Scaling Ellipsoidal-Inclusion Microstructure Generation to Larger Structures

Develop a computationally feasible method for scaling microstructure generation with ellipsoidal inclusions to larger structures, where existing approaches have not yet overcome the associated computational and optimization difficulties.

Background

The paper presents a differentiable optimization framework for generating two-dimensional representative volume elements of anisotropic magnetorheological elastomers using circular inclusions and the two-point correlation function. The authors note that replacing circles with ellipses would provide a closer geometric approximation of the true particle shapes, but that evaluating minimum surface distances between ellipses requires iterative computations within the optimization loop.

Although prior work has demonstrated the generation of small cells containing ellipsoidal inclusions, extending that approach to larger microstructures is identified as an unresolved computational challenge. Solving this problem would enable more geometrically faithful and potentially three-dimensional microstructure reconstruction at practically useful scales.