This repository includes main notebook of the code for our proposed RCGAN: Inverse Structural Design of Graphene/Boron Nitride Hybrids by Regressional and Conditional GAN [https://arxiv.org/abs/1908.07959]
train.pickle has been uploaded, containing preprocessed training data.
Design of materials with desired properties is currently laborious and heavily relies on intuition of researchers through a trial-and-error process. The massive combinational spaces due to the constituent elements and their structural configurations are too overwhelming to be all searched even by high-throughput computations. To tackle this challenge, we have proposed a novel regressional and conditional generative adversarial network (RCGAN) for inverse design of representative two-dimensional materials, graphene and boron-nitride (BN) hybrids. A significant novelty of the proposed RCGAN is that it incorporates a modified supervised regressor network, thus overcoming the common technical barrier in the traditional unsupervised GANs, which cannot generate data when given continuous quantitative labels. The proposed RCGAN enables to autonomously generate graphene/BN hybrids given any targeted bandgap, which are continuous quantitative labels. The generated structures are distinguished from the ones used for training. Moreover, they exhibit high fidelity, yielding bandgaps within ~ 10% MAEF of the desired bandgaps as validated by density functional theory (DFT) calculations. Further analysis by the principle component analysis (PCA) and modified locally linear embedding (MLLE) on the latent features encoded by the regressor reveals that the generator has successfully generated structures that follow the statistical distribution of the real structures. It implies the possibility of the RCGAN in recognizing physical rules hidden in the high-dimensional data. The novel strategy for designing regressional GAN architecture together with the successful application to inverse design of materials would inspire further exploration for research fields beyond materials.