Beyond Graphs: Sparse Structures in Emerging Computational Challenges

Real-world data and systems are becoming increasingly complex, involving rich forms of interaction and changes over time. While graphs have long served as the canonical model for representing relationships in data -- leading to powerful algorithmic techniques and deep theoretical insights -- they are not always sufficient to capture this full complexity. Since numerous real-world applications in machine learning, network science, and high-performance computing increasingly demand richer models, researchers are turning to a broader family of sparse structures that enable more expressive modeling -- such as hypergraphs, temporal networks, and tensors.

This workshop brings together researchers from computer science, mathematics, and network science to explore the theory, algorithms, and applications of sparse structures beyond ordinary graphs. Several converging trends make this the right moment to initiate a dialogue between the relevant communities: (i) the rapid growth of AI and machine learning requires scalable sparse methods beyond static graphs, (ii) scientific computing increasingly relies on sparse multilinear algebra (tensors), and (iii) complex systems in physics, biology, and the social sciences cannot be fully understood without higher-order and/or temporal models. We therefore aim to identify unifying principles across domains, develop algorithmic concepts and common computational principles for the sparse structures in our focus. By bringing together diverse perspectives, the workshop aims to strengthen existing research efforts and foster new collaborations across disciplines.