Matching works

6 works

math.NT — Number Theory

Every even integer greater than 2 is a prime plus a product of at most 18 primes

Contributed by Wren Aquila

We prove that every even integer N≥4N \ge 4 can be written as N=p+ηN = p + \eta with pp prime and η≥2\eta \ge 2 a product of at most 1818 primes, counted with multiplicity. The best previously stated unconditional value we have located is due to Johnston and Starichkova: 395395 in the latest version of their paper, which revised the 369369 of its first two arXiv versions. Assuming that no Dirichlet LL-function vanishes in the half-plane Re⁡s>7/8\operatorname{Re} s > 7/8, the same holds with at most 1414 primes, and under the generalised Riemann hypothesis for Dirichlet LL-functions with at most 99. The improvement comes from counting rather than bounding: an explicit lower bound for the number of primes pp with N−pN - p free of small primes is played against explicit Hardy--Ramanujan-type upper bounds for the integers with many prime factors, both in the range below the sieve threshold and inside the sieved sets. These upper bounds weight by Chebyshev's identity log⁡n=∑d∣nΛ(d)\log n = \sum_{d \mid n} \Lambda(d), use the exact power-series coefficients of a majorant of a finite Euler product and the Mertens density of rough primes, and pre-sieve modulo 11551155, so that the integers to be excluded inherit the residue classes of the primes that produce them. All numerical steps are certified in interval or ball arithmetic, except some discrete convolutions with nonnegative terms, which are evaluated in binary64 under an explicit rounding-error bound and were also recomputed independently in ball arithmetic.

Primarily AI-generated textHuman understanding: no parts
Nov 13, 202619 pages
PDFView work

math.NT — Number Theory

Cross-Energy of Dirichlet Partial Sums. Part 33: Optimised mollifiers and the bound Λ≤0.174\Lambda\le0.174 for the de Bruijn--Newman constant

Contributed by Pedro Caceres

The Polymath project reduced upper bounds for the de Bruijn--Newman constant Λ\Lambda to three verifiable hypotheses: the Riemann hypothesis up to height X/2X/2, a zero-free region for Ht0H_{t_0} to the right of XX, and a barrier at XX. With the verification of the Riemann hypothesis up to 3⋅10123\cdot10^{12} by Platt and Trudgian, this gave Λ≤0.2\Lambda\le0.2. The bottleneck is the zero-free region, which Polymath established with Euler-product mollifiers. We replace the Euler product by a mollifier ∑d≤Dλdd−s∗\sum_{d\le D}\lambda_d d^{-s_*} whose weights are chosen by solving an ℓ1\ell^1-regression problem, so that any choice of weights still yields a valid lower bound. At the same height X≈6⋅1012X\approx6\cdot10^{12} and t0=0.155t_0=0.155, y0=0.2y_0=0.2, the lower bound for the mollified sum improves from −0.12-0.12 (our Lemma~\ref{lem:moll} evaluated with the Euler mollifier on the primes up to 1313) to +0.05+0.05 with D=120D=120; with D=240D=240 it is still +0.04+0.04 at t0=0.154t_0=0.154. We cover all x≥Xx\ge X by a short chunking argument together with a tail estimate, verify the barrier with Polymath's Taylor scheme, and obtain \[ \Lambda\le 0.174 . \] The certificates are computer-assisted. They use Arb ball arithmetic for all transcendental functions and outward-rounded floating-point arithmetic for the large sums. The code, the weights and the logs are included, and the certificate was reproduced on a second platform. We also report that at this height the weights we found with D=240D=240 do not reach 0.1730.173, and we compare with Polymath's table of conditional results. We do not prove the Riemann hypothesis.

Author comments: Revised 10 Nov 2026 after external examination (see CHANGES_v2.md). 20 pages, 3 figures. The ancillary scripts are those of the original part (ancillary zip in the part folder). Part 33 of the book-series Cross-Energy of Dirichlet Partial Sums (47 parts, Parts 0-46); formerly Part 13. Renumbered 2 Nov 2026; mathematics unchanged. The paper states once, in its discussion, that it does not prove the Riemann hypothesis.

A mix of human-written and AI-generated textHuman understanding: all partsArb interval arithmeticRiemann xi-functioncomputer-assisted proofde Bruijn-Newman constantheat flowmollifier
Nov 13, 202620 pages
PDFView work

math.NT — Number Theory

Cumulative Prime Sums and Riemann Zeros: A fine grid, its ripple, and a family of prime races

Contributed by Antonio Mennillo

Summing the primes one after another and taking logarithms gives an increasing sequence of real numbers, σm=log⁡Spm\sigma_m=\log S_{p_m} with SN=1+∑p≤NpS_N = 1+\sum_{p\le N}p, which becomes exponentially dense. The ordinates γn\gamma_n of the zeros of the Riemann zeta function fall on this grid with tiny errors: 1.4⋅10−31.4\cdot10^{-3} at the first zero, 2.4⋅10−72.4\cdot10^{-7} at the fifth, 3⋅10−173\cdot10^{-17} at the twentieth. Part I explains why. The cell of the grid containing a height tt has width (1+o(1))t e−t/2(1+o(1))\sqrt t\,e^{-t/2} (Step Theorem), and every real t>log⁡3t>\log 3 has a node above it within that width, by an argument going back to Bertrand's postulate (Covering Theorem). The precision observed at the zeros is therefore the resolution of the grid, and the same precision holds at every height (Genericity Principle). We compute the first-crossing primes NnN_n with certification up to n=20n=20 and test, in a protocol fixed before the computation, whether the zeros sit closer to the nodes than a generic height. Neither the mean distance to the nearest node (0.2460.246 against 0.2500.250, p=0.92p=0.92) nor the distribution of the position within the cell (Kolmogorov–Smirnov, p=0.79p=0.79) departs detectably from the generic case, and an integer grid used as control behaves in the same way. We also correct the half-exponent law by a factor t′/t\sqrt{t'/t}. Part II shows where the zeros are actually visible. They are not in the position of the nodes but in the ripple: the deviation of the grid from its mean curve B(x)=1+li(x2)−li(4)B(x)=1+\mathrm{li}(x^2)-\mathrm{li}(4). This ripple is a sum of pure oscillations, one per zero; its spectrum shows a peak at each of the first 13 ordinates, with the amplitude predicted by the explicit formula. Under RH the ripple has size e−t/4e^{-t/4} on the grid, much larger than the cell e−t/2e^{-t/2}, and RH is equivalent to this decay rate. The ripple is not centred: on [103,108][10^3,10^8] the sum of the primes stays below li(x2)\mathrm{li}(x^2) for about 97% of logarithmic time, and it first exceeds it at x=109x=109. We place this in a one-parameter family of prime races, ∑p≤xpa\sum_{p\le x}p^a against li(xa+1)\mathrm{li}(x^{a+1}), and prove that under RH the normalized error has mean −1/(2a+1)-1/(2a+1), coming from the squares of the primes. Under RH and LI the limiting logarithmic density of the set where the race is ahead is given by a Rubinstein–Sarnak formula; it moves from 2.6⋅10−72.6\cdot10^{-7} (counting the primes, a=0a=0) to 0.060.06 (summing them, a=1a=1) and tends to 1/21/2 as a→∞a\to\infty, at the rate 12−δa∼1/(2πAlog⁡(A/2π))\tfrac12-\delta_a\sim1/\bigl(2\sqrt{\pi A\log(A/2\pi)}\bigr) with A=a+12A=a+\tfrac12. At a=−1/2a=-1/2 the bias becomes the 12log⁡log⁡x\tfrac12\log\log x of the weighted counting functions of Aoki and Koyama. All theorems are formalized in Lean 4, with the classical results they use as explicit hypotheses.

A mix of human-written and AI-generated textHuman understanding: some partsChebyshev biasLean 4Riemann zerosexplicit formulaprime races
Nov 06, 202629 pages
PDFView work

math.NT — Number Theory

Ramanujan's prime counting inequality holds for all x ≥ 38,358,837,683: ranges from the zeros, a spectral barrier, and trade-offs

Contributed by Sayed Barez

Ramanujan proved that π(x)2<exlog⁡xπ(x/e)\pi(x)^2 < \frac{ex}{\log x}\pi(x/e) for all sufficiently large xx. Let NRN_R be the least integer such that this holds for every real x≥NRx \ge N_R. Dudek and Platt showed that the Riemann hypothesis implies NR=38,358,837,683N_R = 38{,}358{,}837{,}683. We prove that NR=38,358,837,683N_R = 38{,}358{,}837{,}683. To our knowledge, this is the first determination of NRN_R without assuming the Riemann hypothesis. The proof uses the zero-free half-plane Res>7/8\mathrm{Re}s > 7/8 of a recent OpenAI preprint. Its main statement for ζ\zeta is formalized in Lean 4 and has been rebuilt independently; its written proof has not been peer reviewed. The deduction is our Theorem B: if ζ(s)≠0\zeta(s) \ne 0 for Res>0.945092\mathrm{Re}s > 0.945092, then NR=38,358,837,683N_R = 38{,}358{,}837{,}683. The proof of Theorem B uses only refereed inputs, together with Bhattacharjee's correction of a theorem of Büthe. Relying only on the peer-reviewed literature, and on preprints where stated, we also prove the following by rigorous computer-assisted interval arithmetic. (A) The inequality holds for 38,358,837,683≤x≤e753.9838{,}358{,}837{,}683 \le x \le e^{753.98}. The previous range ended at e103e^{103}. (C) It holds for x≥e3035.01x \ge e^{3035.01}. This replaces Axler's threshold e3158.442e^{3158.442}, whose proof applies a lemma of Dudek and Platt in a case where that lemma's proof is defective. (B′, D) Either of the following also suffices: a zero-free region β≤1−1/(1.7373log⁡∣γ∣)\beta \le 1 - 1/(1.7373\log|\gamma|) above height 3⋅10123\cdot10^{12}, or RH verified to height 4.1602⋅10144.1602\cdot10^{14}. We tabulate the trade-off between the two. (E) Arguments that use only non-negativity, analytic continuation and the poles up to height TT cannot go beyond e754.40e^{754.40} when T=3⋅1012T = 3\cdot10^{12}. So (A) is optimal among such arguments to within 0.41 in log⁡x\log x. Building on Bhattacharjee's discovery that the constant E2E_2 in Büthe's Theorem 1 lacks a factor log⁡(3c)\log(3c), we show that the affected rows of the Broadbent–Kadiri–Lumley–Ng–Wilk table inherit the error, by 1.3–2.2% after re-optimization. We also trace the consequences in the literature.

Primarily AI-generated textHuman understanding: all partsChebyshev functionRamanujan's inequalitycomputer-assisted proofexplicit estimatesinterval arithmeticprime-counting functionzero-density estimateszero-free half-planezero-free region
Oct 20, 202628 pages
PDFView work

math.NT — Number Theory

Optimal windowed test functions for bounding analytic ranks of elliptic curves

Contributed by Sabrina Hackfort

Assuming the Generalized Riemann Hypothesis, the explicit formula for the L-function of an elliptic curve E/Q bounds the analytic rank of E in terms of the traces a_p for p < e^(2L), for any admissible test function supported on a window [-L,L]. In rank computations this test function is usually fixed in advance, typically to a Fejer-type kernel; optimising it for an individual L-function goes back to Booker, and an earlier test on quadratic twists of rank about 7 found the gain to be small. We choose, for each individual curve, the test function that minimises the resulting bound within a finite-dimensional space supported on [-L,L], certify the result, and measure the gain systematically, up to the current record curves. The minimiser is the solution of a linear system with the Gram matrix of the Weil quadratic form of E, and the optimal bound equals the reciprocal of the corresponding reproducing kernel at the central point, which we call the radix Riemanniana of E. Because the Fejer-type kernel lies in this space, the optimised bound is never weaker. We certify every bound with ball arithmetic, check the prime-side evaluation against the zero side computed independently from the zeros of L(E,s), and benchmark against the fixed kernel on 1642 curves from Cremona's tables, 2500 curves from the Stein-Watkins database and a set of curves with conductors up to 3*10^16. Depending on the sample and on the stopping criterion, the number of primes required to pin down the analytic rank decreases by 8-23% in aggregate. For individual curves of large conductor the decrease ranges from 0% to 43%. Within each sample the gain is largest at rank 0 and decreases with the rank; it grows again, and by a much larger factor, in the regime of large rank and large conductor where the computation is most expensive. On the smallest-conductor curves of rank 28-31 in the ICARM leaderboard the optimised kernel needs 84-85% fewer primes to certify the rank. For the current record, the rank-32 curve #4239 of Elkies, Klagsbrun and Li, it certifies r_an<=32 under GRH with 2.53*10^9 primes - the first upper bound for this curve - and hence r_an=32 under GRH and BSD, while the fixed kernel, certified at L=12.40 with 2 481 819 193 primes, is still 0.39 above the target. On all 395 leaderboard curves of rank 24-31, with primes below 10^10, the optimised kernel certifies r_an<= r on 379 curves and the fixed kernel on 372; where both succeed the fixed kernel needs a median of 4.0 times as many primes. We validate the fixed-kernel baseline by reproducing every published value of Bober and of Klagsbrun, Sherman and Weigandt with an independent implementation.

Primarily human-written textHuman understanding: all parts
Oct 19, 202614 pages
PDFView work

math.NT — Number Theory

Linnik's constant is at most 3.993.99

Contributed by Eric Naslund

Let P(a,q)P(a,q) denote the least prime congruent to aa modulo qq, where (a,q)=1(a,q)=1. We give a computer-assisted proof that P(a,q)≪q3.99P(a,q)\ll q^{3.99}, improving the exponent 55 of Xylouris. More precisely, P(a,q)<q3.99P(a,q)<q^{3.99} for every sufficiently large qq, uniformly in aa. The main new ingredient is a graded near-density estimate for zeros of Dirichlet LL-functions, a weighted form of Heath-Brown's Lemma~12.1 of the kind he asked for in 1992. We combine it with the zero-location estimates and far-density method of Heath-Brown and Xylouris. The second novelty is the scale of the case analysis. Heath-Brown and Xylouris closed their final case analyses with 1414 and 2121 main cases, each by a chain of inequalities evaluated in floating point. In this paper, with the help of advanced AI models, we can push this much further. The middle range of the first zero is divided into 44534453 root cases and 47884788 terminal cases, each closed by its own linear program. On 4,196,8794{,}196{,}879 threshold boxes these programs give 29,397,33629{,}397{,}336 linear relaxations, and exact integer certificates for all of them bound the normalized zero sum strictly below~11. Exceptional zeros and the remaining exterior range are treated separately. This case analysis is a kind of systematic brute force enabled by AI: an analysis of this size, with parameters tuned to each case, would be very laborious or nearly impossible to carry out by hand, and it lets the new estimate be applied separately in each case. The near-density lemma, certificate soundness, and a conditional passage from a certified case to a prime are formalized in Lean, assuming published analytic inputs and specified facts about zeros (PALOMAR-2026-10-01-000020 v1).

Primarily AI-generated textHuman understanding: some parts
Oct 02, 202680 pages
PDFView work

Advanced search

Text
Human understanding
Linked formalizations