MLatom 3: A Platform for Machine Learning-Enhanced Computational Chemistry Simulations and Workflows

Journal of Chemical Theory and Computation, Год журнала: 2024, Номер 20(3), С. 1193 - 1213

Опубликована: Янв. 25, 2024

Machine learning (ML) is increasingly becoming a common tool in computational chemistry. At the same time, rapid development of ML methods requires flexible software framework for designing custom workflows. MLatom 3 program package designed to leverage power enhance typical chemistry simulations and create complex This open-source provides plenty choice users who can run with command-line options, input files, or scripts using as Python package, both on their computers online XACS cloud computing service at XACScloud.com. Computational chemists calculate energies thermochemical properties, optimize geometries, molecular quantum dynamics, simulate (ro)vibrational, one-photon UV/vis absorption, two-photon absorption spectra ML, mechanical, combined models. The choose from an extensive library containing pretrained models mechanical approximations such AIQM1 approaching coupled-cluster accuracy. developers build own various algorithms. great flexibility largely due use interfaces many state-of-the-art packages libraries.

Язык: Английский

---

Recommendations for Velocity Adjustment in Surface Hopping

Journal of Chemical Theory and Computation, Год журнала: 2024, Номер 20(2), С. 614 - 624

Опубликована: Янв. 11, 2024

This study investigates velocity adjustment directions after hopping in surface dynamics. Using fulvene and a protonated Schiff base (PSB4) as case studies, we investigate the population decay reaction yields of different sets dynamics with adjusted either nonadiabatic coupling, gradient difference, or momentum directions. For latter, addition to conventional algorithm, investigated performance reduced kinetic energy reservoir approach recently proposed. Our evaluation also considered approximate coupling vectors. While results for are susceptible approach, PSB4 is not. We correlated this dependence topography near conical intersections. When vectors unavailable, difference direction best option. If semiempirical vector becomes an excellent option prevent artificial excess back hoppings. The precise less crucial describing than reservoir's size.

Язык: Английский

---

What Controls the Quality of Photodynamical Simulations? Electronic Structure Versus Nonadiabatic Algorithm

Journal of Chemical Theory and Computation, Год журнала: 2023, Номер 19(22), С. 8273 - 8284

Опубликована: Ноя. 8, 2023

The field of nonadiabatic dynamics has matured over the last decade with a range algorithms and electronic structure methods available at moment. While community currently focuses more on developing benchmarking new algorithms, underlying controls outcome simulations. Yet, electronic-structure sensitivity analysis is typically neglected. In this work, we present cyclopropanone to algorithms. particular, compare wave function-based CASSCF, FOMO-CASCI, MS- XMS-CASPT2, density-functional REKS, semiempirical MRCI-OM3 Landau-Zener surface hopping, fewest switches ab initio multiple spawning informed stochastic selection results clearly demonstrate that choice significantly influences accuracy for even when potential energy surfaces exhibit qualitative quantitative similarities. Thus, selecting solely basis mapping can be misleading. Conversely, observe no discernible differences in performance across various methods. Based above results, discuss present-day practice computational photodynamics.

Язык: Английский

---

Prediction Challenge: Simulating Rydberg photoexcited cyclobutanone with surface hopping dynamics based on different electronic structure methods

The Journal of Chemical Physics, Год журнала: 2024, Номер 160(15)

Опубликована: Апрель 16, 2024

This research examines the nonadiabatic dynamics of cyclobutanone after excitation into n → 3s Rydberg S2 state. It stems from our contribution to Special Topic Journal Chemical Physics test predictive capability computational chemistry against unseen experimental data. Decoherence-corrected fewest-switches surface hopping was used simulate with full and approximated couplings. Several simulation sets were computed different electronic structure methods, including a multiconfigurational wavefunction [multiconfigurational self-consistent field (MCSCF)] specially built describe dissociative channels, multireference semiempirical approach, time-dependent density functional theory, algebraic diagrammatic construction, coupled cluster. MCSCF predicts slow deactivation state (10 ps), followed by an ultrafast population transfer S1 S0 (<100 fs). CO elimination (C3 channel) dominates over C2H4 formation (C2 channel). These findings radically differ other which predicted lifetimes 10-250 times shorter C2 channel predominance. results suggest that routine methods may hold low power for outcome dynamics.

Язык: Английский

---

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

---

Effect of Initial Conditions Sampling on Surface Hopping Simulations in the Ultrashort and Picosecond Time Range. Azomethane Photodissociation as a Case Study

Journal of Chemical Theory and Computation, Год журнала: 2023, Номер 19(9), С. 2430 - 2445

Опубликована: Апрель 18, 2023

We tested the effect of different ways sampling initial conditions in surface hopping simulations, with a focus on energy distributions and treatment zero point (ZPE). As test case, we chose gas phase photodynamics azomethane, which features processes occurring overlapping time scales: geometry relaxation excited state, internal conversion, photoisomerization, fast slow dissociation. The based semiempirical method, had sufficiently long duration (10 ps) to encompass all above processes. several variants methods quantum mechanical (QM) nuclear coordinates q momenta p, yield, at least average over large set, correct QM energy, namely ZPE when starting from ground vibrational state. compared samplings classical Boltzmann (CB) distribution obtained by thermostated trajectory, whereby thermal effects are taken into account, but is utterly ignored. found that most CB approaches yield similar results as short dynamics decay lifetimes, whereas rate state dissociation reaction CH3NNCH3 → CH3NN + CH3 sharply affected method. With fraction trajectories dissociate promply (<1 after rates order 10-1 ps-1 first ps. Instead, much smaller prompt dissociations lower times. provided evidence "leaks" high frequency modes reactive ones (N-C bond elongations), therefore unphysically increasing samplings. show an effective way take account avoid "leaking" problem add potential surfaces function relevant coordinates. Then, can be done usual, so this approach suitable also for condensed dynamics. In tests present here, correction method yields intermediate between uncorrected

Язык: Английский

---

Surface hopping modeling of charge and energy transfer in active environments

Physical Chemistry Chemical Physics, Год журнала: 2023, Номер 25(12), С. 8293 - 8316

Опубликована: Янв. 1, 2023

An active environment changes the photophysics of an excited chromophore. Surface hopping simulations are crucial to understanding how.

Язык: Английский

---

Generalized Semiclassical Ehrenfest Method: A Route to Wave Function-Free Photochemistry and Nonadiabatic Dynamics with Only Potential Energies and Gradients

Journal of Chemical Theory and Computation, Год журнала: 2024, Номер 20(11), С. 4396 - 4426

Опубликована: Май 31, 2024

We reconsider recent methods by which direct dynamics calculations of electronically nonadiabatic processes can be carried out while requiring only adiabatic potential energies and their gradients. show that these understood in terms a new generalization the well-known semiclassical Ehrenfest method. This is convenient because it eliminates need to evaluate electronic wave functions matrix elements along mixed quantum-classical trajectories. The approximations procedures enabling this advance are curvature-driven approximation time-derivative coupling, generalized method, gradient correction scheme called (TDM) scheme. When spin–orbit coupling present, one carry fully basis using gradients calculated without plus elements. Even when neglected, method useful allows structure for vectors unavailable. In order place considerations context, article starts with review background material on trajectory surface hopping, scheme, incorporating decoherence. consider both internal conversion intersystem crossing. also several examples from our group successful applications approximation.

Язык: Английский

---

AI in computational chemistry through the lens of a decade-long journey

Chemical Communications, Год журнала: 2024, Номер 60(24), С. 3240 - 3258

Опубликована: Янв. 1, 2024

This article gives a perspective on the progress of AI tools in computational chemistry through lens author's decade-long contributions put wider context trends this rapidly expanding field. over last decade is tremendous: while ago we had glimpse what was to come many proof-of-concept studies, now witness emergence AI-based that are mature enough make faster and more accurate simulations increasingly routine. Such turn allow us validate even revise experimental results, deepen our understanding physicochemical processes nature, design better materials, devices, drugs. The rapid introduction powerful rise unique challenges opportunities discussed too.

Язык: Английский

---

Prediction challenge: First principles simulation of the ultrafast electron diffraction spectrum of cyclobutanone

The Journal of Chemical Physics, Год журнала: 2024, Номер 160(13)

Опубликована: Апрель 4, 2024

Computer simulation has long been an essential partner of ultrafast experiments, allowing the assignment microscopic mechanistic detail to low-dimensional spectroscopic data. However, ability theory make a priori predictions experimental results is relatively untested. Herein, as part community challenge, we attempt predict signal upcoming photochemical experiment using state-of-the-art in context preexisting Specifically, employ ab initio Ehrenfest with collapse block mixed quantum-classical simulations describe real-time evolution electrons and nuclei cyclobutanone following excitation 3s Rydberg state. The gas-phase electron diffraction (GUED) simulated for direct comparison at Stanford Linear Accelerator Laboratory. Following initial ring-opening, dissociation via two distinct channels observed: C3 channel, producing cyclopropane CO, C2 CH2CO C2H4. Direct calculations GUED indicate how ring-opened intermediate, products, products can be discriminated signal. We also report analysis anticipated errors our predictions: without knowledge result, which features spectrum do feel confident have predicted correctly, might wrong?

Язык: Английский

---