{"id":"c6877f96-c9c9-4c8c-b3b6-f9bd0101cd06","arxiv_id":"2601.11052","paper_version":1,"verdict":"REJECT","confidence":"HIGH","novelty_score":4.0,"correctness_risk":"high","formal_verification":"none","parameter_count":1,"one_line_summary":"The paper's claimed Riemann-hypothesis bound for the analytic part of the divisor summatory error term contradicts its own definition of that analytic part.","lead":"This preprint defines an 'analytic part' for the error term of the summatory divisor function and claims a conditional bound of order x^{1/2} exp(log x / log log x) under the Riemann Hypothesis. The claimed bound is inconsistent with the paper's own explicit formula, which contains an x log x term.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1.2.1 contradicts the definition of E^AN_{σ1} in (2.1.15): the term x(log x+2γ−1)/2 alone exceeds the claimed upper bound for every δ<1.","rationale":"The reader's verdict is REJECT with high confidence, and I agree the paper should be rejected. However, I identify a more basic defect than the one listed as the weakest assumption. The reader's weakest assumption focuses on the proof silently dropping an O(x^{3+δ}) term in §4.1. That is a real and fatal flaw, but the theorem fails even before the proof begins: the defined analytic part contains an explicit x log x term, making the claimed O(x^{δ'} exp(log x/log log x)) bound impossible for any δ' < 1. This is not a subtle gap or a missing justification; it is an inconsistency between the statement of Theorem 1.2.1 and equation (2.1.15), the paper's own definition. I also note the proof has the additional independent error of dropping the O(x^{3+δ}) term. In good faith, I checked whether the paper might intend E^AN_{σ1} to denote only the g2 part, but (2.1.15) explicitly includes the log term and the surrounding text calls it the analytic part. The manuscript also contains a Remark 4.2.1 that says the RH-equivalence of Theorem 1.1.1 does not seem to extend to σ1(n); that is consistent with the theorem being false. Therefore no numerical experiment or further computation is needed to reject; the inconsistency is analytically decisive. The concrete test I propose is a direct asymptotic comparison using the lower bound from the definition, which settles the matter without relying on any of the later contour-shifting estimates.","tokens_in":9418,"tokens_out":3116,"duration_ms":29487,"concrete_test":"Analytic check: from (2.1.15), use g2(x) ≥ 0 to obtain E^AN_{σ1}(x) ≥ (x/2)(log x + 2γ − 1). Let U(x) = C x^{δ'} exp(log x / log log x) be the right-hand side of (1.2.1). Then E^AN_{σ1}(x)/U(x) ≥ (x^{1−δ'} log x)/(2C exp(log x/log log x)) = (x^{1−δ'−1/log log x} log x)/(2C). Since δ' < 1 and 1/log log x → 0, the exponent 1−δ'−1/log log x is positive for all large x, so the ratio tends to infinity. This proves the claimed bound is false by the definition alone. To check the second flaw, rerun the §4.1 contour shift while keeping the O(x^{3+δ}) term in the final inequality; the stated estimate cannot be recovered.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim fails immediately from the paper's own definition. In (2.1.15), E^AN_{σ1}(x) = (1/2)g2(x) + (x/2)(log x + 2γ − 1), where g2(x) = ∑_{n=1}^∞ {x/n}² ≥ 0. Therefore E^AN_{σ1}(x) ≥ (x/2)(log x + 2γ − 1). For x large this is ≫ x log x. Theorem 1.2.1 asserts E^AN_{σ1}(x) ≪ x^{δ'} exp(log x / log log x) with δ' = max{1/2, δ} and 0 < δ < 1, so δ' < 1. For any fixed δ' < 1, x^{1−δ'} log x · exp(−log x/log log x) → ∞, so the lower bound is incompatible with the claimed upper bound by a factor that diverges as x → ∞. This requires no estimates of g2 and no RH input; it is a self-inconsistency in the statement of the theorem. Independently, the proof in §4.1 ends with an explicit O(x^{3+δ}) term after the contour shift and then silently drops it. Since x^{3+δ} dominates all displayed terms, the proof is also invalid as written. The paper's own Remark 4.2.1 concedes the RH-equivalence obtained for φ(n) is not available for σ1(n), but that does not affect the direct contradiction from (2.1.15). Thus rejection is warranted.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies an 'analytic part' E^AN_{σ1}(x) for the error term in the summatory function of σ_1(n), using a Volterra integral-equation framework previously developed by the author. The main result (Theorem 1.2.1) claims that, under the Riemann Hypothesis, E^AN_{σ1}(x) ≪ x^{δ'} exp(log x / log log x) for x ≥ e^e, with δ' = max{1/2, δ} and 0 < δ < 1 arbitrary. A corollary (Theorem 1.2.2/1.2.3) is derived as E^AN_{σ1}(x) ≪ x^{δ'+ε}. The derivation consists of an explicit formula for E^AN_{σ1}(x), Mellin-transform identities, contour deformation, and estimates of ζ(s)ζ(s−1) under the Riemann Hypothesis.","tokens_in":9826,"tokens_out":2441,"duration_ms":23497,"significance":"If the main theorem were correct, it would provide a conditional bound analogous to the known result for the Euler-totient analytic part, which is a known equivalent of the Riemann Hypothesis. The author explicitly notes (Remark 4.2.1) that no such equivalence is obtained for σ_1(n). The paper's explicit decomposition and Mellin transforms are clearly presented; however, the central claim is immediately falsified by the paper's own definition of E^AN_{σ1}(x), making the result untenable as stated. The proof also contains an unresolved dominating error term in the final displayed estimate.","major_comments":[{"comment":"The definition in (2.1.15) gives E^AN_{σ1}(x) = (1/2)g2(x) + (x/2)(log x + 2γ − 1), with g2(x) ≥ 0. Therefore E^AN_{σ1}(x) ≥ (x/2)(log x + 2γ − 1), which is ≫ x log x. Theorem 1.2.1 asserts E^AN_{σ1}(x) ≪ x^{δ'} exp(log x/log log x) with δ' = max{1/2, δ} < 1 for every 0 < δ < 1. For any such δ', the ratio x^{1−δ'} log x · exp(−log x/log log x) tends to infinity as x → ∞, so the claimed upper bound is impossible. This is a direct contradiction from the paper's own equations and does not depend on the later estimates or on the Riemann Hypothesis.","section":"§2.1, Eq. (2.1.15) vs. Theorem 1.2.1"},{"comment":"The proof of Theorem 1.2.1 ends with E^AN_{σ1}(x) ≪ x^δ exp(log x/log log x) + x^{1/2} exp(log x/log log x) + O(x^{3+δ}). The term O(x^{3+δ}) is retained in this equation and then silently omitted when the theorem is stated. Since x^{3+δ} dominates the claimed bound x^{δ'} exp(log x/log log x) for every 0 < δ < 1 and all large x, the proof as written does not establish the theorem. This is a load-bearing gap independent of the contradiction in (2.1.15).","section":"§4.1, display after (4.1.13)"},{"comment":"The corollary is stated as Theorem 1.2.3 but its proof is headed 'Proof of Theorem 1.2.2'. This is a numbering inconsistency that should be corrected if the paper is revised, though it does not affect the mathematical content.","section":"§4.2 vs. Theorem 1.2.3"}],"minor_comments":[{"comment":"The abstract says 'explicit bounds', but the constants in Theorem 1.2.1 are implicit; the dependence on δ is also left implicit. Clarify whether the bound is meant to be uniform in δ or valid for each fixed δ.","section":"Abstract/Introduction"},{"comment":"The decomposition E_{σ1}(x) = E^AR_{σ1}(x) + E^AN_{σ1}(x) + O(x^{1/2}) in (2.1.14) contains an error term, so the names 'arithmetic part' and 'analytic part' are only approximate. This is worth stating explicitly, especially since the paper draws an analogy with the exact decomposition (1.1.6).","section":"§2.1, Eq. (2.1.14)"},{"comment":"In the line before (4.1.6), the bound for ζ(s)ζ(s−1) is written with a factor |t|^{1−η} and then replaced by |t| exp(log|t|/log log|t|); the transition is valid but slightly compressed. A sentence explaining the inequality would improve readability.","section":"§4.1, Fact 4.1.1 bound"}],"recommendation":"reject","confidential_remarks":"This is a clear-cut case: the main theorem is directly contradicted by the paper's own definition of the object it studies, and the proof's final estimate has a dominating error term that is dropped. No amount of revision within the manuscript's scope can repair the central claim. The author may wish to reconsider whether a corrected and more modest statement can be salvaged, but the current version should not be published. The report is grounded entirely in internal inconsistencies; no external judgement about the framework is needed."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThe headline: Theorem 1.2.1 is false as stated, and the proof doesn't survive contact with its own definition. The analytic part E^AN_{σ1}(x) is defined in (2.1.15) as (1/2)g_2(x) + (x/2)(log x + 2γ − 1). Since g_2 ≥ 0, E^AN ≥ (x log x)/2, which for any δ′<1 is larger than x^{δ′} exp(log x/log log x) once x is large. That alone kills the claimed bound; no RH input is needed. The stress-test note is correct.\n\nThe paper does have a legitimate framework: the decomposition of the divisor-summatory error term into an arithmetic part x f_2(x) and an analytic part follows Kaczorowski–Wiertelak’s Volterra approach, and the author correctly observes that the RH equivalence for Euler’s totient does not carry over to σ_1. The Mellin transform in Lemma 3.1.1 is standard, and the writing is generally clear.\n\nBut the problems are not minor. The proof of Theorem 1.2.1 in §4.1 ends with an O(x^{3+δ}) term after the contour shift. That term dominates x^{δ′} exp(log x/log log x) for any δ<1, and the proof silently drops it. There’s also an unexplained discrepancy in the residue calculation in Lemma 3.1.2. The final theorem is therefore unsupported even if the definition were altered.\n\nI’d reject this as it stands. It deserves a serious referee only if the author fixes the definition/theorem mismatch; right now the contradiction is too immediate. I wouldn’t cite it, and the reading-group value is mainly as a caution about checking a claimed bound against the object’s explicit formula.","headline":"Theorem 1.2.1 contradicts the paper's own definition of E^AN_{σ1}, and the proof drops a dominant error term; reject.","tokens_in":10288,"tokens_out":3470,"would_cite":false,"duration_ms":32046,"reading_group":"maybe","serious_thinker":"no","would_accept_peer_review":false},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11M26","11N37","11A25"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper claims a Riemann Hypothesis bound for the analytic part of the divisor-sum error term, analogous to the Euler totient case.","keywords":["divisor function","summatory function","analytic part","arithmetic part","Riemann Hypothesis","Mellin transform","Volterra integral equation","error term"],"falsifier":"Compute E^AN_{σ1}(x) = 1/2∑_{n≥1}{x/n}² + x/2(log x+2γ−1). Since the square series is nonnegative, E^AN_{σ1}(x) ≥ x/2(log x+2γ−1). For x large this term is ≫ x log x, which grows faster than x^{δ'} exp(log x / log log x) for any δ'<1. For instance, at x=10^6 the explicit x log x term alone exceeds any such claimed bound by a factor that diverges as x^{1/2}/exp(o(log x)).","tokens_in":9296,"feed_emoji":"🧮","tokens_out":13662,"duration_ms":123169,"temperature":0.7,"pith_summary":"The paper attempts to prove that, under the Riemann Hypothesis, the analytic part E^AN_{σ1}(x) of the error term in the asymptotic formula for the summatory function of the divisor function σ_1(n) satisfies E^AN_{σ1}(x) ≪ x^{δ'} exp(log x / log log x), where δ'=max{1/2,δ} for any 0<δ<1. This would extend a known RH-based bound from the Euler totient function to the divisor function. The author also argues that, unlike the totient case, this bound is one-way: the analytic part for σ_1 cannot be used to characterize the Riemann Hypothesis. The proof uses a Mellin transform identity and a contour shift under RH.","feed_headline":"Divisor-sum error's analytic part claimed near square root under RH","feed_subtitle":"If right, it caps a piece of the divisor-sum error at near the square-root scale, though it cannot certify RH.","key_machinery":"The central object is the analytic part E^AN_{σ1}(x)=1/2 g_2(x)+x/2(log x+2γ−1), with g_2(x)=∑_{n≥1}{x/n}², arising from the Volterra integral equation decomposition of the divisor-summatory error. The carrying identity is the Mellin transform ∫_1^∞ E^AN_{σ1}(x)x^{-s-1}dx = π²/12·1/(s−2)+ζ(s)ζ(s−1)/(s(1−s))+O(1), which after inverse Mellin transform and contour shift reduces the bound to estimating ζ(s)ζ(s−1) in a strip using RH. The distinguishing feature is the factor ζ(s)ζ(s−1) with no ζ(s) in the denominator, which the author argues prevents the equivalence reversal seen for the totient function.","core_discovery":"In the paper's own terms, the central discovery is the bound (1.2.1): under the Riemann Hypothesis, for every 0<δ<1 and x≥e^e, |E^AN_{σ1}(x)| ≪ x^{δ'} exp(log x / log log x) with δ'=max{1/2,δ}. Here E^AN_{σ1}(x) = 1/2∑_{n≥1}{x/n}² + x/2(log x+2γ−1) is the analytic part of the error term E_{σ1}(x)=∑_{n≤x}σ_1(n)−(π²/12)x², obtained through the Volterra integral equation decomposition. The proof derives the bound from the Mellin transform of E^AN_{σ1}, which is ζ(s)ζ(s−1)/(s(1−s)) plus pole terms, by shifting the contour to the critical line and applying RH bounds for ζ(s).","pith_inferences":["The definition of E^AN_{σ1}(x) explicitly includes the term x/2(log x+2γ−1), which alone grows like x log x; since g_2(x) is nonnegative, E^AN_{σ1}(x) ≥ x/2(log x+2γ−1) for large x, so the claimed sublinear bound is inconsistent with the paper's own definition regardless of the analytic machinery.","The proof discards an O(x^{3+δ}) error term from the inverse Mellin transform in Lemma 3.1.2; this remainder is larger than the bound being proved, so the argument as written would not establish the theorem even if the x log x term were absent.","The paper's negative observation about the absence of an RH-equivalence for σ_1 is plausible as a structural remark: the Mellin transform involves ζ(s)ζ(s−1) rather than ζ(s−1)/ζ(s), so the pole at s=2 is not cancelled by a denominator ζ(s)."],"forward_implications":["If the bound holds, the analytic part of the divisor-summatory error would be O(x^{δ'} exp(log x / log log x)), placing it below the full error's leading terms.","A direct corollary would be the bound E^AN_{σ1}(x) ≪_ε x^{δ'+ε} for every ε>0.","Unlike the totient case, no converse implication to the Riemann Hypothesis would follow from this analytic part.","For arithmetic functions in the Volterra framework whose Dirichlet series has ζ(s) in the numerator but not the denominator, the analytic part would not yield an RH characterization."],"fun_headline_variants":["Under RH, divisor-sum error's analytic part squeezed to near sqrt","Divisor-sum analytic term: RH forces near-square-root bound","RH yields near-sqrt bound for divisor-sum analytic error","Divisor function: analytic error part bounded by x^{1/2} under RH"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The proof rests on the unstated assumption that the O(x^{3+δ}) remainder from the inverse Mellin transform in Lemma 3.1.2 can be discarded in the final estimate; that remainder, if kept, overwhelms the bound the theorem claims.","fun_headline_variants_meta":{"raw":{"variants":["Under RH, divisor-sum error's analytic part squeezed to near sqrt","Divisor-sum analytic term: RH forces near-square-root bound","RH yields near-sqrt bound for divisor-sum analytic error","Divisor function: analytic error part bounded by x^{1/2} under RH"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00068,"raw_usage":{"total_tokens":2876,"prompt_tokens":644,"completion_tokens":2232,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":388,"completion_tokens_details":{"reasoning_tokens":2156}},"tokens_in":388,"tokens_out":2232,"duration_ms":15114,"temperature":1.0,"reasoning_tokens":2156,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T10:07:40.161750+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute E^AN_{σ1}(x) = 1/2∑_{n≥1}{x/n}² + x/2(log x+2γ−1). Since the square series is nonnegative, E^AN_{σ1}(x) ≥ x/2(log x+2γ−1). For x large this term is ≫ x log x, which grows faster than x^{δ'} exp(log x / log log x) for any δ'<1. For instance, at x=10^6 the explicit x log x term alone exceeds any such claimed bound by a factor that diverges as x^{1/2}/exp(o(log x)).","supporting_citations":[],"review_version":1}