Beyond MD17: the reactive xxMD dataset

System specific neural force fields (NFFs) have gained popularity in computational chemistry. One of the most popular datasets as a bencharmk to develop NFFs models is the MD17 dataset and its subsequent extension. These datasets comprise geometries from the equilibrium region of the ground electronic state potential energy surface, sampled from direct adiabatic dynamics. However, many chemical reactions involve significant molecular geometrical deformations, for example, bond breaking. Therefore, MD17 is inadequate to represent a chemical reaction. To address this limitation in MD17, we introduce a new dataset, called Extended Excited-state Molecular Dynamics (xxMD) dataset. The xxMD dataset involves geometries sampled from direct non-adiabatic dynamics, and the energies are computed at both multireference wavefunction theory and density functional theory. We show that the xxMD dataset involves diverse geometries which represent chemical reactions. Assessment of NFF models on xxMD dataset reveals significantly higher predictive errors than those reported for MD17 and its variants. This work underscores the challenges faced in crafting a generalizable NFF model with extrapolation capability.

Note:

19 pages, many figures. Data available at https://github.com/zpengmei/xxMD

References(37)

Figures(10)

[1]

Machine learning of accurate energy-conserving molecular force fields. Sci. advances 3, e1603015

S. Chmiela

[2]

On the role of gradients for machine learning of molecular energies and forces

A.S. Christensen
,
O.A. Von Lilienfeld

- Mach.Learn.Sci.Tech. 1 (2020) 045018

[3]

Trajectory surface hopping approach to nonadiabatic molecular collisions: The reaction of h+ with d2. The J

J.C. Tully
,
R.K. Preston

- Chem.Phys. 55 (1971) 562-572

[4]

Trajectory-surface-hopping study of Na(3p 2P) + H2 -> Na(3s 2S) + H2(v’, j’, θ). The J

N.C. Blais
,
D.G. Truhlar

- Chem.Phys. 79 (1983) 1334-1342

[5]

Nonadiabatic semiclassical scattering. i. analysis of generalized surface hopping procedures. The J

M.F. Herman

- Chem.Phys. 81 (1984) 754-763

[6]

Molecular dynamics with electronic transitions. The

J.C. Tully

- J.Chem.Phys. 93 (1990) 1061-1071

[7]

Diabolical conical intersections

David R. Yarkony
(
- Johns Hopkins U.
)

- Rev.Mod.Phys. 68 (1996) 985-1013
•
DOI:
- 10.1103/RevModPhys.68.985

[8]

Conical intersections at the nanoscale: Molecular ideas for materials

B.G. Levine

- Ann.Rev.Phys.Chem. 70 (2019) 21

[9]

Pinheiro Jr, M

S. Zhang
,
P.O. Dral
,
M. Barbatti

- Sci.Data 10 (2023) 95

[10]

E (3)-equivariant graph neural networks for data-efficient and accurate interatomic potentials

S. Batzner

- Nature Commun. 13 (2022) 2453

[11]

Mace: Higher order equivariant message passing neural networks for fast and accurate force fields

I. Batatia
,
D.P. Kovacs
,
G. Simm
,
C. Ortner
,
G. Csányi

- Adv.Neural Inf.Process.Syst. 35 (2022) 11423-11436

[12]

The design space of e (3)-equivariant atom-centered interatomic potentials. arXiv preprint

I. Batatia

•
e-Print:
- 2205.06643

[13]

Equivariant message passing for the prediction of tensorial properties and molecular spectra. In

K. Schütt
,
O. Unke
,
M. Gastegger

- Int.Conf.Machine Learning 9377 (2021) 9388

[14]

Spherical message passing for 3d graph networks. arXiv preprint

Y. Liu

•
e-Print:
- 2102.05013

[15]

complete active space scf method (casscf) using a density matrix formulated super-ci approach

B.O. Roos
,
P.R. Taylor
,
P.E.M.A Sigbahn

- Chem.Phys. 48 (1980) 157

[16]

The m06 suite of density functionals for main group thermochemistry, thermochemical kinetics, noncovalent interactions, excited states, and transition elements: two new functionals and systematic testing of four m06-class functionals and 12 other functionals. Theor. chemistry accounts 120, 215-241

Y. Zhao
,
D.G. Truhlar

[17]

Nonadiabatic dyanmics with trajectory surface hopping method

M. Barbatti

- WIREs Comput.Mol.Sci. 1 (2011) 620

[18]

Ko, C., Quenneville, J. & Martinez, T. J

B.G. Levine

- Mol.Phys. 104 (2006) 1039-1051

[19]

Dual-functional tamm-dancoff approximation: A convenient density functional method that correctly describes s1/s0 conical intersections

Y. Shu
,
K.A. Parker
,
D.G. Truhlar

- J.Phys.Chem.Lett. 8 (2017) 2107-2112

[20]

Dynamics algorithms with only potential energies and gradients: Curvature-driven coherent switching with decay of mixing and curvature-driven trajectory surface hopping. J. Chem

Y. Shu

- Comput. 18 (2022) 1320

[21]

Josene, Pinheiro Jr., M. & Barbatti, M

M.T. do Casal
,
M. Toldo

[22]

Nonadiabatic dynamics of 1,3-cyclohexadiene by curvature-driven coherent switching with decay of mixing. J. Chem

L. Zhang

- Comput. 18 (2022) 7073

[23]

Xu, X. & Truhlar, D. G

X. Zhao
,
Y. Shu
,
L. Zhang

- Comput. 19 (2023) 1672

[24]

Nonadiabatic coupling in trajectory surface hopping: Accurate time derivative coupling by the curvaturedriven approximation. J. Chem

X. Zhao

- Comput. 19 (2023) 6577

[25]

Diabatization by machine intelligence. J. Chem

Y. Shu
,
D.G. Truhlar

- Comput. 16 (2020) 6456-6464

1-25 of 37
1
2
25 / page