Primes of the form n² + 1
Are there infinitely many prime numbers that are one more than a perfect square?
Summary
Why it is open
In 1912 Edmund Landau listed four problems about primes that he considered "unattackable at the present state of science" [1]. This is one of them, numbered first in János Pintz's survey and fourth in the Wikipedia article on Landau's problems; the others are Goldbach's conjecture, the twin prime conjecture and Legendre's conjecture [1, 2].
It is the simplest open case of a broader expectation. Conjectures of Bunyakovsky, of Schinzel and, with a predicted frequency, of Paul Bateman and Roger Horn say that an irreducible integer polynomial with positive leading coefficient and no fixed prime divisor takes infinitely many prime values [3, 4]. Iwaniec wrote in 1978 that the only polynomials known to take infinitely many prime values have degree one, by Dirichlet's theorem on primes in arithmetic progressions [3], and in 2009 Pintz wrote that "there is no single non-linear polynomial for which we know the answer" [1].
The main obstacle is the parity problem, a heuristic principle formulated by Selberg in the 1950s: sieve methods, which count the numbers left after removing multiples of primes, cannot distinguish integers with an even number of prime factors from those with an odd number [1]. Pintz calls the resulting gap in Chen's theorem on twin primes, where p + 2 is shown to be a prime or a product of two primes, "the most significant particular case" of the problem [1]. The best sieve result for n² + 1 stops at the same point [1, 3].
A resolution would be a proof that primes n² + 1 never run out, or that they stop. This page's status follows the Wikipedia article on Landau's problems, which says that as of 2026 all four problems are unresolved [2].
What is known
Almost primes (1978–2012). Earlier sieve results showed that quadratic polynomials take values with at most three prime factors infinitely often [1]. In 1978 Henryk Iwaniec proved that if G(n) = an² + bn + c is irreducible, with a > 0 and c odd, then G(n) has at most two prime factors, counted with multiplicity, for infinitely many n; for large x more than (1/77)·Γ·x/log x of the n ≤ x qualify, where Γ is the constant in the Bateman–Horn prediction [3]. In particular n² + 1 is a product of at most two primes infinitely often [1]. In 2012 Robert Lemke Oliver extended the result to every irreducible quadratic with positive leading coefficient whose constant Γ is not zero [4].
Large prime factors (1967–2026). A second approximation asks how large the greatest prime factor of n² + 1 can be; Landau's question asks for it to equal n² + 1 infinitely often [5]. Christopher Hooley proved in 1967 that it exceeds n^1.1001 for infinitely many n; J.-M. Deshouillers and Iwaniec raised the exponent to 1.2024 in 1982, Régis de la Bretèche and Sary Drappeau to 1.2182 in 2020, and Jori Merikoski to 1.279 in a paper published in 2023 [6, 7]. Merikoski also reached 1.312 assuming Selberg's eigenvalue conjecture [6]. In 2026 Alexandru Pascadi published the exponent 1.3, unconditionally, using new estimates (large sieve inequalities for exceptional Maass forms) [7]. A 2025 preprint by Lasse Grimmelt and Merikoski, with no journal reference on 15 October 2026, claims 1.312 without the eigenvalue conjecture [8].
Thin sequences (1997–2026). John Friedlander and Iwaniec call a sequence thin if it has fewer than N^θ members up to N for some θ < 1, as the values of n² + 1 do, with θ = 1/2 [9]. In 1997 they wrote that for thin sequences "there is no polynomial for which this has hitherto been done", meaning a proof of infinitely many prime values, and announced an asymptotic count of primes of the form x² + y⁴, a sequence with θ = 3/4, using a sieve that breaks the parity problem [9]. Roger Heath-Brown obtained the same for x³ + 2y³ [4]. In a 2024 preprint, accepted by Acta Mathematica, Ben Green and Mehtaab Sawhney proved that p² + nq² is prime for infinitely many primes p and q when n is 0 or 4 modulo 6 [10]. They note that, for parity reasons, the same cannot be expected for x² + y² with x and y prime short of proving that infinitely many primes have the form x² + 4 with x prime [10].
Conjecture and computation. G. H. Hardy and J. E. Littlewood conjectured in 1923 that the number of primes n² + 1 up to x is asymptotic to C√x/log x, with C = 1.3728… [1, 5]. In 2025 Jon Grantham and Hester Graves listed every such prime below 6.25 × 10²⁸. There are 5,342,656,862,803, and the ratio of this count to the refined prediction (C/2)·li(√x), where li is the logarithmic integral, is 0.99999976 [5].
Main approaches
Weighted sieves. Sieve weights bound the number of prime factors of n² + 1 directly [3]. Strongest evidence: Iwaniec's theorem and its extension to all suitable quadratics [3, 4]. Main difficulty: the parity problem, which leaves sieve arguments unable to tell one prime factor from two [1].
Large prime factors. Chebyshev's method compares the total size of the numbers n² + 1 with the contributions of the primes that divide them, which requires counting the solutions of n² + 1 ≡ 0 modulo primes [6]. Strongest evidence: the published exponent 1.3, or 1.312 in the 2025 preprint [7, 8]. Main difficulty: the exponent must reach 2 [5]. Each gain since 1967 has drawn on estimates for certain exponential sums (Kloosterman sums) or towards Selberg's eigenvalue conjecture [7], and Merikoski writes that "the exact limit of the method is hard to determine" [6].
Parity-breaking sieves in two variables. Combine a sieve with the algebraic structure of a polynomial in two variables [9]. Strongest evidence: asymptotic counts of primes x² + y⁴ and x³ + 2y³, and the Green–Sawhney theorem [4, 9, 10]. Main difficulty: the successes are for polynomials in two variables, as Lemke Oliver notes [4]; n² + 1 has one variable and is thinner than x² + y⁴ [9].
Primes just above squares. Landau's conjecture would follow from showing that the fractional part of √p is below c/√p, for a suitable constant c, for infinitely many primes p [1]. Strongest evidence: Balog and Harman each proved in 1983 that it is below p^(−1/4+ε) infinitely often [1]. Main difficulty: the exponent needed is −1/2, and the proved exponents have moved from about −1/15 in 1940 to −1/4 [1].
Open sub-questions
- Iwaniec's constant. How far can the factor 1/77 in Iwaniec's lower bound for the number of n ≤ x with n² + 1 a product of at most two primes be raised, and can the bound be made fully explicit [3]?
- The limit of the exponent. What exponent does Harman's sieve give with the arithmetic information of Pascadi and of Grimmelt and Merikoski when the sieve is optimised numerically, as Merikoski suggests [6, 7, 8]?
- Uniform shifts. Can the hypothesis on real characters that Grimmelt and Merikoski need for uniformity in the shift h be removed, or is it, as they suggest, a fundamental obstruction [8]?
- Beyond 6.25 × 10²⁸. Does the count of primes n² + 1 stay within one part in a million of (C/2)·li(√x) at larger x, and how often does the difference change sign [5]?
- Thinner two-variable sequences. Can a parity-breaking sieve count primes in a two-variable polynomial sequence thinner than x² + y⁴ and x³ + 2y³ [4, 9]?
- Fractional parts of √p. Can the exponent −1/4 of Balog and Harman for the fractional part of √p be improved for primes p [1]?
Work on this in Papex
If you have a testable answer to one of the sub-questions above, you can draft it as a private hypothesis in Papex, or prepare the draft from Claude or ChatGPT with the Papex connector. You review the exact version, sign it and decide whether to publish it; a public record that names this problem is listed on this page. A Papex record shows who signed which version and when. It is not peer review and does not mean the hypothesis is correct.
Key references
- Pintz, J. (2009). Landau’s problems on primes. Journal de Théorie des Nombres de Bordeaux 21(2), 357–404. 10.5802/jtnb.676 · Checked 15 Oct 2026
- Wikipedia. Landau’s problems (revision of 3 June 2026). Source · Checked 15 Oct 2026
- Iwaniec, H. (1978). Almost-primes represented by quadratic polynomials. Inventiones Mathematicae 47(2), 171–188. 10.1007/BF01578070 · Checked 15 Oct 2026
- Lemke Oliver, R. J. (2012). Almost-primes represented by quadratic polynomials. Acta Arithmetica 151(3), 241–261. 10.4064/aa151-3-2 · Checked 15 Oct 2026
- Grantham, J. & Graves, H. (2025). Primes of the form m² + 1 and Goldbach’s ‘other other’ conjecture. Computational Methods in Science and Technology 31(1–3), 49–54. Source · Checked 15 Oct 2026
- Merikoski, J. (2023). On the largest prime factor of n² + 1. Journal of the European Mathematical Society 25(4), 1253–1284 (published online 2022). 10.4171/jems/1216 · Checked 15 Oct 2026
- Pascadi, A. (2026). Large sieve inequalities for exceptional Maass forms and the greatest prime factor of n² + 1. Forum of Mathematics, Pi 14, e8. 10.1017/fmp.2026.10025 · Checked 15 Oct 2026
- Grimmelt, L. & Merikoski, J. (2025). On the greatest prime factor and uniform equidistribution of quadratic polynomials. Preprint, arXiv:2505.00493 (version 2, 30 May 2025), not peer reviewed. Source · Checked 15 Oct 2026
- Friedlander, J. & Iwaniec, H. (1997). Using a parity-sensitive sieve to count prime values of a polynomial. Proceedings of the National Academy of Sciences 94(4), 1054–1058. 10.1073/pnas.94.4.1054 · Checked 15 Oct 2026
- Green, B. & Sawhney, M. (2024). Primes of the form p² + nq². Preprint, arXiv:2410.04189 (version 3, 6 June 2026), accepted for publication in Acta Mathematica according to the arXiv comment. Source · Checked 15 Oct 2026
Source
Open, as listed by Wikipedia · Landau’s problems. Checked 08 Oct 2026.
The status is the source’s. Papex does not decide whether a problem is solved.
Records on this problem
No public record addresses this problem yet.