Transferable Machine-Learning Model of the Electron Density

Andrea Grisafi^1,², Alberto Fabrizio^2,³, Benjamin Meyer^2,³, David Wilkins¹, Clemence Corminboeuf^2,³, Michele Ceriotti^1,^2*

1 Laboratory of Computational Science and Modeling, IMX, École Polytechnique Fédérale de Lausanne, 1015 Lausanne, Switzerland

2 National Centre for Computational Design and Discovery of Novel Materials (MARVEL), École Polytechnique Fédérale de Lausanne, 1015 Lausanne, Switzerland

3 Laboratory for Computational Molecular Design, Institute of Chemical Sciences and Engineering, École Polytechnique Fédérale de Lausanne, CH-1015 Lausanne, Switzerland

* Corresponding authors emails: michele.ceriotti@epfl.ch

DOI10.24435/materialscloud:2019.0076/v1 [version v1]

Publication date: Oct 28, 2019

How to cite this record

Andrea Grisafi, Alberto Fabrizio, Benjamin Meyer, David Wilkins, Clemence Corminboeuf, Michele Ceriotti, Transferable Machine-Learning Model of the Electron Density, Materials Cloud Archive 2019.0076/v1 (2019), doi: 10.24435/materialscloud:2019.0076/v1.

Description

The electronic charge density plays a central role in determining the behavior of matter at the atomic scale, but its computational evaluation requires demanding electronic-structure calculations. We introduce an atom-centered, symmetry-adapted framework to machine-learn the valence charge density based on a small number of reference calculations. The model is highly transferable, meaning it can be trained on electronic-structure data of small molecules and used to predict the charge density of larger compounds with low, linear-scaling cost. Applications are shown for various hydrocarbon molecules of increasing complexity and flexibility, and demonstrate the accuracy of the model when predicting the density on octane and octatetraene after training exclusively on butane and butadiene. This transferable, data-driven model can be used to interpret experiments, accelerate electronic structure calculations, and compute electrostatic interactions in molecules and condensed-phase systems.

Materials Cloud sections using this data

No Explore or Discover sections associated with this archive record.

Files

File name	Size	Description
coords_ethene_1000.xyz MD5md5:812de0655ecb63efe5ed2c46bf0626da	711.4 KiB	Coordinates of 1000 ethene molecules in .xyz format.
coords_ethane_1000.xyz MD5md5:d46091fec866ac5773c22820e05bb1f8	360.6 KiB	Coordinates of 1000 ethane molecules in .xyz format.
coords_butadiene_1000.xyz MD5md5:50376f1eda5a01e88781ca9095310895	701.2 KiB	Coordinates of 1000 butadiene molecules in .xyz format.
coords_butane_1000.xyz MD5md5:8bc7d8f90af420b79c629e35c440960e	912.1 KiB	Coordinates of 1000 butane molecules in .xyz format.
coords_octatetraene_100.xyz MD5md5:04dcf208d2dd3f31b1156378216daad6	115.8 KiB	Coordinates of 100 octatetraene molecules in .xyz format.
coords_octane_100.xyz MD5md5:75f593da7c24649be491839640c39c77	194.5 KiB	Coordinates of 100 octane molecules in .xyz format.
ethene_new_6AA.tar.gz MD5md5:c858fb815155e96e8a7a9e1d4d4e66aa	9.6 GiB	Pseudo-valence electron densities of the ethene dataset.
butane_new_6AA.tar.gz MD5md5:2b479164c6fddba2bb509255c09ba851	22.5 GiB	Pseudo-valence electron densities of the butane dataset.
ethane_new_6AA.tar.gz MD5md5:e4e4ba5768dbf56e3dff8bd950fc5901	12.7 GiB	Pseudo-valence electron densities of the ethane dataset.
butadiene_new_6AA.tar.gz MD5md5:7d4fff70be4d92bbdb9d8518d2da1fd8	16.0 GiB	Pseudo-valence electron densities of the butadiene dataset.
octatetraene_new_6AA.tar.gz MD5md5:4c171fc59c02aa4e7066d2f06a112e6c	2.9 GiB	Pseudo-valence electron densities of the octatetraene dataset.
octane_6AA_new.tar.gz MD5md5:ac8a4a0460202cdf655aac6200b5d4af	4.2 GiB	Pseudo-valence electron densities of the octane dataset.
README.txt MD5md5:24b17a3a22abc378ee6620c289b32288	888 Bytes	README file for data content and format.

License

Files and data are licensed under the terms of the following license: Creative Commons Attribution 4.0 International.
Metadata, except for email addresses, are licensed under the Creative Commons Attribution Share-Alike 4.0 International license.

External references

Journal reference (Paper in which the method is described)

A. Grisafi, A. Fabrizio, B. Meyer, D. Wilkins, C. Corminboeuf, M. Ceriotti, ACS Cent. Sci. 5, 57-64 (2019) doi:10.1021/acscentsci.8b00551

Keywords

MARVEL/DD1 machine learning electron density

Version history:

2019.0076/v1 (version v1) [This version]	Oct 28, 2019	DOI10.24435/materialscloud:2019.0076/v1

Indexed by

materialscloud:2019.0076/v1