REVIEW 1 major objections 4 minor 12 references
On the size of $h$-fold sumsets
T0 review · 1 major / 4 minor · reviewed 2026-08-03 · deepseek-v4-flash
Pith's one-line read This paper proves an exact binomial-coefficient formula for the size of the h-fold sumset of any set {0,1,...,s,a,b}, and shows the formula works for every h exactly when r = 0 or qs + r ≥ a.
desk verdict Clear, mostly correct paper on a Nathanson problem — real progress in Theorem 2 — but the printed sufficiency proof has a floor/ceiling error in (3.8) that must be fixed; the theorem almost certainly stands. 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 the shortest representation function ℓ_A(n), the minimum number of elements (counted with multiplicity) needed to represent n as a sum from A. Its behavior on residue classes modulo the largest element b, coupled with the bracket polynomials P_{s,N} and their coefficients p_{N,e}, carries the whole argument. The bridge is the identity ℓ_A(tb+v) = t + ℓ_B(v), which holds when ℓ_B(n+b) − ℓ_B(n) ≥ 1 for all n; the proof of this inequality via a residue-case analysis in x = n mod a is exactly what produces the condition qs + r ≥ a. The polynomials P_{s,N} encode the distribution of ℓ_B on an interval modulo a and translate into the truncated-binomial coefficients appearing
What would settle it
Enumerate A = {0,1,2,5,7} (s=2, a=5, b=7, q=1, r=2, qs+r=4<5) and compute |2A| directly by listing all pairwise sums; compare it with the right-hand side of (1.2). The paper predicts a strict inequality |2A| > |U_2|, since the point b + (a−r) = 9 = (q+1)a is in 2A but not in U_2.
Extended reading notes
Core claim
The central discovery is the exact generating-function evaluation behind |hA|. For B = {0,1,...,s,a} and A = B ∪ {b} with b = qa + r, the paper defines the polynomial P_{s,N}(z) = 1 + (s−1)z + (τ_{s,N} − s)z^{L_{s,N}+1} − τ_{s,N}z^{L_{s,N}+2} and shows that if ℓ_B(n+b) − ℓ_B(n) ≥ 1 for every n ≥ 0, then ℓ_A(tb+v) = t + ℓ_B(v), so the h-fold sumset hA coincides with the explicit set U_h. Summing the generating functions over residues v mod b then yields the closed formula (1.2). The decisive fact is that the gap inequality holds exactly when r = 0 or qs + r ≥ a; outside that range the equality hA = U_h fails for h = q + 1. Thus the formula is not merely an upper bound: it is the exact answer
Load-bearing premise
The whole closed formula rests on the gap inequality ℓ_B(n+b) − ℓ_B(n) ≥ 1 holding for every n ≥ 0; if this single inequality failed for some residue class modulo b, the count |U_h| would no longer equal |hA| and the formula would be wrong.
Editorial extensions
If this is right
- For s = 1, Corollary 1 gives an explicit six-term truncated-binomial formula for |h{0,1,a,b}| valid for every h exactly when r = 0 or q + r ≥ a, extending Nathanson's family.
- When qs + r < a, the formula fails at h = q + 1; the proof exhibits a concrete element b + (a − r) that belongs to (q+1)A but not to U_{q+1}.
- Theorem 1 gives an exact count of |hA| for every finite normalized set via record values of the shortest representation function, so any specific set outside the regular range can still be evaluated exactly.
- The necessary and sufficient condition means that the closed binomial formula is not an accident of small examples: the threshold qs + r ≥ a is exactly the boundary where the representation gap inequality holds for all residues.
Reading between the lines
- The single missing point at h = q+1 in the irregular case hints that the correction for larger h may come from a systematic count of elements along the residue class x = a − r modulo a; a closed form for the irregular range might be obtainable by tracking how many such points enter hA layer by layer.
- The same record-value framework could be applied to sets with several large elements, replacing the two-summand structure {0,...,s,a,b} with a longer chain; the threshold condition would then become a system of gap inequalities, offering a way to decide when closed binomial formulas exist.
- Because Theorem 1 makes |hA| computable in finite time for any normalized set, a direct algorithm that enumerates all four-element sets for a given h would let one compute the full set R(h,4) and test whether the gaps between consecutive sizes have a simple description.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper studies the size of the h-fold sumset hA for finite sets of integers. It first introduces a shortest-representation function \ell_A(n) and proves a general counting formula for |hA| (Theorem 1) by decomposing N_0 into residue classes modulo the largest element of A and recording, for each residue, the finitely many strict changes of a certain minimal value function M_t(v). The main result is Theorem 2, which specializes to sets A={0,1,\ldots,s,a,b} with 1\le s<a<b, b=qa+r. The authors prove that an explicit formula for |hA| in terms of truncated binomial coefficients and coefficients of a polynomial P_{s,N} holds for every positive integer h if and only if r=0 or qs+r\ge a. For s=1, the formula reduces to a six-term identity (Corollary 1), which is shown to contain Nathanson's tetrahedral-difference family as a special case. The necessity direction exhibits failure at h=q+1 when r>0 and qs+r<a.
Significance. If the paper is correct, it resolves exactly the h-fold sumset size for a natural five-element family {0,\ldots,s,a,b}, including a sharp description of the boundary between a 'regular' range, where a closed-form formula is valid, and an irregular range, where the short formula fails but the general Theorem 1 still applies. The proof is self-contained and does not rely on unstated assumptions. The generating function identities and coefficient extraction are verified by direct computation, and the claimed generalization of Nathanson's result is credible. The general counting formula in Theorem 1 is also a potentially useful tool. Provided the proof defect discussed below is corrected, the paper makes a solid contribution to the exact sumset-size problem for structured small sets.
major comments (1)
- [§3, Eq. (3.8)] The displayed equality in (3.8) is incorrect as printed. Lemma 6 gives \ell_B([x+r]_a)=\lceil [x+r]_a/s\rceil, but (3.8) writes \lfloor [x+r]_a/s\rfloor. This is not merely a notational slip: the equality is false. For example, with s=5, a=7, r=2, q=1, and x=1, the left side is \ell_B(10)-\ell_B(1)=2-1=1, while the printed right side is 1+0+\lfloor 3/5\rfloor-\lceil 1/5\rceil=0. Consequently, the subsequent case analysis that replaces \lfloor [x+r]_a/s\rfloor by \lfloor (x+r)/s\rfloor in the case x<a-r is also invalid; for instance, s=6, a=10, r=4, q=1, x=1 gives a value 0 where the true difference is 1. Since inequality (3.10) is the load-bearing bridge that leads from Lemma 5 to the identity hA=U_h and hence to formula (1.2), the sufficiency proof is not valid as printed. The defect is repairable: replace the floor with a ceiling, and in the two cases use \lceil (x+r)/s\rceil and q+1+\
minor comments (4)
- [§3, Lemma 6] The step '\min_{0\le j\le t} \lceil (x+j(a-s))/s\rceil = \lceil x/s\rceil' is correct but terse; since a>s, each term with j\ge1 is at least 1+\lceil x/s\rceil. A one-line justification would help the reader.
- [§3, Eq. (3.7)] The terms 'zq' and 'Pa,eze' should be superscripted as z^q and P_{a,e}z^e in the display; as typeset they are easy to misread.
- [Remark 1] There is a stray space in 'Nathanson 's result'; also in the proof of Lemma 7, 'Nbe positive integers' should read 'N be positive integers'.
- [Theorem 1] The notation u_{v,r_v+1}=+\infty is used implicitly in the final summand; it may be worth stating explicitly after the record definitions.
Circularity Check
No significant circularity: the closed form is derived from generating functions and then matched to hA under a proven condition; nothing is fitted, renamed, or imported from self-citation.
full rationale
The derivation chain is self-contained. Lemma 6 proves ℓ_B(ta+x)=t+ceil(x/s) from Lemma 4 and the representation function of B'={0,...,s}; Lemma 8 computes the generating function of the auxiliary union U_h, obtaining (1.2) as a coefficient identity from (1-z^q)F_{s,a}(z)/(1-z)^3 + z^qF_{s,r}(z)/(1-z)^2. This identity is not the theorem's conclusion: it counts U_h, not hA. The sufficiency half of Theorem 2 then proves, under exactly the condition r=0 or qs+r≥a, the inequality ℓ_B(n+b)-ℓ_B(n)≥1, hence by Lemma 5 that ℓ_A(tb+v)=t+ℓ_B(v), so hA=U_h; the formula follows. The necessity half constructs an explicit element b+x=(q+1)a that lies in (q+1)A but not in U_{q+1} when r>0 and qs+r<a, so the formula fails independently. No parameter is fitted to |hA|, no self-citation is used as evidence, and no ansatz is borrowed: Nathanson's tetrahedral family is recovered by the direct substitution a=h+1, r=0 in Corollary 1, not by a citation. A possible floor/ceiling typo in (3.8) would be a correctness defect, not a circularity: the lower-bound bridging inequality is proven rather than assumed, and even if the typesetting error required repair, the argument is not equivalent to its input by construction.
Assumptions & free parameters
assumptions (5)
- standard math ℓ_{B'}(n) = ⌈n/s⌉ for B' = {0,...,s}
- standard math Ceiling subadditivity: ⌈u⌉+⌈v⌉ ≥ ⌈u+v⌉
- domain assumption n ∈ hA ⟺ ℓ_A(n) ≤ h, with ℓ_A the minimal total coefficient sum over representations
- standard math Formal identity z^α/(1−z)^{j+1} = Σ_{h≥0} ⟨h−α+j choose j⟩ z^h
- domain assumption Affine invariance |hA| = |h(uA+v)| for u≠0, so sets normalize to {0 = a_0 < ... < a_k = m} with gcd(a_1,...,a_k) = 1
Cite this review
Pith. "Pith review of On the size of $h$-fold sumsets." pith.science (2026). https://pith.science/paper/U44UG6IU
@misc{pith2026260729535,
author = {Pith},
title = {Pith review of: On the size of $h$-fold sumsets},
year = {2026},
howpublished = {\url{https://pith.science/paper/U44UG6IU}},
note = {Machine review of arXiv:2607.29535}
}
abstract
Let $h$ be a positive integer, and let $A$ be a finite set of integers. We derive an exact formula for $|hA|$. Furthermore, let $A=\{0,1,\ldots,s,a,b\}$, $1\leq s<a<b$, and write $b=qa+r$ with $0\leq r<a$. By using generating function, we prove that $|hA|$ equals a definite explicit formula expressed in terms of certain truncated binomial coefficients for all positive integers $h$ if and only if $r=0$ or $qs+r\geq a$. This generalizes a result of Nathanson.
Reference graph
Works this paper leans on
-
[1]
Eliahou and M
S. Eliahou and M. Kervaire, Minimal sumsets in infinite abelian groups , J. Algebra 287 (2005), 449–457
2005
-
[2]
Granville and G
A. Granville and G. Shakan, The Frobenius postage stamp problem, and beyond , Acta Math. Hungar. 161 (2020), 700–718
2020
-
[3]
Granville and A
A. Granville and A. Walker, A tight structure theorem for sumsets , Proc. Amer. Math. Soc. 149 (2021), 4073–4082
2021
-
[4]
Gyarmati, M
K. Gyarmati, M. Matolcsi and I. Z. Ruzsa, A superadditivity and submultiplicativity property for cardinalities of sumsets , Combinatorica 30 (2010), 163–174
2010
-
[5]
V. F. Lev, Structure theorem for multiple addition and the Frobenius problem, J. Num- ber Theory 58 (1996), 79–88
1996
-
[6]
V. F. Lev, The structure of higher sumsets , Proc. Amer. Math. Soc. 150 (2022), 5165– 5177
2022
-
[7]
M. B. Nathanson, Sums of finite sets of integers , Amer. Math. Monthly 79 (1972), 1010–1012. 15
1972
-
[8]
M. B. Nathanson, Problems in additive number theory, VI: sizes of sumsets of finite sets, Acta Math. Hungar. 176 (2025), 498–521
2025
Show all 12 references
-
[9]
M. B. Nathanson, Bh-sets of real and complex numbers, Canad. Math. Bull. 69 (2026), 137–141
2026
-
[10]
M. B. Nathanson, Triangular and tetrahedral number differences of sumset sizes in additive number theory , arXiv:2506.15015, 2025
2025 arXiv
-
[11]
M. B. Nathanson, Additive sumset sizes with tetrahedral differences, arXiv:2507.08646, 2025
2025 arXiv
-
[12]
O’Bryant, On Nathanson ’s triangular number phenomenon, arXiv:2506.20836, 2025
K. O’Bryant, On Nathanson ’s triangular number phenomenon, arXiv:2506.20836, 2025. 16
2025 arXiv
Reviewed August 3, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.