A unified family of vision transformers equivariant to arbitrary discrete subgroups of O(2), with embedding and expressivity theorems, a D6 construction using hexagonal patches, and experiments on aerial images in low-data regimes.
hub Mixed citations
Mailoa, Mordechai Kornbluth, Nicola Molinari, Tess E
Mixed citation behavior. Most common role is background (50%).
26
Pith papers citing it
1,470
external citations · Crossref
Background
50% of classified citations
hub tools
citation-role summary
background 4
baseline 2
citation-polarity summary
representative citing papers
DSpinGNN combines equivariant GNN structural dynamics with a Goodenough-Kanamori-biased MLP to predict exchange in dynamically strained CrI3, yielding domain wall width 1.7 nm and oscillation period 0.27 ps from 3200-atom simulations.
Stable size extrapolation in local score models requires the receptive field to cover the quasi-locality range of the Gaussian-smoothed score, formalized via a size-uniform comparison theorem and validated on the new FDLF benchmark.
LSD extends speculative sampling to second-order Langevin dynamics, achieving 3-9x speedup in MD while exactly sampling from the target distribution without relative error.
DeepPolaron ML-MD simulations show rutile electrons form Ti-localized polarons hopping along [001] with 39 meV barrier and 4.4e-2 cm2/Vs mobility, while anatase holes form O-localized polarons hopping to second neighbors with 139 meV barrier and 1.4e-3 cm2/Vs mobility.
On a C_n-symmetric task, wrong-group symmetry priors reduce performance versus no prior (CI excludes zero), augmentation matches equivariant models, and measured symmetry-data exchange rate is 1.28 (wide CIs include zero).
Proposes generative pseudo-force fields trained on quadratic pseudo-potentials from noisy equilibria as a time-step-agnostic diffusion variant for efficient molecular conformation generation with high validity on QM9.
Chem-GMNet uses sphere-native embeddings, DualSKA attention, and SH-FFN layers to match or beat ChemBERTa-2 on MoleculeNet tasks with fewer parameters and sometimes no pretraining.
PiGGO integrates a learned graph neural ODE as the continuous-time dynamics model within an extended Kalman filter to enable online virtual sensing and uncertainty-aware state estimation for nonlinear dynamic systems with unknown model form and sparse sensing.
Skala is a neural XC functional trained on wavefunction data that beats state-of-the-art hybrids on main-group chemistry benchmarks at semi-local computational cost.
MatMind is a unified LLM-based generative model for crystals that reports lowest MAE on energy above hull, bulk modulus and band gap while achieving 65.3% S.U.N. rate on unconditional generation.
Force-aware Neural Tangent Kernels combined with chunked acquisition provide scalable and distribution-robust active learning for MLIPs, outperforming baselines on OC20 and remaining competitive on other benchmarks.
Molecular dynamics simulations find that both I and MA defects in MAPbI3 diffuse rapidly at room temperature with barriers of 0.15-0.20 eV, with MA interstitials moving via concerted mechanisms and no MA vacancy migration observed.
DenSNet learns the Hohenberg-Kohn map to electron density with equivariant networks and delta-learning, then maps density to energy, producing stable MD trajectories whose infrared spectra match experiment and DFT on ethanol, ethanethiol, resorcinol, and polythiophene oligomers.
QCOF ML potentials tuned on COF data outperform general MACE models for defective systems and reveal higher thermal defect sensitivity in CTF-1 versus COF-LZU1 with nearly invariant low-strain mechanics.
A symmetry-equivariant generative model trained on thermal snapshots of a 1D O(3) spin glass recovers the microscopic Hamiltonian parameters with 99.7% cosine similarity to ground truth through linear inversion of its learned score field.
Bayesian E(3)-equivariant MLPs with joint energy-force NLL loss achieve competitive accuracy while enabling uncertainty-guided active learning, OOD detection, and calibration.
Fine-tuning the MACE-MPA-0 foundation model on 5-10 60-atom DFT configurations reproduces the barocaloric phase transformation in ammonium sulfate, while training from scratch fails at these sizes.
Synthetic pre-training on ML-generated tensor data followed by fine-tuning on ground-truth calculations improves data efficiency for graph models of solid-state NMR parameters when the pre-training and fine-tuning domains match.
Loss-guided adaptive scale refinement on NaCl aqueous system reduces overall force MAE from 399.65 to 381.23 by discovering intermediate scales from initial anchors.
DeltaDiff is a physics-guided inference method that predicts mutant protein structures from a baseline diffusion model without retraining, tested on three systems with nonlocal changes.
Spatial statistics on voxelized structures using FFT correlations and PCA yield low-dimensional convex features that support accurate predictions with as few as 10 training samples.
GROMACS now runs multi-GPU DeePMD inference for molecular dynamics, reaching 40-66% strong scaling efficiency up to 32 devices on a 15k-atom protein system with over 90% time in inference.
mlip v2 is a new software release that integrates API redesign, e3j backend, eSEN model, improved charge modeling, and expanded simulation capabilities to support larger-scale molecular modeling.
citing papers explorer
-
A Unified Framework for Vision Transformers Equivariant to Discrete Subgroups of $\mathrm{O}(2)$
A unified family of vision transformers equivariant to arbitrary discrete subgroups of O(2), with embedding and expressivity theorems, a D6 construction using hexagonal patches, and experiments on aerial images in low-data regimes.
-
DSpinGNN: A Physics-Informed Equivariant Graph Neural Network for Dynamic Magnetic Exchange Prediction in Strain-Deformed Monolayer CrI$_3$
DSpinGNN combines equivariant GNN structural dynamics with a Goodenough-Kanamori-biased MLP to predict exchange in dynamically strained CrI3, yielding domain wall width 1.7 nm and oscillation period 0.27 ps from 3200-atom simulations.
-
When Do Local Score Models Extrapolate Across Size? A Diagnostic Theory and Benchmark
Stable size extrapolation in local score models requires the receptive field to cover the quasi-locality range of the Gaussian-smoothed score, formalized via a size-uniform comparison theorem and validated on the new FDLF benchmark.
-
Speculative Sampling For Faster Molecular Dynamics
LSD extends speculative sampling to second-order Langevin dynamics, achieving 3-9x speedup in MD while exactly sampling from the target distribution without relative error.
-
Polaron Transport in TiO$_{2}$ from Machine Learning Molecular Dynamics
DeepPolaron ML-MD simulations show rutile electrons form Ti-localized polarons hopping along [001] with 39 meV barrier and 4.4e-2 cm2/Vs mobility, while anatase holes form O-localized polarons hopping to second neighbors with 139 meV barrier and 1.4e-3 cm2/Vs mobility.
-
Measuring the Symmetry--Data Exchange Rate
On a C_n-symmetric task, wrong-group symmetry priors reduce performance versus no prior (CI excludes zero), augmentation matches equivariant models, and measured symmetry-data exchange rate is 1.28 (wide CIs include zero).
-
Generative Pseudo-Force Fields for Molecular Generation
Proposes generative pseudo-force fields trained on quadratic pseudo-potentials from noisy equilibria as a time-step-agnostic diffusion variant for efficient molecular conformation generation with high validity on QM9.
-
Chem-GMNet: A Sphere-Native Geometric Transformer for Molecular Property Prediction
Chem-GMNet uses sphere-native embeddings, DualSKA attention, and SH-FFN layers to match or beat ChemBERTa-2 on MoleculeNet tasks with fewer parameters and sometimes no pretraining.
-
PiGGO: Physics-Guided Learnable Graph Kalman Filters for Virtual Sensing of Nonlinear Dynamic Structures under Uncertainty
PiGGO integrates a learned graph neural ODE as the continuous-time dynamics model within an extended Kalman filter to enable online virtual sensing and uncertainty-aware state estimation for nonlinear dynamic systems with unknown model form and sparse sensing.
-
Accurate and scalable exchange-correlation with deep learning
Skala is a neural XC functional trained on wavefunction data that beats state-of-the-art hybrids on main-group chemistry benchmarks at semi-local computational cost.
-
MatMind: A Structure-Activity Knowledge-Driven Generative Foundation Model for Materials Science
MatMind is a unified LLM-based generative model for crystals that reports lowest MAE on energy above hull, bulk modulus and band gap while achieving 65.3% S.U.N. rate on unconditional generation.
-
Force-Aware Neural Tangent Kernels for Scalable and Robust Active Learning of MLIPs
Force-aware Neural Tangent Kernels combined with chunked acquisition provide scalable and distribution-robust active learning for MLIPs, outperforming baselines on OC20 and remaining competitive on other benchmarks.
-
A Unified microscopic picture of cation and anion migration in MAPbI$_3$
Molecular dynamics simulations find that both I and MA defects in MAPbI3 diffuse rapidly at room temperature with barriers of 0.15-0.20 eV, with MA interstitials moving via concerted mechanisms and no MA vacancy migration observed.
-
Enhancing molecular dynamics with equivariant machine-learned densities
DenSNet learns the Hohenberg-Kohn map to electron density with equivariant networks and delta-learning, then maps density to energy, producing stable MD trajectories whose infrared spectra match experiment and DFT on ethanol, ethanethiol, resorcinol, and polythiophene oligomers.
-
Data-Driven Thermal and Mechanical Modeling of Defective Covalent Organic Frameworks
QCOF ML potentials tuned on COF data outperform general MACE models for defective systems and reveal higher thermal defect sensitivity in CTF-1 versus COF-LZU1 with nearly invariant low-strain mechanics.
-
Autonomous Emergence of Hamiltonian in Deep Generative Models
A symmetry-equivariant generative model trained on thermal snapshots of a 1D O(3) spin glass recovers the microscopic Hamiltonian parameters with 99.7% cosine similarity to ground truth through linear inversion of its learned score field.
-
Bayesian E(3)-Equivariant Interatomic Potential with Iterative Restratification of Many-body Message Passing
Bayesian E(3)-equivariant MLPs with joint energy-force NLL loss achieve competitive accuracy while enabling uncertainty-guided active learning, OOD detection, and calibration.
-
Barocaloric phase transformation from data efficient fine-tuning of machine learned interatomic potentials
Fine-tuning the MACE-MPA-0 foundation model on 5-10 60-atom DFT configurations reproduces the barocaloric phase transformation in ammonium sulfate, while training from scratch fails at these sizes.
-
Synthetic pre-training of graph-network models for predicting solid-state NMR parameters
Synthetic pre-training on ML-generated tensor data followed by fine-tuning on ground-truth calculations improves data efficiency for graph models of solid-state NMR parameters when the pre-training and fine-tuning domains match.
-
Loss-Guided Adaptive Scale Refinement for Molecular Force Prediction
Loss-guided adaptive scale refinement on NaCl aqueous system reduces overall force MAE from 399.65 to 381.23 by discovering intermediate scales from initial anchors.
-
DeltaDiff: Training-Free, Physics-Guided Machine Learning for Predicting Mutant Protein Structures
DeltaDiff is a physics-guided inference method that predicts mutant protein structures from a baseline diffusion model without retraining, tested on three systems with nonlocal changes.
-
Spatial statistics for screening molecular structures
Spatial statistics on voxelized structures using FFT correlations and PCA yield low-dimensional convex features that support accurate predictions with as few as 10 training samples.
-
Making Room for AI: Multi-GPU Molecular Dynamics with Deep Potentials in GROMACS
GROMACS now runs multi-GPU DeePMD inference for molecular dynamics, reaching 40-66% strong scaling efficiency up to 32 devices on a 15k-atom protein system with over 90% time in inference.
-
Machine Learning Interatomic Potentials: Advancing Open-Source Software for Efficient and Scalable Molecular Simulation
mlip v2 is a new software release that integrates API redesign, e3j backend, eSEN model, improved charge modeling, and expanded simulation capabilities to support larger-scale molecular modeling.
-
SMC-AI: Scaling Monte Carlo Simulation to Four Trillion Atoms with AI Accelerators
SMC-AI scales Monte Carlo simulations to 4 trillion atoms on AI hardware clusters, achieving 32 times larger systems and 1.3 times higher throughput than prior records while decoupling ML models from the simulation core.
-
Six Open Questions in Machine-Learned Interatomic Potential Foundation Models
This perspective article develops a definition of foundational MLIPs and poses six open questions that the authors believe will define future research in machine-learned interatomic potentials.