Automated Analysis of Continuum Fields from Atomistic Simulations Using Statistical Machine Learning. Issue 12 (16th September 2022)

• Record Type: Journal Article
• Authors: Prakash, Aruna; Sandfeld, Stefan
• Is Part Of: Advanced engineering materials. Volume 24:Issue 12(2022)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2022-09-16
• Subjects: atomistic simulations -- clustering -- data mining -- Gaussian mixture model -- machine learning -- nanocrystalline materials -- statistical distribution functions; Materials -- Periodicals; 620.11
• DOI: 10.1002/adem.202200574
• Languages: English
• ISSNs: 1438-1656

Abstract

Abstract : Atomistic simulations of the molecular dynamics/statics kind are regularly used to study small‐scale plasticity. Contemporary simulations are performed with tens to hundreds of millions of atoms, with snapshots of these configurations written out at regular intervals for further analysis. Continuum scale constitutive models for material behavior can benefit from information on the atomic scale, in particular in terms of the deformation mechanisms, the accommodation of the total strain, and partitioning of stress and strain fields in individual grains. Herein, a methodology is developed using statistical data mining and machine learning algorithms to automate the analysis of continuum field variables in atomistic simulations. Three important field variables are focused on: total strain, elastic strain, and microrotation. The results show that the elastic strain in individual grains exhibits a unimodal lognormal distribution, while the total strain and microrotation fields evidence a multimodal distribution. The peaks in the distribution of total strain are identified with a Gaussian mixture model and methods to circumvent overfitting problems are presented. Subsequently, the identified peaks are evaluated in terms of deformation mechanisms in a grain, which, e.g., helps to quantify the strain for which individual deformation mechanisms are responsible. The overall statistics of the distributions over all grains are an important input for higher scale models, which ultimately also helps to be able to quantitatively discuss the implications for information transfer to phenomenological models. Abstract : Statistical data mining and machine learning algorithms are used to develop a methodology to automate the analysis of continuum field variables, like, e.g., elastic and total strain, in atomistic simulations of polycrystalline materials. The methodology facilitates the determination of the functional form of unimodal distributions and identifies individual peaks in a multimodal distribution and correlates then with mechanisms of deformation.