English

Composing Behaviors of Networks

Category Theory 2021-05-28 v1

Abstract

This thesis aims to develop a compositional theory for the operational semantics of networks. The networks considered are described by either internal or enriched graphs. In the internal case we focus on Q\mathsf{Q}-nets, a generalization of Petri nets based on a Lawvere theory Q\mathsf{Q}. Q\mathsf{Q}-nets include many known variants of Petri nets including pre-nets, integer nets, elementary net systems, and bounded nets. In the enriched case we focus on graphs enriched in a quantale RR regarded as matrices with entries in RR. These RR-matrices represent distance networks, Markov processes, capacity networks, non-deterministic finite automata, simple graphs, and more. The operational semantics of Q\mathsf{Q}-nets is constructed as an adjunction between Q\mathsf{Q}-nets and categories internal to the category of models of Q\mathsf{Q}. Similarly, the operational semantics of RR-matrices is constructed as an adjunction between RR-matrices and categories enriched in RR. The left adjoint of this adjunction sends an RR-matrix MM to the RR-category FR(M)F_R(M) whose hom-objects are solutions of the algebraic path problem: a generalization of the shortest path problem to graphs weighted in RR. For both Q\mathsf{Q}-nets and RR-matrices we use the theory of structured cospans to study the compositionality of the above operational semantics. For each type of network we construct a double category whose morphisms are "open networks", i.e. networks with certain vertices designated as input or output. We introduce the black-boxing of an open network, a profunctor describing the externally observable behavior of an open network. We introduce a class of open networks called "functional open networks" for which black-boxing preserves composition.

Keywords

Cite

@article{arxiv.2105.12905,
  title  = {Composing Behaviors of Networks},
  author = {Jade Master},
  journal= {arXiv preprint arXiv:2105.12905},
  year   = {2021}
}

Comments

101 pages

R2 v1 2026-06-24T02:30:42.304Z