Molecular excited states through a machine learning lens

Nature Reviews Chemistry, Journal Year: 2021, Volume and Issue: 5(6), P. 388 - 405

Published: May 20, 2021

Language: Английский

---

Machine Learning of Reactive Potentials

Annual Review of Physical Chemistry, Journal Year: 2024, Volume and Issue: 75(1), P. 371 - 395

Published: June 28, 2024

In the past two decades, machine learning potentials (MLPs) have driven significant developments in chemical, biological, and material sciences. The construction training of MLPs enable fast accurate simulations analysis thermodynamic kinetic properties. This review focuses on application to reaction systems with consideration bond breaking formation. We development MLP models, primarily neural network kernel-based algorithms, recent applications reactive (RMLPs) at different scales. show how RMLPs are constructed, they speed up calculation dynamics, facilitate study trajectories, rates, free energy calculations, many other calculations. Different data sampling strategies applied building also discussed a focus collect structures for rare events further improve their performance active learning.

Language: Английский

---

Excited state non-adiabatic dynamics of large photoswitchable molecules using a chemically transferable machine learning potential

Nature Communications, Journal Year: 2022, Volume and Issue: 13(1)

Published: June 15, 2022

Light-induced chemical processes are ubiquitous in nature and have widespread technological applications. For example, photoisomerization can allow a drug with photo-switchable scaffold such as azobenzene to be activated light. In principle, photoswitches desired photophysical properties like high isomerization quantum yields identified through virtual screening reactive simulations. practice, these simulations rarely used for screening, since they require hundreds of trajectories expensive methods account non-adiabatic excited state effects. Here we introduce diabatic artificial neural network (DANN), based on states, accelerate derivatives. The is six orders magnitude faster than the chemistry method training. DANN transferable molecules outside training set, predicting unseen species that correlated experiment. We use model virtually screen 3100 hypothetical molecules, identify novel predicted yields. predictions confirmed using high-accuracy dynamics. Our results pave way fast accurate photoactive compounds.

Language: Английский

---

A transferable active-learning strategy for reactive molecular force fields

Chemical Science, Journal Year: 2021, Volume and Issue: 12(32), P. 10944 - 10955

Published: Jan. 1, 2021

Predictive molecular simulations require fast, accurate and reactive interatomic potentials. Machine learning offers a promising approach to construct such potentials by fitting energies forces high-level quantum-mechanical data, but doing so typically requires considerable human intervention data volume. Here we show that, leveraging hierarchical active learning, Gaussian Approximation Potential (GAP) models can be developed for diverse chemical systems in an autonomous manner, requiring only hundreds few thousand energy gradient evaluations on reference potential-energy surface. The uses separate intra- inter-molecular fits employs prospective error metric assess the accuracy of We demonstrate applications range with relevance computational organic chemistry: ranging from bulk solvents, solvated metal ion metallocage onwards reactivity, including bifurcating Diels-Alder reaction gas phase non-equilibrium dynamics (a model S

Language: Английский

---

Potential Application of Machine-Learning-Based Quantum Chemical Methods in Environmental Chemistry

Environmental Science & Technology, Journal Year: 2022, Volume and Issue: 56(4), P. 2115 - 2123

Published: Jan. 27, 2022

It is an important topic in environmental sciences to understand the behavior and toxicology of chemical pollutants. Quantum methodologies have served as useful tools for probing pollutants recent decades. In years, machine learning (ML) techniques brought revolutionary developments field quantum chemistry, which may be beneficial investigating However, ML-based methods (ML-QCMs) only scarcely been used studies so far. To promote applications promising methods, this Perspective summarizes progress ML-QCMs focuses on their potential that could hardly achieved by conventional methods. Potential challenges predicting degradation networks pollutants, searching global minima atmospheric nanoclusters, discovering heterogeneous or photochemical transformation pathways well environmentally relevant end points with wave functions descriptors are introduced discussed.

Language: Английский

---

In Silico Chemical Experiments in the Age of AI: From Quantum Chemistry to Machine Learning and Back

Advanced Materials, Journal Year: 2024, Volume and Issue: 36(30)

Published: May 25, 2024

Abstract Computational chemistry is an indispensable tool for understanding molecules and predicting chemical properties. However, traditional computational methods face significant challenges due to the difficulty of solving Schrödinger equations increasing cost with size molecular system. In response, there has been a surge interest in leveraging artificial intelligence (AI) machine learning (ML) techniques silico experiments. Integrating AI ML into increases scalability speed exploration space. remain, particularly regarding reproducibility transferability models. This review highlights evolution from, complementing, or replacing energy property predictions. Starting from models trained entirely on numerical data, journey set forth toward ideal model incorporating physical laws quantum mechanics. paper also reviews existing their intertwining, outlines roadmap future research, identifies areas improvement innovation. Ultimately, goal develop architectures capable accurate transferable solutions equation, thereby revolutionizing experiments within materials science.

Language: Английский

---

Analogies between photochemical reactions and ground-state post-transition-state bifurcations shed light on dynamical origins of selectivity

Nature Chemistry, Journal Year: 2024, Volume and Issue: 16(4), P. 615 - 623

Published: Jan. 12, 2024

Language: Английский

---

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

Journal of Chemical Theory and Computation, Journal Year: 2024, Volume and Issue: 20(12), P. 5043 - 5057

Published: June 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.

Language: Английский

---

Machine learning photodynamics uncover blocked non-radiative mechanisms in aggregation-induced emission

Chem, Journal Year: 2024, Volume and Issue: 10(7), P. 2295 - 2310

Published: May 14, 2024

Language: Английский

---

A digital twin to overcome long-time challenges in photovoltaics

Joule, Journal Year: 2024, Volume and Issue: unknown

Published: Jan. 1, 2024

Language: Английский

---