REVIEW 7 minor 14 references
Observations on robust diffusive stability and common Lyapunov functions
T0 review · 0 major / 7 minor · reviewed 2026-08-07 · deepseek-v4-flash
Pith's one-line read Robust diffusive stability for a pair of coupled stable discrete-time positive systems is certified by the existence of a common diagonal Lyapunov function, or by a joint linear copositive Lyapunov function when one system matrix is…
desk verdict Solid, modest extension of the CLCLF approach to robust diffusive stability; correct proofs with two presentational fixes needed. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The main argument is a contradiction via the Perron-Frobenius theorem. If an admissible coupling produced a matrix $M$ with spectral radius 1, there would be a nonnegative eigenvector $(u,w)$ for eigenvalue 1. Rearranging the block equations gives $(A-I)u = D(u-w)$ and $(B-I)w = D(w-u)$; adding them and testing against the JLCLF vector $v$ yields $v^T(A-I)u + v^T(B-I)w = 0$. Because $v^T(A-I) \le 0$ and $v^T(B-I) \le 0$ with strict negativity in the sum, the supports of $u$ and $w$ must be disjoint, and the coupling equations then force all cross-block entries of one matrix to vanish, proving that matrix is reducible. Thus, if either $A$ or $B$ is irreducible, no destabilizing diagonal $D$ can exist. For the diagonal Lyapunov result, the machinery is the block Lyapunov inequality: with a common diagonal $E \succ 0$, the matrix $\mathrm{diag}(E,E)$ makes $\mathrm{diag}(E,E)(M-I) + (M-I)^T \mathrm{diag}(E,E)$ negative definite.
What would settle it
Search over small Schur-stable nonnegative pairs $(A,B)$ with at least one irreducible matrix, find a vector $v \gg 0$ satisfying the JLCLF inequalities, and then test every diagonal $D$ with $A-D$ and $B-D$ nonnegative; if any such triple has $\rho(M) > 1$, where $M = \begin{bmatrix} A-D & D \\ D & B-D \end{bmatrix}$, Theorem 3 would be false. A second check targets the boundary case: because the proof shows failure of RDS forces both $A$ and $B$ to be reducible, a reducible pair admitting a JLCLF would determine whether the irreducibility assumption is essential rather than merely technical.
Extended reading notes
Core claim
The central claim is that for Schur-stable nonnegative matrices (having spectral radius below 1) $A$ and $B$, and for diagonal coupling matrices $D$ with $A-D$ and $B-D$ nonnegative, the coupled matrix $M = \begin{bmatrix} A-D & D \\ D & B-D \end{bmatrix}$ has spectral radius below 1 for every admissible $D$ if the two systems admit a common diagonal Lyapunov function (Theorem 2) or, when at least one of $A$ or $B$ is irreducible, a joint linear copositive Lyapunov function $v \gg 0$ with $v^T A \le v^T$, $v^T B \le v^T$, and $v^T(A+B) \ll 2v^T$ (Theorem 3). The JLCLF condition is strictly weaker than the common linear copositive condition used in earlier work, so it certifies robust diffusive stability in cases where previous sufficient conditions do not apply, as Examples 1 and 2 show. For extended Leslie matrices, the paper proves that dispersal restricted to a single age class cannot cause instability (Theorem 4) and gives a sufficient condition for general Leslie-type dispersal based on row-selection matrices (Proposition 1).
Load-bearing premise
The load-bearing assumption is that at least one of the two system matrices is irreducible, meaning it cannot be permuted into block-triangular form with two independent subpopulations; if both are reducible, the contradiction argument collapses and the JLCLF condition is not shown to imply robust diffusive stability.
Editorial extensions
If this is right
- Any pair of Schur-stable nonnegative matrices sharing a diagonal solution of the Stein inequalities is robustly diffusively stable against all diagonal couplings.
- A joint linear copositive Lyapunov function, which only asks the average of the two systems to contract, is a sufficient RDS certificate whenever one system matrix is irreducible, making the earlier CLCLF condition less conservative.
- For extended Leslie matrices, movement from a single age class can never destabilize a coupled two-patch system.
- If the row-maximum Leslie matrices $S_1$ and $S_2$ in Proposition 1 are Schur-stable, then the coupled system is robustly diffusively stable for every admissible Leslie-type dispersal matrix.
- Example 3 shows that allowing dispersal from two age classes can already produce dispersal-driven growth, so the one-class restriction in Theorem 4 is not merely technical.
Reading between the lines
- Implicit extension: the proof shows that failure of RDS forces both $A$ and $B$ to be reducible; this suggests the irreducibility assumption is exactly what separates the JLCLF certificate from the reducible case, and reducible pairs may need support-based conditions or a richer certificate.
- Testable extension: since JLCLF existence is a linear-programming feasibility problem, one could screen large Leslie-matrix metapopulation models for robust diffusive stability with a single LP solve, catching cases that a CLCLF test would miss.
- Neighbouring problem: an analogous joint condition for more than two coupled systems, requiring the sum of the maps to contract, would likely follow from the same averaging idea, though the support partition argument would need a multi-block version.
- Applied reading: Theorem 4 matches the ecological intuition that dispersal restricted to one age class is safe for stable age-structured populations, while multi-class dispersal can create the dispersal-driven growth described in the earlier literature.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper studies robust diffusive stability (RDS) for a pair of coupled, Schur-stable nonnegative matrices A and B in discrete time, with admissible diagonal diffusion matrices D such that A-D and B-D are nonnegative. It proves three sufficient conditions for RDS. Theorem 2 shows that a common diagonal Lyapunov function, in either the Stein form or the continuous-time-type Lyapunov form, implies RDS. Theorem 3 shows that a joint linear copositive Lyapunov function (JLCLF), together with irreducibility of at least one of A or B, implies RDS. Theorem 4 and Proposition 1 give results for extended Leslie matrices: RDS holds when diffusion is confined to one nonzero row, and a sufficient condition based on row-selection matrices is given. Examples illustrate that these conditions are distinct from common linear copositive Lyapunov functions and that RDS can fail when diffusion acts through more than one row.
Significance. The paper's main value is broadening the class of sufficient certificates for robust diffusive stability beyond the common linear copositive Lyapunov function conditions of [7,8]. The JLCLF result under an irreducibility assumption is a clean and useful observation. The proofs are elementary, self-contained, and, as far as I checked, correct: the algebra in Theorem 2, the reducibility argument in Theorem 3, and the Leslie arguments in Theorem 4 and Proposition 1 are valid. The numerical examples are appropriate and illustrate the differences between the conditions. The results should interest researchers in positive systems and stage-structured population dynamics. The paper is generally honest about the scope of its results, with the exception of an overly terse abstract that omits the irreducibility hypothesis.
minor comments (7)
- [Abstract] The abstract states that the weaker condition of joint linear copositive function existence is sufficient for RDS without mentioning the irreducibility hypothesis of Theorem 3; please qualify the statement (e.g., 'when at least one of A and B is irreducible') so that it matches the theorem.
- [Section 4, Theorem 3 proof] The proof begins by assuming there exists D with ρ(M)=1, but it does not justify why failure of RDS yields such a D. Add the standard continuity argument: for any admissible D, the segment tD lies in D_{A,B} for t∈[0,1], ρ(M(0))=max(ρ(A),ρ(B))<1, and if some D has ρ(M)>1, then by continuity there is a t with ρ(M)=1. The same argument is already used implicitly in Theorem 4 and should be stated here.
- [Section 4, Example 2] The example says that v^T=(1 1) satisfies v^T(A+B) ≤ 2v^T, but the actual value is v^T(A+B)=(1.75,1.75) ≪ (2,2), so the strict inequality required by Definition 1 holds; please correct the displayed inequality.
- [Section 5, Theorem 4 proof] The proof states 'as we have already shown that v > 0, w > 0', but only v≠0 and w≠0 were shown. The contradiction still follows because for a Schur-stable nonnegative matrix A, Av ≥ v with v≥0 and v≠0 is impossible; please revise the wording to reflect the weaker condition actually used.
- [Section 5, paragraph before Theorem 4] The second bullet in the motivation for L^1_{A,B} says that one-nonzero-row diffusion corresponds to movement from the two oldest age classes, but L^1_{A,B} consists of matrices with one nonzero row; movement from two age classes would require two nonzero rows, so this bullet appears to be a mistake and should be corrected.
- [Section 3, Example 1] When ruling out a solution to (3), the argument considers only E = diag(1,ε); this is without loss of generality because the inequalities are homogeneous under positive scaling, but the scaling step should be stated explicitly for clarity.
- [Section 4, Theorem 3 proof] There is a typo in the sentence 'bij = 0 for all i ∈ Iu ∪ Iu, j ∈ Iw'; the first union should read Iu ∪ I0.
Circularity Check
No significant circularity; the theorems are proved from standard matrix facts and the cited prior results are not the target conclusion.
full rationale
The paper's derivation chain is self-contained and does not reduce its claims to its inputs. Theorem 2 is proved by direct algebra from the Schur-stability characterizations in Theorem 1 and the Lyapunov inequality; the coupled-system conclusion follows from a block-diagonal negative-definite construction, not from assuming RDS. Theorem 3 begins by assuming a joint linear copositive Lyapunov function and then argues by contradiction: if some admissible coupling destabilizes the system, the Perron-Frobenius eigenvector forces disjoint supports and, in turn, forces both A and B to be reducible. This is a genuine proof, not a restatement of the JLCLF definition or a hidden fit. The cited prior results are background or standard support: Theorem 5, used in Proposition 1, is attributed jointly to [5] and [14], where [14] is an independent external source, and the relevant theorem does not contain the RDS conclusion. The self-citations [2,3,5] are not invoked as unverified uniqueness claims or as substitutes for proof. There are no fitted parameters, no prediction called by another name, and no renaming of a known empirical pattern. The abstract's omission of the irreducibility hypothesis in Theorem 3 is a presentational overreach, not a circular step. A minor typo in the proof ('Iu \cup Iu' where context requires 'Iu \cup I0') is clerical and does not affect the argument. Overall, the central results are derived, not assumed, and any self-citation is marginal rather than load-bearing.
Assumptions & free parameters
assumptions (5)
- standard math Schur stability of a nonnegative matrix is equivalent to the existence of a positive vector v with v^T A << v^T and to diagonal positive definite solutions of the Stein and Lyapunov inequalities (Theorem 1).
- standard math Perron-Frobenius theorem: a nonnegative matrix with spectral radius 1 has a nonnegative nonzero eigenvector.
- standard math Spectral radius is a continuous function of the entries of a matrix.
- standard math For nonnegative matrices, C <= S entrywise with rho(S)<1 implies rho(C)<1.
- domain assumption Admissible coupling sets such as DA,B and L1_A,B contain the ray segment tD for 0<=t<=1 whenever D is admissible.
Cite this review
Pith. "Pith review of Observations on robust diffusive stability and common Lyapunov functions." pith.science (2026). https://pith.science/paper/4SSMW6ST
@misc{pith2026250604863,
author = {Pith},
title = {Pith review of: Observations on robust diffusive stability and common Lyapunov functions},
year = {2026},
howpublished = {\url{https://pith.science/paper/4SSMW6ST}},
note = {Machine review of arXiv:2506.04863}
}
read the original abstract
We consider the problem of robust diffusive stability (RDS) for a pair of coupled stable discrete-time positive linear-time invariant (LTI) systems. We first show that the existence of a common diagonal Lyapunov function is sufficient for RDS and highlight how this condition differs from recent results using linear copositive Lyapunov functions. We also present an extension of these results, showing that the weaker condition of \emph{joint} linear copositive function existence is also sufficient for RDS. Finally, we present two results on RDS for extended Leslie matrices arising in population dynamics.
Reference graph
Works this paper leans on
- [8]
- [1]
-
[2]
O. Mason, Diagonal Riccati stability and positive time-delay systems, Systems & Control Letters 61 (1) (2012) 6–10 13
work page 2012
-
[3]
A. Yu. Aleksandrov, O. Mason, Diagonal Lyapunov?Krasovskii func- tionals for discrete-time positive systems with delay, Systems & Control Letters 63 (2014) 63–67
work page 2014
-
[4]
E. Fornasini, M. E. Valcher, Stability and Stabilizability Criteria for Discrete-Time Positive Switched Systems, IEEE Transactions on Auto- matic Control 57 (5) (2012) 1208–1221
work page 2012
- [5]
-
[6]
Z. Wu, Y. Sun, Common joint linear copositive Lyapunov functions for positive switched systems, Journal of the Franklin Institute 351 (5) (2014) 2504–2514
work page 2014
-
[7]
P. De Leenheer, Stability of diffusively coupled linear systems with an invariant cone, Linear Algebra and its Applications 580 (2019) 396–416
work page 2019
Show all 14 references
-
[9]
R. A. Horn, C. R. Johnson, Matrix Analysis , Cambridge University Press, 1985
1985
-
[10]
Kaszkurewicz, A
E. Kaszkurewicz, A. Bhaya, Matrix Diagonal Stability in Systems and Computation, Birkh¨ auser, Boston, 2000
2000
-
[11]
J.M Cushing, Matrix Models for Population, Disease, and Evolutionary Dynamics, American Mathematical Society, 2024
2024
-
[12]
MATLAB version: 24.1.0 (R2024a), Natick, Massachusetts: The Math- Works Inc.; 2024
2024
-
[13]
McGrane-Corrigan, Coupled Models of Structured Ecological Systems: Patch Dynamics, Population Demography and Stochastic Interactions , PhD Thesis, Maynooth University, 2025
B. McGrane-Corrigan, Coupled Models of Structured Ecological Systems: Patch Dynamics, Population Demography and Stochastic Interactions , PhD Thesis, Maynooth University, 2025
2025
-
[14]
Y. Song, M. Seetharama Gowda, G. Ravindran, On some properties of p-matrix sets, Linear Algebra and its Applications 290 (13) (1999) 237–246 14
1999
Reviewed August 7, 2026 · model on record in the stance chip above.
Discussion (0). Sign in to comment.