Experiments in Agentic AI for Science

arXiv:2605.26305v1 Announce Type: new Abstract: This paper details two novel frameworks for developing autonomous, agentic AI in scientific workflows. Both systems leverage a hybrid Local

arXiv:2405.09748v2 Announce Type: replace
Abstract: Endothelial cells form the linchpin of vascular and lymphatic systems, creating intricate networks that are pivotal for angiogenesis, controlling vessel permeability, and maintaining tissue homeostasis. Despite their critical roles, there is no rigorous mathematical framework to represent the connectivity structure of endothelial networks. Here, we develop a pioneering mathematical formalism called $pi$-graphs to model the multi-type junction connectivity of endothelial networks. We define $pi$-graphs as abstract objects consisting of endothelial cells and their junction sets, and introduce the key notion of $pi$-isomorphism that captures when two $pi$-graphs have the same connectivity structure. We prove several propositions relating the $pi$-graph representation to traditional graph-theoretic representations, showing that $pi$-isomorphism implies isomorphism of the corresponding unnested endothelial graphs, but not vice versa. We also introduce a temporal dimension to the $pi$-graph formalism and explore the evolution of topological invariants in spatial embeddings of $pi$-graphs. Finally, we outline a topological framework to represent the spatial embedding of $pi$-graphs into geometric spaces. The $pi$-graph formalism provides a novel tool for quantitative analysis of endothelial network connectivity and its relation to function, with the potential to yield new insights into vascular physiology and pathophysiology.

Subscribe for Updates

Copyright 2025 dijee Intelligence Ltd.   dijee Intelligence Ltd. is a private limited company registered in England and Wales at Media House, Sopers Road, Cuffley, Hertfordshire, EN6 4RY, UK registration number 16808844