Publications

Results 26–50 of 106
Skip to search filters

Machine learning models of plastic flow based on representation theory

CMES - Computer Modeling in Engineering and Sciences

Jones, R.E.; Templeton, Jeremy A.; Sanders, Clay M.; Ostien, Jakob O.

We use machine learning (ML) to infer stress and plastic flow rules using data from representative polycrystalline simulations. In particular, we use so-called deep (multilayer) neural networks (NN) to represent the two response functions. The ML process does not choose appropriate inputs or outputs, rather it is trained on selected inputs and output. Likewise, its discrimination of features is crucially connected to the chosen input-output map. Hence, we draw upon classical constitutive modeling to select inputs and enforce well-accepted symmetries and other properties. In the context of the results of numerous simulations, we discuss the design, stability and accuracy of constitutive NNs trained on typical experimental data. With these developments, we enable rapid model building in real-time with experiments, and guide data collection and feature discovery.

More Details
TYPE Journal Article YEAR 2018
ScopusOSTIDOI

A Cartesian parametrization for the numerical analysis of material instability

International Journal for Numerical Methods in Engineering

Mota, A.; Chen, Q.; Foulk, James W.; Ostien, Jakob O.; Lai, Z.

We examine four parametrizations of the unit sphere in the context of material stability analysis by means of the singularity of the acoustic tensor. We then propose a Cartesian parametrization for vectors that lie a cube of side length two and use these vectors in lieu of unit normals to test for the loss of the ellipticity condition. This parametrization is then used to construct a tensor akin to the acoustic tensor. It is shown that both of these tensors become singular at the same time and in the same planes in the presence of a material instability. The performance of the Cartesian parametrization is compared against the other parametrizations, with the results of these comparisons showing that in general, the Cartesian parametrization is more robust and more numerically efficient than the others. Copyright © 2016 John Wiley & Sons, Ltd.

More Details
TYPE Journal Article YEAR 2016
ScopusOSTIDOI
Results 26–50 of 106
Results 26–50 of 106