Descriptor-based reconstruction of three-dimensional microstructures through gradient-based optimization

Microstructure reconstruction is an important cornerstone to the inverse materials design concept. In this work, a general algorithm developed reconstruct three-dimensional microstructure from given descriptors. Based on two-dimensional (2D) micrographs, allows valuable insight through spatial visualization of and in silico studies structure-property linkages. The formulation ensures computational efficiency by casting as gradient-based optimization problem. Herein, descriptors can be chosen freely, such correlations or Gram matrices, long they are differentiable with respect microstructure. Because real samples commonly available 2D microscopy images only, desired for process prescribed orthogonal slices. This adds source noise, which handled new, superior interpretable manner. applicability demonstrated various numerical experiments.