The Cutoff Profile for Random Transpositions on Repeated Cards in the Full Range of Parameters
Pith reviewed 2026-05-08 05:13 UTC · model grok-4.3
The pith
The cutoff profile for random transpositions on repeated cards is asymptotically Gaussian when repetitions grow.
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
The limiting profile is asymptotically Gaussian, with different explicit forms in the regimes m fixed and m = ω(1). The proof first combines the Fourier-analytic framework for the many-urn Bernoulli-Laplace model with an approximation of the original measure by an explicitly tractable auxiliary measure directly on the repeated-card quotient space; this comparison step uses new estimates for Kostka numbers. The profile problem is then reduced to quotient fixed-point statistics, which are analyzed via Hoeffding-type combinatorial central limit theorems.
What carries the argument
Quotient fixed-point statistics analyzed via Hoeffding-type combinatorial central limit theorems, after approximation to an auxiliary measure on the repeated-card quotient space.
If this is right
- When m is held fixed the profile takes one explicit Gaussian form.
- When m grows to infinity the profile takes a different explicit Gaussian form.
- The full range of parameters m and l is now covered by combining this Gaussian result with the Poisson-type profile known for fixed l.
- The new Kostka-number estimates enable direct comparison on the quotient space.
Where Pith is reading between the lines
- The Gaussian shape indicates that a central-limit effect governs the fluctuations of the relevant statistics inside the cutoff window.
- The same approximation-plus-CLT strategy could be tested on other mean-field shuffling models that admit a quotient description.
- Moderate-size simulations with growing l could directly check the rate at which the profile approaches the predicted Gaussian.
Load-bearing premise
The repeated-card transposition chain is equivalent to the many-urn mean-field Bernoulli-Laplace model and admits the required new bounds on Kostka numbers when l grows.
What would settle it
Numerical computation of the total variation distance for sequences with l growing to infinity and either fixed m or m also growing, checking whether the normalized distance inside the window matches the predicted explicit Gaussian curve.
read the original abstract
The random transposition shuffle on repeated cards induces a Markov chain on the quotient space of arrangements with multiplicities, and is equivalent to the many-urn mean-field Bernoulli-Laplace model introduced by Scarabotti. Writing $n=ml$, where there are $m$ card types and each type appears $l$ times, we determine the limiting profile for the total variation distance to stationarity at times $t=\frac{n}{2}\left(\log n-\frac{1}{2}\log l+c\right)$, under the assumption $l=\omega(1)$. Scarabotti previously established that this process exhibits cutoff at time $\frac{n}{2}(\log n-\frac{1}{2}\log l)$; our result refines this by identifying the precise asymptotic shape of convergence inside the cutoff window. We show that the limiting profile is asymptotically Gaussian, with different explicit forms in the regimes $m$ fixed and $m=\omega(1)$. Together with our previous work on the fixed-$l$ regime, where the limiting profile is of Poisson type, this yields the cutoff profile for the random transposition shuffle on $n=ml$ repeated cards for the full range of parameters $m$ and $l$. Our argument has two main steps. First, we combine Scarabotti's Fourier-analytic framework for the many-urn Bernoulli-Laplace model with the approximation method of Jain-Sawhney (arXiv:2410.23944). More precisely, we compare the original shuffling measure with an explicitly tractable auxiliary measure directly on the repeated card quotient space, rather than passing through an intermediate comparison on the full symmetric group; this step relies in particular on our new estimates for Kostka numbers. Second, we reduce the limiting-profile problem to quotient fixed-point statistics and analyze them via Hoeffding-type combinatorial central limit theorems.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper establishes the limiting cutoff profile for the total variation distance to stationarity of the random transposition shuffle on repeated cards (n=ml with l=ω(1)) at times t=n/2 (log n - 1/2 log l + c). It shows this profile is asymptotically Gaussian, with explicit forms depending on whether m is fixed or m=ω(1). The argument combines Scarabotti's Fourier analysis on the many-urn Bernoulli-Laplace quotient with a Jain-Sawhney-style comparison to an auxiliary measure (relying on new Kostka number estimates) and reduces the profile to fixed-point statistics analyzed via combinatorial CLTs. This completes the picture begun in prior work for the fixed-l regime (Poisson-type profiles).
Significance. If the central claims hold, the result gives a complete description of cutoff profiles for this family of chains over the full range of m and l, refining Scarabotti's cutoff time with precise window asymptotics. The two-step strategy (Fourier comparison on the quotient plus reduction to Hoeffding-type CLTs) and the new Kostka bounds represent a technical advance in mixing-time analysis for symmetric-group quotients.
major comments (2)
- [Abstract (first main step) and the measure-comparison argument] The new upper/lower bounds on Kostka numbers K_{λ,μ} (invoked in the first main step to control the distance between the original transposition measure and the auxiliary measure on the repeated-card quotient) are not accompanied by explicit error terms that remain uniform over partition shapes when m ranges from fixed to ω(1) while l→∞. Because the total-variation profile inside the window t=n/2 (log n - 1/2 log l + c) is obtained by showing the auxiliary measure is close enough that the two profiles coincide asymptotically, the absence of such uniform bounds leaves open whether the o(1) discrepancy is absorbed into the claimed Gaussian limit.
- [Section on reduction to fixed-point statistics] In the reduction of the limiting profile to quotient fixed-point statistics (second main step), the manuscript invokes Hoeffding-type combinatorial CLTs but does not verify that the requisite moment or variance conditions hold uniformly when m=ω(1). This verification is load-bearing for the explicit Gaussian form in that regime.
minor comments (1)
- Notation for the auxiliary measure and the precise statement of the Jain-Sawhney comparison on the quotient (rather than on the full symmetric group) could be clarified with a short diagram or explicit definition of the measures involved.
Simulated Author's Rebuttal
We thank the referee for the careful reading and for recognizing the significance of the work in completing the cutoff profile picture across all regimes of m and l. The two major comments concern uniformity of error terms and conditions in our estimates; we address each below and will strengthen the manuscript accordingly.
read point-by-point responses
-
Referee: [Abstract (first main step) and the measure-comparison argument] The new upper/lower bounds on Kostka numbers K_{λ,μ} (invoked in the first main step to control the distance between the original transposition measure and the auxiliary measure on the repeated-card quotient) are not accompanied by explicit error terms that remain uniform over partition shapes when m ranges from fixed to ω(1) while l→∞. Because the total-variation profile inside the window t=n/2 (log n - 1/2 log l + c) is obtained by showing the auxiliary measure is close enough that the two profiles coincide asymptotically, the absence of such uniform bounds leaves open whether the o(1) discrepancy is absorbed into the claimed Gaussian limit.
Authors: We agree that explicit uniformity statements are needed for full rigor. The Kostka bounds in Section 3 are proved via generating-function comparisons and majorization inequalities that depend only on the total size n = ml and the multiplicity l; the resulting multiplicative error is 1 + O(1/l) with the implied constant independent of the partition shape λ, μ and of m (for m ranging from fixed to o(l)). This error is absorbed into the o(1) term inside the cutoff window because the auxiliary-measure comparison is performed at distance o(1) uniformly in the window parameter c. We will add an explicit corollary (new Corollary 3.4) stating the uniform error bound and its consequence for the total-variation distance, together with a short proof sketch confirming independence of m. revision: yes
-
Referee: [Section on reduction to fixed-point statistics] In the reduction of the limiting profile to quotient fixed-point statistics (second main step), the manuscript invokes Hoeffding-type combinatorial CLTs but does not verify that the requisite moment or variance conditions hold uniformly when m=ω(1). This verification is load-bearing for the explicit Gaussian form in that regime.
Authors: We thank the referee for highlighting this gap in presentation. The Hoeffding-type CLT (Theorem 4.2) is applied to the vector of fixed-point counts on the quotient; the required moment bounds follow from the fact that the underlying permutation statistics are sums of indicators with dependency graph of bounded degree (at most 2l per coordinate). When m = ω(1), the variance of the normalized linear combination is bounded below by a positive constant times (log m)/m which remains positive and the fourth-moment ratio is O(1) uniformly in m. We will insert a new lemma (Lemma 4.5) that explicitly checks these conditions under the standing assumption l = ω(1) and m = o(n), confirming that the Gaussian limit holds with the stated variance in both regimes. revision: yes
Circularity Check
Minor self-citation to fixed-l regime; central derivation uses new Kostka bounds and external CLTs
full rationale
The paper's derivation combines Scarabotti's external Fourier framework with a Jain-Sawhney-style comparison on the quotient space, inserts newly derived upper/lower bounds on Kostka numbers K_{λ,μ} into the character-ratio formula, and reduces the profile to quotient fixed-point statistics analyzed by Hoeffding-type combinatorial CLTs. The only self-reference is to the author's prior fixed-l work, which is invoked only to complete the full-parameter statement and is not used in any load-bearing step of the l=ω(1) argument. No fitted parameter is renamed as a prediction, no ansatz is smuggled via self-citation, and no quantity is shown to equal its input by construction. The new Kostka estimates constitute independent combinatorial content rather than a recycling of prior fitted quantities.
Axiom & Free-Parameter Ledger
axioms (2)
- standard math Fourier-analytic framework for the many-urn Bernoulli-Laplace model on the quotient space
- domain assumption Existence of Hoeffding-type combinatorial central limit theorems for fixed-point statistics on the quotient
Reference graph
Works this paper leans on
-
[1]
Sami Assaf, Persi Diaconis, and Kannan Soundararajan. Riffle shuffles of a deck with repeated cards.Discrete Mathematics & Theoretical Computer Science, 2009
work page 2009
-
[2]
Trailing the dovetail shuffle to its lair.The Annals of Applied Probability, 2(2):294–313, 1992
Dave Bayer and Persi Diaconis. Trailing the dovetail shuffle to its lair.The Annals of Applied Probability, 2(2):294–313, 1992
work page 1992
-
[3]
Mixing times for random k-cycles and coalescence- fragmentation chains.Annals of Probability, 2011
Nathanaël Berestycki, Oded Schramm, and Ofer Zeitouni. Mixing times for random k-cycles and coalescence- fragmentation chains.Annals of Probability, 2011
work page 2011
-
[4]
Riffle shuffles of decks with repeated cards.The Annals of Probability, 2006
Mark Conger and Divakar Viswanath. Riffle shuffles of decks with repeated cards.The Annals of Probability, 2006
work page 2006
-
[5]
Generating a random permutation with random transpositions
Persi Diaconis and Mehrdad Shahshahani. Generating a random permutation with random transpositions. Zeitschrift für Wahrscheinlichkeitstheorie und verwandte Gebiete, 57(2):159–179, 1981
work page 1981
-
[6]
Cutoff for generalised bernoulli-laplace urn models
Ritesh Goenka, Jonathan Hermon, and Dominik Schmid. Cutoff for generalised bernoulli-laplace urn models. arXiv preprint arXiv:2511.10630, 2025
-
[7]
A combinatorial central limit theorem.The Annals of Mathematical Statistics, pages 558–566, 1951
Wassily Hoeffding. A combinatorial central limit theorem.The Annals of Mathematical Statistics, pages 558–566, 1951
work page 1951
-
[8]
Bob Hough. The random k cycle walk on the symmetric group.Probability Theory and Related Fields, 165(1):447–482, 2016
work page 2016
-
[9]
Hitting time mixing for the random transposition walk.arXiv preprint arXiv:2410.23944, 2024
Vishesh Jain and Mehtaab Sawhney. Hitting time mixing for the random transposition walk.arXiv preprint arXiv:2410.23944, 2024
-
[10]
Hubert Lacoin. Mixing time and cutoff for the adjacent transposition shuffle and the simple exclusion.The Annals of Probability, 44(2):1426–1487, 2016
work page 2016
-
[11]
Evita Nestoridi. Comparing limit profiles of reversible markov chains.Electronic Journal of Probability, 29:1–14, 2024
work page 2024
-
[12]
Evita Nestoridi and Sam Olesker-Taylor. Limit profiles for reversible Markov chains.Probability Theory and Related Fields, 182(1):157–188, 2022
work page 2022
-
[13]
Limit profile for projections of random walks on groups.Electronic Journal of Probability, 2024
Evita Nestoridi and Sam Olesker-Taylor. Limit profile for projections of random walks on groups.Electronic Journal of Probability, 2024
work page 2024
-
[14]
The Annals of Applied Probability , year =
Sam Olesker-Taylor and Dominik Schmid. Limit profile for the bernoulli–laplace urn.arXiv preprint arXiv:2409.07900, 2024
-
[15]
Sam Olesker-Taylor, Lucas Teyssier, and Paul Thévenin. Sharp character bounds and cutoff for symmetric groups.arXiv preprint arXiv:2503.12735, 2025
-
[16]
Sagan.The symmetric group, volume 203 ofGraduate Texts in Mathematics
Bruce E. Sagan.The symmetric group, volume 203 ofGraduate Texts in Mathematics. Springer-Verlag, New York, second edition, 2001. Representations, combinatorial algorithms, and symmetric functions
work page 2001
-
[17]
Fabio Scarabotti. Time to reach stationarity in the bernoulli–laplace diffusion model with many urns.Ad- vances in Applied Mathematics, 18(3):351–371, 1997
work page 1997
-
[18]
Thek-cycle shuffling with repeated cards.arXiv preprint arXiv:2603.27433, 2026
Jiahe Shen. Thek-cycle shuffling with repeated cards.arXiv preprint arXiv:2603.27433, 2026
-
[19]
Limit profile for random transpositions.The Annals of Probability, 48(5):2323–2343, 2020
Lucas Teyssier. Limit profile for random transpositions.The Annals of Probability, 48(5):2323–2343, 2020
work page 2020
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.