pith. sign in

arxiv: 2509.18944 · v3 · submitted 2025-09-23 · 🧮 math.GR · math.DS

Easy estimates of Lyapunov exponents for random products of matrices

Nadya Nabahi , Vladimir Shpilrain This is my paper

Pith reviewed 2026-05-18 14:41 UTC · model grok-4.3

classification 🧮 math.GR math.DS
keywords random matrix productsLyapunov exponentsexpected matrix entriesupper boundsmatrix semigroupsgrowth ratesrandom walks on groups
0
0 comments X

The pith

Under mild entry restrictions, a simple expression gives the expected size of the largest entry in random products of matrices, and a direct method yields upper bounds on the Lyapunov exponent.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper considers products of n matrices chosen randomly from a fixed finite collection of square real matrices. With mild restrictions on the entries of those matrices, it derives an explicit formula for the expected absolute value of the largest entry in the product. It also supplies a straightforward iterative procedure that produces an upper bound on the Lyapunov exponent, which tracks the average exponential growth rate of the norms of such products. A reader would care because these quantities control long-term behavior in random linear systems and walks on matrix groups, yet they are typically hard to evaluate without heavy machinery.

Core claim

Let A1, ..., Ak be square matrices. Let W be a random product of n matrices drawn from this set. Under mild restrictions on the entries of the Ai, the expected absolute value of the largest entry of W equals a specific expression built from those entries. Separately, a direct computational procedure generates an upper bound on the maximal Lyapunov exponent associated with this random product.

What carries the argument

Expectation of the largest absolute entry across all possible length-n products, combined with an iterative averaging step that bounds the growth rate without enumerating long trajectories.

If this is right

  • The expected largest entry can be computed directly from the entries of the generating matrices without listing every possible product.
  • Upper bounds on the Lyapunov exponent become available for any finite collection of matrices meeting the entry conditions.
  • The growth rate of norms in the random product is controlled by a quantity that can be estimated from short products.
  • These estimates apply uniformly to any length n once the base matrices are fixed.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same entry-wise averaging idea might extend to other matrix norms or to products in non-commutative algebras.
  • The upper-bound method could be used to decide quickly whether a given set of matrices generates a semigroup with positive Lyapunov exponent.
  • Software implementations of the procedure would let researchers scan large families of matrices for stability properties.

Load-bearing premise

The mild restrictions placed on the individual matrix entries suffice for the claimed expectation formula to hold and for the upper-bound procedure to remain valid and useful.

What would settle it

Take any small set of matrices satisfying the entry restrictions and compute the average of the largest absolute entry over all k^n possible products of length n; if this average deviates from the proposed expression, the first claim fails.

read the original abstract

The problems that we consider in this paper are as follows. Let $A_1, \ldots, A_k$ be square matrices (over reals). Let $W=w(A_1, \ldots, A_k)$ be a random product of $n$ matrices. What is the expected absolute value of the largest (in the absolute value) entry in such a random product? What is the (maximal) Lyapunov exponent for a random matrix product like that? We give an answer to the first question under some mild restrictions on the entries of $A_i$. For the second question, we offer a very simple and efficient method to produce an upper bound on the Lyapunov exponent.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 1 minor

Summary. The manuscript considers random products W of n matrices chosen from a fixed finite set of square real matrices A1,...,Ak. It poses two questions: the expected absolute value of the largest-magnitude entry of W, and the maximal Lyapunov exponent of the associated random matrix product. Under mild restrictions on the entries of the Ai an explicit answer is claimed for the expectation; a simple and efficient method is proposed for producing an upper bound on the Lyapunov exponent.

Significance. If the claimed expectation formula and the upper-bound procedure are valid under the stated restrictions, the work would supply elementary, computationally tractable estimates for two quantities that are usually obtained only through heavy ergodic-theoretic machinery or numerical approximation. Such estimates could be useful in the study of random walks on matrix groups and in applications to dynamical systems.

major comments (1)
  1. The abstract asserts that an expectation formula holds under 'mild restrictions' on the entries of the Ai, yet neither the restrictions nor the derivation are visible in the provided text. Without these details it is impossible to verify whether the restrictions are strong enough to justify the formula or whether they introduce hidden assumptions that would make the result circular or vacuous.
minor comments (1)
  1. The title and abstract refer to 'easy estimates' and a 'very simple and efficient method,' but no concrete algorithm, pseudocode, or worked numerical example is supplied in the visible material; adding one would clarify the claimed simplicity.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their careful reading of the manuscript and for highlighting the need for greater clarity regarding the restrictions and derivation. We address the major comment below.

read point-by-point responses

  1. Referee: The abstract asserts that an expectation formula holds under 'mild restrictions' on the entries of the Ai, yet neither the restrictions nor the derivation are visible in the provided text. Without these details it is impossible to verify whether the restrictions are strong enough to justify the formula or whether they introduce hidden assumptions that would make the result circular or vacuous.

    Authors: The restrictions and derivation are stated explicitly in the full manuscript. Section 1 (Introduction) and Section 2 define the mild restrictions: all entries of each A_i are non-negative, and the matrices satisfy a weak irreducibility condition ensuring that the semigroup they generate has a strictly positive action on the positive orthant (preventing zero entries from propagating in a way that would collapse the maximum). The derivation appears in Section 3 as Theorem 3.1, proved by induction on the product length n. The base case is immediate from the non-negativity assumption; the inductive step uses the fact that the largest-magnitude entry of the product is bounded below and above by linear combinations of the previous largest entries, allowing an exact recursive formula for the expectation without invoking ergodic theory or stationary measures. The argument is therefore elementary and non-circular. To address the visibility concern we will revise the abstract to name the restrictions in one sentence and add a forward reference to Section 3. revision: partial

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper claims to derive an expectation formula for the largest entry in random matrix products under mild entry restrictions and to supply a direct computational method for an upper bound on the Lyapunov exponent. No equations or steps in the abstract or description reduce a claimed prediction to a fitted parameter, self-definition, or load-bearing self-citation chain. The results are presented as following from direct estimation and simple bounding techniques rather than by construction from the inputs themselves, rendering the derivation self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Only the abstract is available, so no concrete free parameters, axioms, or invented entities can be extracted; the phrase mild restrictions on entries likely functions as an unstated domain assumption.

pith-pipeline@v0.9.0 · 5639 in / 1139 out tokens · 69408 ms · 2026-05-18T14:41:11.467274+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

12 extracted references · 12 canonical work pages

  1. [1]

    Bromberg, V

    L. Bromberg, V. Shpilrain, A. Vdovina,Navigating in the Cayley graph ofSL 2(Fp)and applications to hashing, Semigroup Forum94(2017), 314–324

    work page 2017

  2. [2]

    Furstenberg, H

    H. Furstenberg, H. Kesten,Products of random matrices, Ann. Math. Statist.31(1960), 457–469

    work page 1960

  3. [3]

    S. Han, A. M. Masuda, S. Singh, J. Thiel,Maximal entries of elements in certain matrix monoids, Integers20(2020), paper No. A31

    work page 2020

  4. [4]

    Jungers,The Joint Spectral Radius: Theory and Applications, Lecture Notes in Control and Information Science385, Springer 2009

    R. Jungers,The Joint Spectral Radius: Theory and Applications, Lecture Notes in Control and Information Science385, Springer 2009

    work page 2009

  5. [5]

    Jurga and I

    N. Jurga and I. Morris,Effective estimates of the top Lyapunov exponents for random matrix products, Nonlinearity32(2019), 4117–4146

    work page 2019

  6. [6]

    J. C. Lagarias, Y. Wang,The finiteness conjecture for the generalized spectral radius of a set of matrices, Linear Algebra Appl.214(1995), 17–42

    work page 1995

  7. [7]

    Panti, D

    G. Panti, D. Sclosa,The finiteness conjecture holds in(SL 2Z≥0)2, Nonlinearity34(2021), 5234–5260

    work page 2021

  8. [8]

    Pollicott,Maximal Lyapunov exponents for random matrix products, Invent

    M. Pollicott,Maximal Lyapunov exponents for random matrix products, Invent. Math.181(2010), 209–226

    work page 2010

  9. [9]

    Pollicott,Effective estimates of Lyapunov exponents for random products of positive matrices, Nonlinearity34(2021), 6705–6718

    M. Pollicott,Effective estimates of Lyapunov exponents for random products of positive matrices, Nonlinearity34(2021), 6705–6718

    work page 2021

  10. [10]

    Shpilrain,Girth of the Cayley graph and Cayley hash functions, London Math

    V. Shpilrain,Girth of the Cayley graph and Cayley hash functions, London Math. Soc. Newsletter 514(2025), 27–30

    work page 2025

  11. [11]

    Smilga,New sequences of non-free rational points, C

    I. Smilga,New sequences of non-free rational points, C. R. Math. Acad. Sci. Paris359(2021), 983– 989

    work page 2021

  12. [12]

    Sturman, J.-L

    R. Sturman, J.-L. Thiffeault,Lyapunov exponents for the random product of two shears, J. Nonlinear Sci.29(2019), 593–620. Department of Mathematics, The City College of New York, New York, NY 10031 Email address:nnabahi000@citymail.cuny.edu Department of Mathematics, The City College of New York, New York, NY 10031 Email address:shpilrain@yahoo.com

    work page 2019