ComMat: Complex Material Datasets and Benchmarks for Graph Machine Learning

Recent research has demonstrated the efficacy of graph learning over a wide spectrum of materials, including molecular graphs, crystals, mechanical metamaterials, and strongly disordered systems. In this work, we draw attention to the broad class of \textit{complex materials}, which combine order and disorder. They fall outside the above categories yet have shown superior properties throughout the materials science literature. We present ComMat (Complex Material Benchmark), which comprises three graph datasets of complex materials from experimental and computational research studies and unifies distinctly developed data-to-graph pipelines under a standardized graph-based representation. In particular, we provide the first publicly available 3D graph dataset of a nanoscale network derived from 3D tomography. We then quantitatively show that these graphs are fundamentally different from existing materials datasets. We design various predictive tasks to advance machine learning (ML) methods, including prediction of experimentally measured properties, simulated mechanical response, and structural completeness. Extensive benchmark experiments are conducted on popular graph learning models, revealing their limitations and the need for further development in handling complex materials. ComMat is openly released to accelerate ML research and innovation in complex material design.