On the occupation measure of evolution models with vanishing mutations

Mario Bravo; Mathieu Faure; Michel Bena\"im

arxiv: 2602.06829 · v2 · submitted 2026-02-06 · 🧮 math.PR

On the occupation measure of evolution models with vanishing mutations

Michel Bena\"im , Mario Bravo , Mathieu Faure This is my paper

Pith reviewed 2026-05-16 06:18 UTC · model grok-4.3

classification 🧮 math.PR

keywords occupation measurevanishing mutationstime-inhomogeneous Markov chainsinvariant distributionenergy barriertree-optimality gapalmost sure convergenceevolution models

0 comments

The pith

If the mutation parameter vanishes at a rate controlled by the energy barrier, the empirical occupation measure converges almost surely to a specific invariant distribution of the limiting Markov chain.

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

The paper establishes that in evolution models with time-decreasing mutation rates, the long-run occupation measure settles almost surely on one particular invariant measure of the zero-mutation limit chain, provided the decay occurs at a sufficiently slow pace. The proof proceeds by embedding the models inside a larger family of finite-state time-inhomogeneous Markov chains whose transition rates are controlled by the energy landscape of the limit process. An explicit rate of convergence is also obtained in terms of the tree-optimality gap of that limit chain. A reader would care because the result describes how gradually vanishing noise can still produce deterministic long-term behavior in stochastic population models.

Core claim

When the mutation parameter vanishes at a rate governed by the energy barrier of the limit process, the occupation measure of the time-inhomogeneous Markov chain converges almost surely to the invariant distribution of the limiting chain; the speed of this convergence is governed by the tree-optimality gap of the same limit.

What carries the argument

The energy barrier of the limit process, which dictates the admissible decay rate of the mutation parameter and thereby controls the convergence of the occupation measure in the larger class of time-inhomogeneous chains.

If this is right

The occupation measure stabilizes almost surely to the invariant distribution of the zero-mutation limit chain.
An explicit convergence rate is given by the tree-optimality gap of the limit process.
The result applies to any finite-state time-inhomogeneous Markov chain whose rate control satisfies the energy-barrier condition.
Almost-sure convergence holds uniformly across the admissible decay rates.

Where Pith is reading between the lines

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

The same barrier-controlled decay may therefore allow numerical schemes to approximate ground-state equilibria of large state spaces without persistent noise.
The tree-optimality gap supplies a quantitative diagnostic that can be checked on any given limit chain before running the inhomogeneous dynamics.

Load-bearing premise

The mutation parameter must vanish at a rate controlled by the energy barrier of the limit process.

What would settle it

Numerical simulation of the finite-state chain with mutation rates decaying slower than the energy-barrier threshold should produce occupation measures that fail to converge to the predicted invariant distribution.

read the original abstract

We study the almost sure convergence of the occupation measure of evolution models where mutation rates decrease over time. We show that if the mutation parameter vanishes at a controlled rate, then the empirical occupation measure converges almost surely to a specific invariant distribution of a limiting Markov chain. Our results are obtained through the analysis of a larger class of time-inhomogeneous Markov chains with finite state space, where the control on the mutation parameter is explained by the energy barrier of the limit process. Additionally, we derive an explicit convergence rate, explained through the tree-optimality gap, that may be of independent interest.

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.

Desk Editor's Note private letter to a colleague

The paper gives almost-sure convergence of occupation measures for finite-state time-inhomogeneous Markov chains when the mutation rate vanishes at a speed set by the energy barrier of the limit chain, together with an explicit rate bound via the tree-optimality gap.

read the letter

The central result is that, for these evolution models, if the mutation parameter goes to zero at a rate controlled by the energy barriers, the empirical occupation measure converges almost surely to one specific invariant measure of the frozen limiting chain. They obtain this by working with a larger class of time-inhomogeneous finite-state chains and then specializing back to the evolutionary setting. The explicit rate expressed through the tree-optimality gap is the clearest new piece; it turns the qualitative convergence into something quantitative that could be used elsewhere in stochastic approximation or population dynamics models. That part looks like a clean addition to the existing literature on occupation measures. The approach follows standard lines for inhomogeneous chains, so the technical steps are plausible once the rate condition is granted. The main soft spot is the precise control on the vanishing rate. The stress-test note points out that tying the rate only to the height of the highest barrier may not automatically make the total perturbation sum converge almost surely when several near-optimal trees exist; the summed effect over lower barriers could still diverge on a positive-probability set. The abstract claims the barrier explains the admissible rates, but without the proofs it is hard to see whether they obtain a uniform bound over paths or just a max-barrier estimate. If the latter, the almost-sure claim would need extra work. No circularity or invented entities appear in the statement. This is for readers already comfortable with Markov-chain approximations in evolutionary models or optimization. Someone working on rates for inhomogeneous processes will find the tree-gap device useful even if the main theorem needs tightening. It is worth sending to referees because the claims are specific, the setting is natural, and the rate result stands on its own.

Referee Report

2 major / 2 minor

Summary. The paper claims that in evolution models with time-decreasing mutation rates, if the mutation parameter vanishes at a rate controlled by the energy barrier of the limiting process, then the empirical occupation measure converges almost surely to a specific invariant distribution of the limiting Markov chain. The argument proceeds by analyzing a broader class of time-inhomogeneous finite-state Markov chains, with the admissible vanishing rates determined by the energy landscape; an explicit convergence rate expressed via the tree-optimality gap is also derived.

Significance. If the central claims hold, the work supplies a quantitative metastability result for occupation measures in time-inhomogeneous chains that arise in evolutionary dynamics. The explicit rate tied to the tree-optimality gap and the embedding into a general class of inhomogeneous chains could be useful for applications in population genetics and stochastic search algorithms. The approach avoids parameter fitting and derives rates directly from the energy barrier structure.

major comments (2)

[main convergence theorem] The main convergence theorem (stated after the energy-barrier condition in the abstract and presumably proved in §3–4): the tree-optimality gap controls the height of the highest barrier but does not automatically guarantee that the summed perturbation over all paths remains finite almost surely when multiple near-optimal trees exist. The manuscript must verify that the Borel-Cantelli-type condition still forces the chain into the absorbing components of the limit on a set of probability one; otherwise the a.s. convergence claim fails.
[§2] §2 (definition of the controlled-rate condition): the admissible vanishing rate is tied directly to the energy barrier, yet the argument does not appear to bound the total variation distance uniformly across all possible paths that cross lower barriers at late times. A concrete estimate showing that the cumulative perturbation sum satisfies the required summability condition under the stated hypothesis is needed.

minor comments (2)

[abstract] Clarify the precise definition of the occupation measure (empirical measure versus time-average) and its relation to the invariant distribution of the frozen chain; the notation is used without an explicit equation reference in the abstract.
[introduction] Add a short comparison paragraph with existing results on occupation measures for inhomogeneous chains (e.g., references to works on metastability with slowly varying parameters) to situate the novelty of the tree-optimality-gap rate.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful reading and constructive suggestions. The two major comments identify places where the exposition of the Borel-Cantelli argument and the uniform control on perturbations can be strengthened. We address each point below and will incorporate the requested clarifications and a new auxiliary lemma in the revised manuscript.

read point-by-point responses

Referee: [main convergence theorem] The main convergence theorem: the tree-optimality gap controls the height of the highest barrier but does not automatically guarantee that the summed perturbation over all paths remains finite almost surely when multiple near-optimal trees exist. The manuscript must verify that the Borel-Cantelli-type condition still forces the chain into the absorbing components of the limit on a set of probability one.

Authors: The finite-state assumption implies only finitely many spanning trees, so the union bound over near-optimal trees multiplies the individual tail probabilities by a constant. Under the controlled vanishing rate (which forces the perturbation probabilities to decay faster than 1/n), the resulting series remains summable. The proof of Theorem 3.1 already applies Borel-Cantelli to the union of these events; we will add an explicit sentence after the theorem statement spelling out the union-bound step and confirming that the a.s. convergence to the absorbing components is preserved. revision: yes
Referee: §2 (definition of the controlled-rate condition): the admissible vanishing rate is tied directly to the energy barrier, yet the argument does not appear to bound the total variation distance uniformly across all possible paths that cross lower barriers at late times. A concrete estimate showing that the cumulative perturbation sum satisfies the required summability condition under the stated hypothesis is needed.

Authors: We agree that an explicit uniform estimate improves readability. In the revised Section 2 we will insert Lemma 2.3, which bounds the total-variation distance for any path crossing a barrier of height strictly less than the tree-optimality gap by an exponentially small term (uniform in the starting point) times the current mutation parameter. Summing these bounds yields a convergent series under the hypothesis on the vanishing rate, thereby verifying the required summability. The new lemma uses only the existing energy-landscape assumptions and does not alter the main statements. revision: yes

Circularity Check

0 steps flagged

Derivation self-contained with no circular reductions

full rationale

The paper analyzes convergence of occupation measures for time-inhomogeneous Markov chains under vanishing mutation rates controlled by energy barriers of the limit process. The control condition and explicit rates via tree-optimality gap are derived from the structure of the frozen chain and standard Borel-Cantelli arguments on perturbations, without reducing to fitted parameters, self-definitions, or load-bearing self-citations. The abstract and claim description present an independent extension to a larger class of chains, with no equations or steps that equate outputs to inputs by construction. This is the expected honest non-finding for a self-contained probabilistic analysis.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The paper relies on standard Markov chain theory for finite state spaces and invariant distributions; no free parameters, invented entities, or ad-hoc axioms are mentioned in the abstract.

axioms (1)

standard math Standard ergodic properties and existence of invariant distributions for finite-state Markov chains
Invoked to identify the limiting distribution of the occupation measure.

pith-pipeline@v0.9.0 · 5388 in / 1127 out tokens · 30669 ms · 2026-05-16T06:18:41.208412+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

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

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

energy barrier e(c) := max Elev(x,y) − V(x) − V(y); tree-optimality gap θ; λ(Pε) ≥ C ε^{e(c)}
IndisputableMonolith/Foundation/RealityFromDistinction.lean reality_from_one_distinction unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

A-vanishing mutation rate with 2A e(c) < 1 implies v_n → π* a.s.

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.