Artificial intelligence for computational granular media

Artificial intelligence (AI) has played a transformative role in accelerating scientific discovery and driving engineering innovations. Here we examine the primary applications of AI in computational granular materials over the past decades, focusing on three key objectives: (i) what machine learning (ML) can do in computational granular mechanics, (ii) how ML is integrated into routine computational simulations of granular media, and (iii) the opportunities and challenges that ML presents in this domain. The review highlights the key objectives of computational granular mechanics and the role of ML in bridging these critical research gaps. It systematically covers three aspects: (i) ML-accelerated computational modelling, (ii) ML-enabled pattern recognition and knowledge discovery, and (iii) ML-assisted inverse analysis in granular mechanics. Pertinent challenges are thoroughly discussed from the perspective of data and models. To promote the development of data-driven computational granular mechanics, we launched "Clear Data Bay", a metadata website tailored for domain data sharing and management. Despite ongoing challenges, data-driven approaches offer great potential in enabling computational granular mechanics models to tackle previously unattainable challenges.