From the Volterra type Lyapunov functions of Rahman-Zou towards a competitive exclusion partition property for rank one models

Rim Adenane, Florin Avram, Andrei-Dan Halanay · 2026

Read From the Volterra type Lyapunov functions of Rahman-Zou towards a competitive exclusion partition property for rank one models by Rim Adenane, Florin Avram, Andrei-Dan Halanay (2026) 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-2605-01755.txt.

Introduction

Adenane Rim, Avram Florin, Miruna Beldiman, and Halanay Andrei-Dan partition the fate of multi-strain epidemics. Their framework covers reaction network models where several strains share one host pool through bilinear infection blocks that do not interact. It asks the global questions: does the disease die out, does one strain exclude the rest, or can strains persist together?

The engine is a Perron-Volterra Lyapunov function, a certificate built from Perron eigenvectors of the boundary dynamics and entropy terms for strains. Five pillars support it: siphon sets that mark invariant faces, Metzler Jacobians on those faces, next-generation matrices from splittings, invasion relays at eigenvalue crossings, and the Lyapunov build itself. The parameter space splits into at most one more region than there are strains, each owned by extinction or one dominant strain, with coexistence confined to tie surfaces of equal reproduction numbers. A Mathematica package automates the construction.

Catalog record

CREATORRim Adenane, Florin Avram, Andrei-Dan Halanay
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE47 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.01755

Browse the full library · About the Tower of Babel

Full text

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

From the Volterra type Lyapunov functions of Rahman-Zou towards a competitive exclusion partition property for rank one models Rim Adenane, Florin Avram, Andrei-Dan Halanay Abstract Fall, Iggidr, Sallet, and Bonzi—see also Earn and McCluskey (2025) and [ 5 ] —have studied a class of bilinear epidemic models with rank-one next-generation matrix (NGM), in which the right Perron eigenvector yields an explicit endemic equilibrium (EE), while the left Perron eigenvector provides Lyapunov functions at both the disease-free equilibrium (DFE) and the EE. Together with the graph-theoretic method of Shuai and van den Driessche (2013), these results establish strong threshold theorems: when the DFE loses stability, a unique EE emerges and becomes globally asymptotically stable, with the transition certified by an explicit global Lyapunov function. A striking extension was obtained by [ 17 ] , who proved analogous results for a two-strain rational rates (saturating) model with multiple boundary equilibria. In this setting, stability is relayed across equilibria with increasing infection levels, inducing a partition of parameter space into regions where exactly one equilibrium is certified to be globally stable by an explicit Lyapunov function. We refer to this as the competitive exclusion partition property (CEPP). We show that these results are governed by a common Perron–Volterra mechanism: the Lyapunov functions are sums of a Volterra entropy on resident variables and a linear functional on invaders, weighted by left Perron eigenvectors of transversal Jacobians. A balance identity eliminates higher-order coupling terms, reducing stability certification to the recursive evaluation of invasion numbers on the siphon lattice. For a two-strain model with concave increasing incidence, we prove the CEPP completely, providing explicit global Lyapunov functions at all four equilibria (Theorem 4 ). For n ≥ 3 n\geq 3 strains, the same mechanism works, as long as all the strains involve only one compartment. Beyond that, we also obtain global Lyapunov functions at all equilibria for models with one irreducible rank one strain, and one (or more) scalar strains in Theorem 5 ; whether further extensions are possible remains an open problem. We have provided an algorithmic implementation in the Mathematica package EpidCRN ( https://github.com/florinav/EpidCRNmodels ). The method organizes the dynamics via the minimal siphon lattice, recursively computes Perron eigenvectors of transversal Jacobians, and constructs candidate Lyapunov functions for all equilibria, producing a parameter partition into regions with a unique locally stable equilibrium and, for n = 2 n=2 , explicit global stability certificates. Glossary 1 Introduction This paper studies multi-strain epidemic models through the lens of chemical reaction network (CRN) theory, with two goals: to unify and extend explicit Lyapunov stability results of [ 10 , 16 , 7 , 8 , 17 ] and [ 19 ] , and to implement the resulting algorithmic structure in the Mathematica package EpidCRN . Our framework is encapsulated by the following definitions: Definition 1 (Positive / non-negative ODE) . An ODE is called positive [ 18 ] or non-negative [ 12 ] if the non-negative orthant \mathbb{R}^{n}_{\geq 0}:=\{x\in\mathbb{R}^{n}:x_{i}\geq 0,\;i=1,\dots,n\} is forward invariant under the flow. Definition 2 (Stoichiometric and chemical representation) . A stoichiometric representation of f is a pair (\Gamma,r) such that f(x)=\Gamma r(x),\qquad r(x)\geq 0, where \Gamma\in\mathbb{R}^{n\times n_{R}} is constant and r:\mathbb{R}^{n}_{\geq 0}\to\mathbb{R}^{n_{R}}_{\geq 0} is locally Lipschitz. An ODE is called chemical if \Gamma_{i\rho}<0\implies r_{\rho}(x)=x_{i}\tilde{r}_{\rho}(x),\qquad\tilde{r}_{\rho}(x)\geq 0, and, moreover, r_{\rho} depends only on variables in \operatorname{supp}(r_{\rho})\subseteq\Gamma^{-}_{i} whenever \Gamma_{i\rho}<0 . Furthermore, we assume that each rate r_{\rho}(x) is monotone non-decreasing with respect to each variable x_{k} for k\in\operatorname{supp}(r_{\rho}) . Definition 3 (rank one ME models) . A chemical ODE is called a rank one (RS) model if the diagonal Jacobians J_{\sigma} of minimal siphons \sigma (which are Metzler by [ 4 , 6 , 5 ] ) admit all regular splittings J_{\sigma}=F_{\sigma}-V_{\sigma} in the sense of [ 21 , 20 ] such that the corresponding next generation matrixs K_{\sigma}=F_{\sigma}V_{\sigma}^{-1} have rank one. We are motivated by the following observation: explicit (rational) results of [ 10 , 16 , 7 , 8 , 19 ] and others share a common structure: they are rank-one models, and exploit the computation of left and right Perron eigenvectors of rank-one next generation matrixs (NGM). Why CRN methods. The first characteristic of most models in mathematical epidemiology (RS), ecology, and immuno-virology is that they are positive ODEs (or stochastic models whose mean field is a positive ODE). This means that important phenomena occur when trajectories reach the boundary of the positive orthant, and that the notion of invariant boundary faces is crucial. While boundary faces have been studied in classical ODE theory, the systematic exploitation of simplifications arising for positive ODEs was initiated by CRN works, see notably [ 13 , 9 ] , who showed that all positive polynomial models admit (non-unique) mass-action representations. Beyond the polynomial case, but assuming a stoichiometric representation, crucial results relating boundary \omega -limit points to invariant boundary faces and to the Petri-net concept of siphons were obtained in [ 2 ] . A recent proof of the folklore result that the Jacobian has a triangular block form on siphon faces was given in [ 4 , 6 ] , and [ 5 ] noted that transversal Jacobians on siphons have the Metzler/cooperativity property. This property further ensures the existence of left and right Perron eigenvectors which, in all cases we have studied, provide Lyapunov functions and escape directions from siphon faces respectively. We pause here to recall a useful result on the spectral structure of Metzler matrices [ 15 ] : Theorem 1 (Frobenius normal form of Metzler matrices) . Let M\in\mathbb{R}^{n\times n} be Metzler (i.e. M_{ij}\geq 0 for i\neq j ). Then there exists a permutation matrix P such that P^{-1}MP=\begin{pmatrix}B_{1}&*&\cdots&*\\ 0&B_{2}&\cdots&*\\ \vdots&&\ddots&*\\ 0&0&\cdots&B_{r}\end{pmatrix}, where each diagonal block B_{j} is an irreducible Metzler matrix. Moreover: 1. \lambda_{\max}(M)=\max_{1\leq j\leq r}\lambda_{\max}(B_{j}) , where \lambda_{\max} denotes spectral abscissa. 2. For each j , the Perron–Frobenius theorem yields u_{j}\gg 0 with B_{j}u_{j}=\lambda_{\max}(B_{j})u_{j} . Extending u_{j} by zeros to \mathbb{R}^{n} gives a nonnegative right eigenvector w_{j}\geq 0 of M satisfying Mw_{j}=\lambda_{\max}(M)w_{j} . 3. Letting B_{j^{*}} denote the block satisfying \lambda_{\max}(B_{j})=\lambda_{\max}(M) , it holds that w_{j^{*}} is the Perron right eigenvector of M . Remark 1 (Role of Frobenius normal form of Metzler matrices in this paper) . In n -strain models, the transversal Jacobian J_{\scriptscriptstyle DFE}^{\perp} at the DFE siphon face has a block- diagonal structure with n blocks J_{\sigma_{j}}^{\perp},j=1,...,n . In the case when all blocks are scalar, Theorem 1 is trivial, but it becomes substantive in the matrix-infection extension, where each block J_{\sigma_{j}}^{\perp} is an irreducible Metzler matrix. Part 1 then requires comparing invasion numbers, and part 3 provides dominant escape directions — the Perron eigenvectors of invading blocks. This is the spectral foundation of the CEP relay. 1.1 Short review of some Volterra type Lyapunov function results in ME and CRN Since Volterra, Lyapunov functions for positive ODEs z^{\prime}=f(z) have been often constructed using the Bregman divergence: D_{i}(z_{i}):=z_{i}-z_{i}^{*}-z_{i}^{*}\,\ln\frac{z_{i}}{z_{i}^{*}}=z_{i}^{*}G(u_{i})\geq 0,\qquad u_{i}:=\frac{z_{i}}{z_{i}^{*}},G(u)=u-1-ln(u) associated with x\mapsto x\ln x . “Candidate Lyapunov functions” include often terms of the form: \mathcal{L}(z):=\sum_{i=1}^{n}l_{i}D_{i}(z_{i}),\qquad l_{i}>0. (1) Note that D_{i}^{\prime}=(1-u_{i}^{-1})z_{i}^{\prime} yields the explicit formula: \mathcal{L}^{\prime}=\sum_{i=1}^{n}l_{i}\!\Bigl(1-\frac{1}{u_{i}}\Bigr)f_{i}(z). (2) Unconditional GAS under complex balance and strong endotacticity in chemical reaction network. Classical CRNT [ 14 ] proves unconditional global asymptotic stability (GAS) of complex-balanced equilibria within their

Read the complete text