REVIEW 7 minor 1 cited by
Staircase Discovered for Hamiltonian Subsets Below Dirac's Threshold
Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →
T0 review · glm-5.2
2026-07-08 01:56 UTC pith:VHPUZX44
load-bearing objection Solid paper. Sharp staircase bound for cyclic subsets below Dirac, with a clean trichotomy and correct proofs. Recommend serious referee.
Tight Staircase Bounds for Cyclic Subsets below Dirac's Threshold
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
The paper identifies a 'staircase' mechanism for the count of Hamiltonian-induced vertex subsets in $d$-regular graphs below the Dirac threshold. The minimum count is exactly $(q-o(1))2^{n/q}$ where $q=⌊n/(d+1)⌋$. The integer $q$ acts as a discrete parameter that controls the exponential rate, remaining constant across intervals of $d$ and jumping when $d$ crosses $n/k$ thresholds. The extremal configuration is the disjoint union of $q$ equal cliques. The proof establishes a reduced-graph trichotomy (connected matching, $q$ balanced components, or $q+1$ near-critical components) to show no other graph structure can produce fewer cyclic subsets.
What carries the argument
Reduced-graph trichotomy (Lemma 3.1) sorting graphs into three cases: (a) large connected matching, (b) $q$ balanced dense components, (c) $q+1$ near-critical components with forced cross-block matchings.
Load-bearing premise
The most delicate part of the argument is in the case of $q+1$ near-critical components, where the proof relies on a global edge-counting inequality to force a large matching between two blocks. If this edge-counting fails to guarantee a sufficiently large cross-block matching, the exponential surplus over the target bound in this case would not be secured.
What would settle it
A $d$-regular graph on $n$ vertices with $d=Ω(n)$ and $d<n/2$ where the number of cyclic subsets is asymptotically smaller than $(q-o(1))2^{n/q}$, or where the exponential rate varies smoothly with $d$ rather than jumping at $n/k$ thresholds.
If this is right
- The staircase phenomenon depends on the global regularity assumption; under minimum-degree conditions alone, the extremal obstruction is not governed by $q$ equal pieces, suggesting a fundamentally different landscape for non-regular settings.
- The exact boundary at $d=n/2$ reaches the maximum exponential rate $2^{(1-o(1))n}$, confirming that the transition to full-scale Hamiltonian subset density occurs precisely at the Dirac threshold.
- The trichotomy proof technique—distinguishing genuinely extremal $q$-block configurations from near-critical $(q+1)$-block cases via forced matchings—may generalize to other enumerative problems in extremal graph theory where regularity constrains component structure.
- Size-sensitive estimates counting cyclic subsets of each fixed cardinality could reveal finer distributional structure within the staircase levels.
Where Pith is reading between the lines
- The discrete staircase implies that small changes in degree $d$ that do not cross an $n/k$ threshold have no effect on the asymptotic minimum of cyclic subsets—a form of phase rigidity in the enumerative behavior.
- The sharpness of the leading coefficient $q$ at clique-union examples suggests these configurations are rigid minimizers, and the $o(1)$ gap might be closeable to an exact theorem for specific parameter families.
- The forced cross-block matching in the $(q+1)$-component case reveals that global regularity creates hidden connectivity that produces an exponential surplus over the naive $q$-block bound, indicating the staircase is a robust structural phenomenon rather than an artifact of the proof method.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. This paper determines the sharp asymptotic lower bound on the number of cyclic subsets (subsets inducing a Hamiltonian subgraph) in regular graphs below the Dirac threshold. The main result (Theorem 1.1) states that for an n-vertex d-regular graph with εn ≤ d < n/2 and q = ⌊n/(d+1)⌋ ≥ 2, one has Cyc(G) ≥ (q − o(1))·2^{n/q}. This bound is tight, including the leading coefficient q, as witnessed by the disjoint union of q equal cliques. A boundary result (Proposition 1.2) gives the optimal exponential rate 2^{(1−o(1))n} at d = n/2. The proof proceeds via a trichotomy for the reduced graph obtained from Szemerédi's Regularity Lemma (Lemma 3.1), followed by three separate counting arguments for each case: large connected matchings (Section 4), q balanced dense components (Section 5), and q+1 near-critical components (Section 6).
Significance. The paper resolves a natural and well-motivated problem in the enumerative theory of Hamiltonicity, extending the recent work of Draganic, Keevash, and Muyesser from the Dirac regime to the entire linear below-Dirac range. The staircase phenomenon — discrete jumps in the optimal exponential rate — is a clean conceptual contribution. The proof is self-contained against external benchmarks: the lower bound is derived from standard tools (Regularity Lemma, Dirac's theorem, Chvatal's theorem, Blow-up Lemma, Chernoff bounds), and the upper bound is witnessed by the explicit construction qK_{n/q}. The sharpness of the leading coefficient q in Case (b), requiring careful handling of exceptional vertices, is a notable technical strength. The edge-counting argument in Lemma 6.2 / Claim 6.4 is intricate and verifies correctly upon careful checking.
minor comments (7)
- Section 1, footnote 1: the term 'cyclic subset' is defined as a subset inducing a Hamiltonian subgraph, but the footnote extends this to include the empty set, singletons, and copies of K_2. This convention should be stated more prominently (e.g., in the definition of Cyc(G) itself) since it affects the counting in Proposition 5.1, where the empty set is explicitly excluded from the disjointness argument.
- Lemma 4.5, proof: the claim states that (A,B) is (6ρ, μ/4)-super-regular with probability 1−o(1), but the minimum-degree verification shows every vertex of A has at least μ|B|/4 neighbours in B. The parameter in the super-regularity conclusion should be stated consistently (μ/4 vs μr/4 where r = |B|).
- Proposition 6.1: the constant C in the statement is used both as a given positive constant and as the bound C·2^{n/(k−1)} for k ≥ 3. In the proof, C is set to q−ξ, but the hierarchy 1/n ≪ θ, ρ, λ ≪ 1/k, 1/C requires C to be fixed. This is fine since q and ξ are fixed, but a brief remark clarifying that C is a fixed constant (not depending on n) would improve readability.
- Figure 1: the axis label 'n/log_2 Cyc(G)' is slightly ambiguous — it would be clearer to write 'n / log_2(Cyc(G))' or specify that this is the asymptotic exponential rate.
- Section 9: the open problem about exact minimizers at clique points is natural. A reference to the exact results of Kim, Liu, Sharifzadeh, and Staden [9] for the minimum-degree setting would provide useful context for what 'exact' means here.
- Proof of Proposition 7.1, Case (b): the bound d_G(z, A_i) ≥ (1/(q(q+1)) − ρ/q − o(1))n is shown to be at least 5α|A_i|. The verification relies on α being sufficiently small relative to 1/q, but the chain of inequalities is compressed; spelling out the final comparison (i.e., that 1/(q(q+1))n ≥ 5α·(1/q + 3qτ)n for the chosen hierarchy) would aid the reader.
- Lemma 6.2: the assumption c ≤ k^{−9} is stronger than what the final contradiction requires (the bound k(k+1)^2 C^2 < n/k uses C < C(k,2)·c√n, so c ≤ k^{−9} is more than sufficient). A brief remark that this is not tight would prevent confusion.
Simulated Author's Rebuttal
We thank the referee for a careful reading and for the positive assessment. The report recommends minor revision but does not list any specific major or minor comments requiring changes. We address the report below.
Circularity Check
No circularity: the lower bound is a first-principles theorem and the upper bound is witnessed by an explicit external construction.
full rationale
The paper proves Theorem 1.1 (a lower bound on Cyc(G)) via a trichotomy on the reduced graph (Lemma 3.1), with each case handled by a self-contained proposition (4.1, 5.1, 6.1) using standard tools (Szemerédi's Regularity Lemma, Dirac's theorem, Chvátal's theorem, the Blow-up Lemma). No proposition's conclusion reduces by construction to its hypotheses. The tightness (upper bound) is witnessed by the explicit construction qK_{n/q}, an independent external benchmark. No fitted parameters are renamed as predictions, no self-citation chain is load-bearing for the central claim, and no ansatz is smuggled in. The derivation is self-contained against external benchmarks.
Axiom & Free-Parameter Ledger
axioms (6)
- standard math Szemerédi's Regularity Lemma (Degree form)
- standard math Dirac's theorem on longest paths
- standard math Chvátal's Hamiltonicity theorem
- standard math Blow-up Lemma by Komlós, Sárközy and Szemerédi
- standard math Chernoff bound for hypergeometric random variables
- standard math Kőnig's theorem
read the original abstract
Let $\operatorname{Cyc}(G)$ denote the number of cyclic subsets in a graph $G$, which are subsets that induce a Hamiltonian subgraph. Dragani\'{c}, Keevash and M\"{u}yesser recently proved that every regular Dirac graph has $\Omega(2^n)$ cyclic subsets, resolving a problem of Erd\H{o}s and Faudree. We determine the sharp asymptotic lower bound throughout the linear range below Dirac's threshold. Let $G$ be an $n$-vertex $d$-regular graph with $d=\Omega(n)$ and $d<n/2$, then $$ \operatorname{Cyc}(G)\ge (q-o(1))2^{n/q}, \quad \text{where } \quad q=\left\lfloor \frac{n}{d+1}\right\rfloor \ge 2. $$ This bound is asymptotically best possible, including the leading coefficient $q$, as witnessed at the staircase levels by the disjoint union of $q$ equal cliques. Consequently, the optimal exponential rate changes by discrete jumps as $d$ crosses the thresholds $n/k$, rather than varying smoothly with $d$. We also prove the optimal exponential rate at the Dirac boundary: every $n$-vertex $n/2$-regular graph satisfies $\operatorname{Cyc}(G)\ge 2^{(1-o(1))n},$ which is sharp up to a subexponential factor by $K_{n/2,n/2}$.
Figures
Forward citations
Cited by 1 Pith paper
-
On the number of factorable induced subgraphs
Random induced subgraphs of dense graphs contain F-factors with asymptotically tight probability 1/(rq), where q is the order of a lattice coset group.
Reference graph
Works this paper leans on
-
[1]
V. Chvátal. On Hamilton’s ideals.J. Combinatorial Theory Ser. B, 12:163–168, 1972
work page 1972
- [2]
-
[3]
W. Cuckler and J. Kahn. Hamiltonian cycles in dirac graphs.Combinatorica, 29(3):299–326, 2009
work page 2009
-
[4]
G. A. Dirac. Some theorems on abstract graphs.Proceedings of the London Mathematical Society, 2:69–81, 1952
work page 1952
-
[5]
N. Draganić, P. Keevash, and A. Müyesser. Cyclic subsets in regular Dirac graphs.International Mathematics Research Notices, 2025(14):rnaf215, 2025
work page 2025
-
[6]
P.Erdős.Someofmyfavoriteproblemsandresults.InThe mathematics of Paul Erdős, I,volume13ofAlgorithms Combin., pages 47–67. Springer, Berlin, 1997
work page 1997
-
[7]
A. Frieze and M. Krivelevich. On two Hamilton cycle problems in random graphs.Israel Journal of Mathematics, 166:221–234, 2008
work page 2008
- [8]
-
[9]
J. Kim, H. Liu, M. Sharifzadeh, and K. Staden. Proof of Komlós’s conjecture on Hamiltonian subsets.Proceedings of the London Mathematical Society, 115(5):974–1013, 2017
work page 2017
- [10]
-
[11]
J. Komlós and M. Simonovits. Szemerédi’s regularity lemma and its applications in graph theory. InCombi- natorics, Paul Erdős is Eighty, Vol. 2, volume 2 ofBolyai Society Mathematical Studies, pages 295–352. János Bolyai Mathematical Society, Budapest, 1996
work page 1996
-
[12]
M. Krivelevich, C. Lee, and B. Sudakov. Robust Hamiltonicity of Dirac graphs.Transactions of the American Mathematical Society, 366(6):3095–3130, 2014
work page 2014
-
[13]
D. Kühn, D. Osthus, and A. Treglown. Hamilton decompositions of regular tournaments.Proceedings of the London Mathematical Society, 101(1):303–335, 2010
work page 2010
-
[14]
R. Montgomery. Hamiltonicity in random graphs is born resilient.Journal of Combinatorial Theory, Series B, 139:316–341, 2019
work page 2019
-
[15]
G. N. Sárközy, S. M. Selkow, and E. Szemerédi. On the number of Hamiltonian cycles in Dirac graphs.Discrete Mathematics, 265(1–3):237–250, 2003
work page 2003
-
[16]
W. Sun, S. Wei, and D. Yang. Clique factors in random samplings of regular graphs.arXiv preprint arXiv:2512.20287, 2025. ECOPRO, Institute for Basic Science, 55 Expo-ro, Yuseong-gu, Daejeon, 34126, Korea Email address:hongliu@ibs.re.kr School of Mathematics and Statistics, Zhengzhou University, Zhengzhou, China, and ECOPRO, Institute for Basic Science, 55...
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.