An overview about neural networks potentials in molecular dynamics simulation

International Journal of Quantum Chemistry, Journal Year: 2024, Volume and Issue: 124(11)

Published: May 21, 2024

Abstract Ab‐initio molecular dynamics (AIMD) is a key method for realistic simulation of complex atomistic systems and processes in nanoscale. In AIMD, finite‐temperature dynamical trajectories are generated by using forces computed from electronic structure calculations. with high numbers components typical AIMD run computationally demanding. On the other hand, machine learning (ML) subfield artificial intelligence that consist set algorithms show experience use input output data where capable analysing predicting future. At present, main application ML techniques atomic simulations development new interatomic potentials to correctly describe potential energy surfaces (PES). This technique constant progress since its inception around 30 years ago. The combine advantages classical methods, is, efficiency simple functional form accuracy first principles this article we review evolution four generations some their most notable applications. focuses on MLPs based neural networks. Also, present state art topic future trends. Finally, report results scientometric study (covering period 1995–2023) about impact applied simulations, distribution publications geographical regions hot topics investigated literature.

Language: Английский

---

A review of displacement cascade simulations using molecular dynamics emphasizing interatomic potentials for TPBAR components

npj Materials Degradation, Journal Year: 2025, Volume and Issue: 9(1)

Published: Jan. 2, 2025

Abstract This review explores molecular dynamics simulations for studying radiation damage in Tritium Producing Burnable Absorber Rod (TPBAR) materials, emphasizing the role of interatomic potentials displacement cascades. Recent machine learning (MLPs), trained on quantum data, enhance prediction accuracy over traditional models like EAM. We highlight temperature, PKA energy, and composition effects evolution TPBAR components, recommending suitable discussing advancements materials extreme environments.

Language: Английский

---

Equivariant graph neural networks for fast electron density estimation of molecules, liquids, and solids

npj Computational Materials, Journal Year: 2022, Volume and Issue: 8(1)

Published: Aug. 23, 2022

Electron density $\rho(\vec{r})$ is the fundamental variable in calculation of ground state energy with functional theory (DFT). Beyond total energy, features and changes distributions are often used to capture critical physicochemical phenomena materials. We present a machine learning framework for prediction $\rho(\vec{r})$. The model based on equivariant graph neural networks electron predicted at special query point vertices that part message passing graph, but only receive messages. tested across multiple data sets molecules (QM9), liquid ethylene carbonate electrolyte (EC) LixNiyMnzCo(1-y-z)O2 lithium ion battery cathodes (NMC). For QM9 molecules, accuracy proposed exceeds typical variability obtained from DFT done different exchange-correlation functionals. all three datasets beyond art computation time orders magnitude faster than DFT.

Language: Английский

---

Thermodynamics and dielectric response of BaTiO3 by data-driven modeling

npj Computational Materials, Journal Year: 2022, Volume and Issue: 8(1)

Published: Sept. 29, 2022

Abstract Modeling ferroelectric materials from first principles is one of the successes density-functional theory and driver much development effort, requiring an accurate description electronic processes thermodynamic equilibrium that drive spontaneous symmetry breaking emergence macroscopic polarization. We demonstrate application integrated machine learning model describes on same footing structural, energetic, functional properties barium titanate (BaTiO 3 ), a prototypical ferroelectric. The uses ab initio calculations as reference achieves yet inexpensive predictions energy polarization time length scales are not accessible to direct modeling. These allow us assess microscopic mechanism transition. presence order-disorder transition for Ti off-centered states main transition, even though coupling between cell distortions determines intermediate, partly-ordered phases. Moreover, we thoroughly probe static dynamical behavior BaTiO across its phase diagram without need introduce coarse-grained Finally, apply calculate dielectric response material in full manner, again reproducing correct qualitative experimental behavior.

Language: Английский

---

Taking advantage of noise in quantum reservoir computing

Scientific Reports, Journal Year: 2023, Volume and Issue: 13(1)

Published: May 31, 2023

The biggest challenge that quantum computing and machine learning are currently facing is the presence of noise in devices. As a result, big efforts have been put into correcting or mitigating induced errors. But, can these two fields benefit from noise? Surprisingly, we demonstrate under some circumstances, be used to improve performance reservoir computing, prominent recent algorithm. Our results show amplitude damping beneficial learning, while depolarizing phase noises should prioritized for correction. This critical result sheds new light physical mechanisms underlying devices, providing solid practical prescriptions successful implementation information processing nowadays hardware.

Language: Английский

---

Machine Learning Interatomic Potentials for Reactive Hydrogen Dynamics at Metal Surfaces Based on Iterative Refinement of Reaction Probabilities

The Journal of Physical Chemistry C, Journal Year: 2023, Volume and Issue: 127(50), P. 24168 - 24182

Published: Dec. 4, 2023

The reactive chemistry of molecular hydrogen at surfaces, notably dissociative sticking and evolution, plays a crucial role in energy storage fuel cells. Theoretical studies can help to decipher underlying mechanisms reaction design, but studying dynamics surfaces is computationally challenging due the complex electronic structure interfaces high sensitivity barriers. In addition, ab initio dynamics, based on density functional theory, too demanding accurately predict or desorption probabilities, as it requires averaging over tens thousands initial conditions. High-dimensional machine learning-based interatomic potentials are starting be more commonly used gas-surface yet robust approaches generate reliable training data assess how model uncertainty affects prediction dynamic observables not well established. Here, we employ ensemble learning adaptively while assessing performance with full quantification (UQ) for probabilities scattering different copper facets. We use this approach investigate two message-passing neural networks, SchNet PaiNN. Ensemble-based UQ iterative refinement allow us expose shortcomings invariant pairwise-distance-based feature representation dynamics.

Language: Английский

---

A deep equivariant neural network approach for efficient hybrid density functional calculations

Nature Communications, Journal Year: 2024, Volume and Issue: 15(1)

Published: Oct. 11, 2024

Hybrid density functional calculations are essential for accurate description of electronic structure, yet their widespread use is restricted by the substantial computational cost. Here we develop DeepH-hybrid, a deep equivariant neural network method learning hybrid-functional Hamiltonian as function material which circumvents time-consuming self-consistent field iterations and enables study large-scale materials with accuracy. Our extensive experiments demonstrate good reliability well effective transferability efficiency method. As notable application, DeepH-hybrid applied to large-supercell Moiré-twisted materials, offering first case on how inclusion exact exchange affects flat bands in magic-angle twisted bilayer graphene. The work generalizes deep-learning structure methods beyond conventional theory, facilitating development deep-learning-based ab initio methods. functionals crucial calculations, application limited Here, authors overcome this bottleneck through learning, enabling hybrid calculations.

Language: Английский

---

Applications of machine‐learning interatomic potentials for modeling ceramics, glass, and electrolytes: A review

Journal of the American Ceramic Society, Journal Year: 2024, Volume and Issue: unknown

Published: June 9, 2024

Abstract The emergence of artificial intelligence has provided efficient methodologies to pursue innovative findings in material science. Over the past two decades, machine‐learning potential (MLP) emerged as an alternative technology density functional theory (DFT) and classical molecular dynamics (CMD) simulations for computational modeling materials estimation their properties. MLP offers more computation compared DFT, while providing higher accuracy CMD. This enables us conduct realistic using models with atoms longer simulation times. Indeed, number research studies utilizing MLPs significantly increased since 2015, covering a broad range structures, ranging from simple complex, well various chemical physical phenomena. As result, there are high expectations further applications field science industrial development. review aims summarize applications, particularly ceramics glass science, fundamental theories facilitate future progress utilization. Finally, we provide summary discuss perspectives on next challenges development application MLPs.

Language: Английский

---

Force-field-enhanced neural network interactions: from local equivariant embedding to atom-in-molecule properties and long-range effects

Chemical Science, Journal Year: 2023, Volume and Issue: 14(44), P. 12554 - 12569

Published: Jan. 1, 2023

We introduce FENNIX (Force-Field-Enhanced Neural Network InteraXions), a hybrid approach between machine-learning and force-fields.

Language: Английский

---

Exploration of the Two-Electron Excitation Space with Data-Driven Coupled Cluster

The Journal of Physical Chemistry A, Journal Year: 2024, Volume and Issue: 128(10), P. 1938 - 1947

Published: Feb. 29, 2024

Computational cost limits the applicability of post-Hartree–Fock methods such as coupled-cluster on larger molecular systems. The data-driven (DDCC) method applies machine learning to predict two-electron amplitudes (t2) using data from second-order perturbation theory (MP2). One major limitation DDCC models is size training sets that increases exponentially with system size. Effective sampling amplitude space can resolve this issue. Five different selection techniques reduce amount used for were evaluated, an approach also prevents model overfitting and portability singles doubles more complex molecules or basis sets. In combination a localized orbital formalism CCSD t2 amplitudes, we have achieved 10-fold error reduction energy calculations.

Language: Английский

---