Mixing time and cutoff phenomenon for the interchange process on dumbbell graphs and the labelled exclusion process on the complete graph

G\'abor Pete; Rich\'ard Patk\'o

REVIEW 3 major objections 4 minor 1 cited by

Mixing time and cutoff phenomenon for the interchange process on dumbbell graphs and the labelled exclusion process on the complete graph

T0 review · 3 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The interchange process on a dumbbell graph has an explicit mixing time, with cutoff exactly when the smaller clique grows.

desk verdict The dumbbell theorem is the real contribution and appears sound; the labelled exclusion proof has a real gap at (5.4), so Theorem 1.3 is not proved as written. read the letter →

arxiv 1908.09406 v1 pith:RF5T72LV submitted 2019-08-25 math.PR math.COmath.GR

Richárd Patkó , Gábor Pete This is my paper · ORCID

classification math.PRmath.COmath.GR MSC 60J1060B15

keywords interchangeprocessmixingtimecutoffphenomenonBernoulli-Laplaceurnmodelexclusiondumbbellgraphrandomtranspositionstotalvariationdistance

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper determines the total variation mixing time of the lazy interchange process on the dumbbell graph formed by two complete graphs of sizes $n$ and $m$, joined by a single edge. It shows that the asymptotic mixing time is $\frac{|E|\,nm}{n+m}\log n$ when $m\ge c\sqrt n$, drops to $2|E|\,m\log m$ when $1\ll m\ll\sqrt n$, and is only of order $|E|$ when $m$ stays bounded. Cutoff holds in the first two regimes precisely when $m\to\infty$, matching the product condition. The same coupling argument proves that the labelled exclusion process on the complete graph with $k$ labelled particles mixes in $n\log k+O_\varepsilon(n)$ for every $1\le k\le n$, settling a conjecture and giving a probabilistic proof of the random transpositions mixing time.

What carries the argument

The load-bearing object is the reduction of the dumbbell process to the Bernoulli-Laplace diffusion model, in which two urns of sizes $n$ and $m$ exchange one uniformly chosen ball at each step; here the 'balls' are the particles, and moves occur only at the rare bridge transpositions. To make the reduced process exactly Markovian, the paper replaces the single bridge edge by $nm$ thin bridges, one for each pair of vertices across the cliques, keeping the total bridge rate unchanged; this symmetrization makes the particle partition uniformly distributed given its size. The proof then tracks the number of original small-clique particles inside the small clique, a birth-and-death chain whose transition rates are in (4.9), and bounds its mixing time from below by Chebyshev estimates on single- and two-particle eigenfunctions of the projected chain, and from above by coupling the count difference to a lazy copycat random walk. For bounded $m$, a half-symmetrized graph keeps only the large clique symmetric, and mixing is controlled by the first time the $m$ particles cross the bridge.

What would settle it

Compute or simulate the reduced Bernoulli-Laplace count chain (transition probabilities in (4.9)) with $n=400, m=20$ and $n=900, m=30$, and measure $t_{\rm mix}(0.25)$ and $t_{\rm mix}(0.75)$. The theorem predicts both are $(1+o(1))|E|m\log m$ with their ratio tending to 1; if the ratio stays bounded away from 1, the claimed cutoff is false. For bounded $m$, say $m=5$, the same computation should return $t_{\rm mix}(\varepsilon)=\Theta(|E|)$ with $t_{\rm mix}(1/4)/t_{\rm mix}(1/2)$ bounded away from 1.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that the slow degree of freedom of the dumbbell interchange process is not the permutation order inside each clique but the number of original particles from the smaller clique that have crossed the bridge. Because bridge transpositions are rare, the clique interiors reach stationarity long before the next bridge transposition typically occurs, so the whole chain is governed by a time-changed Bernoulli-Laplace urn model: at each bridge step, one uniformly random particle from each clique is exchanged. Solving that urn model at the right time scale yields Theorem 1.1, with a phase transition at $m\asymp\sqrt n$: for larger $m$ mixing is driven by Gaussian fluctuations of the particle count, while for smaller growing $m$ it is driven by the waiting time until every particle from the small clique has visited the large one. For bounded $m$, a half-symmetrized chain is used instead and gives mixing time of order $|E|$ without cutoff. The same coupling then gives Theorem 1.3: the labelled exclusion process on $K_n$ has $\varepsilon$-mixing time $n\log k+O_\varepsilon(n)$ for every $1\le k\le n$.

Load-bearing premise

The argument rests on a clean separation of time scales: the permutation inside each clique must become essentially uniform before the next bridge transposition typically occurs; if internal clique mixing were as slow as bridge crossing, the derived formulas and the phase transition would not hold.

Editorial extensions

If this is right

For every $0<\varepsilon<1/2$, the $\varepsilon$-mixing time on the dumbbell is $(1+o(1))\,|E|\,nm/(n+m)\,\log n$ when $m\ge c\sqrt n$, and $(2+o(1))\,|E|\,m\log m$ when $1\ll m\ll\sqrt n$.
Cutoff occurs in this family exactly when $m(n)\to\infty$, which is exactly when the product condition holds; when $m$ is bounded, the mixing time is of order $|E|$ and cutoff fails.
The labelled exclusion process on the complete graph with $k$ labelled particles has $\varepsilon$-mixing time $n\log k+O_\varepsilon(n)$ for every $1\le k\le n$.
Taking $k=n$ recovers the random transpositions mixing time $(1+o(1))n\log n$ through a coupling argument rather than representation theory.
The paper's general conjecture is that interchange-process cutoff on any graph sequence is equivalent to the product condition, with failure caused exactly by bounded bad bottlenecks; one direction is proved.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

A testable extension the paper leaves implicit is to replace the two cliques by two arbitrary dense graphs joined by one edge; the same symmetrization should reduce mixing to a two-colour count process whenever bridge moves stay rare, predicting the same phase transition in terms of volumes.
The phase transition at $m\asymp\sqrt n$ separates a Gaussian-fluctuation mechanism from a rare-event waiting-time mechanism; this suggests that for labelled exclusion on other dense graphs the mixing time is the larger of a count-mixing time and a first-exit time, which could be checked numerically on star-like or complete bipartite graphs.
If the stated conjecture is right, any bounded-degree graph sequence whose interchange process satisfies the product condition must exhibit cutoff, so testing two expanders connected by a single edge would separate the role of local bottlenecks from the role of graph size.
The proof technique for labelled exclusion might extend to the non-lazy chain with small parity-breaking laziness, giving the sharp constant $\frac12 n\log n$ for $k=n$ rather than an $O_\varepsilon(n)$ window.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

The dumbbell theorem is the real contribution here, and it looks right. The first exact mixing-time asymptotics for the interchange process on a family with a phase transition, the m ≍ √n threshold, and the cutoff criterion (cutoff exactly when m → ∞) are all new and substantive. The proof via the symmetrized graph and the Bernoulli–Laplace reduction is careful, and the time-scale separation that is the most natural worry is addressed head-on by Propositions 3.1 and 3.2. Wilson's method lower bound is also concrete. This part of the paper deserves serious attention.

The labelled exclusion theorem, however, is not proved as written. The gap is in Section 5. Equation (5.4) contains a false algebraic step: (1 − e^{−a})^k is not exp(−a + log k), and the claimed lower bound can actually exceed 1 (e.g. k = n/2, ε > 0). The correct asymptotics are exp(−k^{1−a}) times lower-order factors, which do not give the bound used later. Separately, the negative correlation of the events {β*_j ≥ 1} is asserted with “one can easily see” but no proof is supplied. That assertion is plausible but it is load-bearing, and it cannot just be waved through. As it stands, the upper bound for k ≥ c√n does not go through, so Theorem 1.3 is unproved. The lower bound comes from Lacoin–Leblond, and the k > n/2 reduction is sketched rather than fully checked, but the main problem is clearly the (5.4) computation and the missing correlation argument.

I have no issue with the citation pattern or the framing; the paper is honest about what is new and what is borrowed. The conjecture on interchange cutoff is speculative but clearly labeled and motivated.

Bottom line: the dumbbell result is likely correct and significant, and the exclusion part is probably repairable, but it needs real work. A serious referee should engage with the paper, but should not accept it until the exclusion proof is fixed. My recommendation: send to review with the expectation of a major revision.

Referee Report

3 major / 4 minor

Summary. The paper studies the lazy interchange process on dumbbell graphs consisting of two cliques K_n and K_m joined by a single edge. Theorem 1.1 gives asymptotic total-variation mixing times in three regimes: m ≥ c√n, 1 ≪ m ≪ √n, and m = O(1), and concludes that cutoff holds exactly when m → ∞, which coincides with the product condition. The proof symmetrizes the bridge edge into many thin bridges, reduces the particle-partition process to a time-changed Bernoulli-Laplace urn chain, and couples the original and symmetrized processes using within-clique mixing on a faster timescale. Section 5 states Theorem 1.3, asserting that the labelled exclusion process on K_n with k labelled particles has tmix(ε) = n log k + O_ε(n) for every 1 ≤ k ≤ n, which would resolve a conjecture of Lacoin and Leblond and give a new probabilistic proof of the Diaconis-Shahshahani random-transposition result.

Significance. The dumbbell result is a substantial and carefully executed contribution to the interchange-process literature: it provides a non-transitive example where cutoff is characterized exactly by the product condition, and the explicit three-regime phase transition is new. The coupling arguments in Sections 3 and 4 are detailed and the claimed asymptotics are supported by explicit computations. If the proof of Theorem 1.3 can be repaired, the paper would also settle the Lacoin-Leblond conjecture for labelled exclusion on the complete graph. As it stands, however, the proof of Theorem 1.3 contains a load-bearing algebraic error and an unproved correlation assertion, so the labelled-exclusion claim is not established. I therefore assess the paper as significant but in need of major revision.

major comments (3)

[Section 5, Eq. (5.4)] Equation (5.4) contains a false algebraic identity. The paper claims that (1 - exp(-(1+ε+o(1)) (n-k)/n log k))^k equals exp(-(1+ε+o(1)) (n-k)/n log k + log k). Writing a = (1+ε)(n-k)/n, the left-hand side is approximately exp(-k e^{-a}) = exp(-k^{1-a}) when e^{-a} = o(1), not exp(-a log k + log k). For k = n/2 and small ε, the claimed right-hand side is of order k^{(1+ε)/2}, which exceeds 1 for large k, while the true left-hand side tends to 0 like exp(-k^{(1-ε)/2}). Since the subsequent bound P(G|B) ≤ P(G)/P(B) depends on this estimate, the proof of the upper bound of Theorem 1.3 for k ≤ n/2 fails at this point.
[Section 5, negative correlation claim] The paragraph before (5.4) asserts that the events {β*_j ≥ 1} are negatively correlated and then concludes P(∀j β*_j ≥ 1) ≥ P(β*_j ≥ 1)^{k−ℓ}. For negatively correlated events, the probability of the intersection is at most the product, not at least the product, so the inequality has the wrong direction under the stated correlation assumption. No proof is supplied for either correlation direction. Because P(B) enters the denominator of the key conditional estimate, this is a load-bearing unsupported assertion and must be either proved or replaced by a different estimate.
[Section 5, use of P(G|B) ≤ P(G)/P(B)] Even if the numerical value of P(B) in (5.4) were corrected, the bound P(G|B) ≤ P(G)/P(B) cannot work uniformly in k: for k comparable to n and ε small, the true probability that a fixed set of k−ℓ particles all cross to K^c is exponentially small in k^{(1-ε)/2}, so the right-hand side diverges. A valid proof of Theorem 1.3 would need a direct estimate of the conditional probability or a different way to handle the particles that have not crossed. This is a structural gap in the argument as written, not merely a typographical issue.

minor comments (4)

[Theorem 1.1(iii)] In the statement of Theorem 1.1(iii), the notation 'A(εm, m)' appears to contain a typo; it should presumably read 'A(ε, m)'.
[Section 4.2, after Eq. (4.18)] The phrase 'end the proof of (1.1) is complete' should be rewritten, for example as 'This ends the proof of (1.1).'
[Section 3.1, Eq. (3.3)] The displayed expression for total variation distance is missing the factor 1/2: for product measures, TV = (1/2) ∑_{σ1,σ2} |f1(σ1)f2(σ2) - π1(σ1)π2(σ2)|. The subsequent inequalities are valid up to a factor of 2, which is absorbed into constants, but the display should be corrected.
[Section 4.4, Figure 4.1] The text refers to 'the Markov chain with 3 states on Figure 4.1', but no figure appears in the supplied manuscript; either include the figure or describe the transition matrix explicitly.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: all load-bearing inputs are external theorems or forward-derived intermediate results; the Section 5 gap is a correctness issue, not a circular step.

full rationale

The paper's central derivations are forward: the dumbbell upper bound reduces the interchange process on G_{n,m} to the symmetrized chain hat G_{n,m} via an explicit coupling (Proposition 3.2), and the symmetrized chain's mixing time is then computed from the Bernoulli-Laplace projection using eigenvalues, Chebyshev's inequality, and coupling arguments in Section 4. The conditional wording in Proposition 3.2 ('assume that we already know (proved later) that t_mix(hat G) ≍ n^2 m log m') is a modular forward reference: Section 4 establishes the hat-G bounds independently before the transfer is applied, so there is no dependency cycle. The within-clique mixing estimate (3.1) is imported from the external complete-graph results of Diaconis-Shahshahani and Levin-Peres-Wilmer, and the small-number-of-particles exclusion upper bound is imported from Lacoin-Leblond [18]; the authors cite no prior work of their own, so no self-citation is load-bearing. Theorem 1.3 uses Theorem 1.1 only for the already-proved Bernoulli-Laplace mixing time and uses [18] for the leftover-particle sub-process, which is parameter-free external support. No parameter is fitted and later renamed a prediction, and no uniqueness theorem from the authors is invoked. The reviewer's concern about Section 5 is a genuine correctness gap: the negative-correlation assertion is unproved and equation (5.4) miscomputes (1 - k^{-a})^k, but this is an algebraic and probabilistic error in the proof of Theorem 1.3, not a circular identification of output with input, and it does not affect Theorem 1.1. Therefore the circularity score is 0.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

The paper introduces no fitted constants and no new physical entities. It relies on standard theorems and one unproved negative-correlation assertion.

assumptions (4)

standard math Mixing time of the lazy interchange process on the complete graph K_k is (1+o(1)) k log k, with exponentially small TV distance after L k log k steps (Eq. 3.1).
Used in Proposition 3.1 to show within-clique mixing is much faster than bridge crossings; cited from Diaconis-Shahshahani and [19, Corollary 8.10].
standard math Theorem 5.1 from Lacoin-Leblond: for the labelled exclusion process, tmix(1-eps) >= n log k - beta n for all k, and tmix(eps) <= n log k + beta n when k = o(sqrt(n)).
Used as a black box in the proof of Theorem 1.3 to cover the lower bound and the small-k upper bound; the present paper proves the remaining k >= c sqrt(n) case.
standard math Standard Markov chain estimates: spectral gap variational formula, Wilson's lower bound, exponential decay after mixing time, hitting time bounds for lazy random walks (Levin-Peres-Wilmer).
Used throughout Sections 2-4 for lower bounds and coupling estimates.
ad hoc to paper The events {beta*_j >= 1} are negatively correlated, asserted in the proof of Theorem 1.3.
Not proven or cited; used to lower-bound P(forall j beta*_j >= 1) in (5.4). This is a gap in the written proof.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Mixing time and cutoff phenomenon for the interchange process on dumbbell graphs and the labelled exclusion process on the complete graph." pith.science (2026). https://pith.science/paper/RF5T72LV

@misc{pith2026190809406,
  author       = {Pith},
  title        = {Pith review of: Mixing time and cutoff phenomenon for the interchange process on dumbbell graphs and the labelled exclusion process on the complete graph},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RF5T72LV}},
  note         = {Machine review of arXiv:1908.09406}
}

abstract

We find the total variation mixing time of the interchange process on the dumbbell graph (two complete graphs, $K_n$ and $K_m$, connected by a single edge), and show that this sequence of chains exhibits the cutoff phenomenon precisely when the smaller size $m$ goes to infinity. The mixing time undergoes a phase transition at $m\asymp \sqrt{n}$. We also state a conjecture on when exactly cutoff holds for the interchange process on general graphs. Our proofs use coupling methods, and they also give the mixing time of the simple exclusion process of $k$ labelled particles in the complete graph $K_n$, for any $k\leq n$, with cutoff, as conjectured by Lacoin and Leblond (2011). In particular, this is a new probabilistic proof for the mixing time of random transpositions, first established by Diaconis and Shahshahani (1981).

Figures

Figures reproduced from arXiv: 1908.09406 by the authors.

**Figure 2.1.** Following a single particle between the cliques [PITH_FULL_IMAGE:figures/full_fig_p006_2_1.png] view at source ↗

**Figure 4.1.** A projected chain for a single particle in the int [PITH_FULL_IMAGE:figures/full_fig_p017_4_1.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

Cutoff for congestion dynamics and related generalized exclusion processes
math.PR 2025-02 accept novelty 6.0 of 10

The labeled and unlabeled versions of a capacity-constrained congestion dynamics both exhibit abrupt convergence (cutoff), at times (1/2)n log n and (1/2)(1-ρ)n log n respectively.

Reference graph

Works this paper leans on

24 extracted references · 24 canonical work pages · cited by 1 Pith paper

[1]

Shuﬄing cards and stopping times

David Aldous and Persi Diaconis. Shuﬄing cards and stopping times . The American Mathe- matical Monthly 93 (1986), 333–348

work page 1986
[2]

Comparing with octopi

Gil Alon and Gady Kozma. Comparing with octopi. arXiv:1811.10537 [math.PR]

work page Pith review arXiv
[3]

Character ization of cutoﬀ for reversible Markov chains

Riddhipratim Basu, Jonathan Hermon, Yuval Peres. Character ization of cutoﬀ for reversible Markov chains. The Annals of Probability 45 (2017), No. 3, 1448–1487

work page 2017
[4]

Mixing time s for random k-cycles and coalescence-fragmentation chains

Nathanael Berestycki, Oded Schramm, Ofer Zeitouni. Mixing time s for random k-cycles and coalescence-fragmentation chains. The Annals of Probability 39 (2011), 1815–1843

work page 2011
[5]

Cutoff for conjugacy-invariant random walks on the permutation group

Nathanael Berestycki, Batı S ¸eng¨ ul. Cutoﬀ for conjugacy-invariant random walks on the per- mutation group. Probability Theory Related Fields , to appear. arXiv:1410.4800 [math.PR]

work page Pith review arXiv
[6]

Liggett, Thomas Richthammer

Pietro Caputo, Thomas M. Liggett, Thomas Richthammer. Proof of Aldous’ spectral gap conjecture. Journal of the American Mathematical Society 23 (2010), 831–851

work page 2010
[7]

Generating a random per mutation with random trans- positions

Persi Diaconis, Mehrdad Shahshahani. Generating a random per mutation with random trans- positions. Zeitschrift f¨ ur Wahrscheinlichkeitstheorie und verwandte Gebiete 57 (1981), No. 2, 159–179

work page 1981
[8]

Time to reach stationarit y in the Bernoulli–Laplace urn model, SIAM Journal on Mathematical Analysis 18 (1987), No

Persi Diaconis, Mehrdad Shahshahani. Time to reach stationarit y in the Bernoulli–Laplace urn model, SIAM Journal on Mathematical Analysis 18 (1987), No. 1, 208–218

work page 1987

Show all 24 references

[9]

Total-variation cutoﬀ in birth-and-death chains, Prob- ability Theory Related Fields 146 (2010), 61–85

Jian Ding, Eyal Lubetzky, Yuval Peres. Total-variation cutoﬀ in birth-and-death chains, Prob- ability Theory Related Fields 146 (2010), 61–85

2010
[10]

Mixing times for neighbour transposition shuﬄ es on graphs, Chalmers Uni- versity of Technology, 2011

Stefan Erikshed. Mixing times for neighbour transposition shuﬄ es on graphs, Chalmers Uni- versity of Technology, 2011. https://core.ac.uk/display/70588470

2011
[11]

Forsstr¨ om and Johan Jonasson

Malin P. Forsstr¨ om and Johan Jonasson. The spectrum and co nvergence rates of exclusion and interchange processes on the complete graph. J. Theor. Probab. 30 (2017), 639–654

2017
[12]

Helfgott

Harald A. Helfgott. Growth in groups: ideas and perspectives. Bull. Amer. Math. Soc. 52 (2015), 357–413

2015
[13]

The exclusion process mix es (almost) faster than in- dependent particles

Jonathan Hermon and Richard Pymar. The exclusion process mix es (almost) faster than in- dependent particles. arXiv:1808.10846 [math.PR]

arXiv
[14]

The interchange process o n high-dimensional products

Jonathan Hermon and Justin Salez. The interchange process o n high-dimensional products. arXiv:1905.02146 [math.PR]

1905 arXiv
[15]

Mixing times for the interchange process, ALEA, Lat

Johan Jonasson. Mixing times for the interchange process, ALEA, Lat. Am. J. Probab. Math. Stat. 9 (2012), 667–683

2012
[16]

Mixing time and cutoﬀ for the adjacent transpo sition shuﬄe and the simple exclusion, The Annals of Probability 44 (2016), No

Hubert Lacoin. Mixing time and cutoﬀ for the adjacent transpo sition shuﬄe and the simple exclusion, The Annals of Probability 44 (2016), No. 2, 1426–1487

2016
[17]

The cutoﬀ proﬁle for the simple exclusion proces s on the circle

Hubert Lacoin. The cutoﬀ proﬁle for the simple exclusion proces s on the circle. The Annals of Probability 44 (2016), No. 5, 3399–3430

2016
[18]

Cutoﬀ phenomenon for the simp le exclusion process on the complete graph

Hubert Lacoin, R´ emi Leblond. Cutoﬀ phenomenon for the simp le exclusion process on the complete graph. ALEA, Lat. Am. J. Probab. Math. Stat. 8 (2011), 285–301

2011
[19]

Levin, Yuval Peres, Elizabeth L

David A. Levin, Yuval Peres, Elizabeth L. Wilmer. Markov chains and Mixing Times, second edition. American Mathematical Society, 2009

2009
[20]

Bounds on Mixing Times in the Bernoulli–La place Dif- fusion Model — Elementary proofs for variation distance and ﬁrst p assage times

Chjan Lim, William Pickering. Bounds on Mixing Times in the Bernoulli–La place Dif- fusion Model — Elementary proofs for variation distance and ﬁrst p assage times. https://homepages.rpi.edu/~limc/BL2017.pdf
[21]

Cutoﬀ on all Ramanujan grap hs

Eyal Lubetzky and Yuval Peres. Cutoﬀ on all Ramanujan grap hs. Geometric and Functional Analysis 26 (2016), 1190–1216

2016
[22]

A strong uniform time for random transposit ions

Peter Matthews. A strong uniform time for random transposit ions. J. Theoret. Probab. 1 (1988), no. 4, 411–423. 20

1988
[23]

Oliveira

Roberto I. Oliveira. Mixing of the symmetric exclusion processes in terms of the corresponding single-particle random walk. The Annals of Probability 41 (2013), No. 2, 871–913

2013
[24]

David B. Wilson. Mixing times of Lozenge tiling and card shuﬄing Mark ov chains. Ann. Appl. Probab. 14 (2004), 274–325. 21

2004

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [1]

Shuﬄing cards and stopping times

David Aldous and Persi Diaconis. Shuﬄing cards and stopping times . The American Mathe- matical Monthly 93 (1986), 333–348

work page 1986

[2] [2]

Comparing with octopi

Gil Alon and Gady Kozma. Comparing with octopi. arXiv:1811.10537 [math.PR]

work page Pith review arXiv

[3] [3]

Character ization of cutoﬀ for reversible Markov chains

Riddhipratim Basu, Jonathan Hermon, Yuval Peres. Character ization of cutoﬀ for reversible Markov chains. The Annals of Probability 45 (2017), No. 3, 1448–1487

work page 2017

[4] [4]

Mixing time s for random k-cycles and coalescence-fragmentation chains

Nathanael Berestycki, Oded Schramm, Ofer Zeitouni. Mixing time s for random k-cycles and coalescence-fragmentation chains. The Annals of Probability 39 (2011), 1815–1843

work page 2011

[5] [5]

Cutoff for conjugacy-invariant random walks on the permutation group

Nathanael Berestycki, Batı S ¸eng¨ ul. Cutoﬀ for conjugacy-invariant random walks on the per- mutation group. Probability Theory Related Fields , to appear. arXiv:1410.4800 [math.PR]

work page Pith review arXiv

[6] [6]

Liggett, Thomas Richthammer

Pietro Caputo, Thomas M. Liggett, Thomas Richthammer. Proof of Aldous’ spectral gap conjecture. Journal of the American Mathematical Society 23 (2010), 831–851

work page 2010

[7] [7]

Generating a random per mutation with random trans- positions

Persi Diaconis, Mehrdad Shahshahani. Generating a random per mutation with random trans- positions. Zeitschrift f¨ ur Wahrscheinlichkeitstheorie und verwandte Gebiete 57 (1981), No. 2, 159–179

work page 1981

[8] [8]

Time to reach stationarit y in the Bernoulli–Laplace urn model, SIAM Journal on Mathematical Analysis 18 (1987), No

Persi Diaconis, Mehrdad Shahshahani. Time to reach stationarit y in the Bernoulli–Laplace urn model, SIAM Journal on Mathematical Analysis 18 (1987), No. 1, 208–218

work page 1987