Research

Molecular Dynamics (MD) simulations are essential for accurately predicting the physical and chemical properties of large molecular systems across various pressure and temperature ensembles. However, the high computational costs associated with All-Atom (AA) MD simulations have led to the development of Coarse-Grained Molecular Dynamics (CGMD), providing a lower-dimensional compression of the AA structure into representative CG beads, offering reduced computational expense at the cost of predictive accuracy. Existing CGMD methods, such as CG-Martini (calibrated against experimental data), aim to generate an embedding of a topology that sufficiently generalizes across a range of structures.

https://arxiv.org/abs/2501.02707

High mix, low volume processes such as additive manufacturing (AM) offer tremendous promise for increasing the customization in manufacturing but are hindered by the lack of efficient methods for identifying process parameters for complex new geometries exhibiting the desired performance. The search over the process space can be automated with analysis tools that can be applied in a time and resource efficient manner such that ambitious print designs are not dissuaded by the cost of process parameter discovery. In this work, we propose an image analysis tool that can classify spanning prints as one of five process-relevant archetypes, invariant of the span dimensions.

Materials and Manufacturing Directorate, Air Force Research Laboratory, Wright-Patterson Air Force Base, OH, USA
https://doi.org/10.1016/j.addma.2021.102191

This article introduces systematically the foundational concepts undergirding the recently formulated AI (artificial intelligence)-based materials knowledge system (AI-MKS) framework. More specifically, these concepts deal with features engineering the heterogeneous material internal structure to obtain low-dimensional representations that can then be combined with machine learning models to establish low computational cost surrogate models for capturing the process–structure–property linkages over a hierarchy of material structure/lengths scales.

Journal of Applied Physics 128, 041103 (2020); https://doi.org/10.1063/5.0011258

This prospective offers a new Bayesian framework that could guide the systematic application of the emerging toolsets of machine learning in the efforts to address two of the central bottlenecks encountered in materials innovation: (i) the capture of core materials knowledge in reduced-order forms that allow one to rapidly explore the vast materials design spaces, and (ii) objective guidance in the selection of experiments or simulations needed to identify the governing physics in the materials phenomena of interest. The framework builds on recent advances in the low-dimensional representation of the statistics describing the material’s
hierarchical structure.

Kalidindi, S.R. A Bayesian framework for materials knowledge systems. MRS Communications 9, 518–531 (2019). https://doi.org/10.1557/mrc.2019.56

This paper lays the foundations to develop automated segmentation workflows. It systematically identifies the various steps and the sequence in which they should be applied to obtain good segmentation results. The proposed framework is validated on diverse collections of micrographs from very different material systems.

Iskakov, A., Kalidindi, S.R. A Framework for the Systematic Design of Segmentation Workflows. Integr Mater Manuf Innov 9, 70–88 (2020). https://doi.org/10.1007/s40192-019-00166-z