A categorical approach to threshold agent networks (Q1430174)

The author considers his threshold agent networks (TANs) as introduced in [Appl. Math. Comput. 142, 521--543 (2003; Zbl 1023.68121)]. These are discretized neural networks with integer thresholds. The paper introduces morphisms between TANs, creating a category \textbf{TAN}. This category \textbf{TAN} is shown to have finite products, but does not have equalizers and coproducts. Therefore an extended category \textbf{GTAN} of generalized TANs is introduced which has \textbf{TAN} as a subcategory. The crucial idea is that the underlying graphs for generalized TANs allow for multiple edges between nodes. And it is shown that \textbf{GTAN} has finite limits.