betaFeedback

Abstract

Let u(U)u(U) count the unordered pairs at distance one in a finite planar set UU. We construct finite sets UjU_j with ∣Uj∣→∞|U_j|\to\infty and u(Uj)/∣Uj∣1.043171→∞u(U_j)/|U_j|^{1.043171}\to\infty. The largest exponent previously claimed, apart from preliminary versions of the present work, is 1.03581.0358, in the author's unpublished manuscript. The method is the number-field construction of OpenAI and Sawin. Here unit distances come from elements of relative norm one in quadratic extensions K/FK/F in which FF has both real and complex places, and the fields come from an infinite pro-22 tower over \Q(\sqrt{241}), ramified only above 22, 33 and 55, with Frobenius conditions at the two primes above 2929 and at one prime above 4141. The relative zeta value is bounded through the dihedral quotients of the tower group: the tower group has 2727 quotients isomorphic to the dihedral group of order~88; fifteen of them detect, by a census, the residue degree of the primes that split completely in the Kummer field up to norm 4⋅10134\cdot10^{13}, and seven of them generate a field EW′′E_{W''} of degree 81928192 contained in every field of the tower. We write ζEW′′\zeta_{E_{W''}} as a product of LL-functions of degrees 22, 44 and~88, evaluate the 576576 of degrees 44 and 88 rigorously at 2121 real points near~11, and combine the values through a kernel inequality with coefficients of both signs. Finite facts and numerical inequalities are certified by exact computation and interval arithmetic. Partial formalization PALOMAR-2026-10-01-000018 v2

  • AI Methodology. This work was done with extensive AI use, with agents developing the mathematics, writing the programs and writing the paper; I supervised, prompted, and coordinated them. The choice of the Frobenius condition at a prime above 41, rather than at the prime 7 (Section 2), was found in an agent run of the Muse Spark system. The refinements of the bound for the relative zeta value (§ 6–§ 10), the general shell weights, the certificates and this paper were produced with Claude Opus 5.5 working as an agent in Claude Code. Preliminary versions of this work, on which the construction builds, were produced with Claude Opus 5.5, ChatGPT Sol 5.6 and ChatGPT Astra 6. The reviews mentioned in Subsection 1.5 are automated reviews by AI agents in fresh contexts, not human peer review. I have had the agents provide a clearer introduction, and a dependency graph on page 4 to help navigate the paper and the proof, and make it easier to read. However I still recommend that the reader navigate this paper with their AI model of choice, and to ask the paper questions.

Introduction

For a finite set U⊂R2U \subset\mathbb{R}^2, let

u(U)=#{{x,y}⊂U:∣x−y∣=1},u(n)=max⁡∣U∣=nu(U).u(U)=\#\{\{x,y\}\subset U: |x-y|=1\},\qquad u(n)=\max_{|U|=n}u(U).

where the pairs are unordered and the distance is Euclidean. The unit distance problem of Erdős [5] asks for the order of growth of u(n)u(n). Our main result is the following.

Theorem 1.1. There are finite sets Uj⊂R2U_j \subset\mathbb{R}^2, with ∣Uj∣→∞|U_j| \to\infty, such that

u(Uj)∣Uj∣1.043171⟶∞.\frac{u(U_j)}{|U_j|^{1.043171}} \longrightarrow\infty.

In particular, u(n)≥n1.043171u(n) \ge n^{1.043171} for arbitrarily large integers nn.

The theorem concerns an unbounded sequence of cardinalities; it asserts nothing for every sufficiently large nn. Its proof uses no unproved hypothesis, such as the generalized Riemann hypothesis, but it is computer-assisted: finite facts about the tower, a census of primes, and many numerical inequalities, among them enclosures of the values of 576 LL-functions near s=1s = 1, are verified by exact computation and interval arithmetic (Subsection 1.8).

The largest exponent previously claimed, apart from preliminary versions of the present work, is 1.0358, in the author’s unpublished manuscript [22]. The method is the number-field construction of OpenAI [26] and Sawin [32]. Subsection 1.4 describes the ingredients that give the present exponent: fields of mixed signature from a tower over Q(241)\mathbb{Q}(\sqrt{241}), the local conditions of that tower, a bound for the relative zeta value through the dihedral quotients of the tower group, and general weights at the selected primes (the primes used for equal-norm choices, Subsection 1.2). The paper is self-contained, apart from the cited standard results and the supplementary programs.

Background

Erdős [5] showed that a suitable section of a scaled integer lattice gives

u(n)≥n1+c0/log⁡log⁡nu(n) \ge n^{1+c_0/\log\log n}

for some c0>0c_0 > 0 and arbitrarily large nn, and he conjectured that this is essentially optimal [5, 6]. The best upper bound,

u(n)=O(n4/3),u(n) = O(n^{4/3}),

is due to Spencer, Szemerédi and Trotter [35]; Székely [37] gave a proof by crossing numbers.

In 2026 a team at OpenAI [26] disproved Erdős’s conjecture. They constructed sets with

u(U)≥∣U∣1+δu(U) \ge|U|^{1+\delta}

for a fixed δ>0\delta> 0 and arbitrarily large ∣U∣|U|, replacing the integer lattice by ideal lattices in number fields of large degree and small root discriminant rd⁡(M)=∣ΔM∣1/[M:Q]\operatorname{rd}(M) = |\Delta_M|^{1/[M:\mathbb{Q}]} (ΔM\Delta_M the discriminant). Alon, Bloom, Gowers, Litt, Sawin, Shankar, Tsimerman, Wang and Wood [1] simplified the argument, and Sawin [32] made every step explicit and reached the exponent 1.014114. Improvements followed in online discussions [19, 36, 20], and the author’s manuscript [22] recorded 1.0358324. The present paper uses a tower over Q(241)\mathbb{Q}(\sqrt{241}) and quadratic extensions K/FK/F of mixed signature, that is, with FF having both real and complex places. Table 1 lists the explicit exponents, and the repository [27] records further bounds posted online.

ExponentSourceMain new inputKind of source
1+c0/log⁡log⁡n1 + c_0/\log\log nErdős [5]integer latticejournal
1+δ1 + \delta, δ>0\delta> 0OpenAI [26]ideal lattices from class field towersmanuscript
1+6⋅10−381 + 6 \cdot10^{-38}Alon et al. [1]simplified argumentpreprint
1.014114Sawin [32]explicit and sharpened criterionpreprint
1.015263Emmerich [3]optimized parameterspreprint
1.031185Emmerich, Cordella [4]extended range of primesonline certificate
1.031849mlewko [19]finite data found by ChatGPTforum post
1.0333487spiderduckpig [36]weighted Zassenhaus countonline post
1.0358324Naslund [22, 20]convolution counting, dyadic ramification, fixed-subfield Euler productsmanuscript, online post
1.043171this papertower over Q(241)\mathbb{Q}(\sqrt{241}), mixed signature, dihedral quotients (Subsection 1.4)

Table 1. Explicit lower bounds u(n)≥n1+δu(n) \ge n^{1+\delta} for arbitrarily large nn, in increasing order. The exponents are rounded down and constant factors are omitted. Apart from Erdős’s, none of these bounds had appeared in a refereed journal when this paper was written.

The construction in outline

The construction carries Erdős’s lattice argument to number fields of large degree. Let KK be such a field, of degree 2d2d, with an involution ι\iota whose fixed field is FF. We take the points of a translated ideal lattice of KK that lie in a bounded region of K⊗RK \otimes\mathbb{R}. A complex embedding in which ι\iota acts as complex conjugation maps them injectively to the plane. It sends every difference of relative norm one over FF to a unit vector, so two points whose difference has relative norm one are at distance one. The exponent is then set by a balance between four quantities, each normalized as d−1d^{-1} times the logarithm of a count. First, equal-norm choices: if a prime p\mathfrak{p} of FF splits in KK as P⋅ιP\mathfrak{P} \cdot\iota\mathfrak{P}, the k+1k+1 ideals Pj(ιP)k−j\mathfrak{P}^{j}(\iota\mathfrak{P})^{k-j}, 0≤j≤k0 \leq j \leq k, all have relative norm pk\mathfrak{p}^{k}; the primes used in this way are the selected primes (they lie above the selected primes of BB of Definition 4.16). Second, ideal classes: keeping the products of such ideals that lie in one class of Cl⁡(K)\operatorname{Cl}(K) modulo the image of Cl⁡(F)\operatorname{Cl}(F) loses at most a factor equal to the order of that quotient (Lemma 8.11). Through the class-number formula, this loss is governed by the root discriminant λ=rd⁡(K)\lambda=\operatorname{rd}(K) of KK (Definition 2.44), which equals that of FF because K/FK/F will be unramified at the finite places, and by the relative zeta value LF(1)L_F(1), where LF(s)=ζK(s)/ζF(s)L_F(s)=\zeta_K(s)/\zeta_F(s). Third, the region: its shape decides how many pairs of points with difference of relative norm one lie in it, and its volume fixes the number of points. Fourth, density: using a prime p\mathfrak{p} with exponent kk makes the ideal lattice denser by the factor Npk\mathrm{N}\mathfrak{p}^{k}, which costs δklog⁡Np\delta k\log\mathrm{N}\mathfrak{p} before division by dd.

Quantitatively, suppose that such fields exist with d→∞d \to\infty, a fixed root discriminant λ\lambda, and selected primes of fixed local types (ramification index and residue degree, Definition 2.45). Let (b,c)(b,c) be the signature of FF, so that d=b+2cd=b+2c, and put θ=c/d\theta=c/d. If CC is an upper bound for d−1log⁡LF(1)d^{-1}\log L_F(1), then the exponent 1+δ1+\delta is attained when the margin

1d∑plog⁡Fδ,p−(12−δ)log⁡λ−C+(1−δ)log⁡2+(1−θ)log⁡π+(1−2θ)JR+θJC(1)\frac{1}{d}\sum_{\mathfrak{p}}\log\mathcal{F}_{\delta,\mathfrak{p}}-\left(\frac{1}{2}-\delta\right)\log\lambda-C+(1-\delta)\log2+(1-\theta)\log\pi+(1-2\theta)J_{\mathbb{R}}+\theta J_{\mathbb{C}} \tag*{(1)}

is bounded below by a positive constant. Here the sum runs over the selected primes of FF, and Fδ,p\mathcal{F}_{\delta,\mathfrak{p}} is the value of the local functional of Definition 9.3 at the weight function placed at p\mathfrak{p}, a shell profile (Definition 9.7); for an unweighted profile used with exponent kk at a prime of norm QQ it is (k+1)Q−δk(k+1)Q^{-\delta k}. The numbers JRJ_{\mathbb{R}} and JCJ_{\mathbb{C}} are functionals of the profiles at the real and complex places of FF (Definition 8.28). This is Proposition 9.26, the geometric transfer with shell profiles; § 17 proves that the margin is positive at δ=0.043171\delta=0.043171.

The fields come from a tower: a pro-22 extension of B=Q(241)B=\mathbb{Q}(\sqrt{241}) with restricted ramification and prescribed local Galois groups. Its Galois group is the tower group GBG_B (Definition 2.23), a pro-22 group; we write (DnGB)n≥1(D_nG_B)_{n\geq1} for its Zassenhaus filtration and gr⁡nGB=DnGB/Dn+1GB\operatorname{gr}_nG_B=D_nG_B/D_{n+1}G_B for its graded pieces (recalled at the start of Section 2). The tower is infinite because its Golod–Shafarevich function PBP_B of (10) takes a negative value [9, 10]. Its fields have λ=29/43615≈286.0\lambda=2^{9/4}\sqrt{3615}\approx286.0, the field FF is the fixed field of one complex conjugation ν1\nu_1, whose conjugacy class in GB/D4GBG_B/D_4G_B has 2152^{15} elements (Lemma 2.30(c)), which forces b/d≤2−16b/d\leq2^{-16} and θ≥θ∗=65535/131072\theta\geq\theta_*=65535/131072 (Theorem 2.46).

The structure of the proof

Figure 1 shows how the main statements depend on each other. The proof has four parts: Parts (II)–(IV) apply to the fields of Part (I); Parts (II) and (III) are independent of each other and meet in the final inequality of Part (IV).

The structure of the proof

Figure 1. The structure of the proof (Subsection 1.3). Each box is a numbered statement, coloured by the part of the proof it belongs to. An arrow goes from a statement to each statement whose proof uses it, directly or through statements not shown, except that arrows implied by a chain of drawn arrows are left out; dashed arrows join different parts. The arrows from Theorem 2.46, which supplies the fields to which Parts (II)–(IV) apply, are also left out. Appendix § 20 lists the citations between all numbered statements; the notation is collected in Subsection 1.7.

  1. (I) The fields (Section 2). A presentation of the Galois group of the maximal pro-22 extension of BB unramified outside 2,3,52,3,5, the local conditions, and the Golod–Shafarevich criterion show that the tower is infinite (Proposition 2.43); the graded pieces of its Zassenhaus filtration (Lemmas 2.29 and 2.30) give the root discriminant, the local types and the signature of its fields (Theorem 2.46).

  1. (II) The relative zeta value (§ 5–§ 10). A limit lemma (Lemma 4.13) and the signed kernel inequality (Proposition 4.19) reduce d−1log⁡LF(1)d^{-1}\log L_F(1) to values of a zeta function at real points σj>1\sigma_j>1 and to floors: lower bounds for the residue degrees of the free primes of BB, those other than the selected ones (Definitions 4.16 and 4.18). The dihedral quotients of GBG_B supply

the floors, through a census of primes (Proposition 5.12), and a field EW′′E_{W^{\prime\prime}} of degree 8192 (Definition 5.7), whose zeta function factors into the LL-functions of Lemma 5.13. Their values (Proposition 6.5) and the kernel (Lemmas 7.2 and 7.3) give the bound Ceff=0.0422764C_{\mathrm{eff}} = 0.0422764 for d−1log⁡LF(1)d^{-1}\log L_F(1) (Theorem 4.2).

(III) The planar sets (§ 11–§ 15). A mass formula for ideal lattices (Proposition 8.5), the class selection (Lemma 8.11) and local and archimedean profiles give the transfer Proposition 9.26: for fields as in (I), the exponent 1+δ1 + \delta is attained when the margin (1.2) is positive.

(IV) The certificate (§ 17). Exact checks and interval enclosures of the local functionals, the profile integrals and the margin at δ=0.043171\delta= 0.043171, with C=CeffC = C_{\mathrm{eff}} from (II) (Theorem 4.2), complete the proof of Theorem 1.1.

The dependency index (Appendix § 20) lists, for each numbered statement, the numbered statements that it or its proof cites and the statements that cite it.

The main ingredients

The base field and the signature. In the construction of Sawin [32], and in [22], the field KK is a CM field. FF is totally real, so θ=0\theta= 0, and the norm-one units are finite. Sawin’s fields come from an unramified pro-2 tower over a real quadratic field ramified at 2 and at the thirteen odd primes up to 43, and their root discriminant is about 1.6⋅1081.6 \cdot10^8. Here the base field is B=Q(241)B = \mathbb{Q}(\sqrt{241}), the real quadratic field of smallest discriminant in which 2, 3 and 5 all split. The tower is ramified only at the six primes of BB above 2, 3 and 5, with prescribed local Galois groups there, and is still infinite (Proposition 2.43); its fields have root discriminant λ≈286\lambda\approx286. The field KK is totally imaginary and Galois over BB, and FF is the fixed field of one complex conjugation, which is not central, so that almost every place of FF is complex and θ≥θ∗\theta\ge\theta_*. Above a complex place of FF the two coordinates of a norm-one element can have different moduli, and the norm-one units, of rank cc, move them along a lattice; averaging over a fundamental domain of that lattice counts the resulting pairs exactly, and the class-number formula cancels the regulator of these units (Proposition 8.5). By (1), mixed signature adds θ(JC−2JR−log⁡π)\theta(J_{\mathbb{C}} - 2J_{\mathbb{R}} - \log\pi) to the margin: with the profiles of § 17, the lower bound for the margin proved there is positive for θ≥θ∗\theta\ge\theta_*, but would be below −0.312-0.312 at θ=0\theta= 0 (Remark 11.13).

References

  1. [1]Noga Alon, Thomas F. Bloom, W. T. Gowers, Daniel Litt, Will Sawin, Arul Shankar, Jacob Tsimerman, Victor Wang, and Melanie Matchett Wood, Remarks on the disproof of the unit distance conjecture, 2026, arXiv:2605.20695v1; https://arxiv.org/abs/2605.20695v1.
  2. [2]Tim Dokchitser, Computing special values of motivic L-functions, Experimental Mathematics 13 (2004), no. 2, 137–149.
  3. [3]Michael T. M. Emmerich, Optimizing explicit unit-distance lower-bound certificates, 2026, arXiv:2606.03419v5, June 9, 2026 (v1 June 2, 2026); certificate with δ = 0.015263 . . .; https://arxiv.org/abs/2606.03419v5.
  4. [4]Michael T. M. Emmerich and Francesco Cordella, Optimized certificate for the unit distance problem with extended prime number range, Zenodo record, June 5, 2026. Certificate and verification code. https://doi.org/10.5281/zenodo.20551478.
  5. [5]Paul Erdős, On sets of distances of n points, The American Mathematical Monthly 53 (1946), no. 5, 248–250, https://doi.org/10.1080/00029890.1946.11991674.
  6. [6]———, Some problems in number theory, combinatorics and combinatorial geometry, Mathematica Pannonica 5 (1994), no. 2, 261–269.
  7. [7]Mikhail Ershov, Golod–Shafarevich groups: a survey, International Journal of Algebra and Computation 22 (2012), no. 5, 1230001, 68 pages.DOI
  8. [8]Ralph H. Fox, Free differential calculus. I: Derivation in the free group ring, Annals of Mathematics 57 (1953), no. 3, 547–560.DOI
  9. [9]E. S. Golod and I. R. Shafarevich, On the class field tower, Izv. Akad. Nauk SSSR Ser. Mat. 28 (1964), no. 2, 261–272 (Russian), English translation: Amer. Math. Soc. Transl. (2) 48 (1965), 91–102; https://www.mathnet.ru/eng/im2955.
  10. [10]Farshid Hajir, Christian Maire, and Ravi Ramakrishna, Cutting towers of number fields, Annales mathématiques du Québec 45 (2021), no. 2, 321–345.arxiv.org/abs/1901.04354
  11. [11]Erich Hecke, Eine neue Art von Zetafunktionen und ihre Beziehungen zur Verteilung der Primzahlen. Zweite Mitteilung, Mathematische Zeitschrift 6 (1920), 11–51, https://doi.org/10.1007/BF01202991.
  12. [12]Nicholas J. Higham, Accuracy and stability of numerical algorithms, second ed., SIAM, Philadelphia, 2002.
  13. [13]S. A. Jennings, The structure of the group ring of a p-group over a modular field, Transactions of the American Mathematical Society 50 (1941), no. 1, 175–185, https://doi.org/10.2307/1989916.
  14. [14]Fredrik Johansson, Rigorous high-precision computation of the Hurwitz zeta function and its derivatives, Numerical Algorithms 69 (2015), no. 2, 253–270.arxiv.org/abs/1309.2877
  15. [15]———, Arb: Efficient arbitrary-precision midpoint-radius interval arithmetic, IEEE Transactions on Computers 66 (2017), no. 8, 1281–1292, https://doi.org/10.1109/TC.2017.2690633.
  16. [16]James S. Milne, Lectures on étale cohomology, 2013, Version 2.21, March 22, 2013; https://www.jmilne.org/math/CourseNotes/LEC.pdf.
  17. [17]———, Algebraic number theory, 2020, Version 3.08, July 19, 2020; https://www.jmilne.org/math/CourseNotes/ANT.pdf.
  18. [18]———, Class field theory, 2020, Version 4.03, August 6, 2020; https://www.jmilne.org/math/CourseNotes/CFT.pdf.
  19. [19]mlewko, Post in the discussion thread of Erdős Problem #90, Erdős Problems forum, May 21, 2026, Reports δ = 0.031849981 . . . in Sawin’s criterion, with finite data found by ChatGPT; accessed September 30, 2026; https://www.erdosproblems.com/forum/thread/90.
  20. [20]Eric Naslund, Answer to “What is the unit distance exponent?”, MathOverflow, answer 511576, 2026, Posted May 23, 2026, last edited June 9, 2026; accessed September 30, 2026; https://mathoverflow.net/a/511576.
  21. [21]———, An exponent of 1.043 for the unit distance problem: Lean formalization at exponent 1.04315, conditional on one zeta inequality, Palomar registry, PALOMAR-2026-10-01-000018, version 2, 2026, https://palomar-registry.org/entry?pid=PALOMAR-2026-10-01-000018&version=2; Lean 4 source at https://github.com/enaslund/unit-distance-bound-0.0427/tree/edd88c0c175ab4d18cf6115e2f6ddf3865d36b12/lean.
  22. [22]———, An improved explicit lower bound for the unit distance problem, Unpublished manuscript, 2026, Records exponent 1.0358324. https://github.com/enaslund/unit-distance-bound-0.0427/tree/8ab94da4ba4faf6f6d8fe43c298f04c3b284e796/papers/0.0358324.
  23. [23]______, Supplementary programs and data for “An exponent of 1.043 for the unit distance problem”, GitHub repository enaslund/unit-distance-bound-0.0427, directories papers/0.043171, papers/0.042901, papers/0.04273/certificates, papers/0.0418235/certificates and research, 2026, https://github.com/enaslund/unit-distance-bound-0.0427.
  24. [24]National Institute of Standards and Technology, NIST Digital Library of Mathematical Functions, 2026, Release 1.2.8, September 15, 2026; accessed September 27, 2026; https://dlmf.nist.gov/.
  25. [25]Jürgen Neukirch, Alexander Schmidt, and Kay Wingberg, Cohomology of number fields, 2 ed., Grundlehren der mathematischen Wissenschaften, vol. 323, Springer, Berlin, Heidelberg, 2008.DOI
  26. [26]OpenAI, Planar point sets with many unit distances, Research manuscript, May 20, 2026, https://cdn.openai.com/pdf/74c24085-19b0-4534-9c90-465b8e29ad73/unit-distance-proof.pdf.
  27. [27]Erdős unit distance exponent, Entry 84a of Optimization Constants in Mathematics, a crowdsourced repository started by Terence Tao, 2026, Accessed September 30, 2026; https://teorth.github.io/optimizationproblems/constants/84a.html.
  28. [28]Bjorn Poonen, Rational points on varieties, Graduate Studies in Mathematics, vol. 186, American Mathematical Society, Providence, RI, 2017, https://math.mit.edu/~poonen/papers/Qpoints.pdf.DOI
  29. [29]Claudio Quadrelli, Pro-p groups with few relations and universal Koszulity, Mathematica Scandinavica 127 (2021), no. 1, 28–42, https://doi.org/10.7146/math.scand.a-123644.
  30. [30]D. G. Quillen, On the associated graded ring of a group ring, Journal of Algebra 10 (1968), no. 4, 411–418, https://doi.org/10.1016/0021-8693(68)90069-0.
  31. [31]J. Barkley Rosser and Lowell Schoenfeld, Approximate formulas for some functions of prime numbers, Illinois Journal of Mathematics 6 (1962), 64–94.
  32. [32]Will Sawin, An explicit lower bound for the unit distance problem, 2026, arXiv:2605.20579v1; https://arxiv.org/abs/2605.20579v1.
  33. [33]Jean-Pierre Serre, A course in arithmetic, Graduate Texts in Mathematics, vol. 7, Springer, New York, 1973.DOI
  34. [34]______, Linear representations of finite groups, Graduate Texts in Mathematics, vol. 42, Springer-Verlag, New York, 1977.DOI
  35. [35]Joel Spencer, Endre Szemerédi, and William T. Trotter, Jr., Unit distances in the Euclidean plane, Graph Theory and Combinatorics (Béla Bollobás, ed.), Academic Press, London, 1984, https://trotter.math.gatech.edu/papers/44.pdf, pp. 293–303.
  36. [36]spiderduckpig, Answer to “What is the unit distance exponent?”, MathOverflow, answer 511531, 2026, Posted May 21, 2026, with 1 + δ ≥ 1.03188306553964…; comment of May 22, 2026 proposing a weighted count in the Zassenhaus filtration in Sawin’s Golod–Shafarevich step, with δ ≥ 0.0333487156372102; last edited June 9, 2026; accessed September 30, 2026; https://mathoverflow.net/a/511531.
  37. [37]László A. Székely, Crossing numbers and hard Erdős problems in discrete geometry, Combinatorics, Probability and Computing 6 (1997), no. 3, 353–358, https://doi.org/10.1017/S0963548397002976.
  38. [38]John T. Tate, Fourier analysis in number fields and Hecke’s zeta-functions, Algebraic Number Theory (J. W. S. Cassels and A. Fröhlich, eds.), Academic Press, London, 1967, Originally a Princeton University doctoral thesis, 1950; https://www.lms.ac.uk/publications/algebraic-number-theory, pp. 305–347.
  39. [39]The FLINT team, FLINT: Fast Library for Number Theory, 2026, Version 3.6.0, used through python-flint 0.9.0; https://flintlib.org.
  40. [40]The mpmath development team, mpmath: A Python library for arbitrary-precision floating-point arithmetic, 2023, Version 1.3.0; https://github.com/mpmath/mpmath/releases/tag/1.3.0.
  41. [41]The PARI Group, PARI/GP, Université de Bordeaux, 2025, Version 2.17.2; https://pari.math.u-bordeaux.fr/.
  42. [42]M. A. Tsfasman and S. G. Vlăduț, Infinite global fields and the generalized Brauer–Siegel theorem, Moscow Mathematical Journal 2 (2002), no. 2, 329–402, https://doi.org/10.17323/1609-4514-2002-2-2-329-402.

Paper details

vlnc:W_64dxrxcs
Version
v2 · latest
Published
Added
License
CC BY 4.0
Usage

views · 21

downloads · 10

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

OpenAI ChatGPT Astra 6, OpenAI ChatGPT Sol 5.6, Anthropic Claude Opus 5.5, Meta Muse Spark 1.3 Contributor

View statement
Looking for related papers…
Contents