Communication: An accurate full 15 dimensional permutationally invariant potential energy surface for the OH + CH4 → H2O + CH3 reaction

Li, Jun; Guo, Hua

doi:10.1063/1.4937570

Cited by 69 publications

(92 citation statements)

References 40 publications

(63 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Another quality control besides the propertybased one proposed by Behler can be obtained by comparing the molecular structures at each time step as done by Dral et al 138,482 and Ceriotti et al 481 A combination of a structurebased and property-based detection of sparsely sampled regions of the PESs has been done by Zhang et al and Guo et al 362,524,[544][545][546] Very recently, an alternative approach has been applied with NNs by Lin et al 402 that does not require MD simulations. It is based on the finding that the negative of the squared difference surface obtained from NNs approaches zero in regions, where no data points are available.…”

Section: Initial Training Setmentioning

confidence: 99%

Machine learning for electronically excited states of molecules

Westermayr,

Marquetand

2020

Preprint

View full text Add to dashboard Cite

Electronically excited states of molecules are at the heart of photochemistry, photophysics, as well as photobiology and also play a role in material science. Their theoretical description requires highly accurate quantum chemical calculations, which are computationally expensive. In this review, we focus on how machine learning is employed not only to speed up such excited-state simulations but also how this branch of artificial intelligence can be used to advance this exciting research field in all its aspects. Discussed applications of machine learning for excited states include excited-state dynamics simulations, static calculations of absorption spectra, as well as many others. In order to put these studies into context, we discuss the promises and pitfalls of the involved machine learning techniques. Since the latter are mostly based on quantum chemistry calculations, we also provide a short introduction into excited-state electronic structure methods, approaches for nonadiabatic dynamics simulations and describe tricks and problems when using them in machine learning for excited states of molecules.

show abstract

Section: Initial Training Setmentioning

confidence: 99%

Machine learning for electronically excited states of molecules

Westermayr,

Marquetand

2020

Preprint

View full text Add to dashboard Cite

show abstract

“…Since the PES-2000 and PES-2014 were calculated based on a simple functional form, the dynamical uncertainties result from the two PESs remain unclear. Very recently, Li and Guo developed a new global full-dimensional PES using the PIP-NN approach based on large number of high-level ab initio points 279 . This PES is considered globally accurate based on the small RMS fitting error and preliminary QCT results.…”

Section: Oh+ch4 → H2o+ch3mentioning

confidence: 99%

Recent advances in quantum scattering calculations on polyatomic bimolecular reactions

Xiao

Zhang

et al. 2017

Chem. Soc. Rev.

View full text Add to dashboard Cite

This review surveys quantum scattering calculations on chemical reactions of polyatomic molecules in the gas phase published in the last ten years. These calculations are useful because they provide highly accurate information on the dynamics of chemical reactions which can be compared in detail with experimental results. They also serve as quantum mechanical benchmarks for testing approximate theories which can more readily be applied to more complicated reactions. This review includes theories for calculating quantities such as rate constants which have many important scientific applications.

show abstract

“…39 Thanks to the universal approximator NN, the use of the PIPs, nonredundant PIPs, or FIs as the NN input can guarantee not only the high fidelity but also the rigorous symmetry property of the molecular system with respect to the permutation of like atoms. Recent reported examples include F/Cl/H + CH3OH, [40][41][42] OH +CH4, 43 etc., for each more than 100000 UCCSD(T)-F12a/AVTZ energies were computed. The root-mean squared error (RMSE) is generally smaller than 0.23 kcal mol -1 (10 meV), namely, well within the highly coveted chemical accuracy, 1 kcal mol -1 .…”

Section: Introductionmentioning

confidence: 99%

Permutation-Invariant-Polynomial Neural-Network-based Δ-Machine Learn-ing Approach: A Case for the HO2 Self-reaction and its Dynamics Study

Liu

2022

Preprint

Self Cite

View full text Add to dashboard Cite

The potential energy surface (PES) plays a central role in chemistry. As the size of the reaction system increases, it would be more and more difficult to develop its globally accurate full-dimensional PES. One unavoidable difficulty is that it is too expensive to calculate electronic energies of ample configurations for complicated reactions. Δ-machine learning is a highly cost-effective method as only a small number of high-level ab initio energies are required to improve a potential energy surface (PES) fit to a large number of low-level points. Here, we propose a permutation-invariant-polynomial neural-network (PIP-NN)-based Δ-machine learning approach to construct full-dimensional accurate PESs for complicated reactions. The approach is applied to the HO2 + HO2 → H2O2 + O2 reaction, a key process in combustion and atmosphere. The full-dimensional triplet state PES is constructed with a large number of density functional theory (DFT) points, which cover all dynamically relevant regions. Only 14% of the DFT dataset are used to successfully bring the DFT PES to the UCCSD(T)-F12a/AVTZ quality. On this PES of high quality, quasi-classical trajectory (QCT) calculations are performed to study the dynamics of the title reaction. A surprising mode-speciﬁc dynamics is observed, in which exciting a spectator mode leads to significant enhancement of the reactivity at low collision energy. This special mechanism can be attributed to increased attraction potential caused by the excited spectator mode. Such mode specificity may be quite prevalent in free radical reactions involving HO2, which is common in combustion and atmosphere.

show abstract

Communication: An accurate full 15 dimensional permutationally invariant potential energy surface for the OH + CH4 → H2O + CH3 reaction

Abstract: Mode specificity in the OH + CHD 3 reaction: Reduced-dimensional quantum and quasi-classical studies on an ab initio based full-dimensional potential energy surface

Cited by 69 publications

References 40 publications

Machine learning for electronically excited states of molecules

Machine learning for electronically excited states of molecules

Recent advances in quantum scattering calculations on polyatomic bimolecular reactions

Permutation-Invariant-Polynomial Neural-Network-based Δ-Machine Learn-ing Approach: A Case for the HO2 Self-reaction and its Dynamics Study

Contact Info

Product

Resources

About