Persistent homology of Morse decomposition in Markov chains based on combinatorial multivector fields

Donald Woukeng · 2025

Read Persistent homology of Morse decomposition in Markov chains based on combinatorial multivector fields by Donald Woukeng (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-22308.txt.

Introduction

Donald Woukeng, writing alone, builds a persistence framework for random processes in a 2025 mathematics paper. His setting is a Markov chain, a system that hops between states with set odds. He studies its Morse decomposition, a split of the system into stable pockets with flow between them. The tool kit comes from combinatorial multivector fields, discrete arrows defined right on the state graph.

In this version, the filtration that orders the analysis runs on transition odds, not on distance in space. Woukeng builds the fields straight from the chain's transition table. As the odds vary, the pockets appear, merge, and vanish in a hierarchy. That record forms a persistence diagram, a map of which structures last. Each pocket is scored by homology and its Conley index. The author offers the scheme as a structured way to read recurrence and stability in finite random systems.

Catalog record

CREATORDonald Woukeng
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE38 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-22308

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Persistent homology of Morse decomposition in Markov chains based on combinatorial multivector fields Donald Woukeng Abstract In this paper, we introduce a novel persistence framework for Morse decompositions in Markov chains using combinatorial multivector fields. Our approach provides a structured method to analyze recurrence and stability in finite-state stochastic processes. In our setting filtrations are governed by transition probabilities rather than spatial distances. We construct multivector fields directly from Markov transition matrices, treating states and transitions as elements of a directed graph. By applying Morse decomposition to the induced multivector field, we obtain a hierarchical structure of invariant sets that evolve under changes in transition probabilities. This structure naturally defines a persistence diagram, where each Morse set is indexed by its topological and dynamical complexity via homology and Conley index dimensions. 1 Introduction 1.1 Motivation The study of dynamics has seen a growing interaction between topological methods and discrete combinatorial approaches, especially in understanding recurrence and stability in finite-state models. One of the key frameworks in this domain is Morse decomposition, which provides a hierarchical structure for understanding long-term system behavior. Persistent homology has been instrumental in tracking topological features across scales, but its direct application to Morse decompositions has remained a challenge. Several works have addressed the persistence of topological invariant for dynamics data or dynamical systems, it is the case for example of [ 5 , 7 , 8 ] where the study of persistence is done for Morse decomposition, Conley index and Conley-Morse graph. However, a key missing component in these works is a stability theorem that ensures that Morse decompositions remain robust under small perturbations in the system’s transition dynamics. In particular, no prior work has established a formal stability result for Morse set persistence in Markov chains. Our work addresses this gap by providing a rigorous foundation for the stability of Morse decompositions in combinatorial multivector fields constructed from Markov chains. Markov chains are widely used to model stochastic processes in various real-world domains, including biological networks , financial markets, climate systems, and reinforcement learning. In these applications, understanding long-term behavior and recurrent structures is crucial for predicting trends, identifying stable states, and classifying different system dynamics. Some traditional methods for analyzing Markov chains focus on stationary distributions, spectral properties, and absorbing states [ 11 , 2 , 12 ] , but they often lack a topological and hierarchical perspective on the structure of state transitions. Some recent work have been done for the analysis of Markov chains using classical persistence [ 13 , 17 , 9 ] , but they do not take into account the potential dynamics hidden in the Markov chains. Our method introduces a new classification approach for Markov chains by applying persistent homology to Morse decompositions, allowing us to track how recurrent structures change as transition probabilities vary. By constructing persistence diagrams indexed by homology and Conley indices, we provide a robust way to classify Markov chains based on their intrinsic recurrence structures rather than just numerical transition probabilities. 1.2 Overview of the result In this paper, we develop a persistence framework for Morse decompositions in Markov chains using combinatorial multivector fields. Our main contributions is as follows: Theorem 1 (Stability of Morse Set Persistence Diagrams) . Let P and P^{\prime} be two transition matrices of a Markov process with n states such that: \|P-P^{\prime}\|_{\infty}<\delta. Then, the bottleneck distance between their Morse set persistence diagrams satisfies: d_{B}(D(P),D(P^{\prime}))<C\delta, (1) where C depends only on the structure of the Markov process. 2 Preliminaries 2.1 Relations and Posets Let X be a set. A binary relation on X is defined as a subset R\subset X\times X . We use the standard notation xRy to indicate that (x,y)\in R . A relation \leq that is reflexive, antisymmetric, and transitive is called a partial order , and the pair (X,\leq) is known as a poset . A subset A\subset X of a poset (X,\leq) is called an upper set if \{z\in X\mid\exists_{x\in A}x\leq z\}\subset A . Similarly, A\subset X is a lower set if \{z\in X\mid\exists_{x\in A}z\leq x\}\subset A . A subset A\subset X is said to be convex in a poset (X,\leq) if for every x,z\in A and y\in X satisfying x\leq y\leq z , it follows that y\in A . Equivalently, a convex set is the intersection of a lower set and an upper set. A relation R is called an equivalence relation if it satisfies reflexivity, symmetry, and transitivity. The equivalence class of an element x under R is given by [x]_{R}:=\{y\in X\mid xRy\} . A partition of X is a collection \mathcal{V} of non-empty subsets of X such that for every distinct A,B\in\mathcal{V} , we have A\cap B=\emptyset and \bigcup\mathcal{V}=X . A partition \mathcal{V} induces an equivalence relation R defined by xRy if there exists some A\in\mathcal{V} such that x,y\in A . We denote the equivalence class of a point x\in X under partition \mathcal{V} as [x]_{\mathcal{V}} . 2.2 Topological Spaces Given a topology \mathcal{T} on X , the pair (X,\mathcal{T}) is referred to as a topological space . When the topology \mathcal{T} is understood, we may simply refer to X as a topological space. The interior, closure, and boundary of a subset A\subset X with respect to \mathcal{T} are denoted by \operatorname{int}_{\mathcal{T}}A , \operatorname{cl}_{\mathcal{T}}A , and \operatorname{bd}_{\mathcal{T}}A , respectively. The mouth of A is defined as \operatorname{mo}_{\mathcal{T}}A:=\operatorname{cl}_{\mathcal{T}}A\setminus A . When the topology is clear from context, we may use the shorthand \operatorname{cl}_{X}A or omit the subscript altogether. For a subset Y\subset X , the induced topology on Y is given by \mathcal{T}_{Y}:=\{U\cap Y\mid U\in\mathcal{T}\} . The closure of A\subset Y with respect to \mathcal{T}_{Y} is denoted as \operatorname{cl}_{\mathcal{T}_{Y}}A or simply \operatorname{cl}_{Y}A . The same applies to the interior, boundary, and mouth. A subset A\subset X is called locally closed if every point x\in A has a neighborhood U in X such that A\cap U is closed in U . Proposition 2.1 . [ 10 , Problem 2.7.1] . A subset A of a topological space X satisfies the following equivalent conditions: (i) A is locally closed, (ii) The mouth \operatorname{mo}A=\operatorname{cl}A\setminus A is closed, (iii) A can be expressed as the difference of two closed sets in X , (iv) A is the intersection of an open set and a closed set in X . Moreover, the finite intersection of locally closed sets remains locally closed. A topological space is called a T_{0} space if for every pair of distinct points in X , at least one of them has a neighborhood that does not contain the other. We are particularly interested in finite T_{0} spaces, which, by Alexandroff’s theorem [ 1 ] , can be associated with posets. Theorem 2.2 . [ 1 ] For a finite poset (P,\leq) , the collection of all upper sets of \leq forms a T_{0} topology \mathcal{T}_{\leq} on P . Conversely, for any finite T_{0} space (X,\mathcal{T}) , defining x\leq_{\mathcal{T}}y by x\in\operatorname{cl}_{\mathcal{T}}\{y\} results in a partial order on X . These two correspondences are mutually inverse. This equivalence allows all topological properties of a finite T_{0} space to be expressed in terms of posets and vice versa. In particular, given a finite T_{0} space (X,\mathcal{T}) with an associated poset (X,\leq) , the closure and

Read the complete text