| Description |
The idea of material discovery has excited and perplexed research scientists for centuries. Several different methods have been employed to find new types of materials, ranging from the arbitrary replacement of atoms in a crystal structure to advanced machine learning methods for predicting entirely new crystal structures. In this work, we investigate and analyze the performance of various Generative Adversarial Network (GAN) architectures to find an innovative and effective way of generating theoretical crystal structures that are synthesizable and stable. Over one hundred thousand Crystallographic Information Files (CIFs) from Pearson's Crystal Data are used for the training of each GAN. The space group number, atomic positions, and lattice parameters are parsed from the CIFs and used to construct an input tensor for each of the different network architectures. Several different GAN layer configurations are designed and analyzed, including Wasserstein GANs with gradient penalty, in order to identify a model that can adequately recognize and discern symmetry patterns that are present in known material crystal structures and are imperceptible to humans. This work will detail the process and techniques that were used in an attempt to generate never-before-seen crystal structures that are both stable and synthesizable, as well as reveal a plethora of guiding questions for future work. |
