The first counterexample to the Mertens conjecture is below exp(3.485·10¹⁸)
Description
Let M(x) be the Mertens function, the sum of the Möbius function μ(n) over n ≤ x. Odlyzko and te Riele (1985) disproved the Mertens conjecture, |M(x)| < √x for all x > 1, without exhibiting a counterexample; |M(n)| < √n has been verified for 1 < n ≤ 10¹⁶. Let x₁ be the least integer n ≥ 1 with |M(n)| > √n.
In this preprint I prove, unconditionally, that x₁ < exp(3.485·10¹⁸); in fact x₁ < 10^(1.51351·10¹⁸). The previous best bound I am aware of is exp(1.96·10¹⁹), by Kim and Nguyen (2025), stated for the least real x > 1 with |M(x)| ≥ √x (my bound also holds for that quantity); the exponent is smaller by a factor of about 5.6. This is an upper bound: it does not locate the first counterexample.
If, in addition, an explicit hypothesis on the zeros of ζ holds (H5 in the paper: it follows from the Riemann hypothesis and can fail only for a zero with |γ| > 3·10¹²), then there is a counterexample x with ln x in an explicit window of width about 1.78·10⁴, and it satisfies M(x) > +√x. I do not claim that this window contains x₁.
The proof is computer-assisted and introduces no new theory: it applies the criterion of Pintz and of Saouter–te Riele with a Gaussian kernel, as Kim and Nguyen do, with other parameters. It uses a sum over the 16 622 627 zeros of ζ with 0 < γ ≤ 7 999 035.172, evaluated in ball arithmetic (Arb), and Table 1 of Saouter–te Riele recomputed in ball arithmetic.
Verification: the deposit contains the data and programs. The zeros and the sum can be recomputed rigorously with public software (Arb, via python-flint), and the inequality of Table 1 can be checked independently from Platt's public list of zeros (rigorous given the accuracy and completeness of that list). The analytic part of the proof can be checked by hand, following Appendix A. The file SHA256SUMS lists the SHA-256 checksum of every file and carries an OpenTimestamps timestamp (file SHA256SUMS.ots). The text of the paper is licensed under CC BY 4.0 and the programs under the MIT License.
Status: preprint, not yet peer reviewed. The English version is the main text; a Spanish translation is included (the English version prevails).
Other (Spanish)
Sea M(x) la función de Mertens, la suma de la función de Möbius μ(n) sobre n ≤ x. Odlyzko y te Riele (1985) refutaron la conjetura de Mertens, |M(x)| < √x para todo x > 1, sin exhibir un contraejemplo; |M(n)| < √n está verificada para 1 < n ≤ 10¹⁶. Sea x₁ el menor entero n ≥ 1 con |M(n)| > √n.
En este preprint demuestro, de manera incondicional, que x₁ < exp(3.485·10¹⁸); de hecho, x₁ < 10^(1.51351·10¹⁸). La mejor cota previa que conozco es exp(1.96·10¹⁹), de Kim y Nguyen (2025), enunciada para el menor real x > 1 con |M(x)| ≥ √x (mi cota también vale para esa cantidad); el exponente es unas 5.6 veces menor. Es una cota superior: no ubica el primer contraejemplo.
Si además se cumple una hipótesis explícita sobre los ceros de ζ (H5 en el paper: se deduce de la hipótesis de Riemann y sólo puede fallar para un cero con |γ| > 3·10¹²), existe un contraejemplo x con ln x en una ventana explícita de ancho de alrededor de 1.78·10⁴, y cumple M(x) > +√x. No afirmo que esa ventana contenga a x₁.
La demostración es asistida por computadora y no introduce teoría nueva: aplica el criterio de Pintz y de Saouter–te Riele con núcleo gaussiano, como Kim y Nguyen, con otros parámetros. Usa una suma sobre los 16 622 627 ceros de ζ con 0 < γ ≤ 7 999 035.172, evaluada en aritmética de bolas (Arb), y la Tabla 1 de Saouter–te Riele recalculada en aritmética de bolas.
Verificación: el depósito contiene los datos y los programas. Los ceros y la suma pueden recalcularse de forma rigurosa con software público (Arb, mediante python-flint), y la desigualdad de la Tabla 1 puede comprobarse de forma independiente con la lista pública de ceros de Platt (de manera rigurosa, dada la exactitud y completitud de esa lista). La parte analítica de la demostración puede comprobarse a mano, siguiendo el Apéndice A. El archivo SHA256SUMS contiene la suma SHA-256 de cada archivo y lleva un sello de tiempo de OpenTimestamps (archivo SHA256SUMS.ots). El texto del paper se publica con licencia CC BY 4.0 y los programas con licencia MIT.
Estado: preprint, todavía sin revisión por pares. El texto principal es la versión en inglés; se incluye una traducción al español (prevalece la versión en inglés).
Files
Gallego_2026_Mertens_bound_EN.pdf
Additional details
Additional titles
- Translated title (Spanish)
- El primer contraejemplo de la conjetura de Mertens es menor que exp(3.485·10¹⁸)
References
- The FLINT team, FLINT: Fast Library for Number Theory, https://flintlib.org (python-flint: https://github.com/flintlib/python-flint).
- G. Hurst, Computations of the Mertens function and improved bounds on the Mertens conjecture, Math. Comp. 87 (2018), 1013–1028. doi:10.1090/mcom/3275. arXiv:1610.08551.
- F. Johansson, Arb: efficient arbitrary-precision midpoint-radius interval arithmetic, IEEE Trans. Comput. 66 (2017), 1281–1292. doi:10.1109/TC.2017.2690633.
- D. R. Johnston, T. Trudgian, A round of Pintz to celebrate oscillations in sums, arXiv:2511.09978 (v4, 22 July 2026).
- S. Kim, P. Q. Nguyen, On counterexamples to the Mertens conjecture, Res. Number Theory 11 (2025), no. 1, article 3. doi:10.1007/s40993-024-00603-9. arXiv:2502.21021.
- T. Kotnik, H. J. J. te Riele, The Mertens conjecture revisited, in: Algorithmic Number Theory (ANTS VII), Lecture Notes in Comput. Sci. 4076, Springer, 2006, 156–167. doi:10.1007/11792086_12.
- T. Kotnik, J. van de Lune, On the order of the Mertens function, Experiment. Math. 13 (2004), no. 4, 473–481. doi:10.1080/10586458.2004.10504556.
- The LMFDB Collaboration, The L-functions and modular forms database, https://www.lmfdb.org (zeros of ζ(s): https://www.lmfdb.org/knowledge/show/rcs.source.zeros.zeta).
- F. Mertens, Ueber eine zahlentheoretische Function, Sitzungsber. Akad. Wiss. Wien, Math.-Naturwiss. Kl., Abt. IIa 106 (1897), 761–830.
- The mpmath development team, mpmath: a Python library for arbitrary-precision floating-point arithmetic, https://mpmath.org.
- A. M. Odlyzko, H. J. J. te Riele, Disproof of the Mertens conjecture, J. reine angew. Math. 357 (1985), 138–160. doi:10.1515/crll.1985.357.138.
- OpenTimestamps, https://opentimestamps.org.
- The PARI Group, PARI/GP, Univ. Bordeaux, https://pari.math.u-bordeaux.fr.
- J. Pintz, An effective disproof of the Mertens conjecture, Astérisque 147–148 (1987), 325–333.
- D. J. Platt, Computing π(x) analytically, Math. Comp. 84 (2015), 1521–1535. doi:10.1090/S0025-5718-2014-02884-6.
- D. J. Platt, Isolating some non-trivial zeros of zeta, Math. Comp. 86 (2017), 2449–2467. doi:10.1090/mcom/3198.
- D. J. Platt, T. Trudgian, The Riemann hypothesis is true up to 3·10¹², Bull. London Math. Soc. 53 (2021), 792–797. doi:10.1112/blms.12460. arXiv:2004.09765.
- J. Rozmarynowycz, S. Kim, A new upper bound on the smallest counterexample to the Mertens conjecture, Exp. Math. 35 (2026), no. 2, 397–400 (published online 18 March 2025). doi:10.1080/10586458.2025.2477565. arXiv:2305.00345.
- The Sage Developers, SageMath, the Sage Mathematics Software System, https://www.sagemath.org.
- Y. Saouter, H. J. J. te Riele, Improved results on the Mertens conjecture, Math. Comp. 83 (2014), 421–433. doi:10.1090/S0025-5718-2013-02716-0.
- T. Trudgian, Improvements to Turing's method, Math. Comp. 80 (2011), 2259–2279. doi:10.1090/S0025-5718-2011-02470-1.