Updates to the DScribe library : New descriptors and derivatives

Jarno Laakso; Lauri Himanen; Henrietta Homm; Eiaki V. Morooka; Marc O.J. Jäger; Milica Todorović; Patrick Rinke

doi:10.1063/5.0151031

Updates to the DScribe library : New descriptors and derivatives

Jarno Laakso^*, Lauri Himanen, Henrietta Homm, Eiaki V. Morooka, Marc O.J. Jäger, Milica Todorović, Patrick Rinke

^*Corresponding author for this work

University of Turku

Research output: Contribution to journal › Article › Scientific › peer-review

38 Citations (Scopus)

15 Downloads (Pure)

Abstract

We present an update of the DScribe package, a Python library for atomistic descriptors. The update extends DScribe’s descriptor selection with the Valle-Oganov materials fingerprint and provides descriptor derivatives to enable more advanced machine learning tasks, such as force prediction and structure optimization. For all descriptors, numeric derivatives are now available in DScribe. For the many-body tensor representation (MBTR) and the Smooth Overlap of Atomic Positions (SOAP), we have also implemented analytic derivatives. We demonstrate the effectiveness of the descriptor derivatives for machine learning models of Cu clusters and perovskite alloys.

Original language	English
Article number	234802
Pages (from-to)	1-8
Number of pages	8
Journal	Journal of Chemical Physics
Volume	158
Issue number	23
DOIs	https://doi.org/10.1063/5.0151031
Publication status	Published - 21 Jun 2023
MoE publication type	A1 Journal article-refereed

Access to Document

10.1063/5.0151031

Updates to the DScribe library - New descriptors and derivativesFinal published version, 5.5 MB

https://arxiv.org/abs/2303.14046

Cite this

@article{f0bb4c2c99864bd8948cddcea6659dae,

title = "Updates to the DScribe library : New descriptors and derivatives",

abstract = "We present an update of the DScribe package, a Python library for atomistic descriptors. The update extends DScribe{\textquoteright}s descriptor selection with the Valle-Oganov materials fingerprint and provides descriptor derivatives to enable more advanced machine learning tasks, such as force prediction and structure optimization. For all descriptors, numeric derivatives are now available in DScribe. For the many-body tensor representation (MBTR) and the Smooth Overlap of Atomic Positions (SOAP), we have also implemented analytic derivatives. We demonstrate the effectiveness of the descriptor derivatives for machine learning models of Cu clusters and perovskite alloys.",

author = "Jarno Laakso and Lauri Himanen and Henrietta Homm and Morooka, {Eiaki V.} and J{\"a}ger, {Marc O.J.} and Milica Todorovi{\'c} and Patrick Rinke",

note = "Funding Information: We acknowledge the funding from the European Union{\textquoteright}s Horizon program under Grant Agreement No. 951786, the Academy of Finland through Project No. 334532, and the Center of Excellence Virtual Laboratory for Molecular Level Atmospheric Transformations (VILMA; Project No. 346377). We further acknowledge the CSC-IT Center for Science, Finland, and the Aalto Science-IT project. | openaire: EC/H2020/951786/EU//NOMAD CoE ",

year = "2023",

month = jun,

day = "21",

doi = "10.1063/5.0151031",

language = "English",

volume = "158",

pages = "1--8",

journal = "Journal of Chemical Physics",

issn = "0021-9606",

publisher = "American Institute of Physics",

number = "23",

}

TY - JOUR

T1 - Updates to the DScribe library : New descriptors and derivatives

AU - Laakso, Jarno

AU - Himanen, Lauri

AU - Homm, Henrietta

AU - Morooka, Eiaki V.

AU - Jäger, Marc O.J.

AU - Todorović, Milica

AU - Rinke, Patrick

N1 - Funding Information: We acknowledge the funding from the European Union’s Horizon program under Grant Agreement No. 951786, the Academy of Finland through Project No. 334532, and the Center of Excellence Virtual Laboratory for Molecular Level Atmospheric Transformations (VILMA; Project No. 346377). We further acknowledge the CSC-IT Center for Science, Finland, and the Aalto Science-IT project. | openaire: EC/H2020/951786/EU//NOMAD CoE

PY - 2023/6/21

Y1 - 2023/6/21

N2 - We present an update of the DScribe package, a Python library for atomistic descriptors. The update extends DScribe’s descriptor selection with the Valle-Oganov materials fingerprint and provides descriptor derivatives to enable more advanced machine learning tasks, such as force prediction and structure optimization. For all descriptors, numeric derivatives are now available in DScribe. For the many-body tensor representation (MBTR) and the Smooth Overlap of Atomic Positions (SOAP), we have also implemented analytic derivatives. We demonstrate the effectiveness of the descriptor derivatives for machine learning models of Cu clusters and perovskite alloys.

AB - We present an update of the DScribe package, a Python library for atomistic descriptors. The update extends DScribe’s descriptor selection with the Valle-Oganov materials fingerprint and provides descriptor derivatives to enable more advanced machine learning tasks, such as force prediction and structure optimization. For all descriptors, numeric derivatives are now available in DScribe. For the many-body tensor representation (MBTR) and the Smooth Overlap of Atomic Positions (SOAP), we have also implemented analytic derivatives. We demonstrate the effectiveness of the descriptor derivatives for machine learning models of Cu clusters and perovskite alloys.

UR - http://www.scopus.com/inward/record.url?scp=85162717051&partnerID=8YFLogxK

U2 - 10.1063/5.0151031

DO - 10.1063/5.0151031

M3 - Article

C2 - 37338028

AN - SCOPUS:85162717051

SN - 0021-9606

VL - 158

SP - 1

EP - 8

JO - Journal of Chemical Physics

JF - Journal of Chemical Physics

IS - 23

M1 - 234802

ER -

Updates to the DScribe library : New descriptors and derivatives

Abstract

Access to Document

Other files and links

Fingerprint

VILMA: Virtual laboratory for molecular level atmospheric transformations

NOMAD CoE: Novel Materials Discovery

LEARNSOLAR: Rinke-LearnSolar

Science-IT

Cite this

Updates to the DScribe library : New descriptors and derivatives

Abstract

Access to Document

Other files and links

Fingerprint

Projects

VILMA: Virtual laboratory for molecular level atmospheric transformations

NOMAD CoE: Novel Materials Discovery

LEARNSOLAR: Rinke-LearnSolar

Equipment

Science-IT

Cite this