{"id":"b9db657a-e059-409b-90d6-cfd0586ca2c8","arxiv_id":"2501.02131","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Ahlfors regular sets of dimension s ≤ 1/2 satisfy Nδ(A+A) + Nδ(AA) ≥ δ^{-4s/3+η}, the fractal analogue of Solymosi's 4/3 bound.","lead":"This short mathematics paper proves a sum-product bound for fractal subsets of the real line, showing that sums and products of points cannot both be small. The main result is the fractal analogue of a classic 4/3 exponent from finite set combinatorics, derived from a recent projection theorem.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Central bound hinges on unrefereed Orponen Theorem 3.2; if the measure-regular definition there differs from Def. 1.1, Proposition 3.6 fails.","rationale":"The reader's weakest assumption identifies exactly the reliance on Orponen's Theorem 3.2, and I agree that this is the most load-bearing point. The internal proof from Lemma 2.8 and Proposition 3.6 is mostly sound: the entropy submodularity step is correct, and the tube-counting in Proposition 3.6 is justified once the external theorem is granted. The typos in the parameter choice and the sharpness section are real but do not affect the central lower bound. The only way the main theorem could fail is if Orponen's theorem is wrong or is not applicable to the measures used here. Since that theorem is an unrefereed preprint and the paper's definition of regular measure is nonstandard (only upper-regular support plus Frostman upper bound), the hypothesis match is a genuine concern that should be checked before the result is considered established. Therefore the reader's CONDITIONAL verdict remains appropriate, and no change is needed.","tokens_in":9027,"tokens_out":35969,"duration_ms":329496,"concrete_test":"Compare Def. 1.1 with the definition of 'Ahlfors regular' in Orponen (arXiv:2410.06872). If Orponen's theorem requires a lower bound μ(B(x,r)) ≥ c r^s, determine whether the proof of Prop. 3.6 can be carried out for measures that are only upper-regular (e.g., by constructing an upper-regular measure with no lower bound and testing the collision estimate). If the lower bound is necessary, Theorem 1.2 as stated is not established, and the paper must either strengthen its hypotheses or provide a new proof.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 1.2 rests on Corollary 3.7, which applies Proposition 3.6 to the product of reciprocal, identity, and negation of μ. Proposition 3.6 ultimately applies Lemma 3.4, which is a re-statement of Orponen's Theorem 3.2 after a projective transformation. The single load-bearing step is that the hypotheses of Theorem 3.2 are satisfied. In this paper, an (s,C)-regular measure is defined (Def. 1.1) merely by having upper Frostman bound and upper-regular support. Orponen's theorem, as quoted in Theorem 3.2, uses the same phrase 'regular measure', but if Orponen's original (arXiv:2410.06872, Thm 1.13) requires the stronger two-sided Ahlfors regularity (c r^s ≤ μ(B(x,r)) ≤ C r^s), then the product measure μ×ν built from two such merely-upper-regular measures need not satisfy the lower bound, and the tube-counting in Prop. 3.6, which uses only upper bounds, cannot be justified by that theorem. The paper does not verify this hypothesis match. Also, δ0 in Theorem 3.2 is stated independent of s, which is likely a typo; if the original depends on s, constants still work but the statement should be corrected. Since the entire entropy lower bound is a corollary of this external result, a missing or incorrect hypothesis in Orponen's theorem would collapse Theorem 1.2.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves a discretised sum-product estimate for (upper) regular subsets of the real line. Theorem 1.2 states that for an (s,C)-regular measure μ supported in [1,2] with 0<s≤1/2 and for i.i.d. samples X,Y from μ, the δ-entropy of X+Y plus twice that of XY is at least (4s−η)log(1/δ)−O_C(1). Theorem 1.3 translates this into a covering-number bound Nδ(A+A)+Nδ(AA) ≥ δ^{-4s/3+η}, the fractal analogue of Solymosi's 4/3 bound. The proof combines a theorem of Orponen on high-multiplicity radial projections (Theorem 3.2) with an entropy submodularity inequality (Lemma 2.8) and a conditional-entropy/tube-counting argument (Proposition 3.6). The paper also claims sharpness via a self-similar construction.","tokens_in":9279,"tokens_out":29016,"duration_ms":235924,"significance":"If the external theorem of Orponen is valid in the needed form, the main inequality is a genuine improvement for regular sets, achieving the expected exponent 4s/3 in the discretised sum-product problem for s≤1/2. The entropy-submodularity framework is elegant and may be of independent interest; in particular, Lemma 2.7 and Lemma 2.8 are clearly and correctly presented. The paper is concise and does not rely on any fitted parameters. However, the argument depends critically on the precise regularity hypotheses in Orponen's theorem, and the sharpness section currently contains errors; both need attention before the claims are established.","major_comments":[{"comment":"The parameter condition '(1−ε)(s−σ) ≥ 2s−η' is impossible for any positive ε, σ when η<s, because the left-hand side is at most s < 2s−η. The inequality needed for the subsequent display is (1−ε)(s−σ) ≥ s−η/2 (equivalently 2(1−ε)(s−σ) ≥ 2s−η). As written, the proof of Theorem 1.2 fails at this step; this is a load-bearing point, though it appears to be a typographical error that is readily fixable.","section":"§3.2, proof of Theorem 1.2"},{"comment":"The sharpness argument contains several incorrect assertions. First, the map x→2x sends [0,1] to [0,2], not into [1,2]. Second, the claimed bound Nδ(A+A) < δ^{-s+η} contradicts the trivial inclusion A+A ⊇ A, since Nδ(A) ≈ δ^{-s} and δ^{-s+η} < δ^{-s} for η>0 and small δ. Third, even if the stated bounds Nδ(A+A) < δ^{-s+η} and Nδ(A1A1) < δ^{-s+η/2} were accepted, they would give log(Nδ(A1+A1)) + 2 log(Nδ(A1A1)) ≤ (3s+O(η)) log(1/δ), not the displayed (4s+η) log(1/δ). Thus the section is internally inconsistent and, as written, would contradict Theorem 1.2 for the very measure being constructed.","section":"§4, Sharpness of Theorem 1.2"},{"comment":"The manuscript's Definition 1.1 of '(s,C)-regular measure' requires only an upper Frostman bound for the measure and upper regularity of its support. Orponen's Theorem 3.2, quoted from [Orp24], is invoked for the product measure μ×ν and its projective image P(μ×ν), but the paper does not verify that these measures satisfy the hypotheses of Orponen's theorem. If Orponen's regularity notion is the standard two-sided Ahlfors-regularity, then the product of merely Frostman measures need not be regular, and the proof of Proposition 3.6, which uses only upper bounds, cannot be justified by Theorem 3.2. This is a load-bearing gap: the authors should either prove that Theorem 3.2 holds under their weaker notion of regularity or restate the main theorems for Ahlfors-regular measures in the standard sense.","section":"§3.1, Lemma 3.4 and Theorem 3.2"}],"minor_comments":[{"comment":"The statement gives δ0 = δ0(C, ε, σ) independent of s; unless this independence is actually proved, δ0 should be allowed to depend on s as well.","section":"§3.1, Theorem 3.2"},{"comment":"Lemma 3.4 is stated for a general line l, but the proof is only supplied for l = {0}×R and refers to [OSW24, Remark 4.13] for the general case. Since only the special case is used later, the lemma statement should be restricted accordingly or the missing cases supplied.","section":"§3.1, Lemma 3.4"},{"comment":"The title and abstract use 'Ahlfors-regular sets', but Definition 1.1 defines only upper (s,C)-regularity for sets and measures. This terminology mismatch could mislead readers about the strength of the hypotheses.","section":"§1, Definition 1.1"},{"comment":"The proof silently uses that a δ-cube is contained in a ball of radius proportional to δ; the O(1) term in the conclusion should account for the dimension and the implicit geometric constant.","section":"§2.2, Lemma 2.6"},{"comment":"The regularity constant C for the self-similar measure depends on N in the construction, and the theorem's O_C(1) may grow with N; the argument should track this dependence to ensure that the upper bound (4s+η) log(1/δ) is uniform in the parameters that are allowed to scale with η.","section":"§4, Sharpness construction"}],"recommendation":"major_revision","confidential_remarks":"The main theorem is entirely conditional on Theorem 3.2 from [Orp24], which is an arXiv preprint that has not yet appeared in a refereed venue. If the journal's standards require the main ingredient to be independently published or verified, this dependency should be weighed. Additionally, the sharpness section, if not corrected, is a substantial overclaim and should be fixed before the paper is accepted. The central entropy-submodularity argument is a nice contribution and appears salvageable."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The main result is real and worth engaging with. Theorem 1.2 gives an entropy lower bound Hδ(X+Y)+2Hδ(XY) ≥ (4s−η) log(1/δ) for (s,C)-regular measures on [1,2], and the covering-number corollary is the promised fractal analogue of Solymosi's 4/3 bound. The proof is a fresh combination: Orponen's high-multiplicity projection theorem, a radial-projection conversion, and the discretized submodularity lemma from Máthé–O'Regan. I checked the chain Proposition 3.6 → Corollary 3.7 → Theorem 1.2 and it is structurally sound. This is a subfield-level advance, not a revolution, and it is honest about resting on Orponen's recent deep theorem.\n\nThe paper as written has three soft spots, in increasing order of seriousness. First, in Section 3.2 the parameter choice says (1−ε)(s−σ) ≥ 2s−η, which is impossible for small ε,σ. The intended inequality is clearly 2(1−ε)(s−σ) ≥ 2s−η, and with that the proof goes through. That is a typo, but it should be fixed.\n\nSecond, Section 4 is more serious. The construction claims Nδ(A+A) < δ^{−s+η}, but A⊂A+A forces Nδ(A+A) ≥ Nδ(A) ≈ δ^{−s}, and δ^{−s+η} is smaller than δ^{−s} for small δ. So the displayed inequality cannot hold. Also, the doubling map sends [0,1] to [0,2], not to [1,2], as claimed. The sharpness section should be corrected or simply removed; the main theorem does not depend on it.\n\nThird, and most importantly, the load-bearing external input is Orponen's Theorem 3.2, from an unreviewed arXiv preprint. The paper never verifies that the word 'regular' in Orponen's theorem means the same as Definition 1.1. If Orponen requires two-sided Ahlfors regularity, then the merely upper-regular Frostman measures used here may not satisfy his hypotheses, and Proposition 3.6 would not follow. The quoted δ0 also appears independent of s, which is likely a typo if Orponen's constant depends on s. The author needs to check this and add a sentence confirming the hypothesis match.\n\nMy overall read: the central lower bound is credible and the method is attractive. The paper deserves a serious referee, but not acceptance in its current form. The right recommendation is to send it to peer review with a request for revision: fix the parameter typo, fix or cut the sharpness section, and resolve the Orponen hypothesis question. Analysts in fractal additive combinatorics will want this result, and the proof is short enough that an expert can verify the main line in an afternoon.","headline":"A genuinely new fractal sum-product bound (Solymosi-type 4/3) derived from Orponen's projection theorem, with a clean entropy argument; the sharpness section and a parameter typo need fixing before publication.","tokens_in":9875,"tokens_out":4476,"would_cite":false,"duration_ms":43956,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["05B99","28A78","28A80"],"pacs":[],"model":"deepseek-v4-flash","headline":"Ahlfors-regular sets of dimension at most 1/2 satisfy a sharp sum-product bound: at every small scale, the sum set and product set cannot both be small.","keywords":["sum-product phenomenon","Ahlfors-regular sets","discretised sum-product","Shannon entropy","fractal dimension","Solymosi bound","radial projections","box-counting numbers"],"falsifier":"Find an $(s,C)$-regular set $A\\subset[1,2]$ with $0<s\\le 1/2$ for which, along arbitrarily small $\\delta$, $N_\\delta(A+A)+N_\\delta(AA)\\le \\delta^{-4s/3-\\eta'}$ for some $\\eta'>0$, or produce a counterexample to the projection-multiplicity theorem (Theorem 3.2) at the parameters the proof needs. A numerical search on self-similar Cantor-type sets of dimension near $1/2$ in $[1,2]$ would be a practical way to look for such a violation.","tokens_in":8776,"feed_emoji":"📏","tokens_out":10097,"duration_ms":85737,"temperature":0.7,"pith_summary":"This paper proves a sum-product theorem for Ahlfors-regular (fractal) sets of dimension at most $1/2$. If $A\\subset[1,2]$ is an $(s,C)$-regular set with $0<s\\le 1/2$, then for every $\\eta>0$ and all sufficiently small $\\delta$ the number of length-$\\delta$ intervals needed to cover the sum set and the product set satisfies $N_\\delta(A+A)+N_\\delta(AA)\\ge \\delta^{-4s/3+\\eta}$. Equivalently, for i.i.d. random variables $X,Y$ drawn from an $(s,C)$-regular measure, $H_\\delta(X+Y)+2H_\\delta(XY)\\ge (4s-\\eta)\\log(1/\\delta)-O_C(1)$. This is the fractal analogue of Solymosi's $4/3$ bound for finite sets of reals, and the paper shows the entropy exponent $4s$ cannot be increased. If correct, it says that regular fractal sets of dimension at most $1/2$ cannot resemble rings at any small scale.","feed_headline":"Fractal sets obey the 4/3 sum-product bound","feed_subtitle":"For any regular set of dimension s≤1/2, covering numbers of A+A and A·A add to at least δ^{-4s/3+η}.","key_machinery":"The mechanism is a two-step entropy argument. A discretised submodularity inequality (Lemma 2.7) is applied to three i.i.d. variables $X,Y,Z$ to obtain $H_\\delta((X+Y)Z)+2H_\\delta(X)\\le H_\\delta(X+Y)+2H_\\delta(XY)+O(1)$. The lower bound on $H_\\delta((X+Y)Z)$ comes from a projection-multiplicity theorem quoted as Theorem 3.2 (from [Orp24]), which bounds the measure of points whose orthogonal projection fibre has multiplicity at least $\\delta^{-\\sigma}$; a projective transformation converts this into a radial-projection estimate (Lemma 3.4), and Proposition 3.6 turns that into the entropic lower bound $H_\\delta((Y-Z)/X\\mid Z)\\ge (1-\\epsilon)(2s-2\\sigma)\\log(1/\\delta)$. Combining the lower and upper bounds yields Theorem 1.2.","core_discovery":"The paper's central claim is that Ahlfors-regularity forces a quantitative sum-product tradeoff at every scale. Theorem 1.2 states that for an $(s,C)$-regular probability measure $\\mu$ supported in $[1,2]$, with $0<s\\le 1/2$ and $X,Y$ i.i.d. with law $\\mu$, the entropy inequality $H_\\delta(X+Y)+2H_\\delta(XY)\\ge (4s-\\eta)\\log(1/\\delta)-O_C(1)$ holds for arbitrarily small $\\eta>0$ and all $\\delta<\\delta_0(C,s,\\eta)$. The immediate covering-number corollary (Theorem 1.3) is $N_\\delta(A+A)+N_\\delta(AA)\\ge \\delta^{-4s/3+\\eta}$ for every $(s,C)$-regular set $A\\subset[1,2]$, and the same statements hold with subtraction or division replacing addition or multiplication. A self-similar construction in Section 4 shows the coefficient $4s$ in the entropy inequality is best possible. The author notes that the proof only uses the projection-multiplicity theorem for dimensions at most $1$, so an extension of that input would give the result for all $0<s<1$.","pith_inferences":["Because the entropy inequality is proved from submodularity and a projection-multiplicity estimate, a Frostman (rather than Ahlfors-regular) version of that estimate would immediately give the same sum-product bound for the broader class of Frostman measures, which the paper does not address.","The coefficient $2$ on $H_\\delta(XY)$ suggests an underlying multiplicative-energy control similar to the discrete proof via sumsets; this hints that a collision-entropy or $L^2$-based version of the $4/3$ bound might hold directly for regular sets, without passing through Shannon entropy.","A concrete check on digit-restricted Cantor sets in $[1,2]$ with dimension just below $1/2$ could test whether $N_\\delta(A+A)+N_\\delta(AA)$ actually approaches $\\delta^{-4s/3}$, providing numerical evidence for the sharpness of Theorem 1.3's exponent."],"forward_implications":["For every $(s,C)$-regular $A\\subset[1,2]$ with $0<s\\le 1/2$ and every $\\eta>0$, at all sufficiently small $\\delta$ either $N_\\delta(A+A)$ or $N_\\delta(AA)$ exceeds $(1/2)\\delta^{-4s/3+\\eta}$, so sums and products cannot both be compressed.","The stronger product inequality $N_\\delta(A+A)N_\\delta(AA)^2\\ge \\delta^{-4s+\\eta}$ also follows, and is sharp in the exponent.","Replacing addition by subtraction or multiplication by division gives the same bounds, so the result is insensitive to the signs in the two operations.","If the projection-multiplicity input is extended to dimensions above $1$, the same proof yields the fractal sum-product bound for the full range $0<s<1$, as Remark 1.6 observes."],"supporting_citations":[{"why":"Supplies the projection-multiplicity theorem (Theorem 3.2) that the proof's lower bound depends on; without it the argument stops.","marker":"[Orp24]"},{"why":"Provides the discretised submodular inequality (Lemma 2.7) that converts the lower bound into a two-term entropy estimate.","marker":"[MO23]"},{"why":"The previous discretised sum-product bound that this paper improves for the restricted class of Ahlfors-regular sets.","marker":"[RW23]"},{"why":"The finite-set sum-product bound whose $4/3$ exponent Theorem 1.3 reproduces in the fractal setting.","marker":"[Sol09]"},{"why":"The submodularity lemma for Shannon entropy used in the proof of the discretised submodular inequality.","marker":"[Tao10]"}],"fun_headline_variants":["Fractal sets satisfy Solymosi's 4/3 sum-product bound","Ahlfors-regular sets meet 4/3 sum-product bound","Sum-product 4/3 bound extended to fractal sets","Solymosi's 4/3 bound now holds for fractal sets","Fractal sum-product: Solymosi's 4/3 bound"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is an external theorem (Theorem 3.2, quoted from [Orp24]) saying that, for regular fractal measures, the set of points whose projecting lines meet the fractal in unusually many places has very small measure; if that estimate is wrong at the parameters used here, the entropy bound and the $4/3$ covering-number bound collapse.","fun_headline_variants_meta":{"raw":{"variants":["Fractal sets satisfy Solymosi's 4/3 sum-product bound","Ahlfors-regular sets meet 4/3 sum-product bound","Sum-product 4/3 bound extended to fractal sets","Solymosi's 4/3 bound now holds for fractal sets","Fractal sum-product: Solymosi's 4/3 bound"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000648,"raw_usage":{"total_tokens":2924,"prompt_tokens":841,"completion_tokens":2083,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":457,"completion_tokens_details":{"reasoning_tokens":1987}},"tokens_in":457,"tokens_out":2083,"duration_ms":16515,"temperature":1.0,"reasoning_tokens":1987,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T22:19:29.593200+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Find an $(s,C)$-regular set $A\\subset[1,2]$ with $0<s\\le 1/2$ for which, along arbitrarily small $\\delta$, $N_\\delta(A+A)+N_\\delta(AA)\\le \\delta^{-4s/3-\\eta'}$ for some $\\eta'>0$, or produce a counterexample to the projection-multiplicity theorem (Theorem 3.2) at the parameters the proof needs. A numerical search on self-similar Cantor-type sets of dimension near $1/2$ in $[1,2]$ would be a practical way to look for such a violation.","supporting_citations":[],"review_version":1}