properties efficiently.
problem Training neural networks for accurate interatomic potentials is computationally expensive.
method Gaussian moment-based neural networks with improved architecture and active learning.
result The new models achieve high accuracy and reduced training times.
Study examines how twisting graphene nanoribbons affects their thermal conductivity.
problem Understanding how twisting affects thermal conductivity in graphene nanoribbons.
method Calculated geometric parameters of TGNRs, including twist and writhe, and used molecular dynamics simulations.
result Twisted graphene nanoribbons require at least two parameters to accurately describe their thermal conductivity.
We present atomistic molecular dynamics simulations of two Polyethylene systems where all entanglements are trapped: a perfect network, and a melt with grafted chain ends. We examine microscopically at what level topological constraints can be considered as a collective entanglement effect, as in tube model theories, o…
Differentiable simulations control molecular Hamiltonians for desired outcomes.
problem Control and learning of molecular Hamiltonians for desired outcomes.
method Differentiable simulations to differentiate Hamiltonians with respect to target observables.
result Control and learning of molecular Hamiltonians for desired outcomes.
Machine learning improves coarse-graining of molecular dynamics models.
problem Creating accurate coarse-grained models for molecular dynamics simulations.
method Reformulated coarse-graining as a supervised machine learning problem using statistical learning theory and deep learning (CGnets).
result CGnets can capture multi-body terms and all-atom explicit-solvent free energy surfaces with fewer coarse-grained beads.
ML4Chem offers a user-friendly platform for developing and deploying machine learning models in chemistry.
problem Developing and deploying machine learning models in chemistry and materials science.
method User-experience design, six core building blocks: data, featurization, models, model optimization, inference, and visualization.
result Ease of use and functionality of the atomistic module for neural networks and kernel ridge regression.
Improved CG force-field learning from all-atom data.
problem Training accurate coarse-grained models from all-atom simulations is challenging.
method Optimized force mapping to improve statistical efficiency of force-field learning.
result Substantially improved CG force-fields can be learned from the same simulation data.
CG-BGs combine flow-based models with PMFs to sample large systems efficiently.
problem Sampling equilibrium molecular configurations from the Boltzmann distribution is challenging.
method Coarse-grained Boltzmann Generators (CG-BGs) use flow-based models and learned PMFs for efficient sampling.
result CG-BGs provide a practical route for sampling larger molecular systems efficiently.
Bayesian regression underestimates parameter uncertainties in noisy models.
problem Parameter uncertainties are underestimated in Bayesian regression for imperfect models.
method Analyzed and designed an ansatz to correct for misspecification in near-deterministic surrogate models.
result Posterior distributions must cover all training points to avoid divergent generalization error.
Transfer learning boosts chemically accurate neural network potentials for organic molecules.
problem Developing accurate interatomic potentials from ab-initio data.
method Discriminative fine-tuning of pre-trained neural networks.
result Fine-tuning with energy labels alone can achieve accurate atomic forces.
Develops neural networks for reductive Lie groups, enhancing symmetry respect.
problem Symmetry respect in neural networks for reductive Lie groups.
method General equivariant neural network architecture for any reductive Lie Group G.
result Demonstrates generality and performance in top quark decay tagging and shape recognition.
Optimizes feature selection for molecular kinetics models.
problem Selecting interpretable features for kinetic models from molecular dynamics simulations.
method Direct optimization of features without constructing a full kinetic model.
result Direct feature optimization leads to more efficient model selection.
EGR refines and assesses protein complex structures.
problem Improving the accuracy of protein complex 3D structures for drug discovery.
method E(3)-equivariant graph neural network (GNN) for multi-task refinement and assessment.
result EGR achieves state-of-the-art performance in refining and assessing protein complexes.
Graph neural networks fail to distinguish certain 3D atom configurations.
problem Graph neural networks (GNN) fail to distinguish certain 3D atom configurations.
method Construction of degenerate 3D atom configurations that are indistinguishable by first-order GNNs.
result First-order GNNs are incomplete for 3D atom configurations.
New method uses neural networks to improve free energy estimation.
problem Estimating free energy differences using FEP is limited by insufficient overlap between distributions.
method Developed a neural network to parameterize a high-dimensional mapping in configuration space.
result Demonstrated substantial variance reduction in free energy estimates.
New method designs antimicrobial peptides with high potency and low toxicity.
problem Designing potent antimicrobial drugs with low toxicity.
method CLaSS method using deep generative autoencoder and atomistic simulations.
result Design and synthesis of two novel AMPs with high potency and low toxicity.
Time-lagged VAE reduces complex dynamics to a single embedding.
problem Interpreting high-dimensional time-series data for nonlinear systems.
method Variational dynamics encoder (VDE) using time-lagged variational autoencoders.
result Captures nontrivial dynamics in various examples, including protein folding.
Unconstrained models learn physical symmetries effectively with simple data augmentation.
problem Ensuring physical symmetries in machine learning models.
method Rigorous metrics to measure symmetry content, data augmentation strategy, architectural analysis.
result Unconstrained models can learn approximate equivariant behavior with simple data augmentation.
New CGMD model predicts non-equilibrium processes better than existing methods.
problem Inconsistency in conditional distribution of unresolved variables.
method Time-lagged independent component analysis to minimize entropy contribution of unresolved variables.
result The model's generalization ability for non-equilibrium processes is significantly improved.
Machine learning generates coarse-grained force fields for molecular dynamics.
problem Creating thermodynamically consistent coarse-grained models for larger systems.
method Hybrid architecture using graph neural networks to learn molecular features.
result Framework reproduces thermodynamics for small biomolecular systems.
Data-driven framework learns coarse-scale PDEs from fine-scale observations.
problem Deriving macroscopic PDEs from microscopic observations is challenging.
method Machine learning algorithms (Gaussian Processes, Artificial Neural Networks, Diffusion Maps) to uncover macroscopic fields and their evolution.
result Identifies multiple macroscopic PDEs approximating fine-scale microscopic models.
DiAMoNDBack models protein backmapping from coarse-grained Cα traces.
problem Restoring all-atom details from coarse-grained protein representations.
method Autoregressive denoising diffusion model for residue-by-residue backmapping.
result Achieves state-of-the-art reconstruction performance in diverse applications.