# Molecular Dynamics

A computational simulation method for modeling molecular systems, including ab initio molecular-dynamics simulations implemented in ORCA.

This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles.

## Stein Kernelized Molecular Dynamics for Active Learning of Interatomic Potentials

DevFeed: [Stein Kernelized Molecular Dynamics for Active Learning of Interatomic Potentials](<https://devfeed.tech/articles/stein-kernelized-molecular-dynamics-for-active-learning-of-interatomic-potentials-141575.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2606.04100>)

Author: Joanna Zou, Fraser Birks, Dallas Foster, Youssef Marzouk

Published: 2026-11-18T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Molecular Dynamics](<https://devfeed.tech/topics/molecular-dynamics.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Dataset development](<https://devfeed.tech/topics/dataset-development.md>), [Fine-tuning](<https://devfeed.tech/topics/fine-tuning.md>), [Neural Network](<https://devfeed.tech/topics/neural-network.md>)

Tags: [acquisition](<https://devfeed.tech/tags/acquisition.md>), [adaptive](<https://devfeed.tech/tags/adaptive.md>), [asynchronous](<https://devfeed.tech/tags/asynchronous.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [dft](<https://devfeed.tech/tags/dft.md>), [diversity](<https://devfeed.tech/tags/diversity.md>), [energy](<https://devfeed.tech/tags/energy.md>), [exploration](<https://devfeed.tech/tags/exploration.md>), [fewer](<https://devfeed.tech/tags/fewer.md>), [fine-tuning](<https://devfeed.tech/tags/fine-tuning.md>), [foundation](<https://devfeed.tech/tags/foundation.md>), [foundation-model](<https://devfeed.tech/tags/foundation-model.md>), [gradient](<https://devfeed.tech/tags/gradient.md>), [gradient-descent](<https://devfeed.tech/tags/gradient-descent.md>), [kernel](<https://devfeed.tech/tags/kernel.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [molecular-dynamics](<https://devfeed.tech/tags/molecular-dynamics.md>), [online](<https://devfeed.tech/tags/online.md>), [oracle](<https://devfeed.tech/tags/oracle.md>), [physics-comp-ph](<https://devfeed.tech/tags/physics-comp-ph.md>), [training](<https://devfeed.tech/tags/training.md>), [training-data](<https://devfeed.tech/tags/training-data.md>)

### AI overview

Stein kernelized molecular dynamics (SKMD) uses interacting-particle dynamics to select informative configurations for fine-tuning machine-learning interatomic potentials. It adapts stochastic Stein variational gradient descent to molecular dynamics with asynchronous updates and a kernel of global atomic descriptors for symmetry-aware configurational similarity. In the appropriate asymptotic limit, its dynamics retain the Boltzmann distribution as a stationary distribution, balancing exploration with attraction to high-probability regions. An adaptive stopping criterion selects non-redundant data online. The abstract reports demonstrations on the Müller-Brown potential, alanine dipeptide with MACE, and the Pandey reconstruction of Si(111) using a MACE foundation model and DFT oracle; compared with active-learning baselines, SKMD achieved higher model accuracy in fewer training iterations.

### Source excerpt

arXiv:2606.04100v2 Announce Type: replace Abstract: Machine learning interatomic potentials (MLIPs) enable efficient and accurate atomistic simulations but depend critically on the quality and diversity of the training data. We introduce Stein kernelized molecular dynamics (SKMD), an enhanced sampling method that uses interacting particle dynamics to acquire informative training configurations for the fine-tuning of MLIPs. SKMD corresponds to a stochastic variant of Stein variational gradient descent that is adapted for molecular dynamics by incorporating asynchronous particle updates and a kernel of global atomic descriptors, which provides a symmetry-aware measure of configurational similarity. Unlike other enhanced samplers used in molecular dynamics, SKMD maintains the Boltzmann distribution as a stationary distribution of its dynamics in the appropriate asymptotic limit. This property enforces a balance between the exploration of diverse configurations and attraction toward high-probability regions of the energy landscape. We further propose an approach to efficient online data acquisition using an adaptive stopping criterion that selects non-redundant training data over the course of simulation. We demonstrate SKMD for the active learning of a neural network model of the M\"uller--Brown potential, a MACE interatomic potential for alanine dipeptide, and a MACE foundation model with a DFT oracle for predicting the Pandey reconstruction of the Si(111) surface. Compared to active learning baselines, our method achieves higher model accuracy in fewer training iterations.

## Generation of protein conformations using a simple inverted transformer

DevFeed: [Generation of protein conformations using a simple inverted transformer](<https://devfeed.tech/articles/generation-of-protein-conformations-using-a-simple-inverted-transformer-137431.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.19321>)

Author: Luigi Leonardo Palese

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [AI Research](<https://devfeed.tech/topics/ai-research.md>), [Molecular Dynamics](<https://devfeed.tech/topics/molecular-dynamics.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [Neural Network](<https://devfeed.tech/topics/neural-network.md>), [Transformer](<https://devfeed.tech/topics/transformer.md>)

Tags: [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [deep-learning](<https://devfeed.tech/tags/deep-learning.md>), [generation](<https://devfeed.tech/tags/generation.md>), [molecular-dynamics](<https://devfeed.tech/tags/molecular-dynamics.md>), [q-bio-bm](<https://devfeed.tech/tags/q-bio-bm.md>), [q-bio-qm](<https://devfeed.tech/tags/q-bio-qm.md>), [transformer](<https://devfeed.tech/tags/transformer.md>)

### AI overview

A simple Transformer variant trained on classical all-atom molecular dynamics conformations of one protein directly generates physically plausible structures resembling both simulation and experimental conformations. Its output sequence quickly converges on the most populated regions of that protein's conformational landscape, suggesting the model may capture the relative statistical weights of feasible conformations. The authors report low computational load and propose the approach as a candidate for generative molecular dynamics; the supplied abstract does not establish its performance across other proteins.

### Source excerpt

arXiv:2611.19321v1 Announce Type: cross Abstract: Molecular dynamics is an extremely powerful approach for studying protein behavior, but it requires enormous computational resources to obtain an adequate sampling of possible conformations. Recently, thanks to successes in predicting three-dimensional protein structures, there has been an explosion of research exploring the opportunities offered by deep learning to address the problem of protein conformational landscape. In this work, an artificial intelligence system based on a variant of the Transformer architecture is introduced and applied to the problem of determining the ensemble of three-dimensional structures that a protein can access and their statistical weight. After being trained on a dataset of conformations of a particular protein obtained using classical all-atom molecular dynamics, this neural network is able to directly generate physically plausible protein conformations similar to those obtained both through molecular dynamics and experimentally. The ordered sequence of conformations obtained as output of the neural network rapidly converges towards the most populated regions of the protein's conformational landscape, suggesting that the system may have learned not only the protein's three-dimensional organization but also the thermodynamic weights of the various feasible conformations. These features, combined with its simplicity and low computational load, make this Transformer variant an ideal candidate for the development of systems capable of performing generative molecular dynamics.

## Mesoscale Neural Operators: Learning Nonlinear Constitutive Laws from Molecular Dynamics for Scalable Fracture Predictions

DevFeed: [Mesoscale Neural Operators: Learning Nonlinear Constitutive Laws from Molecular Dynamics for Scalable Fracture Predictions](<https://devfeed.tech/articles/mesoscale-neural-operators-learning-nonlinear-constitutive-laws-from-molecular-dynamics-for-scalable-fracture-predictions-138906.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.19658>)

Author: Yue Yu, Jihong Wang, Stewart Silling, Ziwei Yang

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Molecular Dynamics](<https://devfeed.tech/topics/molecular-dynamics.md>)

Tags: [cond-mat-mtrl-sci](<https://devfeed.tech/tags/cond-mat-mtrl-sci.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [data-driven](<https://devfeed.tech/tags/data-driven.md>), [horizon](<https://devfeed.tech/tags/horizon.md>), [kernel](<https://devfeed.tech/tags/kernel.md>), [laws](<https://devfeed.tech/tags/laws.md>), [learning](<https://devfeed.tech/tags/learning.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [molecular-dynamics](<https://devfeed.tech/tags/molecular-dynamics.md>), [neural](<https://devfeed.tech/tags/neural.md>), [operators](<https://devfeed.tech/tags/operators.md>), [physics](<https://devfeed.tech/tags/physics.md>), [predictions](<https://devfeed.tech/tags/predictions.md>), [simulations](<https://devfeed.tech/tags/simulations.md>), [synthetic-data](<https://devfeed.tech/tags/synthetic-data.md>), [training](<https://devfeed.tech/tags/training.md>), [workflow](<https://devfeed.tech/tags/workflow.md>)

### AI overview

The mesoscale neural operator (MNO) learns a nonlocal fracture model from molecular-dynamics loading and displacement data, then applies it in continuum simulations on larger domains. To preserve physical consistency across scales, it uses a normalized kernel to keep strain energy density invariant as the horizon changes and rescales critical bond stretch to maintain energy release rate. Synthetic-data tests show the workflow recovers the nonlocal kernel and nonlinear constitutive relation. On coarse-grained MD data, the learned model matches strain energy density and critical energy release rate within 3%, and predicts peak transmitted forces within 6% in out-of-distribution tension tests. In a plate with an initial defect and 4,000 times the training specimens' area, it reproduces reported features of dynamic brittle fracture, including branching and cascading branches.

### Source excerpt

arXiv:2611.19658v1 Announce Type: new Abstract: Molecular dynamics (MD) simulations can capture the nanoscale physics of fracture at the atomistic level, but their computational cost prohibits predictions at the length scales and time scales needed for many applications. Machine learning approaches, on the other hand, have been proposed as heuristic surrogates for material fracture, yet they cannot guarantee the physical consistency of the predicted response. To address this shortcoming, we propose the mesoscale neural operator (MNO): a data-driven nonlocal constitutive model that is learned directly from MD loading and displacement data. After training, it can be used in continuum fracture simulations on much larger domains and grid sizes. The key idea of the MNO is to exploit the built-in characteristic length of a nonlocal model, called the horizon, to travel across scales. Physical consistency across scales is obtained by building two structural ingredients into the MNO: a normalized kernel that renders the strain energy density invariant under horizon rescaling, and a rescaled critical bond stretch that keeps the energy release rate consistent across scales. Through tests on synthetic data, we show that the MNO and its learning workflow indeed discover the nonlocal continuum multiscale model in the form of the nonlocal kernel and the nonlinear constitutive relation. We then apply the framework to coarse-grained MD data and find that in a test problem, the learned model matches the MD strain energy density and critical energy release rate within 3%. It also predicts the peak transmitted forces within 6% in out-of-distribution tension tests. As a further demonstration, the model is applied to a plate with 4000 times the area of the training specimens and containing an initial defect. In this problem, the model reproduces the main features of dynamic brittle fracture, including branch and cascading branch networks.

## Accelerate Molecular Dynamics Simulations with Accurate, Flexible Primitives Using NVIDIA cuPotential

DevFeed: [Accelerate Molecular Dynamics Simulations with Accurate, Flexible Primitives Using NVIDIA cuPotential](<https://devfeed.tech/articles/accelerate-molecular-dynamics-simulations-with-accurate-flexible-primitives-using-nvidia-cupotential-136676.md>)

Original publisher: [Read original article](<https://developer.nvidia.com/blog/accelerate-molecular-dynamics-simulations-with-accurate-flexible-primitives-using-nvidia-cupotential/>)

Author: Yutong Zhao

Published: 2026-11-16T22:30:00Z

Content type: tutorial

Language: en

Sources: [NVIDIA Developer](<https://devfeed.tech/sources/nvidia-developer.md>), [NVIDIA Technical Blog](<https://devfeed.tech/sources/nvidia-technical-blog.md>)

Topics: [Molecular Dynamics](<https://devfeed.tech/topics/molecular-dynamics.md>), [Computational Chemistry / Materials Science](<https://devfeed.tech/topics/computational-chemistry-materials-science.md>), [GPU](<https://devfeed.tech/topics/gpu.md>), [RTX PRO](<https://devfeed.tech/topics/rtx-pro.md>)

Tags: [accelerate](<https://devfeed.tech/tags/accelerate.md>), [api](<https://devfeed.tech/tags/api.md>), [biology](<https://devfeed.tech/tags/biology.md>), [chemistry](<https://devfeed.tech/tags/chemistry.md>), [computational-chemistry-materials-science](<https://devfeed.tech/tags/computational-chemistry-materials-science.md>), [cuda-x](<https://devfeed.tech/tags/cuda-x.md>), [developer-tools-techniques](<https://devfeed.tech/tags/developer-tools-techniques.md>), [materials-science](<https://devfeed.tech/tags/materials-science.md>), [molecular-dynamics](<https://devfeed.tech/tags/molecular-dynamics.md>), [nvidia](<https://devfeed.tech/tags/nvidia.md>), [physics](<https://devfeed.tech/tags/physics.md>), [sc26](<https://devfeed.tech/tags/sc26.md>), [science](<https://devfeed.tech/tags/science.md>), [simulation-modeling-design](<https://devfeed.tech/tags/simulation-modeling-design.md>), [simulations](<https://devfeed.tech/tags/simulations.md>)

### AI overview

NVIDIA cuPotential provides modular GPU primitives for molecular dynamics, targeting long-range electrostatics and dispersion bottlenecks. Its OpenMM plugin replaces the nonbonded force and can also supply a Langevin integrator and Monte Carlo barostat; the excerpt gives Linux build steps using Conda or Micromamba, CUDA 13.1, and OpenMM 8.4.0. Reported NPT benchmarks show over 3x end-to-end speedups on selected RTX PRO 6000 Blackwell systems and gains often around 2x or more on other tested GPUs, for standard non-alchemical systems. The text also reports reduced force error and energy drift under specified comparisons. The plugin targets solvated, roughly uniform-density systems in rectangular periodic boxes; it does not support LJPME, triclinic boxes, or particle and exception parameter offsets. These results and instructions are from the supplied excerpt.

### Source excerpt

Molecular dynamics (MD) is an indispensable tool across physics, chemistry, biology, and materials science. Owing to its arithmetic structure, particularly the...

## Kolmogorov-Chapman Maps: Trajectory-Free Learning for Simulating Molecular Dynamics

DevFeed: [Kolmogorov-Chapman Maps: Trajectory-Free Learning for Simulating Molecular Dynamics](<https://devfeed.tech/articles/kolmogorov-chapman-maps-trajectory-free-learning-for-simulating-molecular-dynamics-125599.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.12880>)

Author: Majdi Hassan, Marta Skreta, Dominique Beaini, Kirill Neklyudov

Published: 2026-11-10T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Molecular Dynamics](<https://devfeed.tech/topics/molecular-dynamics.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [Neural Network](<https://devfeed.tech/topics/neural-network.md>)

Tags: [biology](<https://devfeed.tech/tags/biology.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [deep-learning](<https://devfeed.tech/tags/deep-learning.md>), [diffusion](<https://devfeed.tech/tags/diffusion.md>), [distribution](<https://devfeed.tech/tags/distribution.md>), [finance](<https://devfeed.tech/tags/finance.md>), [gpu](<https://devfeed.tech/tags/gpu.md>), [kernel](<https://devfeed.tech/tags/kernel.md>), [learning](<https://devfeed.tech/tags/learning.md>), [molecular-dynamics](<https://devfeed.tech/tags/molecular-dynamics.md>), [neural-network](<https://devfeed.tech/tags/neural-network.md>), [physics](<https://devfeed.tech/tags/physics.md>), [sde](<https://devfeed.tech/tags/sde.md>), [simulation](<https://devfeed.tech/tags/simulation.md>), [training](<https://devfeed.tech/tags/training.md>)

### AI overview

Kolmogorov-Chapman Maps (KC-Maps) learn an SDE's transition kernel without long trajectories, using trajectory-free marginal samples and known drift and diffusion coefficients. For a given initial time and state, the model predicts the process distribution at a later time. Its architecture enforces boundary conditions, while the Kolmogorov-Chapman equation supplies the training objective. The authors report that KC-Maps learn peptide molecular dynamics and outperform classical molecular dynamics in GPU-hours; the excerpt gives no quantitative results or further comparison details.

### Source excerpt

arXiv:2611.12880v1 Announce Type: new Abstract: Stochastic differential equations (SDEs) describe stochastic processes across a wide range of scientific domains, including physics, biology, and finance. Although individual numerical integration steps are inexpensive, observing meaningful events or transitions can require millions of sequential steps, resulting in substantial wall-clock simulation time and poor hardware utilization. Deep learning offers an alternative by amortizing many integration steps into a single neural network forward pass, potentially enabling much faster simulation. However, existing supervised approaches typically require long simulated trajectory data, which restricts their applicability. In this work, we introduce Kolmogorov-Chapman Maps (KC-Maps), a method for learning the transition kernel of an SDE that relies only on the trajectory-free marginal samples together with the known drift and diffusion coefficients. Given an initial time and state of a process, KC-Maps models its distribution integrated to any given time, corresponding to the weak solution of the SDE. To recover the correct dynamics, KC-Maps satisfies two key properties: (i) it satisfies the boundary conditions, which we incorporate directly into the neural network architecture, and (ii) it respects the Kolmogorov--Chapman equation, which provides an efficient training objective. We empirically demonstrate that KC-Maps learns to integrate the dynamics of peptide molecular systems and outperforms classical molecular dynamics in terms of GPU-hours.

## Static Recovery Is Not Dynamic Stability: Dynamics-Aware Benchmarking of Protein Motif Scaffolding

DevFeed: [Static Recovery Is Not Dynamic Stability: Dynamics-Aware Benchmarking of Protein Motif Scaffolding](<https://devfeed.tech/articles/static-recovery-is-not-dynamic-stability-dynamics-aware-benchmarking-of-protein-motif-scaffolding-120656.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.07723>)

Author: Emile de Bruyn, Moritz Sch\\"affler, Jannik Schneider, Sebastian R. Schmidt, Chanwoong Hwang, Aleena Siji, Utkarsh Upadhyay, Stefan Bauer, Alexander Schug, Stefan Kesselheim, Artur Yakimovich, Alina Ba

Published: 2026-11-06T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Molecular Dynamics](<https://devfeed.tech/topics/molecular-dynamics.md>), [Computer science](<https://devfeed.tech/topics/computer-science.md>)

Tags: [benchmarking](<https://devfeed.tech/tags/benchmarking.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [molecular-dynamics](<https://devfeed.tech/tags/molecular-dynamics.md>), [q-bio-bm](<https://devfeed.tech/tags/q-bio-bm.md>), [recovery](<https://devfeed.tech/tags/recovery.md>), [static](<https://devfeed.tech/tags/static.md>)

### AI overview

The benchmark finds that static motif recovery does not establish dynamical stability of the full protein scaffold. BILBY-PRO expands evaluation to 220 cases from 174 PROSITE motif-pattern entries. Across six scaffold-generation models, RFDiffusion and Proteina solve more targets, while ESM3 yields more unique successful scaffolds on the targets it solves. BILBY-MD compares statically solved MotifBench designs with reference molecular-dynamics simulations using motif and scaffold drift, motif-contact preservation, and motif-dihedral free-energy landscapes. Of 55 statically solved model-problem pairs, 31 meet the MD-supported criterion and 8 pass all checks. Compact, self-contained motifs are often supported; interface-like or context-dependent motifs fail more frequently. The authors frame evaluation around target coverage, unique-solution yield, and MD-supported dynamical fidelity.

### Source excerpt

arXiv:2611.07723v1 Announce Type: new Abstract: Protein motif scaffolding benchmarks assess whether generative models can build scaffolds around functionally important residues. Their conclusions depend on which motifs are tested and which criteria define success. Existing benchmarks typically use small hand-curated motif sets and static refolding criteria, but protein function often depends on conformational ensembles and transitions. Here, we address both benchmark scope and success criteria. We introduce BILBY-PRO, a systematically derived benchmark of structurally conserved functional motifs from PROSITE-linked experimental structures, comprising 220 cases from 174 PROSITE motif-pattern entries. Evaluating six recent scaffold-generation models reveals distinct patterns in target coverage and unique-solution yield: RFDiffusion and Proteina solve more targets, whereas ESM3 produces more unique successful scaffolds on the targets it solves. We further introduce BILBY-MD, a reference-based all-atom molecular dynamics evaluation framework. Applied to statically solved designs from the original MotifBench benchmark, it compares motif drift, scaffold drift, motif-contact preservation, and motif-dihedral free-energy landscapes against reference simulations. Of 55 statically solved model-problem pairs, 31 satisfy our MD-supported criterion and only 8 pass all checks. Compact, self-contained motifs are often MD-supported, whereas interface-like or context-dependent motifs fail more frequently. Static refolding success therefore provides useful evidence of local motif recovery but does not establish dynamical fidelity of the full scaffold. BILBY highlights three complementary axes of motif-scaffolding evaluation: target coverage, unique-solution yield, and MD-supported dynamical fidelity.

## Neural Stochastic Forecasting for Lateral Molecular Diffusion Simulations

DevFeed: [Neural Stochastic Forecasting for Lateral Molecular Diffusion Simulations](<https://devfeed.tech/articles/neural-stochastic-forecasting-for-lateral-molecular-diffusion-simulations-120533.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.06157>)

Author: Jacob Ede Levine, Mario Hidalgo-Soria, Chen Yun Wen, Yun Lyna Luo, Sai Chandra Kosaraju

Published: 2026-11-06T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Molecular Dynamics](<https://devfeed.tech/topics/molecular-dynamics.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [Simulation](<https://devfeed.tech/topics/simulation.md>)

Tags: [collective](<https://devfeed.tech/tags/collective.md>), [computer-science](<https://devfeed.tech/tags/computer-science.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [diffusion](<https://devfeed.tech/tags/diffusion.md>), [forecasting](<https://devfeed.tech/tags/forecasting.md>), [lateral](<https://devfeed.tech/tags/lateral.md>), [md-trajectories](<https://devfeed.tech/tags/md-trajectories.md>), [membrane](<https://devfeed.tech/tags/membrane.md>), [modeling](<https://devfeed.tech/tags/modeling.md>), [molecular-dynamics](<https://devfeed.tech/tags/molecular-dynamics.md>), [multiple](<https://devfeed.tech/tags/multiple.md>), [networks](<https://devfeed.tech/tags/networks.md>), [neural](<https://devfeed.tech/tags/neural.md>), [proteins](<https://devfeed.tech/tags/proteins.md>), [recurrent](<https://devfeed.tech/tags/recurrent.md>), [simulations](<https://devfeed.tech/tags/simulations.md>), [transformer](<https://devfeed.tech/tags/transformer.md>)

### AI overview

The neural stochastic model forecasts long-time collective diffusion of membrane proteins from coarse-grained molecular dynamics trajectories. It combines a transformer representation of protein configurations, recurrent networks at multiple temporal scales, a stochastic latent process with learned memory, and a cross-attention decoder that generates correlated displacements while accounting for system geometry and periodic boundaries. In simulations of seven interacting PIEZO channels, it reproduces the MD anomalous diffusion exponent (α ~ 1), unlike a previously developed variational autoregressive model that shows superdiffusive behavior. It generates trajectories up to 40,000 frames and captures distinct PIEZO and TREK diffusion in a mixed seven-PIEZO, ten-TREK system. The reported time-averaged mean-square displacements, Van Hove distributions, and non-Gaussian parameters retain key statistical features of the simulated dynamics.

### Source excerpt

arXiv:2611.06157v2 Announce Type: replace Abstract: Predicting the long-time diffusion of membrane proteins is challenging because molecular dynamics (MD) simulations are computationally expensive, while experimental single-particle tracking is often limited by finite observation times. Here, we introduce a neural stochastic (NS) model for forecasting the collective diffusion of membrane proteins from coarse-grained MD trajectories. The model combines a transformer-based representation of protein configurations, recurrent networks operating at multiple temporal scales, and a stochastic latent process with learned memory. A cross-attention decoder then generates correlated protein displacements while accounting for the instantaneous system geometry and periodic boundary conditions. We evaluate the model in two membrane systems: seven interacting PIEZO channels and a mixed system containing seven PIEZO and ten TREK channels. For the seven PIEZO system, the NS model reproduces the anomalous diffusion exponent measured in MD simulations $\alpha \sim 1$, substantially improving on a previously developed variational autoregressive model that exhibits superdiffusive behavior. The model also generates trajectories extending to 40,000 frames and captures the distinct diffusion characteristics of PIEZO and TREK populations in the mixed system. Comparisons of time-averaged mean-square displacements, Van Hove displacement distributions, and non-Gaussian parameters show that the model preserves key statistical features of the simulated dynamics. These results establish neural stochastic modeling as a promising approach for generating long trajectories of multi-protein membrane diffusion and extending the statistical analysis of molecular dynamics simulations.

## Ab Initio Accuracy Beyond the Cutoff: Distributed Message-Passing Equivariant Attention at Tens-of-Millions-of-Atom Scale

DevFeed: [Ab Initio Accuracy Beyond the Cutoff: Distributed Message-Passing Equivariant Attention at Tens-of-Millions-of-Atom Scale](<https://devfeed.tech/articles/ab-initio-accuracy-beyond-the-cutoff-distributed-message-passing-equivariant-attention-at-tens-of-millions-of-atom-scale-118425.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.06122>)

Author: Allex Kan, Ji Zhang, Hao Yin, Hao Li, Owen Fu, Xunxing Chen, Lin Huang, Zion Wang, Lucas Li, Jiajun Cheng, Arthur Jiang, Junping Zhao, Jia Zhang

Published: 2026-11-05T05:00:00Z

Content type: news

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>)

Topics: [Molecular Dynamics](<https://devfeed.tech/topics/molecular-dynamics.md>), [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>), [High-Performance Computing](<https://devfeed.tech/topics/high-performance-computing.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>)

Tags: [atom](<https://devfeed.tech/tags/atom.md>), [attention](<https://devfeed.tech/tags/attention.md>), [communication](<https://devfeed.tech/tags/communication.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [cuda](<https://devfeed.tech/tags/cuda.md>), [distributed](<https://devfeed.tech/tags/distributed.md>), [efficiency](<https://devfeed.tech/tags/efficiency.md>), [gpus](<https://devfeed.tech/tags/gpus.md>), [graph](<https://devfeed.tech/tags/graph.md>), [kernels](<https://devfeed.tech/tags/kernels.md>), [linear](<https://devfeed.tech/tags/linear.md>), [message](<https://devfeed.tech/tags/message.md>), [molecular-dynamics](<https://devfeed.tech/tags/molecular-dynamics.md>), [physics-bio-ph](<https://devfeed.tech/tags/physics-bio-ph.md>), [salt](<https://devfeed.tech/tags/salt.md>), [scale](<https://devfeed.tech/tags/scale.md>), [scaling](<https://devfeed.tech/tags/scaling.md>), [search](<https://devfeed.tech/tags/search.md>), [serial](<https://devfeed.tech/tags/serial.md>), [shapes](<https://devfeed.tech/tags/shapes.md>), [simulations](<https://devfeed.tech/tags/simulations.md>), [skin](<https://devfeed.tech/tags/skin.md>)

### AI overview

Tessera enables equivariant message-passing Transformer simulations at tens-of-millions-of-atom scale by keeping distributed workloads' shapes fixed between neighbor-list rebuilds. It assigns equal-count atom shards using a space-filling curve, uses a conservative neighbor skin to preserve topology between updates, and replays each rebuild interval as a CUDA Graph; fused kernels, three-level neighbor search, and overlapped All-to-All handle computation and communication. Four model layers can reach up to 23 Å. On 512 GPUs, Tessera reports 30.7 million atoms at 89.7% weak-scaling efficiency, with per-step time increasing 1.11-1.29x as machine and problem size grow 16x. Graph reuse reduces full-step time 1.87x; across an 8x device increase, neighbor-search cost grows 1.7x versus 3.0x for Morton order. A 10.24-million-atom salt simulation reports low NVE energy drift and structure matching a serial reference. A 10.06-million-atom Norovirus capsid ran 100 ps of NVT at about 8 ps/day on 256 A100 GPUs, with no reported broken bonds or anomalously close pairs over 200,000 steps.

### Source excerpt

arXiv:2611.06122v1 Announce Type: cross Abstract: Molecular dynamics predicts molecular motion from atomic forces and trajectory integration. Density functional theory (DFT) provides first-principles forces, but its $O(N^3)$ cost limits simulations to thousands of atoms; empirical force fields scale to millions but miss bond breaking and charge transfer. Equivariant message-passing Transformers learn the ab initio potential at linear cost. Stacked layers propagate information beyond a single cutoff, giving four layers a receptive field up to 23{\AA}. At this frontier, scale is limited by communication: layers exchange boundary features across devices, force backpropagation repeats the exchange, and moving atoms change neighbor graphs and tensor shapes. We present Tessera, a distributed runtime that fixes workload shapes between rebuilds. It sorts atoms along a space-filling curve into equal-count shards, ensuring identical atom counts and tensor shapes per rank. Peer sets scale as $K^{0.5}$, not $K$. A conservative neighbor skin preserves topology across updates, enabling each rebuild interval to replay as one CUDA Graph. Fused kernels, three-level neighbor search, and overlapped All-to-All complete it. On 512 GPUs, Tessera sustains 30.7M atoms at 89.7% weak-scaling efficiency (77--90% across three per-GPU loads); per-step time rises 1.11--1.29x as machine and problem grow 16x. Graph reuse cuts full-step time $1.87\times$. Across an $8\times$ device increase, neighbor-search cost grows $1.7\times$, versus $3.0\times$ for Morton order. In a live $10.24$M-atom salt solution, NVE energy drift is $0.42\,\mu$eV/atom/ps over $10$ ps, and structure matches a serial reference. A 10.06M-atom Norovirus capsid sustains 100 ps of NVT at $\approx 8$ ps/day on 256 A100 GPUs. Over 200,000 steps, it has no broken bonds or anomalously close pairs and uses three compiled variants, with no graph breaks or eager fallbacks.

## Beyond Unit-Circle Eigenvalues: Invariant Bases for Stable State Space Dynamics

DevFeed: [Beyond Unit-Circle Eigenvalues: Invariant Bases for Stable State Space Dynamics](<https://devfeed.tech/articles/beyond-unit-circle-eigenvalues-invariant-bases-for-stable-state-space-dynamics-114659.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.02174>)

Author: Jiaqian Zhu, Yang Zhang, Junhua Ding, Xiaowei Yu

Published: 2026-11-03T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Mamba](<https://devfeed.tech/topics/mamba.md>), [Molecular Dynamics](<https://devfeed.tech/topics/molecular-dynamics.md>)

Tags: [benchmarks](<https://devfeed.tech/tags/benchmarks.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [eigenvalues](<https://devfeed.tech/tags/eigenvalues.md>), [graph](<https://devfeed.tech/tags/graph.md>), [invariant](<https://devfeed.tech/tags/invariant.md>), [molecular-dynamics](<https://devfeed.tech/tags/molecular-dynamics.md>), [momentum](<https://devfeed.tech/tags/momentum.md>), [power-grid](<https://devfeed.tech/tags/power-grid.md>), [quantity](<https://devfeed.tech/tags/quantity.md>), [stable](<https://devfeed.tech/tags/stable.md>), [state-space-models](<https://devfeed.tech/tags/state-space-models.md>)

### AI overview

The Kronecker-Cayley transition (KC) is designed to preserve layered conserved quantities in diagonal state-space dynamics, since unit-modulus eigenvalues alone preserve only the outermost layer. Its generator is a graph Laplacian over object subspaces, H = L ⊗ I_m, advanced by a Cayley transform; the step remains diagonal with stated cost O(N^3 + N^2m). The construction preserves an exact 2m²-dimensional family of quadratic forms at every parameter value. Across four conservation benchmarks, KC keeps total mass near the reference floor while diagonal baselines lose it by orders of magnitude. Two mass correctors fail when the underlying prediction diverges. A free modal basis trained with a conservation penalty conserves only by freezing its spectrum, while KC allows all but one mode to rotate. The abstract also reports preserved linear momentum in three N-body systems and operator-level conserved quantities in several physical and applied domains.

### Source excerpt

arXiv:2611.02174v1 Announce Type: new Abstract: Diagonal state space models process long sequences at linear cost, yet the same transition is applied at every step of a rollout, so an unconstrained recurrence holds no quantity fixed and the state decays or grows without bound. Current designs constrain one quantity, and preserving one does not guarantee preservation of the others, so a model can hold its norm to machine precision while momentum drifts by hundreds of percent. We claim that the conserved quantities of a diagonal recurrence are layered, that each layer requires a strictly stronger condition on the operator, and that a unit-modulus spectrum reaches only the outermost. Acting on the claim, we enforce the conditions required for every layer by construction. The generator of the Kronecker-Cayley transition (KC) is a graph Laplacian over object subspaces, $\mathbf{H}=\mathbf{L}\otimes I_m$, advanced by a Cayley transform, and the step stays diagonal at $O(N^3+N^2m)$ cost. The conserved set is exactly a $2m^2$-dimensional family of quadratic forms, held at every parameter value, so the quantities and the dynamics the construction reaches are fixed before training. On four conservation benchmarks, KC holds total mass at or near the reference floor while every diagonal baseline loses it by orders of magnitude. We further compare against mass correction and find that two correctors fail once the underlying prediction diverges, which shows that assembly rather than correction is necessary for a guarantee that holds across the rollout. A free modal basis trained under a conservation penalty conserves only by freezing its spectrum, whereas the assembled operator conserves with every mode but one free to rotate. Beyond these benchmarks, KC holds the linear momentum of three $N$-body systems and, at the operator level, the conserved quantity in molecular dynamics, chemical kinetics, pharmacokinetics and power-grid swing dynamics.

## Where the Error Lives: Spectral Representations of Learned Kohn-Sham Hamiltonians

DevFeed: [Where the Error Lives: Spectral Representations of Learned Kohn-Sham Hamiltonians](<https://devfeed.tech/articles/where-the-error-lives-spectral-representations-of-learned-kohn-sham-hamiltonians-107973.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.37843>)

Author: Fang Sun, Hekai Cheng, Ruoyan Li, Yuanzhou Chen, Wei Wang, Jason Cong, Yizhou Sun

Published: 2026-10-30T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Fine-tuning](<https://devfeed.tech/topics/fine-tuning.md>), [Molecular Dynamics](<https://devfeed.tech/topics/molecular-dynamics.md>)

Tags: [alignment](<https://devfeed.tech/tags/alignment.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [dft](<https://devfeed.tech/tags/dft.md>), [distribution](<https://devfeed.tech/tags/distribution.md>), [eigenvalues](<https://devfeed.tech/tags/eigenvalues.md>), [eigenvectors](<https://devfeed.tech/tags/eigenvectors.md>), [electron](<https://devfeed.tech/tags/electron.md>), [energy](<https://devfeed.tech/tags/energy.md>), [fine-tuning](<https://devfeed.tech/tags/fine-tuning.md>), [gap](<https://devfeed.tech/tags/gap.md>), [hamiltonian](<https://devfeed.tech/tags/hamiltonian.md>), [local](<https://devfeed.tech/tags/local.md>), [machine](<https://devfeed.tech/tags/machine.md>), [matching](<https://devfeed.tech/tags/matching.md>), [molecular-dynamics](<https://devfeed.tech/tags/molecular-dynamics.md>), [networks](<https://devfeed.tech/tags/networks.md>), [observable](<https://devfeed.tech/tags/observable.md>), [physics-chem-ph](<https://devfeed.tech/tags/physics-chem-ph.md>), [physics-informed](<https://devfeed.tech/tags/physics-informed.md>), [signal](<https://devfeed.tech/tags/signal.md>), [space](<https://devfeed.tech/tags/space.md>), [training](<https://devfeed.tech/tags/training.md>)

### AI overview

Entry-wise Hamiltonian accuracy does not guarantee accurate orbital energies or eigenvectors. The abstract argues that wavefunction-alignment and electron-density losses each miss parts of prediction error; in particular, shifting predicted orbital energies leaves the density unchanged. It proposes cumulative local density of states (cumulative LDOS)--the electron count at each spatial point below an energy threshold--as a complete training signal that responds to all error components, uniquely determines the Hamiltonian, and retains gradients for large orbital-energy errors. Fine-tuning three Hamiltonian networks on QH9 with this approach, called FermiScope, reduces HOMO-LUMO gap error without increasing element-wise error. The reported gains extend to out-of-distribution and molecular-dynamics splits, where wavefunction alignment helps less.

### Source excerpt

arXiv:2610.37843v1 Announce Type: cross Abstract: Kohn-Sham density functional theory (DFT) is the standard method for computing the electronic structure of molecules and materials, but its self-consistent-field (SCF) iterations make it expensive for large systems and high-throughput screening. Machine-learned Hamiltonian models offer a shortcut: an equivariant network predicts the Kohn-Sham Hamiltonian matrix in one forward pass, and diagonalizing the prediction yields orbital energies, the HOMO-LUMO gap, molecular orbitals and the electron density. These models are trained to match the reference matrix entry by entry, yet a small entry-wise error does not guarantee accurate eigenvalues or eigenvectors. Two physics-informed alternatives go partway: a recently proposed wavefunction-alignment loss adds a spectral penalty without diagonalizing the prediction, and matching the electron density of the diagonalized prediction supervises it directly. Neither is complete: by representing the prediction error in the eigenbasis of the reference Hamiltonian, we show that each sees only part of the error, and that shifting the predicted orbital energies leaves the electron density exactly unchanged. The same analysis identifies a complete training signal: the cumulative local density of states (cumulative LDOS), the number of electrons at each point in space with energy below a given threshold. The cumulative LDOS is a single physical observable that contains the electron density as a special case, responds to every component of the error, determines the Hamiltonian uniquely, and yields gradients that do not vanish for large orbitalenergy errors. Fine-tuning three Hamiltonian networks on QH9 in this way, which we call FermiScope, reduces the HOMO-LUMO gap error without increasing the element-wise error. The gains carry over to out-of-distribution and molecular-dynamics splits, where the wavefunction-alignment loss helps less.

## SKG in Practice: OOD Generalization Cases Across FourDomains

DevFeed: [SKG in Practice: OOD Generalization Cases Across FourDomains](<https://devfeed.tech/articles/skg-in-practice-ood-generalization-cases-across-fourdomains-99600.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.32314>)

Author: Jian An, Chaoshi Wu, Xinpan Luo

Published: 2026-10-26T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Neural Network](<https://devfeed.tech/topics/neural-network.md>), [Molecular Dynamics](<https://devfeed.tech/topics/molecular-dynamics.md>), [Transformers](<https://devfeed.tech/topics/transformers.md>)

Tags: [case-studies](<https://devfeed.tech/tags/case-studies.md>), [cases](<https://devfeed.tech/tags/cases.md>), [comparison](<https://devfeed.tech/tags/comparison.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [detection](<https://devfeed.tech/tags/detection.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [framework](<https://devfeed.tech/tags/framework.md>), [id](<https://devfeed.tech/tags/id.md>), [implementation](<https://devfeed.tech/tags/implementation.md>), [improvement](<https://devfeed.tech/tags/improvement.md>), [inference](<https://devfeed.tech/tags/inference.md>), [invariant](<https://devfeed.tech/tags/invariant.md>), [methodology](<https://devfeed.tech/tags/methodology.md>), [metric](<https://devfeed.tech/tags/metric.md>), [modeling](<https://devfeed.tech/tags/modeling.md>), [molecular-dynamics](<https://devfeed.tech/tags/molecular-dynamics.md>), [neural](<https://devfeed.tech/tags/neural.md>), [ood](<https://devfeed.tech/tags/ood.md>), [paper](<https://devfeed.tech/tags/paper.md>), [parameter](<https://devfeed.tech/tags/parameter.md>)

### AI overview

The paper reports SKG results across four out-of-distribution settings: lower OOD error than neural baselines for a dissipative mechanical system and sine-to-square sequence modeling, improved real-machine target-detection scores without real-machine retraining, and lower ADE than ESTAG on MD17. The sequence result averages five seeds (MSE 0.0296 vs. 0.0641; p=0.024); the detection result is a single run on 32 images, and the MD17 comparison uses one seed. The paper also outlines pre-validation, task-invariant construction, three-stage and residual training, and supporting theory. It reports variance and limitations, including negative updates from parameterization mismatch, and treats engineering thresholds as diagnostic signals rather than hard constraints.

### Source excerpt

arXiv:2610.32314v1 Announce Type: new Abstract: This paper presents the engineering implementation methods of the SKG inference framework and provides complete case studies across four domains. Case 1, 2D Dissipative Mechanical System: OOD MSE is 1.69 to 4.12 times lower than the best neural baseline; the state-dependent metric yields a 10^8-order improvement under controlled conditions. Case 2, Target Detection Sim-to-Real: Real-machine mAP increases from 0.2564 to 0.4964 (+24.00 points, single-run experiment, 32 images, point estimate), and AP75 increases from 0.1736 to 0.5765 (+40.29 points), with no real-machine retraining required. Case 3, Sequence Modeling sine to square: OOD MSE is 0.0296 versus Transformer 0.0641 (mean of 5 seeds, p=0.024). Case 4, Molecular Dynamics MD17: ID@100 ADE is 0.0285 versus ESTAG 0.1467; OOD@100 is 0.0316 versus 0.1411 (single seed). For each case, we present the objective, applied SKG theory, parameter settings, baseline comparison protocol, data results, and sources of advantage. Following the cases, we provide the complete methodology: pre-validation, task invariant construction, three-stage training, residual training, and the four theorems and one proposition supporting these methods. This paper not only demonstrates effective configurations of SKG across four domains but also honestly reports the variance of single-run experiments, limitations of single-seed results, negative updates caused by parameterization mismatch, and practical experience of using engineering thresholds as diagnostic signals rather than hard constraints.

## SoLiD26: A First Principles Solid-Liquid Interface Dataset for Machine-learned Interatomic Potentials

DevFeed: [SoLiD26: A First Principles Solid-Liquid Interface Dataset for Machine-learned Interatomic Potentials](<https://devfeed.tech/articles/solid26-a-first-principles-solid-liquid-interface-dataset-for-machine-learned-interatomic-potentials-94757.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2609.28013>)

Author: Jonas Busk, Emil J. P. Frost, Yogeshwaran Krishnan, Henrik H. Kristoffersen, August E. G. Mikkelsen, Xueping Qin, Xin Yang, Heine A. Hansen, Arghya Bhowmik, Tejs Vegge

Published: 2026-10-22T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Dataset development](<https://devfeed.tech/topics/dataset-development.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Molecular Dynamics](<https://devfeed.tech/topics/molecular-dynamics.md>)

Tags: [aimd](<https://devfeed.tech/tags/aimd.md>), [cond-mat-mtrl-sci](<https://devfeed.tech/tags/cond-mat-mtrl-sci.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [dataset](<https://devfeed.tech/tags/dataset.md>), [dft](<https://devfeed.tech/tags/dft.md>), [dispersion](<https://devfeed.tech/tags/dispersion.md>), [elements](<https://devfeed.tech/tags/elements.md>), [first](<https://devfeed.tech/tags/first.md>), [functional](<https://devfeed.tech/tags/functional.md>), [interface](<https://devfeed.tech/tags/interface.md>), [interfaces](<https://devfeed.tech/tags/interfaces.md>), [liquid](<https://devfeed.tech/tags/liquid.md>), [mace](<https://devfeed.tech/tags/mace.md>), [machine](<https://devfeed.tech/tags/machine.md>), [materials](<https://devfeed.tech/tags/materials.md>), [metal](<https://devfeed.tech/tags/metal.md>), [modelling](<https://devfeed.tech/tags/modelling.md>), [models](<https://devfeed.tech/tags/models.md>), [molecular-dynamics](<https://devfeed.tech/tags/molecular-dynamics.md>), [physics-chem-ph](<https://devfeed.tech/tags/physics-chem-ph.md>), [physics-comp-ph](<https://devfeed.tech/tags/physics-comp-ph.md>), [principles](<https://devfeed.tech/tags/principles.md>), [reference](<https://devfeed.tech/tags/reference.md>), [solid](<https://devfeed.tech/tags/solid.md>)

### AI overview

SoLiD26 is a dataset designed to address limited training data for machine-learned interatomic potentials at solid-liquid interfaces. It contains 15.4 million first-principles structures, with systems up to 576 atoms and 15 elements. Most structures come from ab initio molecular dynamics; examples include aqueous coinage-metal interfaces, electrode-electrolyte systems, and selected bulk references. They were calculated using VASP with the PBE functional and D3 dispersion corrections. The authors demonstrate use of the dataset with MACE models on a simple training, validation, and test split. The supplied text does not report model performance details.

### Source excerpt

arXiv:2609.28013v2 Announce Type: replace-cross Abstract: Machine-learned interatomic potentials (MLIPs) enable simulations of larger atomistic systems over longer timescales than traditional ab initio methods but require large, diverse training data. Existing datasets lack comprehensive sampling of solid-liquid interfaces, which exhibit complex phenomena such as solvent layering and surface reconstruction critical for energy and catalytic materials. Here, we present SoLiD26, a curated solid-liquid interface dataset of 15.4 million first-principles structures with up to 576 atoms and 15 chemical elements for training and evaluating MLIPs. The structures were compiled from density functional theory (DFT) calculations, with most configurations originating from ab initio molecular dynamics (AIMD) simulations. SoLiD26 includes aqueous coinage metal interfaces, electrode-electrolyte systems, and selected bulk reference structures, calculated with VASP using the PBE functional and D3 dispersion corrections. The application of SoLiD26 for training and evaluating MLIPs is demonstrated with a suite of MACE models on a simple training, validation and test split. This dataset provides a valuable resource for advancing MLIP development targeted at modelling solid-liquid interfaces in electrochemical and catalytic systems.

## From structures to dynamics: scaling AI for molecular dynamics in drug discovery

DevFeed: [From structures to dynamics: scaling AI for molecular dynamics in drug discovery](<https://devfeed.tech/articles/from-structures-to-dynamics-scaling-ai-for-molecular-dynamics-in-drug-discovery-84854.md>)

Original publisher: [Read original article](<https://lambda.ai/blog/eginterpolator-molecular-dynamics-drug-discovery>)

Author: Jianwen Xie

Published: 2026-09-21T17:26:34Z

Content type: article

Language: en

Sources: [The Lambda deep learning blog](<https://devfeed.tech/sources/the-lambda-deep-learning-blog.md>)

Topics: [Molecular Dynamics](<https://devfeed.tech/topics/molecular-dynamics.md>), [AI for science](<https://devfeed.tech/topics/ai-for-science.md>), [BioNeMo](<https://devfeed.tech/topics/bionemo.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-for-drug-discovery](<https://devfeed.tech/tags/ai-for-drug-discovery.md>), [biotech](<https://devfeed.tech/tags/biotech.md>), [diffusion-models](<https://devfeed.tech/tags/diffusion-models.md>), [drug-discovery](<https://devfeed.tech/tags/drug-discovery.md>), [generative-ai](<https://devfeed.tech/tags/generative-ai.md>), [gpu-computing](<https://devfeed.tech/tags/gpu-computing.md>), [md-trajectories](<https://devfeed.tech/tags/md-trajectories.md>), [molecular-dynamics](<https://devfeed.tech/tags/molecular-dynamics.md>), [molecular-simulation](<https://devfeed.tech/tags/molecular-simulation.md>), [molecular-structure-prediction](<https://devfeed.tech/tags/molecular-structure-prediction.md>), [scaling-ai](<https://devfeed.tech/tags/scaling-ai.md>), [structure-pretraining](<https://devfeed.tech/tags/structure-pretraining.md>)

### AI overview

EGInterpolator uses a two-stage approach to generate molecular dynamics trajectories: it first pretrains a diffusion model on abundant molecular conformer data, then fine-tunes a trajectory interpolator on scarcer molecular dynamics data. On the DRUGS benchmark, it outperformed GeoTDM on reported measures of molecular geometry and motion. The article describes the work as a way to make AI-driven molecular dynamics more data-efficient for drug discovery.

### Source excerpt

Why molecular dynamics matters for drug discovery Molecular dynamics (MD) is an important tool in modern drug discovery. While structure prediction and molecular docking can show how a potential drug might fit into a protein, they provide only a snapshot. Molecules are constantly moving. MD simulations allow researchers to see how a drug and its target behave over time: whether the interaction remains stable, how the protein changes shape, and how the molecular system evolves.

## Resilient HPC and ML on AWS: Running Tightly Coupled Workloads on Spot Instances

DevFeed: [Resilient HPC and ML on AWS: Running Tightly Coupled Workloads on Spot Instances](<https://devfeed.tech/articles/resilient-hpc-and-ml-on-aws-running-tightly-coupled-workloads-on-spot-instances-50156.md>)

Original publisher: [Read original article](<https://aws.amazon.com/blogs/hpc/resilient-hpc-and-ml-on-aws-running-tightly-coupled-workloads-on-spot-instances/>)

Author: Santosh Kumar

Published: 2026-08-05T00:00:55Z

Content type: tutorial

Language: en

Sources: [AWS HPC Blog](<https://devfeed.tech/sources/aws-hpc-blog.md>)

Topics: [High-Performance Computing](<https://devfeed.tech/topics/high-performance-computing.md>), [Amazon EC2](<https://devfeed.tech/topics/amazon-ec2.md>), [Amazon Web Services](<https://devfeed.tech/topics/aws.md>), [Computational Fluid Dynamics (CFD)](<https://devfeed.tech/topics/cfd.md>), [Simulation](<https://devfeed.tech/topics/simulation.md>), [GROMACS](<https://devfeed.tech/topics/gromacs.md>), [Molecular Dynamics](<https://devfeed.tech/topics/molecular-dynamics.md>), [Training AI Models](<https://devfeed.tech/topics/training-ai-models.md>)

Tags: [amazon-ec2](<https://devfeed.tech/tags/amazon-ec2.md>), [amazon-elastic-file-system-efs](<https://devfeed.tech/tags/amazon-elastic-file-system-efs.md>), [amazon-fsx-for-lustre](<https://devfeed.tech/tags/amazon-fsx-for-lustre.md>), [artificial-intelligence](<https://devfeed.tech/tags/artificial-intelligence.md>), [availability](<https://devfeed.tech/tags/availability.md>), [aws](<https://devfeed.tech/tags/aws.md>), [aws-fault-injection-service-fis](<https://devfeed.tech/tags/aws-fault-injection-service-fis.md>), [aws-parallel-computing-service](<https://devfeed.tech/tags/aws-parallel-computing-service.md>), [aws-parallelcluster](<https://devfeed.tech/tags/aws-parallelcluster.md>), [best-practices](<https://devfeed.tech/tags/best-practices.md>), [capacity](<https://devfeed.tech/tags/capacity.md>), [cfd](<https://devfeed.tech/tags/cfd.md>), [computational-fluid-dynamics](<https://devfeed.tech/tags/computational-fluid-dynamics.md>), [compute](<https://devfeed.tech/tags/compute.md>), [computing](<https://devfeed.tech/tags/computing.md>), [cost](<https://devfeed.tech/tags/cost.md>), [developer-tools](<https://devfeed.tech/tags/developer-tools.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [gromacs](<https://devfeed.tech/tags/gromacs.md>), [high-performance](<https://devfeed.tech/tags/high-performance.md>), [high-performance-computing](<https://devfeed.tech/tags/high-performance-computing.md>), [hpc](<https://devfeed.tech/tags/hpc.md>), [ml](<https://devfeed.tech/tags/ml.md>), [research-computing](<https://devfeed.tech/tags/research-computing.md>), [scientific-computing](<https://devfeed.tech/tags/scientific-computing.md>), [simulations](<https://devfeed.tech/tags/simulations.md>), [slurm](<https://devfeed.tech/tags/slurm.md>), [storage](<https://devfeed.tech/tags/storage.md>), [technical-how-to](<https://devfeed.tech/tags/technical-how-to.md>)

### AI overview

This AWS post presents an architecture for running tightly coupled, long-running HPC and machine learning workloads on Amazon EC2 Spot Instances. It addresses interruption risk through checkpointing and demonstrates the approach with CFD and GROMACS workloads.

### Source excerpt

This post references AWS ParallelCluster. Check out AWS Parallel Computing Service (AWS PCS), our new managed Slurm service for running HPC and AI workloads on AWS. This post was contributed by Santosh Kumar, Bhagyaraju Kasina, Dr. Sandeep Sovani and Dr. Max Starr Researchers and engineering teams running High Performance Computing (HPC) jobs face a constant [...]

## How Aionics accelerates chemical formulation and discovery with AWS Parallel Computing Service

DevFeed: [How Aionics accelerates chemical formulation and discovery with AWS Parallel Computing Service](<https://devfeed.tech/articles/how-aionics-accelerates-chemical-formulation-and-discovery-with-aws-parallel-computing-service-50149.md>)

Original publisher: [Read original article](<https://aws.amazon.com/blogs/hpc/how-aionics-accelerates-chemical-formulation-and-discovery-with-aws-parallel-computing-service/>)

Author: Moe Elshazly

Published: 2025-11-26T15:35:53Z

Content type: article

Language: en

Sources: [AWS HPC Blog](<https://devfeed.tech/sources/aws-hpc-blog.md>)

Topics: [cloud-infrastructure](<https://devfeed.tech/topics/cloud-infrastructure.md>), [Panic Playdate SDK](<https://devfeed.tech/topics/playdate-sdk.md>), [Simulation](<https://devfeed.tech/topics/simulation.md>), [GPU](<https://devfeed.tech/topics/gpu.md>), [Molecular Dynamics](<https://devfeed.tech/topics/molecular-dynamics.md>), [AI Models](<https://devfeed.tech/topics/ai-models.md>), [DSQL](<https://devfeed.tech/topics/dsql.md>)

Tags: [accelerated](<https://devfeed.tech/tags/accelerated.md>), [aerospace](<https://devfeed.tech/tags/aerospace.md>), [aws](<https://devfeed.tech/tags/aws.md>), [aws-parallel-computing-service](<https://devfeed.tech/tags/aws-parallel-computing-service.md>), [batteries](<https://devfeed.tech/tags/batteries.md>), [computing](<https://devfeed.tech/tags/computing.md>), [elastic-fabric-adapter](<https://devfeed.tech/tags/elastic-fabric-adapter.md>), [gpu](<https://devfeed.tech/tags/gpu.md>), [gpu-accelerated](<https://devfeed.tech/tags/gpu-accelerated.md>), [high-performance](<https://devfeed.tech/tags/high-performance.md>), [high-performance-computing](<https://devfeed.tech/tags/high-performance-computing.md>), [hpc](<https://devfeed.tech/tags/hpc.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [machine](<https://devfeed.tech/tags/machine.md>), [models](<https://devfeed.tech/tags/models.md>), [parallel](<https://devfeed.tech/tags/parallel.md>), [research-computing](<https://devfeed.tech/tags/research-computing.md>), [scientific-computing](<https://devfeed.tech/tags/scientific-computing.md>), [simulation](<https://devfeed.tech/tags/simulation.md>), [simulations](<https://devfeed.tech/tags/simulations.md>), [slurm](<https://devfeed.tech/tags/slurm.md>), [storage](<https://devfeed.tech/tags/storage.md>)

### AI overview

This AWS post explains how Aionics uses AWS Parallel Computing Service to build flexible high-performance computing infrastructure for computational chemistry. It describes plane-wave DFT simulations with MPI parallelization and GPU-accelerated molecular dynamics using machine-learned interatomic potentials to support battery electrolyte discovery.

### Source excerpt

This post was contributed by Mohamed K. Elshazly, PhD, Kareem Abdol-Hamid, Sam Bydlon, PhD, Aarabhi Achanta, and Mark Azadpour The decarbonization of our modern economy depends on solving a defining scientific challenge: developing batteries that are both safe and high performing. From electrical grids to vehicles and aviation, these energy storage devices must provide power [...]