Recently, a novel diabatization scheme has been proposed [Shu, Y.; Truhlar, D. G. J. Chem. Theory Comput. 2020, 16, 6456-6464] using artificial neural networks. Most importantly, the method almost exclusively requires the knowledge of adiabatic energies, which are routinely obtained from ab initio calculations. However, many questions related to the favorable performance of the method remain unanswered. In the present paper, some of these questions are considered for selected one-dimensional models with one configurational variable. In particular, various activation functions are tested, including nonlinear ones in the output layer, the effect of the regularization term in the loss function is analyzed, and computationally cheap extensions of training sets are proposed. Significant improvements of the performance of the original method have been achieved.

Department of Applied Mathematics Faculty of Electrical Engineering and Computer Science VŠB─Technical University of Ostrava 17 listopadu 2172 15 708 00 Ostrava Poruba Czech Republic

Université de Toulouse CNRS UMR5626 Laboratoire de Chimie et Physique Quantiques 118 Route de Narbonne 31062 Toulouse Cedex 09 France

Zobrazit více v PubMed

Tully, J. C. Modern Methods for Multidimensional Dynamics Computations in Chemistry; Thompson, D. L. , Ed.; World Scientific: Singapore, 1998; pp 34–72.

Shu Y., Varga Z., Kanchanakungwankul S., Zhang L., Truhlar D. G.. Diabatic States of Molecules. J. Phys. Chem. A. 2022;126:992–1018. doi: 10.1021/acs.jpca.1c10583. PubMed DOI

Behler J.. Constructing High-Dimensional Neural Network Potentials: A Tutorial Review. Int. J. Quantum Chem. 2015;115:1032–1050. doi: 10.1002/qua.24890. DOI

Unke O. T., Koner D., Patra S., Käser S., Meuwly M.. Machine learning and excited-state molecular dynamics. Mach. Learn. Sci. Technol. 2020;1:013001. doi: 10.1088/2632-2153/ab5922. DOI

Westermayr J., Marquetand P.. Machine learning and excited-state molecular dynamics. Mach. Learn. Sci. Technol. 2020;1:043001. doi: 10.1088/2632-2153/ab9c3e. DOI

Lenzen T., Manthe U.. Neural network based coupled diabatic potential energy surfaces for reactive scattering. J. Chem. Phys. 2017;147:084105. doi: 10.1063/1.4997995. PubMed DOI

Williams D. M. G., Eisfeld W.. Neural network diabatization: A new ansatz for accurate high-dimensional coupled potential energy surfaces. J. Chem. Phys. 2018;149:204106. doi: 10.1063/1.5053664. PubMed DOI

Li C., Hou S., Xie C.. Three-dimensional diabatic potential energy surfaces of thiophenol with neural networks. Chin. J. Chem. Phys. 2021;34:825–832. doi: 10.1063/1674-0068/cjcp2110196. DOI

Hou S., Li C., Han H., Xie C.. EOM-CCSD-based neural network diabatic potential energy matrix for 1πσ*-mediated photodissociation of thiophenol. Chin. J. Chem. Phys. 2022;35:461–470. doi: 10.1063/1674-0068/cjcp2201016. DOI

Zhao X., Xie G., Li W., Zhao Y., Yang C.. New diabatic potential energy surfaces of KH2 system constructed using neural network method. Mol. Phys. 2024;122:e2287685. doi: 10.1080/00268976.2023.2287685. DOI

Yuan J., He D., Wang S., Chen M., Han K.. Diabatic potential energy surfaces of MgH2 + and dynamic studies for the Mg+(3p) + H2– – >MgH+ + H reaction. Phys. Chem. Chem. Phys. 2018;20:6638–6647. doi: 10.1039/C7CP08679B. PubMed DOI

Xie C., Zhu X., Yarkony D. R., Guo H.. Permutation invariant polynomial neural network approach to fitting potential energy surfaces. IV. Coupled diabatic potential energy matrices. J. Chem. Phys. 2018;149:144107. doi: 10.1063/1.5054310. PubMed DOI

Guan Y., Fu B., Zhang D. H.. Construction of diabatic energy surfaces for LiFH with artificial neural networks. J. Chem. Phys. 2017;147:224307. doi: 10.1063/1.5007031. PubMed DOI

Yin Z., Guan Y., Fu B., Zhang D. H.. Two-state diabatic potential energy surfaces of ClH2 based on nonadiabatic couplings with neural networks. Phys. Chem. Chem. Phys. 2019;21:20372–20383. doi: 10.1039/C9CP03592C. PubMed DOI

Guan Y., Zhang D. H., Guo H., Yarkony D. R.. Representation of coupled adiabatic potential energy surfaces using neural network based quasi-diabatic Hamiltonians: 1,2 2A’ states of LiFH. Phys. Chem. Chem. Phys. 2019;21:14205–14213. doi: 10.1039/C8CP06598E. PubMed DOI

Sršeň Š., von Lillienfeld O. A., Slavíček P.. Fast and accurate excited states predictions: machine learning and diabatization. Phys. Chem. Chem. Phys. 2024;26:4306–4319. doi: 10.1039/D3CP05685F. PubMed DOI PMC

Shu Y., Truhlar D. G.. Diabatization by Machine Intelligence. J. Chem. Theory Comput. 2020;16:6456–6464. doi: 10.1021/acs.jctc.0c00623. PubMed DOI

Varga Z., Shu Y., Ning J., Truhlar D. G.. Diabatic potential energy surfaces and semiclassical multi-state dynamics for fourteen coupled 3A’ states of O3 . Electron. Struct. 2022;4:047002. doi: 10.1088/2516-1075/ac94ac. DOI

Akher, F. B. ; Shu, Y. ; Varga, Z. ; Bhaumik, S. ; Truhlar, D. G. . Post-Activation Function for Managing the Behavior of a Neural Network Potential with a Low-Dimensional Potential. 2023. PubMed

Shu Y., Akher F. B., Guo H., Truhlar D. G.. Parametrically Managed Activation Functions for Improved Global Potential Energy Surfaces for Six Coupled 5A’ States and Fourteen Coupled 3A’ States of O + O2 . J. Phys. Chem. A. 2024;128:1207–1217. doi: 10.1021/acs.jpca.3c06823. PubMed DOI

Shu Y., Varga Z., de Oliveira-Filho A. G. S., Truhlar D. G.. Permutationally Restrained Diabatization by Machine Intelligence. J. Chem. Theory Comput. 2021;17:1106–1116. doi: 10.1021/acs.jctc.0c01110. PubMed DOI

Tikhonov A. N.. Solution of incorrectly formulated problems and the regularization method. Sov. Dok. 1963;4:1035–1038.

Tikhonov, A. N. ; Arsenin, V. Y. . Solutions of Ill-Posed Problems; John Wiley: New York, 1977.

Levenberg K.. A Method for the Solution of Certain Non-Linear Problems in Least Squares. Q. Appl. Math. 1944;2:164–168. doi: 10.1090/qam/10666. DOI

Marquardt D. W.. An Algorithm for Least-Squares Estimation of Nonlinear Parameters. J. Soc. Ind. Appl. Math. 1963;11:431–441. doi: 10.1137/0111030. DOI

Glorot, X. ; Bengio, Y. . Understanding the difficulty of training deep feedforward neural networks. Proceedings of the 13

Hendrycks, D. ; Gimpel, K. ; Gaussian, E. . Gaussian Error Linear Units (GELUs). 2023, arXiv:1606.08415. arXiv.org e-Printarchive. https://arxiv.org/abs/1606.08415.

Truhlar, D. G. Private Communication. 2025.

Zhang L., Truhlar D. G., Sun S.. Full-dimensional three-state potential energy surfaces and state couplings for photodissociation of thiophenol. J. Chem. Phys. 2019;151:154306. doi: 10.1063/1.5124870. PubMed DOI