Microstructure Characterization and Reconstruction in Python: MCRpy

Abstract Microstructure characterization and reconstruction (MCR) is an important prerequisite for empowering accelerating integrated computational materials engineering. Much progress has been made in MCR recently; however, the absence of a flexible software platform it difficult to use ideas from other researchers develop them further. To address this issue, work presents MCRpy as easy-to-use, extensible open-source platform. can be used program with graphical user interface, command line tool Python library. The central idea that microstructure formulated modular optimization problem. In way, arbitrary descriptors loss functions combining minimized using optimizers reconstructing random heterogeneous media. With stochastic optimizers, leads variations well-known Yeong–Torquato algorithm. Furthermore, features automatic differentiation, enabling utilization gradient-based optimizers. work, after brief introduction underlying concepts, capabilities are demonstrated by exemplarily applying typical tasks. Finally, shown how extend defining new descriptor readily without additional implementation effort.