Comparing machine learning potentials for water: Kernel-based regression and Behler–Parrinello neural networks

The Journal of Chemical Physics, Journal Year: 2024, Volume and Issue: 160(11)

Published: March 20, 2024

In this paper, we investigate the performance of different machine learning potentials (MLPs) in predicting key thermodynamic properties water using RPBE + D3. Specifically, scrutinize kernel-based regression and high-dimensional neural networks trained on a highly accurate dataset consisting about 1500 structures, as well smaller dataset, half size, obtained only on-the-fly learning. This study reveals that despite minor differences between MLPs, their agreement observables such diffusion constant pair-correlation functions is excellent, especially for large training dataset. Variations predicted density isobars, albeit somewhat larger, are also acceptable, particularly given errors inherent to approximate functional theory. Overall, emphasizes relevance database over fitting method. Finally, underscores limitations root mean square need comprehensive testing, advocating use multiple MLPs enhanced certainty, when simulating complex may not be fully captured by simpler tests.

Language: Английский

---

The Potential of Neural Network Potentials

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

---

Electroplating of Wear‐ and Corrosion‐Resistant CrCoNi Medium‐Entropy Alloys beyond Hard Chromium Coatings

Advanced Functional Materials, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 19, 2025

Abstract High‐entropy alloys (HEAs) and medium‐entropy (MEAs) are a new class of that attract attention because their mechanical properties. The application such for coating is highly desired; however, the number technologies remains limited. Although electrodeposition expected to be an environmentally friendly energy‐saving technology, neither CrMnFeCoNi HEA nor any its derivatives successfully electrodeposited difficulty in controlling composition alloying Cr, key component, as crystal. Here, successful CrCoNi MEA demonstrated using mixture ionic liquid aqueous solution containing metal salts. resultant layer exhibits high wear corrosion resistance superior conventional hard Cr coatings prepared toxic Cr(VI) ions. Mesoscopic phase separation shown MEA. strong potential substitute coatings; thus, it believed important advancement anticorrosion surface coatings.

Language: Английский

---

Density isobar of water and melting temperature of ice: Assessing common density functionals

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

Published: Oct. 3, 2024

We investigate the density isobar of water and melting temperature ice using six different functionals. Machine-learning potentials are employed to ensure computational affordability. Our findings reveal significant discrepancies between various base Notably, even choice damping can result in substantial differences. Overall, outcomes obtained through functional theory not entirely satisfactory across most utilized All functionals exhibit deviations either or equilibrium volume, with them predicting an incorrect volume difference water. heuristic analysis indicates that a hybrid 25% exact exchange van der Waals averaged zero Becke–Johnson dampings yields closest agreement experimental data. This study underscores necessity for further enhancements treatment interactions and, more broadly, enable accurate quantitative predictions molecular liquids.

Language: Английский

---

Sparse Gaussian process based machine learning first principles potentials for materials simulations: Application to batteries, solar cells, catalysts, and macromolecular systems

Chemical Physics Reviews, Journal Year: 2024, Volume and Issue: 5(4)

Published: Nov. 27, 2024

To design new materials and understand their novel phenomena, it is imperative to predict the structure properties of that often rely on first-principles theory. However, such methods are computationally demanding limited small systems. This topical review investigates machine learning (ML) approaches, specifically non-parametric sparse Gaussian process regression (SGPR), model potential energy surface (PES) materials, while starting from basics ML for a comprehensive review. SGPR can efficiently represent PES with minimal ab initio data, significantly reducing computational costs by bypassing need inverting massive covariance matrices. rank reduction accelerates density functional theory calculations orders magnitude, enabling accelerated simulations. An optimal adaptive sampling algorithm utilized on-the-fly molecular dynamics, extending interatomic potentials through scalable formalism. Through merging quantum mechanics methods, universal SGPR-based create digital-twin capable predicting phenomena arising static dynamic changes as well inherent collective characteristics materials. These techniques have been applied successfully solid electrolytes, lithium-ion batteries, electrocatalysts, solar cells, macromolecular systems, reproducing structures, energetics, properties, phase-changes, performance, device efficiency. discusses built-in library potential, showcasing its applications successes, offering insights into development future in advanced catering both educational expert readers.

Language: Английский

---

Random Sampling Versus Active Learning Algorithms for Machine Learning Potentials of Quantum Liquid Water

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

---

A Perspective on Multiscale Modeling of Explicit Solvation-Enabled Simulations of Catalysis at Liquid–Solid Interfaces

ACS Catalysis, Journal Year: 2025, Volume and Issue: unknown, P. 7448 - 7457

Published: April 21, 2025

Language: Английский

---

Hybrid Quantum Mechanical, Molecular Mechanical, and Machine Learning Potential for Computing Aqueous-Phase Adsorption Free Energies on Metal Surfaces

Journal of Chemical Theory and Computation, Journal Year: 2024, Volume and Issue: unknown

Published: Sept. 10, 2024

Performing reliable computer simulations of elementary processes occurring at metal-water interfaces is pivotal for novel catalyst design in sustainable energy applications. Computational hinges on the ability to reliably and efficiently compute potential surface (PES) system. Due large system sizes needed studying liquid water-metal interfaces, these systems can currently not be described using density functional theory (DFT). In this work, we used a hybrid quantum mechanical, molecular machine learning adsorption behavior phenol, atomic hydrogen, 2-butanol, 2-butanone (0001) facet Ru under reducing conditions when oxidized. Specifically, describe adsorbate surrounding metal atoms DFT level theory. Here, also considered electrostatic field effect water molecules adsorbate-metal interactions. Next, water-water water-adsorbate interactions, established classical force fields. Finally, water-Ru interaction, which no fields have been published, Behler-Parrinello high-dimensional neural network potentials (HDNNPs). Employing setup, our explicit solvation (eSMS) approach aqueous-phase low-coverage selected Ru. agreement with previous experimental computational studies oxygenated over transition facets, found that destabilizes tested adsorbates Ru(0001). Interestingly, findings indicate are less affected by presence an aqueous phase than other metals (e.g., Pt), highlighting necessity investigations Ru-based catalytic water.

Language: Английский

---

Slowly quenched, high pressure glassy B2O3 at DFT accuracy

The Journal of Chemical Physics, Journal Year: 2025, Volume and Issue: 162(4)

Published: Jan. 24, 2025

Modeling inorganic glasses requires an accurate representation of interatomic interactions, large system sizes to allow for intermediate-range structural order, and slow quenching rates eliminate kinetically trapped motifs. Neither first principles-based nor force field-based molecular dynamics (MD) simulations satisfy these three criteria unequivocally. Herein, we report the development a machine learning potential (MLP) classic glass, B2O3, which meets goals well. The MLP is trained on condensed phase configurations whose energies forces atoms are obtained using periodic quantum density functional theory. Deep MD based this accurately predict equation state densification glass with slower from melt. At ambient conditions, larger than 1011 K/s shown lead artifacts in structure. Pressure-dependent x-ray neutron structure factors compare excellently experimental data. High-pressure show varied coordination geometries boron oxygen, concur observations.

Language: Английский

---

Machine learning in molecular biophysics: Protein allostery, multi-level free energy simulations, and lipid phase transitions

Biophysics Reviews, Journal Year: 2025, Volume and Issue: 6(1)

Published: Feb. 12, 2025

Machine learning (ML) techniques have been making major impacts on all areas of science and engineering, including biophysics. In this review, we discuss several applications ML to biophysical problems based our recent research. The topics include the use identify hotspot residues in allosteric proteins using deep mutational scanning data analyze how mutations these hotspots perturb co-operativity framework a statistical thermodynamic model, improve accuracy free energy simulations by integrating from different levels potential functions, determine phase transition temperature lipid membranes. Through examples, illustrate unique value extracting patterns or parameters complex sets, as well remaining limitations. By implementing approaches context physically motivated models computational frameworks, are able gain deeper mechanistic understanding better convergence numerical simulations. We conclude briefly discussing introduced can be further expanded tackle more problems.

Language: Английский

---