Quotient toposes of discrete dynamical systems

TL;DR

This paper classifies classes of discrete dynamical systems closed under certain limits and colimits, linking them to ideals of a product poset, and addresses an open problem in topos theory using a novel framework.

Contribution

It provides a classification of discrete dynamical systems related to topos theory and introduces a new framework for analyzing quotient toposes.

Findings

Classes correspond bijectively to ideals of a product poset

Non-periodic behaviors are characterized by injective epimorphisms from 

Provides a non-trivial example addressing Lawvere's open problem

Abstract

This paper gives a classification of classes of discrete dynamical systems (a set equipped with an endofunction) closed under finite limits and small colimits. The conclusion is simple: they bijectively correspond to the ideals of the product poset $\mathbb{N} \times \mathbb{N}$, where the first $\mathbb{N}$ is ordered by the usual order and the second is by the divisibility. Our method is based on a detailed analysis of the behaviors of states, especially non-periodic behaviors, in discrete dynamical systems. Specifically, extending the fundamental quantity, time until entering a loop, to even those states that do not enter a loop plays a crucial role. Our classification is closely related to epimorphisms from $\mathbb{N}$ in the category of monoids. There are countably many injective epimorphisms from $\mathbb{N}$, including $\mathbb{N} \to \mathbb{Z}$. Those injective epimorphisms correspond to the non-periodic behaviors of states of discrete dynamical systems. We discuss this point at the end of the paper. This fun puzzle is motivated by an open problem in topos theory. Lawvere left open problems in topos theory on his webpage, and the first problem is called quotient toposes. The main theorem of this paper provides a non-trivial example of this problem, which is not implied by any known results. This paper also provides a theoretical framework to address the open problem. We define a preorder among the objects of a Grothendieck topos (which we have named generative order), which enables us to reduce calculations of quotient toposes to calculations of objects. Our method is its application to the topos of discrete dynamical systems.