Inferring colloidal interaction from scattering by machine learning

Chi Huan Tung; Shou Yi Chang; Ming Ching Chang; Jan Michael Carrillo; Bobby G. Sumpter; Changwoo Do; Wei Ren Chen

doi:10.1016/j.cartre.2023.100252

Inferring colloidal interaction from scattering by machine learning

Chi Huan Tung, Shou Yi Chang, Ming Ching Chang, Jan Michael Carrillo, Bobby G. Sumpter, Changwoo Do, Wei Ren Chen

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

4 Scopus citations

Abstract

A machine learning solution for the potential inversion problem in elastic scattering is outlined. The inversion scheme consists of two major components, a generative network featuring a variational autoencoder which extracts the targeted static two-point correlation functions from experimentally measured scattering cross sections, and a Gaussian process framework which probabilistically infers the relevant structural parameters from the inverted correlation functions. Via a case study of charged colloidal suspensions, the feasibility of this approach for quantitative study of molecular interaction is critically benchmarked and its merit over existing deterministic approaches, in terms of numerical accuracy and computationally efficiency, is demonstrated.

Original language	English
Article number	100252
Journal	Carbon Trends
Volume	10
DOIs	https://doi.org/10.1016/j.cartre.2023.100252
State	Published - Mar 2023

Funding

This research was performed at the Spallation Neutron Source and the Center for Nanophase Materials Sciences, which are DOE Office of Science User Facilities operated by Oak Ridge National Laboratory. Molecular dynamics simulations used resources of the Oak Ridge Leadership Computing Facility, which is supported by DOE Office of Science under Contract DE-AC05-00OR22725. Application of machine learning to soft matter was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences Data, Artificial Intelligence and Machine Learning at DOE Scientific User Facilities Program under Award Number 34532. M.-C.C. thanks the support provided by the University at Albany - SUNY. This research was performed at the Spallation Neutron Source and the Center for Nanophase Materials Sciences, which are DOE Office of Science User Facilities operated by Oak Ridge National Laboratory. Molecular dynamics simulations used resources of the Oak Ridge Leadership Computing Facility, which is supported by DOE Office of Science under Contract DE-AC05-00OR22725 . Application of machine learning to soft matter was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences Data, Artificial Intelligence and Machine Learning at DOE Scientific User Facilities Program under Award Number 34532. M.-C.C. thanks the support provided by the University at Albany - SUNY.

Funders	Funder number
Center for Nanophase Materials Sciences
Office of Basic Energy Sciences Data, Artificial Intelligence and Machine Learning
U.S. Department of Energy	34532
Office of Science	DE-AC05-00OR22725
Oak Ridge National Laboratory
University at Albany

Keywords

Large-scale simulations
Machine learning
Neutron scattering
Soft matter

Access to Document

10.1016/j.cartre.2023.100252

Cite this

@article{bb71660e18994814b19eedbe4a197def,

title = "Inferring colloidal interaction from scattering by machine learning",

abstract = "A machine learning solution for the potential inversion problem in elastic scattering is outlined. The inversion scheme consists of two major components, a generative network featuring a variational autoencoder which extracts the targeted static two-point correlation functions from experimentally measured scattering cross sections, and a Gaussian process framework which probabilistically infers the relevant structural parameters from the inverted correlation functions. Via a case study of charged colloidal suspensions, the feasibility of this approach for quantitative study of molecular interaction is critically benchmarked and its merit over existing deterministic approaches, in terms of numerical accuracy and computationally efficiency, is demonstrated.",

keywords = "Large-scale simulations, Machine learning, Neutron scattering, Soft matter",

author = "Tung, {Chi Huan} and Chang, {Shou Yi} and Chang, {Ming Ching} and Carrillo, {Jan Michael} and Sumpter, {Bobby G.} and Changwoo Do and Chen, {Wei Ren}",

note = "Publisher Copyright: {\textcopyright} 2023 The Author(s)",

year = "2023",

month = mar,

doi = "10.1016/j.cartre.2023.100252",

language = "English",

volume = "10",

journal = "Carbon Trends",

issn = "2667-0569",

publisher = "Elsevier",

}

TY - JOUR

T1 - Inferring colloidal interaction from scattering by machine learning

AU - Tung, Chi Huan

AU - Chang, Shou Yi

AU - Chang, Ming Ching

AU - Carrillo, Jan Michael

AU - Sumpter, Bobby G.

AU - Do, Changwoo

AU - Chen, Wei Ren

PY - 2023/3

Y1 - 2023/3

N2 - A machine learning solution for the potential inversion problem in elastic scattering is outlined. The inversion scheme consists of two major components, a generative network featuring a variational autoencoder which extracts the targeted static two-point correlation functions from experimentally measured scattering cross sections, and a Gaussian process framework which probabilistically infers the relevant structural parameters from the inverted correlation functions. Via a case study of charged colloidal suspensions, the feasibility of this approach for quantitative study of molecular interaction is critically benchmarked and its merit over existing deterministic approaches, in terms of numerical accuracy and computationally efficiency, is demonstrated.

AB - A machine learning solution for the potential inversion problem in elastic scattering is outlined. The inversion scheme consists of two major components, a generative network featuring a variational autoencoder which extracts the targeted static two-point correlation functions from experimentally measured scattering cross sections, and a Gaussian process framework which probabilistically infers the relevant structural parameters from the inverted correlation functions. Via a case study of charged colloidal suspensions, the feasibility of this approach for quantitative study of molecular interaction is critically benchmarked and its merit over existing deterministic approaches, in terms of numerical accuracy and computationally efficiency, is demonstrated.

KW - Large-scale simulations

KW - Machine learning

KW - Neutron scattering

KW - Soft matter

UR - http://www.scopus.com/inward/record.url?scp=85148543880&partnerID=8YFLogxK

U2 - 10.1016/j.cartre.2023.100252

DO - 10.1016/j.cartre.2023.100252

M3 - Article

AN - SCOPUS:85148543880

SN - 2667-0569

VL - 10

JO - Carbon Trends

JF - Carbon Trends

M1 - 100252

ER -

Inferring colloidal interaction from scattering by machine learning

Abstract

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this