MLatom Software Ecosystem for Surface Hopping Dynamics in Python with Quantum Mechanical and Machine Learning Methods

Journal of Chemical Theory and Computation, Год журнала: 2024, Номер 20(12), С. 5043 - 5057

Опубликована: Июнь 5, 2024

We present an open-source MLatom@XACS software ecosystem for on-the-fly surface hopping nonadiabatic dynamics based on the Landau–Zener–Belyaev–Lebedev algorithm. The can be performed via Python API with a wide range of quantum mechanical (QM) and machine learning (ML) methods, including ab initio QM (CASSCF ADC(2)), semiempirical methods (e.g., AM1, PM3, OMx, ODMx), many types ML potentials KREG, ANI, MACE). Combinations also used. While user build their own combinations, we provide AIQM1, which is Δ-learning used out-of-the-box. showcase how AIQM1 reproduces isomerization yield trans-azobenzene at low cost. example scripts that, in dozens lines, enable to obtain final population plots by simply providing initial geometry molecule. Thus, those perform optimization, normal mode calculations, condition sampling, parallel trajectories propagation, analysis, result plotting. Given capabilities MLatom training different models, this seamlessly integrated into protocols building models dynamics. In future, deeper more efficient integration Newton-X will vast functionalities dynamics, such as fewest-switches hopping, facilitate similar workflows API.

Язык: Английский

---

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

Journal of Chemical Theory and Computation, Год журнала: 2024, Номер 20(8), С. 2992 - 3007

Опубликована: Апрель 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

Язык: Английский

---

Ag–Bi Charge Redistribution Creates Deep Traps in Defective Cs₂AgBiBr₆: Machine Learning Analysis of Density Functional Theory

The Journal of Physical Chemistry Letters, Год журнала: 2022, Номер 13(16), С. 3645 - 3651

Опубликована: Апрель 18, 2022

Lead-free double perovskites hold promise for stable and environmentally benign solar cells; however, they exhibit low efficiencies because defects act as charge recombination centers. Identifying trap-assisted loss mechanisms developing defect passivation strategies constitute an urgent goal. Applying unsupervised machine learning to density functional theory nonadiabatic molecular dynamics, we demonstrate that negatively charged Br vacancies in Cs2AgBiBr6 create deep hole traps through redistribution between the adjacent Ag Bi atoms. Vacancy electrons are first accepted by then shared with Ag, trap transforms from shallow deep. Subsequent losses promoted motions perpendicular rather than along Ag-Bi axis, can be expected. In contrast, pristine correlates most displacements of Cs atoms Br-Br-Br angles. Doping replace at vacancy maintains keeps shallow.

Язык: Английский

---

Defect modeling and control in structurally and compositionally complex materials

Nature Computational Science, Год журнала: 2023, Номер 3(3), С. 210 - 220

Опубликована: Март 31, 2023

Язык: Английский

---

Nonadiabatic Dynamics with Exact Factorization: Implementation and Assessment

Journal of Chemical Theory and Computation, Год журнала: 2024, Номер 20(12), С. 5022 - 5042

Опубликована: Июнь 5, 2024

In this work, we report our implementation of several independent-trajectory mixed-quantum-classical (ITMQC) nonadiabatic dynamics methods based on exact factorization (XF) in the Libra package for and excited-state dynamics. Namely, surface hopping (SHXF), mixed quantum-classical (MQCXF), mean-field (MFXF) are introduced. Performance these is compared to that traditional schemes, such as fewest-switches (FSSH), branching-corrected (BCSH), simplified decay mixing (SDM), well conventional Ehrenfest (mean-field, MF) method. Based a comprehensive set 1D model Hamiltonians, find ranking SHXF ≈ MQCXF > BCSH SDM FSSH ≫ MF, with sometimes outperforming XF terms describing coherences. Although MFXF method can yield reasonable populations coherences some cases, it does not conserve total energy therefore recommended. We also branching correction auxiliary trajectories important accurate However, worsen quality conservation MQCXF. Finally, using time-dependent Gaussian width approximation used computing decoherence improve The parameter-free scheme Subotnik widths found deliver best performance situations where known priori.

Язык: Английский

---

Designing semiconductor materials and devices in the post-Moore era by tackling computational challenges with data-driven strategies

Nature Computational Science, Год журнала: 2024, Номер 4(5), С. 322 - 333

Опубликована: Май 23, 2024

Язык: Английский

---

Nonadiabatic Field: A Conceptually Novel Approach for Nonadiabatic Quantum Molecular Dynamics

Journal of Chemical Theory and Computation, Год журнала: 2025, Номер unknown

Опубликована: Апрель 7, 2025

Reliable trajectory-based nonadiabatic quantum dynamics methods at the atomic/molecular level are critical for practical understanding and rational design of many important processes in real large/complex systems, where dynamical behavior electrons that nuclei coupled. The paper reports latest progress field (NaF), a conceptually novel approach with independent trajectories. Substantially different from mainstreams Ehrenfest-like surface hopping methods, nuclear force NaF involves arising coupling between electronic states, addition to adiabatic contributed by single state. is capable faithfully describing interplay motion broad regime, which covers relevant states keep coupled wide range or all time bifurcation characteristic essential. derived exact generalized phase space formulation coordinate-momentum variables, constraint (CPS) employed discrete electronic-state degrees freedom (DOFs) infinite Wigner used continuous DOFs. We propose efficient integrators equations both diabatic representations. Since formalism CPS not unique, can principle be implemented various representations correlation function (TCF) time-dependent property. They applied suite representative gas-phase condensed-phase benchmark models numerically results available comparison. It shown relatively insensitive representation TCF will potential tool reliable simulations mechanical transition systems.

Язык: Английский

---

Simulations of molecular photodynamics in long timescales

Philosophical Transactions of the Royal Society A Mathematical Physical and Engineering Sciences, Год журнала: 2022, Номер 380(2223)

Опубликована: Март 28, 2022

Nonadiabatic dynamics simulations in the long timescale (much longer than 10 ps) are next challenge computational photochemistry. This paper delimits scope of what we expect from methods to run such simulations: they should work full nuclear dimensionality, be general enough tackle any type molecule and not require unrealistic resources. We examine main methodological challenges venture advance field, including costs electronic structure calculations, stability integration methods, accuracy nonadiabatic algorithms software optimization. Based on designed shed light each these issues, show how machine learning may a crucial element for time-scale dynamics, either as surrogate calculations or aiding parameterization model Hamiltonians. that conventional integrating classical equations adequate extended up 1 ns surface hopping agrees semiquantitatively with wave packet propagation weak-coupling regime. also describe our optimization Newton-X program reduce overheads data processing storage. article is part theme issue ‘Chemistry without Born–Oppenheimer approximation’.

Язык: Английский

---

A comparative study of different machine learning methods for dissipative quantum dynamics

Machine Learning Science and Technology, Год журнала: 2022, Номер 3(4), С. 045016 - 045016

Опубликована: Окт. 14, 2022

Abstract It has been recently shown that supervised machine learning (ML) algorithms can accurately and efficiently predict long-time population dynamics of dissipative quantum systems given only short-time dynamics. In the present article we benchmarked 22 ML models on their ability to a two-level system linearly coupled harmonic bath. The include uni- bidirectional recurrent, convolutional, fully-connected feedforward artificial neural networks (ANNs) kernel ridge regression (KRR) with linear most commonly used nonlinear kernels. Our results suggest KRR kernels serve as inexpensive yet accurate way simulate in cases where constant length input trajectories is appropriate. Convolutional gated recurrent unit model found be efficient ANN model.

Язык: Английский

---

Libra: A modular software library for quantum nonadiabatic dynamics

Software Impacts, Год журнала: 2022, Номер 14, С. 100445 - 100445

Опубликована: Ноя. 17, 2022

Язык: Английский

---