Discovering rare designs of metamaterials with unusual combinations of material properties, such as double-auxeticity and high elastic moduli, is a challenging task. This research uses computational models and deep learning algorithms to identify such rare designs in three types of planar lattices with random distributions of hard and soft phases. A mapping from design parameters to mechanical properties significantly reduces computational time and allows for parallelization. Ten designs were 3D printed, mechanically tested, and characterized using digital image correlation, validating the computational models' accuracy.

Helda Pahlavani, Muhamad Amani, Mauricio Cruz Saldívar, Jie Zhou, Mohammad J. Mirzaali, Amir A. Zadpoor