AnisoGNN: Graph neural networks generalizing to anisotropic properties of polycrystals

Guangyu Hu; Marat I. Latypov

doi:10.1016/j.commatsci.2024.113121

AnisoGNN: Graph neural networks generalizing to anisotropic properties of polycrystals

Guangyu Hu, Marat I. Latypov

Materials Science and Engineering

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

3 Scopus citations

Abstract

We present AnisoGNNs — graph neural networks (GNNs) that generalize predictions of anisotropic properties of polycrystals in arbitrary testing directions without the need in excessive training data. To this end, we develop GNNs with a physics-inspired combination of node attributes and aggregation function. We demonstrate the excellent generalization capabilities of AnisoGNNs in predicting anisotropic elastic and inelastic properties of two alloys.

Original language	English (US)
Article number	113121
Journal	Computational Materials Science
Volume	243
DOIs	https://doi.org/10.1016/j.commatsci.2024.113121
State	Published - Jul 2024

Keywords

Computational homogenization
Graph neural networks
Mechanical properties
Polycrystals

ASJC Scopus subject areas

General Computer Science
General Chemistry
General Materials Science
Mechanics of Materials
General Physics and Astronomy
Computational Mathematics

Access to Document

10.1016/j.commatsci.2024.113121

Cite this

@article{e2c148412dc74454887de905498bf16d,

title = "AnisoGNN: Graph neural networks generalizing to anisotropic properties of polycrystals",

abstract = "We present AnisoGNNs — graph neural networks (GNNs) that generalize predictions of anisotropic properties of polycrystals in arbitrary testing directions without the need in excessive training data. To this end, we develop GNNs with a physics-inspired combination of node attributes and aggregation function. We demonstrate the excellent generalization capabilities of AnisoGNNs in predicting anisotropic elastic and inelastic properties of two alloys.",

keywords = "Computational homogenization, Graph neural networks, Mechanical properties, Polycrystals",

author = "Guangyu Hu and Latypov, \{Marat I.\}",

note = "Publisher Copyright: {\textcopyright} 2024 Elsevier B.V.",

year = "2024",

month = jul,

doi = "10.1016/j.commatsci.2024.113121",

language = "English (US)",

volume = "243",

journal = "Computational Materials Science",

issn = "0927-0256",

publisher = "Elsevier B.V.",

}

TY - JOUR

T1 - AnisoGNN

T2 - Graph neural networks generalizing to anisotropic properties of polycrystals

AU - Hu, Guangyu

AU - Latypov, Marat I.

PY - 2024/7

Y1 - 2024/7

N2 - We present AnisoGNNs — graph neural networks (GNNs) that generalize predictions of anisotropic properties of polycrystals in arbitrary testing directions without the need in excessive training data. To this end, we develop GNNs with a physics-inspired combination of node attributes and aggregation function. We demonstrate the excellent generalization capabilities of AnisoGNNs in predicting anisotropic elastic and inelastic properties of two alloys.

AB - We present AnisoGNNs — graph neural networks (GNNs) that generalize predictions of anisotropic properties of polycrystals in arbitrary testing directions without the need in excessive training data. To this end, we develop GNNs with a physics-inspired combination of node attributes and aggregation function. We demonstrate the excellent generalization capabilities of AnisoGNNs in predicting anisotropic elastic and inelastic properties of two alloys.

KW - Computational homogenization

KW - Graph neural networks

KW - Mechanical properties

KW - Polycrystals

UR - https://www.scopus.com/pages/publications/85194301397

UR - https://www.scopus.com/inward/citedby.url?scp=85194301397&partnerID=8YFLogxK

U2 - 10.1016/j.commatsci.2024.113121

DO - 10.1016/j.commatsci.2024.113121

M3 - Article

AN - SCOPUS:85194301397

SN - 0927-0256

VL - 243

JO - Computational Materials Science

JF - Computational Materials Science

M1 - 113121

ER -

AnisoGNN: Graph neural networks generalizing to anisotropic properties of polycrystals

Abstract

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this