Robust prediction of force chains in jammed solids using graph neural networks

Rituparno Mandal; Corneel Casert; Peter Sollich

doi:10.1038/s41467-022-31732-3

Robust prediction of force chains in jammed solids using graph neural networks

Rituparno Mandal^*, Corneel Casert, Peter Sollich

^*Corresponding author for this work

Disordered Systems

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

12 Citations (Scopus)

Abstract

Force chains are quasi-linear self-organised structures carrying large stresses and are ubiquitous in jammed amorphous materials like granular materials, foams or even cell assemblies. Predicting where they will form upon deformation is crucial to describe the properties of such materials, but remains an open question. Here we demonstrate that graph neural networks (GNN) can accurately predict the location of force chains in both frictionless and frictional materials from the undeformed structure, without any additional information. The GNN prediction accuracy also proves to be robust to changes in packing fraction, mixture composition, amount of deformation, friction coefficient, system size, and the form of the interaction potential. By analysing the structure of the force chains, we identify the key features that affect prediction accuracy. Our results and methodology will be of interest for granular matter and disordered systems, e.g. in cases where direct force chain visualisation or force measurements are impossible.

Original language	English
Article number	4424
Journal	Nature Communications
Volume	13
Issue number	1
DOIs	https://doi.org/10.1038/s41467-022-31732-3
Publication status	Published - Dec 2022

Access to Document

10.1038/s41467-022-31732-3

Cite this

@article{fd9976c095164457a048e10135fcefdf,

title = "Robust prediction of force chains in jammed solids using graph neural networks",

abstract = "Force chains are quasi-linear self-organised structures carrying large stresses and are ubiquitous in jammed amorphous materials like granular materials, foams or even cell assemblies. Predicting where they will form upon deformation is crucial to describe the properties of such materials, but remains an open question. Here we demonstrate that graph neural networks (GNN) can accurately predict the location of force chains in both frictionless and frictional materials from the undeformed structure, without any additional information. The GNN prediction accuracy also proves to be robust to changes in packing fraction, mixture composition, amount of deformation, friction coefficient, system size, and the form of the interaction potential. By analysing the structure of the force chains, we identify the key features that affect prediction accuracy. Our results and methodology will be of interest for granular matter and disordered systems, e.g. in cases where direct force chain visualisation or force measurements are impossible.",

author = "Rituparno Mandal and Corneel Casert and Peter Sollich",

note = "Funding Information: We are grateful to Bulbul Chakraborty for her valuable comments on the manuscript. R.M. and C.C. would like to thank Mehdi Bouzid and Pritam Kumar Jana for their help in setting up the LAMMPS simulation for frictional particles. R.M. acknowledges funding from the European Union{\textquoteright}s Horizon 2020 research and innovation program under the Marie Sk{\l}odowska-Curie grant agreement No 893128. Computational resources (Stevin Supercomputer Infrastructure) and services used in this work were provided by the VSC (Flemish Supercomputer Center), and the Flemish Government—department EWI We acknowledge support by the Open Access Publication Funds of the G{\"o}ttingen University. Publisher Copyright: {\textcopyright} 2022, The Author(s).",

year = "2022",

month = dec,

doi = "10.1038/s41467-022-31732-3",

language = "English",

volume = "13",

journal = "Nature Communications",

issn = "2041-1723",

publisher = "Nature Publishing Group",

number = "1",

}

TY - JOUR

T1 - Robust prediction of force chains in jammed solids using graph neural networks

AU - Mandal, Rituparno

AU - Casert, Corneel

AU - Sollich, Peter

N1 - Funding Information: We are grateful to Bulbul Chakraborty for her valuable comments on the manuscript. R.M. and C.C. would like to thank Mehdi Bouzid and Pritam Kumar Jana for their help in setting up the LAMMPS simulation for frictional particles. R.M. acknowledges funding from the European Union’s Horizon 2020 research and innovation program under the Marie Skłodowska-Curie grant agreement No 893128. Computational resources (Stevin Supercomputer Infrastructure) and services used in this work were provided by the VSC (Flemish Supercomputer Center), and the Flemish Government—department EWI We acknowledge support by the Open Access Publication Funds of the Göttingen University. Publisher Copyright: © 2022, The Author(s).

PY - 2022/12

Y1 - 2022/12

N2 - Force chains are quasi-linear self-organised structures carrying large stresses and are ubiquitous in jammed amorphous materials like granular materials, foams or even cell assemblies. Predicting where they will form upon deformation is crucial to describe the properties of such materials, but remains an open question. Here we demonstrate that graph neural networks (GNN) can accurately predict the location of force chains in both frictionless and frictional materials from the undeformed structure, without any additional information. The GNN prediction accuracy also proves to be robust to changes in packing fraction, mixture composition, amount of deformation, friction coefficient, system size, and the form of the interaction potential. By analysing the structure of the force chains, we identify the key features that affect prediction accuracy. Our results and methodology will be of interest for granular matter and disordered systems, e.g. in cases where direct force chain visualisation or force measurements are impossible.

AB - Force chains are quasi-linear self-organised structures carrying large stresses and are ubiquitous in jammed amorphous materials like granular materials, foams or even cell assemblies. Predicting where they will form upon deformation is crucial to describe the properties of such materials, but remains an open question. Here we demonstrate that graph neural networks (GNN) can accurately predict the location of force chains in both frictionless and frictional materials from the undeformed structure, without any additional information. The GNN prediction accuracy also proves to be robust to changes in packing fraction, mixture composition, amount of deformation, friction coefficient, system size, and the form of the interaction potential. By analysing the structure of the force chains, we identify the key features that affect prediction accuracy. Our results and methodology will be of interest for granular matter and disordered systems, e.g. in cases where direct force chain visualisation or force measurements are impossible.

UR - http://www.scopus.com/inward/record.url?scp=85135168441&partnerID=8YFLogxK

U2 - 10.1038/s41467-022-31732-3

DO - 10.1038/s41467-022-31732-3

M3 - Article

C2 - 35908018

AN - SCOPUS:85135168441

SN - 2041-1723

VL - 13

JO - Nature Communications

JF - Nature Communications

IS - 1

M1 - 4424

ER -

Robust prediction of force chains in jammed solids using graph neural networks

Abstract

Access to Document

Other files and links

Fingerprint

Cite this