Vitelli Group


Information-processing agents — including humans, bots, cells, and even individual molecules — all possess the capacity for adaptive, intelligent behavior. Our research centers on a fundamental question: what laws emerge when many such agents interact? To uncover the dynamics of these agentic many-body systems, we combine data-driven, interpretable machine learning with principled tools from theoretical physics, notably generalized symmetries and conservation laws. The robustness of our mathematical models is baked directly into their underlying geometric and topological architecture. We put these abstract frameworks to test through minimalist in-house experiments and active collaborations with laboratories worldwide.

Selected Recent Publications