Search

Abstract Obtaining microscopic structure-property relationships for grain boundaries is challenging due to their complex atomic structures. Recent efforts use machine learning to derive these relationships, but the way the atomic grain boundary structure is represented can have a significant impact on the predictions. Key steps for property prediction common to grain boundaries and other variable-sized atom clustered structures include: (1) describing the atomic structure as a feature matrix, (2) transforming the variable-sized feature matrix to a fixed length common to all structures, and (3) applying a machine learning algorithm to predict properties from the transformed matrices. We examine how these steps and different combinations of engineered features impact the accuracy of grain boundary energy predictions using a database of over 7000 grain boundaries. Additionally, we assess how different engineered features support interpretability, offering insights into the physics of the structure-property relationships.

Abstract Observations of microstructural coarsening at cryogenic temperatures, as well as numerous simulations of grain boundary motion that show faster migration at low temperature than at high temperature, have been troubling because they do not follow the expected Arrhenius behavior. This work demonstrates that classical equations, that are not simplified, account for all these oddities and demonstrate that non-Arrhenius behavior can emerge from thermally activated processes. According to this classical model, this occurs when the intrinsic barrier energies of the processes become small, allowing activation at cryogenic temperatures. Additional thermal energy then allows the low energy process to proceed in reverse, so increasing temperature only serves to frustrate the forward motion. This classical form is shown to reconcile and describe a variety of diverse grain boundary migration observations.

Measurement(s) chemical composition of aluminum alloys • mechanical properties of aluminum alloys Technology Type(s) natural language processing

We present a method for performing efficient barycentric interpolation for large grain boundary octonion point sets which reside on the surface of a hypersphere. This method includes removal of degenerate dimensions via singular value decomposition (SVD) transformations and linear projections, determination of intersecting facets via nearest neighbor (NN) searches, and interpolation. This method is useful for hyperspherical point sets for applications such as grain boundaries structure-property models, robotics, and specialized neural networks. We provide a case study of the method applied to the 7-sphere. We provide 1-sphere and 2-sphere visualizations to illustrate important aspects of these dimension reduction and interpolation methods. A MATLAB implementation is available at github.com/sgbaird-5dof/interp. • Barycentric interpolation is combined with hypersphere facet intersections, dimensionality reduction, and linear projections to reduce computational complexity without loss of information • A max nearest neighbor threshold is used in conjunction with facet intersection determination to reduce computational runtime.

One of the limitations of atomistic simulations is that many of the computational tools used to extract structural information from atomic trajectories provide metrics that are not directly compatible with experiments for validation. In this work, to bridge between simulation and experiment, a method is presented to produce simulated Kikuchi diffraction patterns using data from atomistic simulations, without requiring a priori specification of the crystal structure or defect periodicity. The Kikuchi pattern simulation is based on the kinematic theory of diffraction, with Kikuchi line intensities computed via a discrete structure factor calculation. Reciprocal lattice points are mapped to Kikuchi lines using a geometric projection of the reciprocal space data. This method is validated using single crystal atomistic models, and the novelty of this approach is emphasized by simulating kinematic Kikuchi diffraction patterns from an atomistic model containing a nanoscale dislocation loop. Deviations in kinematic Kikuchi line intensities are explained considering the displacement field of the dislocation loop, as is done in diffraction contrast theory. Method: Simulated kinematic Kikuchi diffraction from atomistic simulation, Keywords: Kikuchi diffraction, Atomistic simulation, Crystal structure, Dislocations

Machine learning: Modelling atomic systems to make property predictions A method for representing atomic systems for machine learning is shown that can provide access to the physical properties of these systems. Machine learning is a powerful tool for finding correlations but when used to look at real-word systems, the complexity of the models often limits the amount of information that can be extracted about the underlying physics. An international team of researchers led by Conrad Rosenbrock from Brigham Young University now present a machine learning-based approach for modelling atomic systems that can provide insight into the physical building blocks that influence them. They demonstrate the power of their approach by examining the predictive performance of several machine learning models, providing connections between the structure and behaviour of grain boundaries in crystalline materials, which could be extended to other systems that involve local structural changes.

The atomic structure of grain boundaries plays a defining but poorly understood role in the properties they exhibit. Due to the complex nature of these structures, machine learning is a natural tool for extracting meaningful relationships and new physical insight. We apply a new structural representation, called the scattering transform, that uses wavelet-based convolutional neural networks to characterize the complete three-dimensional atomic structure of a grain boundary. The machine learning to predict GB energy, mobility, and shear coupling using the scattering transform representation is compared and contrasted with learning using a smooth overlap of atomic positions (SOAP) based representation. While predictions using the scattering transform are not as good as those of SOAP, other factors suggest that the scattering transform may yet play an important role in GB structure learning. These factors include the ability of the scattering transform to learn well on larger datasets, in a process similar to deep learning, as well as their ability to provide physically interpretable information about what aspects of the GB structure contribute to the learning through an inverse scattering transform.