The importance of sampling the dynamical modes: Reevaluating benchmarks for invariant and equivariant features of machine learning potentials for simulation of free energy landscapes DOI
Gustavo R. Pérez-Lemus, Yinan Xu,

Yezhi Jin

et al.

The Journal of Chemical Physics, Journal Year: 2024, Volume and Issue: 161(24)

Published: Dec. 23, 2024

Machine learning interatomic potentials (MLIPs) are rapidly gaining interest for molecular modeling, as they provide a balance between quantum-mechanical level descriptions of atomic interactions and reasonable computational efficiency. However, questions remain regarding the stability simulations using these potentials, well extent to which learned potential energy function can be extrapolated safely. Past studies have encountered challenges when MLIPs applied classical benchmark systems. In this work, we show that some related characteristics training datasets, particularly inefficient exploration dynamical modes inclusion rigid constraints. We demonstrate long in with achieved by generating unconstrained datasets unbiased simulations, provided important correctly sampled. addition, emphasize order achieve precise predictions, it is resort enhanced sampling techniques dataset generation, safe extrapolation depends on judicious choices system’s underlying free landscape symmetry features embedded within machine models.

Language: Английский

The Potential of Neural Network Potentials DOI Creative Commons
Timothy T. Duignan

ACS Physical Chemistry Au, Journal Year: 2024, Volume and Issue: 4(3), P. 232 - 241

Published: March 21, 2024

In the next half-century, physical chemistry will likely undergo a profound transformation, driven predominantly by combination of recent advances in quantum and machine learning (ML). Specifically, equivariant neural network potentials (NNPs) are breakthrough new tool that already enabling us to simulate systems at molecular scale with unprecedented accuracy speed, relying on nothing but fundamental laws. The continued development this approach realize Paul Dirac's 80-year-old vision using mechanics unify physics providing invaluable tools for understanding materials science, biology, earth sciences, beyond. era highly accurate efficient first-principles simulations provide wealth training data can be used build automated computational methodologies, such as diffusion models, design optimization scale. Large language models (LLMs) also evolve into increasingly indispensable literature review, coding, idea generation, scientific writing.

Language: Английский

Citations

14
From Ab Initio to Instrumentation: A Field Guide to Characterizing Multivalent Liquid Electrolytes DOI
Glenn Pastel, Travis P. Pollard,

Oleg Borodin

et al.

Chemical Reviews, Journal Year: 2025, Volume and Issue: unknown

Published: March 10, 2025

In this field guide, we outline empirical and theory-based approaches to characterize the fundamental properties of liquid multivalent-ion battery electrolytes, including (i) structure chemistry, (ii) transport, (iii) electrochemical properties. When detailed molecular-scale understanding multivalent electrolyte behavior is insufficient use examples from well-studied lithium-ion electrolytes. recognition that coupling techniques highly effective, but often nontrivial, also highlight recent characterization efforts uncover a more comprehensive nuanced underlying structures, processes, reactions drive performance system-level behavior. We hope insights these discussions will guide design future studies, accelerate development next-generation batteries through modeling with experiments, help avoid pitfalls ensure reproducibility results.

Language: Английский

Citations

1
Random Sampling Versus Active Learning Algorithms for Machine Learning Potentials of Quantum Liquid Water DOI
Nore Stolte, János Daru, Harald Forbert

et al.

Journal of Chemical Theory and Computation, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 14, 2025

Training accurate machine learning potentials requires electronic structure data comprehensively covering the configurational space of system interest. As construction this is computationally demanding, many schemes for identifying most important structures have been proposed. Here, we compare performance high-dimensional neural network (HDNNPs) quantum liquid water at ambient conditions trained to sets constructed using random sampling as well various flavors active based on query by committee. Contrary common understanding learning, find that a given set size, leads smaller test errors not included in training process. In our analysis, show can be related small energy offsets caused bias added which overcome instead correlations an error measure invariant such shifts. Still, all HDNNPs yield very similar and structural properties water, demonstrates robustness procedure with respect algorithm even when few 200 structures. However, preliminary potentials, reasonable initial avoid unnecessary extension covered configuration less relevant regions.

Language: Английский

Citations

0
Investigating Ionic Diffusivity in Amorphous LiPON using Machine-Learned Interatomic Potentials DOI Creative Commons

Aqshat Seth,

Rutvij Pankaj Kulkarni,

Gopalakrishnan Sai Gautam

et al.

ACS Materials Au, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 5, 2025

Language: Английский

Citations

0
Machine Learning Accelerated Interfacial Fluxionality in Ni-Supported Metal Nitride Ammonia Synthesis Catalysts DOI
Pranav Roy, Brandon C. Bukowski

Published: Jan. 1, 2025

Language: Английский

Citations

0
Application of Machine Learning Interatomic Potentials in Heterogeneous Catalysis DOI

Gbolagade Olajide,

Khagendra Baral, Sophia Ezendu

et al.

Published: Jan. 1, 2025

Language: Английский

Citations

0
Simulation of lithium hydroxide decomposition using deep potential molecular dynamics DOI
Dina Kussainova, Athanassios Z. Panagiotopoulos

The Journal of Chemical Physics, Journal Year: 2024, Volume and Issue: 161(13)

Published: Oct. 1, 2024

Chemical reactions and vapor–liquid equilibria for molten lithium hydroxide (LiOH) were studied using molecular dynamics simulations a deep potential (DP) model. The neural network the model was trained on quantum density functional theory data range of conditions. DP allows over timescales hundreds ns, which provide equilibrium compositions systems interest. Single-phase NPT liquid show decomposition LiOH into oxide (Li2O) dissolved water (H2O). These results validated by direct ab initio that confirmed accuracy with respect to reaction kinetics properties melt. reactive behavior this system subsequently coexistence interfacial simulations. Partial pressures H2O in vapor are found be close agreement available experimental measurements. By fitting temperature-dependent expressions Henry’s law constants, composition any given initial temperature can quantitatively modeled. For high concentrations Li2O or H2O, mixtures + Li2O/H2O undergo phase separation. present study illustrates how DP-based used quantitative modeling multiphase underlying chemical methods.

Language: Английский

Citations

1
Machine-Learning-Backed Evolutionary Exploration of Ti-rich SrTiO3(110) Surface Reconstructions DOI Creative Commons
Ralf Wanzenböck, Esther Heid, Michele Riva

et al.

Published: June 26, 2024

The investigation of inhomogeneous surfaces, where various local structures co-exist, is crucial for understanding interfaces technological interest, yet it presents significant challenges. Here, we study the atomic configurations (2 × m) Ti-rich surfaces at (110)-oriented SrTiO3 by bringing together scanning tunneling microscopy and transferable neural-network force fields combined with evolutionary exploration. We leverage an active learning methodology to iteratively extend training data as needed different configurations. Training on only small well-known reconstructions are able extrapolate complicated diverse overlayers encountered in regions heterogeneous SrTiO3(110)-(2×m) surface. Our machine-learning-backed approach generates several new candidate structures, good agreement experiment verified using density functional theory.

Language: Английский

Citations

0
Exploring Inhomogeneous Surfaces: Ti-rich SrTiO3(110) Reconstructions via Active Learning DOI Creative Commons
Ralf Wanzenböck, Esther Heid, Michele Riva

et al.

Digital Discovery, Journal Year: 2024, Volume and Issue: unknown

Published: Jan. 1, 2024

The investigation of inhomogeneous surfaces, where various local structures coexist, is crucial for understanding interfaces technological interest, yet it presents significant challenges. Here, we study the atomic configurations (2 ×

Language: Английский

Citations

0
The importance of sampling the dynamical modes: Reevaluating benchmarks for invariant and equivariant features of machine learning potentials for simulation of free energy landscapes DOI
Gustavo R. Pérez-Lemus, Yinan Xu,

Yezhi Jin

et al.

The Journal of Chemical Physics, Journal Year: 2024, Volume and Issue: 161(24)

Published: Dec. 23, 2024

Machine learning interatomic potentials (MLIPs) are rapidly gaining interest for molecular modeling, as they provide a balance between quantum-mechanical level descriptions of atomic interactions and reasonable computational efficiency. However, questions remain regarding the stability simulations using these potentials, well extent to which learned potential energy function can be extrapolated safely. Past studies have encountered challenges when MLIPs applied classical benchmark systems. In this work, we show that some related characteristics training datasets, particularly inefficient exploration dynamical modes inclusion rigid constraints. We demonstrate long in with achieved by generating unconstrained datasets unbiased simulations, provided important correctly sampled. addition, emphasize order achieve precise predictions, it is resort enhanced sampling techniques dataset generation, safe extrapolation depends on judicious choices system’s underlying free landscape symmetry features embedded within machine models.

Language: Английский

Citations

0