{"id":"42a8cb64-a5c0-4875-9fc0-000c7b9dec30","arxiv_id":"2507.08296","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A new zero-density estimate for Dirichlet L-functions with exponent 7/3, improving Huxley's 12/5, is proven by extending the Guth-Maynard large value method to character-twisted Dirichlet polynomials.","lead":"This paper proves a sharper bound on how many zeros of Dirichlet L-functions can lie close to the vertical line with real part 1, improving the previous best exponent from 12/5 to 7/3. The proof extends the 2024 Guth-Maynard large value method to character-twisted Dirichlet polynomials and derives new results on primes in arithmetic progressions and the least Goldbach number.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Value threshold σ/6 vs σ is internally inconsistent at Lemma 4.1/Proposition 10.1; with the stated weaker threshold the SVD and energy bounds do not produce the claimed exponent.","rationale":"The reader identified the Fourier decay check for Proposition 8.1 as the weakest point. That concern is reasonable but appears repairable: the convolution theorem plus ∫|R(u,−b)|^2 du≲|W|^2≤(qT)^{2+o(1)} leaves room under the stated (qT)^3 factor. The more load-bearing issue is the inconsistent value threshold. Lemma 4.1's stated hypothesis N^{σ/6} cannot yield the N^{1−2σ} factor that Proposition 4.6 and the final exponent depend on; Proposition 10.1 repeats the same threshold while Lemma 10.3's proof uses N^σ. If the weak threshold were taken literally, the energy bound degrades to E(W)≲N^{−σ/3}Σ|R|^3, and the S3 bound would only imply |W| of size about (qT)^{1+o(1)} in the critical range, instead of the claimed (qT)^{4(1−σ)/(1+σ)}. Because Section 12.3 applies the argument with |D_N|⪆N^σ and Lemma 10.3 uses N^σ, the intended assumption is almost certainly N^σ, making this a systematic typo rather than a conceptual gap. That is why the verdict should remain CONDITIONAL: the proof is plausible and the fix is clear, but the manuscript as written cannot be verified without correction. No judgment is made on the author; the issue is purely internal consistency of the argument.","tokens_in":45736,"tokens_out":21962,"duration_ms":215880,"concrete_test":"Re-derive Lemma 4.1 from the singular value decomposition: for a common threshold V, |W|V^2≤s1(M)^2Σ|b_n|^2, hence |W|≲N V^{-2}s1(M)^2. Substitute V=N^{σ/6} and V=N^σ, and check which version is used in Proposition 4.6 and in the final simplification of Section 11.2. Independently, in Proposition 10.1, replace N^{σ/6} by N^σ in Lemmas 4.1, 10.3, and Propositions 3.1/4.6, then re-run Sections 10–11; if the S3 bound and the final exponent 7/3 are reproduced, the manuscript needs only a systematic typo correction; if the bound fails, the central claim lacks proof.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Lemma 4.1 states that if |S_N(t,χ)|≥N^{σ/6}, then |W|≲N^{1−2σ}s1(M)^2. The SVD argument gives |W|≲N V^{-2}s1(M)^2; with V=N^{σ/6} this is N^{1−σ/3}s1(M)^2, not N^{1−2σ}s1(M)^2, so Proposition 4.6's N^{1−2σ}(ΣI)^{1/3} term is not justified. The same threshold N^{σ/6} appears in Propositions 3.1 and 10.1, while Lemma 10.3's proof begins with |D_N(t,χ)|≥N^σ. If the weaker threshold were intended, the energy bound becomes E(W)≲N^{−σ/3}Σ|R|^3, and the S3 term qT N|W|^{1/2}E(W)^{1/2} is at least qT N^{3/2−σ/6}|W|^2; at N=(qT)^{3/4}, σ≈0.7, the resulting |W| bound is of size (qT)^{1+o(1)}, far worse than the claimed (qT)^{4(1−σ)/(1+σ)}. Thus either all occurrences of N^{σ/6} in the reduction must be N^σ, or the central large-value estimate fails. Section 12.3 explicitly uses |D_N|⪆N^σ, so the fix appears intended, but it must be made systematically and consistently throughout Sections 3, 4, 10, and 11.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper adapts the Guth-Maynard large-value method to Dirichlet polynomials twisted by primitive Dirichlet characters and obtains a zero-density estimate for Dirichlet L-functions: summing N(σ,T,χ) over χ mod q gives exponent 7/3, improving Huxley's 12/5. The proof follows the standard framework: a matrix SVD reduction, bounds for the trace terms S1, S2, S3, an affine-GCD-twist sum estimate, moments and energy estimates for the character-weighted function R, and a final zero-detection argument. Two arithmetic applications are stated: an improved bound for the least prime in an arithmetic progression modulo a prime power and an improved bound for the least Goldbach number modulo a prime.","tokens_in":46108,"tokens_out":14290,"duration_ms":147555,"significance":"If correct, this is a substantial advance: it is the first adaptation of the Guth-Maynard machinery to character-twisted Dirichlet polynomials and yields the best known zero-density exponent in the q-aspect as well as in T. A notable strength is that the argument is parameter-free: the bounds are derived from analytic lemmas (Heath-Brown's double zeta sum bound, Guth-Maynard's matrix lemmas) rather than tuned to the final exponent. The claimed improvement from 12/5 to 7/3, together with the arithmetic corollaries, makes this a consequential contribution to multiplicative number theory, provided the normalization and Fourier-decay issues described below are resolved.","major_comments":[{"comment":"The threshold N^{σ/6} is internally inconsistent with the SVD argument and with the rest of the paper. In Lemma 4.1 the hypothesis |S_N(t,χ)| ≥ N^{σ/6} and the inequality ∑_{(t,χ)} |S_N(t,χ)|^2 ≤ s_1(M)^2 ∑_n |b_n|^2 give |W| N^{σ/3} ≤ N s_1(M)^2, hence |W| ≤ N^{1−σ/3} s_1(M)^2, not |W| ≤ N^{1−2σ} s_1(M)^2. For σ = 4/5 the claimed right-hand side N^{−3/5} s_1(M)^2 is typically below 1, which cannot bound a nonempty set W. Lemma 10.3's proof and Section 12.3 both use the condition |D_N(t,χ)| ≥ N^σ, and Proposition 10.1 is used with the same normalization in Section 11. The intended hypothesis is evidently V ≈ N^σ, up to an absolute constant (e.g. N^σ/6, not N^{σ/6}). The occurrences in Lemma 4.1, Propositions 3.1, 4.6, and 10.1, and in the S3/energy bounds of Sections 10–11, must be corrected systematically, and the reduction in Section 3 from |D_N| ≥ N^σ to the three parts D^{(i)} each inheriting ≥ N^σ/3 should be restated with the corrected '≥ c N^σ' form.","section":"Sections 3, 4, 10; Lemma 10.3; Section 12.3"},{"comment":"The verification that f_b(u) = ψ(u)|\\tilde R_{M2}(u,-b)|^2 satisfies the Fourier-decay hypothesis \\hat f_b(ξ) ≤ (qT)^3 (T/|ξ|)^j is compressed into a single sentence. A complete proof needs to use the convolution theorem for \\tilde R_{M2}^2 as K * |R|^2, then handle the product with ψ by splitting the convolution integral for \\hat ψ * \\widehat{|\\tilde R|^2} at |η| ≈ |ξ|/2, and it should quote the correct bound on ∫|R|^2 from Lemma 9.1 (which is ≲ ϕ(q)|W|, hence ≤ (qT)^{O(1)}, rather than literally |W|^2). Since Proposition 8.1 is the key input to the S3 bound, the omitted derivation should be written out.","section":"Section 11.1, Proposition 8.1 hypothesis"}],"minor_comments":[{"comment":"The statement writes 'W ≲_ϵ ...' where it should write '|W| ≲_ϵ ...'.","section":"Proposition 4.6"},{"comment":"In the splitting argument, the third part is said to give the same bound for '|W2|' again; it should be '|W3|'.","section":"Section 3, proof of Theorem 1.3"},{"comment":"The reference to Guth-Maynard's work appears as '[8]' in Theorem 1.3, but in the bibliography [8] is Forti-Viola and [9] is Guth-Maynard; the citation should be [9].","section":"Sections 1 and 3"},{"comment":"When applying the first part of Theorem 1.3 to D_N^2, the first term should be N^{4−4σ} rather than N^{2−2σ}; the subsequent domination by the second term is unaffected, but the displayed formula should be corrected.","section":"Section 12.3, first range"},{"comment":"There are numerous typos and formatting issues, including 'V ALUE' in the title, 'Specificallly', 'definded', 'triangel', 'Cauchy-Schwaz', 'innner sum', 'Airth', and 'Moebius'; these should be cleaned up before publication.","section":"Throughout"},{"comment":"The sentence 'Applying the convolution theorem and noting ∫|R(u,−b)|^2 1_{[1/2,2]}(u) du ≲ |W|^2 ≤ (qT)^2' should be replaced by the precise Lemma 9.1 bound ≲ φ(q)|W|, with a short explanation of why this is sufficient for the Fourier-decay hypothesis.","section":"Section 11.1"}],"recommendation":"major_revision","confidential_remarks":"The threshold issue looks like a notational slip rather than a fundamental flaw: Lemma 10.3 and Section 12.3 already use the correct N^σ normalization, and the SVD argument works with V = N^σ/6. If the author confirms this correction and supplies the missing Fourier-decay details for Proposition 8.1, the main theorem is plausible. I do not see circularity or fitted parameters. The paper is a serious contribution but needs the consistency fix and a fuller S3 verification before I can recommend acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this paper extends the Guth–Maynard large value machinery to Dirichlet characters and claims a zero-density exponent of 7/3. The new content is real—the affine-transformation/GCD-twist sum in Proposition 8.1 is a genuine adaptation of the method—and the overall architecture follows the Guth–Maynard template cleanly. If the main estimates hold, it improves Huxley's 12/5, Jutila's, and Forti–Viola's exponents, and the applications to the least prime and Goldbach numbers are reasonable corollaries.\n\nThe problem is a threshold error that runs through the core lemmas. Lemma 4.1, Proposition 3.1, Proposition 4.6, and Proposition 10.1 all state |S_N(t,χ)| ≥ N^{σ/6}, but the SVD argument gives |W| ≲ N^{1−σ/3} s1(M)^2, not N^{1−2σ} s1(M)^2. The stress-test note is right: with the weaker threshold, the energy bound and the final large-value estimate collapse. The intended threshold is clearly N^σ—Section 12.3 uses |D_N| ⪆ N^σ—so this looks like a systematic typo rather than a conceptual gap. Still, it has to be fixed all through Sections 3, 4, 10, and 11, and the exponents re-checked at each step. A referee should demand that.\n\nThe other soft spot is the verification of the Fourier decay hypothesis for Proposition 8.1 in Section 11. They assert the decay in a single sentence using the convolution theorem and the |W|² bound. That step needs the derivative estimates written out; without them, S3 and the whole theorem depend on an unproven hypothesis. It may be routine, but it is not routine from what is printed.\n\nThe rest of the paper looks credible. The S1 and S2 bounds follow the established pattern with Heath-Brown's double zeta sum bound, and the energy argument is structurally sound. There are some citation slips (e.g., Corollary 1.6 cites [7] for Jutila, which is Davies), but those are cosmetic.\n\nBottom line: this is a serious paper with a plausible main theorem, but not ready as written. I would send it to a strong referee and ask for the threshold fix and the Fourier-decay details before recommending acceptance. It deserves the referee's time.","headline":"A serious Guth-Maynard extension to Dirichlet characters with a plausible 7/3 exponent, but a systematic N^{σ/6} vs N^σ threshold typo must be fixed before the estimates as written are trustworthy.","tokens_in":46637,"tokens_out":4845,"would_cite":false,"duration_ms":48986,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11M06","11M26","11N05","11N13"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves the zero-density bound (qT)^{7(1-σ)/3+ε} for Dirichlet L-functions summed over characters modulo q, improving the earlier exponent 12/5 and yielding new bounds on least primes and Goldbach numbers.","keywords":["large value estimates","Guth-Maynard method","zero-density estimates","Dirichlet L-functions","Dirichlet polynomials","distribution of primes","primes in arithmetic progressions","Goldbach number"],"falsifier":"Take an explicit small modulus $q$, a separated set $W$ of pairs $(t,\\chi)$, and the functions $f_b(u)=\\psi(u)|\\widetilde R_{M_2}(u,-b)|^2$ of Section 11, then compute $|\\hat f_b(\\xi)|$ at frequencies $|\\xi|$ from $T$ up to $(qT)^2$: if the decay is ever weaker than $(qT)^3(T/|\\xi|)^j$, the hypothesis of Proposition 8.1 fails and the $S_3$ bound — hence the exponent $7/3$ — collapses.","tokens_in":45481,"feed_emoji":"🔢","tokens_out":24172,"duration_ms":222973,"temperature":0.7,"pith_summary":"This paper proves a new zero-density estimate for Dirichlet $L$-functions: summed over all characters $\\chi$ modulo $q$, the number of zeros $\\rho=\\beta+it$ of $L(s,\\chi)$ with $\\beta\\ge\\sigma$ and $|t|\\le T$ is bounded by $(qT)^{7(1-\\sigma)/3+\\epsilon}$, improving the earlier exponent $12/5$. The proof extends a recent large-value method for long Dirichlet polynomials to polynomials twisted by primitive characters, and its key new ingredient is a sharp bound for sums over affine transformations of smooth functions carrying a GCD twist. From the new density bound the paper derives two arithmetic consequences: a bound on the least prime in an arithmetic progression modulo a prime power, and a bound on the least Goldbach number in a progression modulo a prime. Zero-density exponents control how many zeros can sit close to $\\sigma=1$, and tighter control translates directly into sharper quantitative statements about primes.","feed_headline":"L-function zero density exponent drops from 12/5 to 7/3","feed_subtitle":"New zero-density bounds for Dirichlet L-functions sharpen results on least primes and Goldbach numbers.","key_machinery":"The load-bearing object is the large-value set $W$ of separated pairs $(t,\\chi)$ on which a character-twisted Dirichlet polynomial $D_N(t,\\chi)=\\sum_{N<n\\le 2N}a_n\\chi(n)n^{it}$ is large. The argument forms the $|W|\\times N$ matrix $M$ with entries $w(n/N)\\chi(n)n^{it}$, bounds its largest singular value through the traces of $MM^*$ and $(MM^*)^3$, and splits the resulting oscillatory sum into $S_1,S_2,S_3$ according to how many of the three Fourier variables vanish. $S_3$, the hardest term, is controlled by the paper's principal new tool, Proposition 8.1: a bound for the integral over $u$ of the square of $\\sum_{b,m_1,m_2,m_3}f_b((m_1u+m_3)/m_2)\\,(am_1+bm_2+m_3,q)$, valid when the smooth functions $f_b$ have Fourier decay $|\\hat f_b(\\xi)|\\lesssim (qT)^3(T/|\\xi|)^j$. This affine-transform-with-GCD-twist estimate is what carries the adaptation of the Guth-Maynard large-value method (a recent technique for bounding how often a long Dirichlet polynomial can be large) into the character setting.","core_discovery":"On the paper's own terms, the central discovery is that a recent large-value method for long Dirichlet polynomials survives the passage to polynomials twisted by primitive Dirichlet characters, provided the character sums are controlled by a new bound for affine transformations with GCD twists. For a separated set $W$ of pairs $(t,\\chi)$ on which a character-twisted polynomial of length $N$ attains values $\\ge V$, Theorem 1.3 gives $|W|\\ll_\\epsilon N^2V^{-2}+(qT)^{4/3}N^2V^{-4}$ when $(qT)^{3/4}\\le N\\le (qT)^{5/6}$, and a further four-term bound for larger $N$. Fed into the zero-detection method, this yields $\\sum_{\\chi\\bmod q}N(\\sigma,T,\\chi)\\ll_\\epsilon (qT)^{4(1-\\sigma)/(1+\\sigma)}$, and the combination with the classical bound for $\\sigma\\le 5/7$ produces the headline estimate $(qT)^{7(1-\\sigma)/3+\\epsilon}$. The paper also shows the new exponent pays off arithmetically: for fixed prime $p$, the least prime $p(p^n,k)$ in the progression $k\\bmod p^n$ is $\\ll_{p,\\epsilon}(p^n)^{7/3+\\epsilon}$, and the least Goldbach number $G(p,k)\\equiv k\\pmod p$ is $\\ll_\\epsilon p^{7/6+\\epsilon}$.","pith_inferences":["Beyond the paper: the Fourier decay asserted in Section 11 is the point where the proof is most exposed; writing out the full derivative estimates — or exhibiting a configuration where they fail — would clarify whether the method still has slack to push below the exponent $7/3$.","Beyond the paper: the GCD-twist affine-sum bound of Proposition 8.1 is also the natural lever for the $Q^2T$-analogue that the introduction points to; averaging over moduli up to $Q$ rather than a single $q$ would likely need a $Q$-dependent version of this proposition.","Beyond the paper: the same architecture of singular values, trace expansion, and affine-transform sums should transfer to other $L$-function families whose characters enter through sums of the same shape, so the improvement need not stop at Dirichlet $L$-functions.","Editorial note: Corollary 1.6 attributes its antecedent to 'Jutila [7]', but entry [7] in the reference list is Davies's Kakeya paper; the Jutila work on the least Goldbach number is listed as [15]. This mismatch does not touch the proof of the corollary."],"forward_implications":["Corollary 1.5: for a fixed prime $p$, every integer $k$ coprime to $p^n$ has a prime in the progression with $p(p^n,k)\\ll_{p,\\epsilon}(p^n)^{7/3+\\epsilon}$.","Corollary 1.6: for an odd prime $p$, the least Goldbach number $G(p,k)\\equiv k\\pmod p$ satisfies $G(p,k)\\ll_\\epsilon p^{7/6+\\epsilon}$.","At the threshold $N=(qT)^{4/5}$ with $V=N^{3/4}$, Theorem 1.3 gives $|W|\\ll(qT)^{8/15}$, strictly better than the $(qT)^{3/5}$ from both the classical mean value theorem and the Halász–Montgomery–Huxley estimates.","In the range $5/7\\le\\sigma\\le 7/9$ the new bound is the strongest available zero-density estimate for all $q,T$ simultaneously, since it beats (1.3) for $\\sigma>5/7$ and (1.4) for $\\sigma<7/9$.","The abstract additionally announces new results on primes in arithmetic progressions in short intervals, in particular for prime-power moduli, as consequences of the density estimate."],"supporting_citations":[{"why":"Supplies the large-value method being adapted: the singular-value matrix reduction, the trace expansion, the energy bound, and the moment estimates for long Dirichlet polynomials.","marker":"[9]"},{"why":"Gives Huxley's large-value estimates, the comparison bound (1.4), and the previous zero-density exponent 12/5 that Theorem 1.4 improves.","marker":"[12]"},{"why":"Montgomery's monograph provides the classical mean value theorem, the zero-detection method used in Section 12, and the large value conjecture motivating the work.","marker":"[19]"},{"why":"Heath-Brown's double zeta sum bound is applied directly in the S2 estimate (Section 6) and in the second/fourth moment and energy bounds of Section 10.","marker":"[10]"},{"why":"Jutila's approximate functional equation for Dirichlet polynomials is the reflection principle used to prove the S2 bound.","marker":"[17]"}],"fun_headline_variants":["Dirichlet L-function: zero density exponent cut to 7/3","Improved zero density bounds for Dirichlet L-functions","New zero density estimate: exponent 7/3 beats Huxley's 12/5","Goldbach and prime progressions improved via L-function zeros"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The $S_3$ bound, and with it the entire improved exponent $7/3$, rests on the Fourier decay assumption that the smoothed functions $f_b(u)=\\psi(u)|\\widetilde R_{M_2}(u,-b)|^2$ satisfy $|\\hat f_b(\\xi)|\\lesssim (qT)^3(T/|\\xi|)^j$ for every $j$; Section 11 asserts this decay in a single sentence via the convolution theorem, and the derivative estimates behind it are not written out.","fun_headline_variants_meta":{"raw":{"variants":["Dirichlet L-function: zero density exponent cut to 7/3","Improved zero density bounds for Dirichlet L-functions","New zero density estimate: exponent 7/3 beats Huxley's 12/5","Goldbach and prime progressions improved via L-function zeros"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000555,"raw_usage":{"total_tokens":2692,"prompt_tokens":1043,"completion_tokens":1649,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":659,"completion_tokens_details":{"reasoning_tokens":1573}},"tokens_in":659,"tokens_out":1649,"duration_ms":11468,"temperature":1.0,"reasoning_tokens":1573,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T18:23:37.310574+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take an explicit small modulus $q$, a separated set $W$ of pairs $(t,\\chi)$, and the functions $f_b(u)=\\psi(u)|\\widetilde R_{M_2}(u,-b)|^2$ of Section 11, then compute $|\\hat f_b(\\xi)|$ at frequencies $|\\xi|$ from $T$ up to $(qT)^2$: if the decay is ever weaker than $(qT)^3(T/|\\xi|)^j$, the hypothesis of Proposition 8.1 fails and the $S_3$ bound — hence the exponent $7/3$ — collapses.","supporting_citations":[{"cited_title":"Huxley, Large values of Dirichlet polynomials","cited_arxiv_id":null,"evidence_quote":"Gives Huxley's large-value estimates, the comparison bound (1.4), and the previous zero-density exponent 12/5 that Theorem 1.4 improves."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Montgomery's monograph provides the classical mean value theorem, the zero-detection method used in Section 12, and the large value conjecture motivating the work."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Heath-Brown's double zeta sum bound is applied directly in the S2 estimate (Section 6) and in the second/fourth moment and energy bounds of Section 10."},{"cited_title":"Jutila, Zero-density estimates for L-functions, Acta Arith","cited_arxiv_id":null,"evidence_quote":"Jutila's approximate functional equation for Dirichlet polynomials is the reflection principle used to prove the S2 bound."}],"review_version":1}