Active learning of neural network potentials for rare events

Digital Discovery, Journal Year: 2024, Volume and Issue: 3(3), P. 514 - 527

Published: Jan. 1, 2024

Developing an automated active learning framework for Neural Network Potentials, focusing on accurately simulating bond-breaking in hexane chains through steered molecular dynamics sampling and assessing model transferability.

Language: Английский

---

Machine-Learned Kohn–Sham Hamiltonian Mapping for Nonadiabatic Molecular Dynamics

Journal of Chemical Theory and Computation, Journal Year: 2024, Volume and Issue: 20(8), P. 2992 - 3007

Published: April 6, 2024

In this work, we report a simple, efficient, and scalable machine-learning (ML) approach for mapping non-self-consistent Kohn-Sham Hamiltonians constructed with one kind of density functional to the nearly self-consistent another functional. This is designed as fast surrogate Hamiltonian calculator use in long nonadiabatic dynamics simulations large atomistic systems. approach, input output features are matrices computed from different levels theory. We demonstrate that developed ML-based method (1) speeds up calculations by several orders magnitude, (2) conceptually simpler than alternative ML approaches, (3) applicable systems sizes can be used arbitrary functionals, (4) requires modest training data, learns fast, generates molecular orbitals their energies accuracy matching conventional calculations, (5) when applied simulation excitation energy relaxation yields corresponding time scales within margin error calculations. Using explore C

Language: Английский

---

The emergence of machine learning force fields in drug design

Medicinal Research Reviews, Journal Year: 2024, Volume and Issue: 44(3), P. 1147 - 1182

Published: Jan. 3, 2024

In the field of molecular simulation for drug design, traditional mechanic force fields and quantum chemical theories have been instrumental but limited in terms scalability computational efficiency. To overcome these limitations, machine learning (MLFFs) emerged as a powerful tool capable balancing accuracy with MLFFs rely on relationship between structures potential energy, bypassing need preconceived notion interaction representations. Their depends models used, quality volume training data sets. With recent advances equivariant neural networks high-quality datasets, significantly improved their performance. This review explores MLFFs, emphasizing design. It elucidates MLFF principles, provides development validation guidelines, highlights successful implementations. also addresses challenges developing applying MLFFs. The concludes by illuminating path ahead outlining to be opportunities harnessed. inspires researchers embrace investigations new perform simulations

Language: Английский

---

No Headache for PIPs: A PIP Potential for Aspirin Runs Much Faster and with Similar Precision Than Other Machine-Learned Potentials

Journal of Chemical Theory and Computation, Journal Year: 2024, Volume and Issue: 20(8), P. 3008 - 3018

Published: April 9, 2024

Assessments of machine-learning (ML) potentials are an important aspect the rapid development this field. We recently reported assessment linear-regression permutationally invariant polynomial (PIP) method for ethanol, using widely used (revised) rMD17 data set. demonstrated that PIP approach outperformed numerous other methods, e.g., ANI, PhysNet, sGDML, and p-KRR, with respect to precision notably speed [Houston et al., J. Chem. Phys. 2022, 156, 044120]. Here, we extend 21-atom aspirin molecule, set, a focus on evaluation. Both energies forces training, several PIPs is examined both. Normal mode frequencies, methyl torsional potential, 1d vibrational OH stretch presented. show achieves level obtained from ML atom-centered neural network linear regression ACE, kernel as by Kovács al. in Theory Comput. 2021, 17, 7696–7711. More significantly, PESs run much faster than all whose timings were evaluated paper. also PES extrapolates well enough describe internal motions aspirin, including stretch.

Language: Английский

---

Unified Approach to Generating a Training Set for Machine Learning Interatomic Potentials: The Case of BCC Tungsten

Materials Today Communications, Journal Year: 2025, Volume and Issue: unknown, P. 111437 - 111437

Published: Jan. 1, 2025

Language: Английский

---