REVIEW 2 major objections 3 minor 8 references
Stationary Distributions for the Voter Model in $d\geq 3$ are Factors of IID
T0 review · 2 major / 3 minor · reviewed 2026-08-14 · deepseek-v4-flash
Pith's one-line read The paper proves that on $\mathbb{Z}^d$ with $d\ge 3$, every extremal stationary distribution $\mu_p$ of the voter model is a Bernoulli shift, constructed explicitly as a factor of IID.
desk verdict The voter model stationary measures in d>=3 are Bernoulli shifts; the explicit coupling is new and the proof holds up under scrutiny. 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 central object is an explicit two-time coupling map $F_t$ built from coalescing simple random walks, the dual process of the voter model. Given a sparse set of walkers with an IID Bernoulli coloring, the map selects each walker's path from a biased mixture of two measures, $W_0$ and $W_1$, whose average is the true law of a simple random walk; this preserves the marginal distribution exactly while making walkers of the same color preferentially coalesce. The construction uses only IID inputs: independent random-walk proposals, uniform random variables, and Bernoulli coloring, grouped into finite sparse sets to define a translation-invariant ordering. The paper concatenates these maps at times $2^k-1$, and controls the total number of color changes via negative-correlation estimates for coalescing walks and coalescence-time estimates for independent walks, proving that the coupled opinions converge almost surely.
What would settle it
Compute, for coalescing simple random walks on $\mathbb{Z}^d$ run until time $2^k-1$, the three-point probability $\mathbb{P}[a_1=\tilde P_{0,k}(2^k-1),\, a_2,a_3\in S_k]$ for three distinct sites and compare it with $2\,\mathbb{P}[a_1=\tilde P_{0,k}(2^k-1)]\,\mathbb{P}[0\in S_k]^2$; a violation at any $k$ would refute the paper's negative-correlation lemma and break the flip-probability bound.
Extended reading notes
Core claim
On its own terms, the paper establishes that for every $0\le p\le 1$ and every $d\ge 3$, the extremal stationary distribution $\mu_p$ of the voter model is isomorphic to a Bernoulli shift. The route is to construct $\mu_p$ as a factor of an IID process on $\mathbb{Z}^d$: a measurable, translation-equivariant function of independent random variables attached to the lattice sites. Because the measure has finite entropy, the classical entropy-isomorphism theorem for amenable group actions turns this factor-of-IID statement into an isomorphism with a finite-state Bernoulli shift. The construction is explicit: a recursively defined coupling of the measures $M_{2^k}\rho_p$ converges almost surely in the product topology, and the limiting configuration has law $\mu_p$ and is a factor of IID.
Load-bearing premise
The whole argument leans on the estimate that coalescing random walkers repel one another: if two or three walkers occupy specified locations, the chance is no larger than a small constant times what independent walkers would give, and without that repulsion the coupled opinions might never settle down.
Editorial extensions
If this is right
- For each $p\in[0,1]$ and $d\ge 3$, the stationary distribution $\mu_p$ is a factor of an IID process on $\mathbb{Z}^d$, with an explicit translation-invariant construction.
- Every extremal stationary distribution of the voter model in $d\ge 3$ is isomorphic to a finite-state Bernoulli shift.
- The dyadic-time coupling gives an almost-sure limit: for each fixed site, the opinion process changes only finitely often as the coupling time grows.
- The result covers the entire family $\{\mu_p\}_{p\in[0,1]}$ uniformly, including the trivial extremal measures at $p=0$ and $p=1$.
Reading between the lines
- The explicit coupling suggests that quantitative rates of convergence to $\mu_p$ should be extractable from the flip-probability bounds, although the paper does not state such rates.
- Because the proof uses only transience plus negative correlation of the coalescing walkers, a similar staged-coupling scheme may apply to other partition-color models on $\mathbb{Z}^d$ whose dual partitions satisfy the same two-point and three-point negative-correlation estimates.
- The dyadic schedule $2^k$ is a natural but not obviously necessary choice; a proof with any geometrically growing sequence of times would likely go through, though the constants in the flip bound would need to be recomputed.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. This paper proves that for the voter model on Z^d with d≥3, every extremal stationary distribution μ_p (0≤p≤1) is isomorphic to a Bernoulli shift. The proof constructs μ_p explicitly as a factor of an IID process on Z^d: the authors build a sequential coupling of the measures M_{2^k}ρ_p by a biased coupling of coalescing random walks, recoloring a small set of remaining walkers at each stage so that each site's color changes only finitely many times almost surely. The flip probabilities are controlled by two- and three-walker intersection estimates (Proposition 2.9 and Appendix B) and by a negative-correlation estimate for occupied sites of coalescing walks (Lemma 3.3 and Appendix A). The resulting limit is a translation-invariant factor of IID with finite entropy, so Ornstein's isomorphism theorem gives the Bernoulli property, answering a question of Steif and Tykesson.
Significance. This is a significant result: it resolves an open question of Steif and Tykesson and provides the first natural divide-and-color example with infinite equivalence classes whose coloring process is nonetheless Bernoulli. The construction is explicit, translation-invariant, and quantitative, with detailed estimates in the body and appendices. If the technical point in Proposition 3.1 is clarified, the paper is a strong contribution to the theory of factors of IID for interacting particle systems and to the ergodic theory of infinite-range random fields.
major comments (2)
- [Section 3, proof of Proposition 3.1] The final displayed summability bound does not follow from the preceding estimates as written. After plugging (3.5) and (3.6) into (3.4), the two summands inside the square root are of order 2^{-k(λ-1/2)} and 2^{-k(2λ-3/2)}. The paper then bounds the whole expression by a constant (displayed with λ and λ^2) times 2^{-k/4}. This is justified only if λ>3/4 (and for the second term, as printed, λ>7/8). Lemma 3.2, however, only asserts the existence of some λ>0 with P[0∈S_k]<2^{-kλ}, and gives no quantitative lower bound. If the Bramson-Griffeath estimate only supplies λ≤3/4, the Borel-Cantelli argument in Proposition 3.1 would not go through. Please state explicitly the quantitative exponent available from [BG80] (for example, that λ can be taken strictly larger than 3/4) and correct the final estimates accordingly.
- [Appendix A, proof of Lemma 3.3] The key coupling assertion in this appendix is compressed: the statement that one can couple {W_i} and {W'_i} so that the path sets satisfy the displayed pointwise inclusion is not proved, and this inclusion underpins both (3.2) and (3.3). Since Lemma 3.3 is a load-bearing nonstandard estimate, the graphical-construction argument should be written out in detail. In addition, the sentence 'using symmetry between a2 and a3' is not immediately justified, because the enumeration {x_i} is fixed and the coalescing-walk law is not invariant under swapping two arbitrary target sites while fixing 0 and a1. The desired bound still follows by applying the two-order inequality (A.1) separately to the orders i2<i3 and i3<i2 and then using (3.2), so the statement of (3.3) is not at risk, but the text should be corrected.
minor comments (3)
- [Lemma 3.2] The strict inequality P[0∈S_k]<2^{-kλ} fails at k=0, because S_0=Z^d and P[0∈S_0]=1. Since Proposition 3.1 only needs a bound up to a constant factor, please restate the lemma for k≥1 or with a leading constant.
- [Proposition 2.9, bound after (2.5)] The text says the expected number of locations visited by the rate-2 walk is bounded by 2t; more precisely it is at most 1+2t (the starting site plus one per jump). The subsequent constant 2t^{-1} should be adjusted to 3t^{-1} for t≥1 (or an equivalent statement), which does not affect the finiteness of the final sums.
- [Notation throughout] Several displayed quantities such as '21−k', '23/2', and '2−kλ' are missing superscript formatting, making the algebra harder to read than it should be. Please ensure all exponents are typeset correctly.
Circularity Check
No circularity: the factor-of-IID construction is self-contained and relies on external results (Ornstein, Ornstein-Weiss, Bramson-Griffeath), not on its conclusion.
full rationale
The paper's central claim, Theorem 1.2, is obtained by explicitly constructing the stationary measure μ_p as a factor of an IID process. The construction in Section 2 defines a coupling from IID randomness and establishes distributional identities in Lemma 2.7: the constructed walks are coalescing random walks and, conditionally on them, the coloring is IID Ber(p). These are exact identities, not fitted parameters, and the target measure μ_p enters only as the weak limit of the correctly distributed marginals M_{2^k-1}ρ_p. Proposition 3.1 supplies almost-sure convergence of the staged coupling, with the summability argument relying on the negative-correlation estimates of Lemma 3.3 and the external coalescing-random-walk decay estimate of Bramson and Griffeath [BG80]. Finally, the inference from factor-of-IID to Bernoulli shift uses the external Ornstein isomorphism theorems and their amenable-group extensions. There are no fitted inputs renamed as predictions, no self-citations carrying the argument, and no claimed uniqueness theorem imported from the authors' prior work. The only potentially vulnerable point, the terse proof of Lemma 3.3 in Appendix A, is a question of proof correctness or completeness, not circularity: nothing in that lemma is assumed from the theorem being proved. The derivation is therefore self-contained against external benchmarks and exhibits no significant circularity.
Assumptions & free parameters
free parameters (2)
- delta(t) (sparsity threshold) =
2^{-1} (sum_{r in Z+} (5r)^{2d} P[R0=r])^{-1}
- M (number of groups in two-time coupling) =
ceil(max{delta(t)^{-1}, t^2})
assumptions (4)
- standard math Ornstein's isomorphism theorem: factors of Bernoulli shifts are Bernoulli shifts, and equal-entropy Bernoulli shifts are isomorphic for Z actions; generalized to amenable groups by Ornstein and Weiss.
- domain assumption Voter model dual representation: M_t rho_p equals the color process of coalescing simple random walks with independent Bern(p) colors per cluster.
- standard math Bramson and Griffeath bound: there exists lambda > 0 such that P[0 in S_k] < 2^{-k lambda} (Lemma 3.2).
- standard math Variational principle for entropy on amenable groups.
Cite this review
Pith. "Pith review of Stationary Distributions for the Voter Model in $d\geq 3$ are Factors of IID." pith.science (2026). https://pith.science/paper/WAEBQRCG
@misc{pith2026190809450,
author = {Pith},
title = {Pith review of: Stationary Distributions for the Voter Model in $d\geq 3$ are Factors of IID},
year = {2026},
howpublished = {\url{https://pith.science/paper/WAEBQRCG}},
note = {Machine review of arXiv:1908.09450}
}
abstract
For the Voter Model on $\mathbb{Z}^d$, $d\geq 3$, we show that the (extremal) stationary distributions are isomorphic to Bernoulli shifts, and answer an open question asked by Steif and Tykesson. The proof gives explicit constructions of the stationary distributions as factors of IID processes on $\mathbb{Z}^d$.
Figures
Reference graph
Works this paper leans on
-
[1]
write newline
" write newline "" before.all 'output.state := FUNCTION fin.entry add.period write newline FUNCTION new.block output.state before.all = 'skip after.block 'output.state := if FUNCTION new.sentence output.state after.block = 'skip output.state before.all = 'skip after.sentence 'output.state := if if FUNCTION not #0 #1 if FUNCTION and 'skip pop #0 if FUNCTIO...
-
[2]
D. Aldous and J. Fill. Reversible markov chains and random walks on graphs. http://www.stat.berkeley.edu/users/aldous/RWG/book.pdf, 2002
work page 2002
-
[3]
M. Bramson and D. Griffeath. Asymptotics for interacting particle systems on Z ^d . Zeitschrift f \"u r Wahrscheinlichkeitstheorie und Verwandte Gebiete , 53(2):183--196, 1980
work page 1980
-
[4]
T. M. Liggett. Interacting particle systems - An introduction . In G. F. Lawler, editor, School and Conference on Probability Theory , volume 17 of ICTP Lecture Notes , pages 1--56. Abdus Salam Int. Cent. Theoret. Phys., Trieste, Italy, 2004
work page 2004
-
[5]
D. S. Ornstein. Bernoulli shifts with the same entropy are isomorphic. Adv. Math. , 4(3):337--352, 1970
work page 1970
-
[6]
D. S. Ornstein. Factors of Bernoulli shifts are Bernoulli shifts . Adv. Math. , 5(3):349--364, 1970
work page 1970
-
[7]
D. S. Ornstein and B. Weiss. Entropy and isomorphism theorems for actions of amenable groups. J. Anal. Math. , 48(1):1--141, 1987
work page 1987
-
[8]
J. E. Steif and J. Tykesson. Generalized divide and color models. 2017, arXiv:1702.04296
work page Pith review arXiv 2017
Reviewed August 14, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.