MathLabs

Open problem, Combinatorics and discrete mathematics, Geometry, posed 1917

No-three-in-line problem

Open

For each integer n≥2n \ge 2, determine the maximum number f(n)f(n) of lattice points that can be selected from the n×nn \times n grid {1,2,…,n}2\{1, 2, \dots, n\}^2 so that no three selected points are collinear on any line of any slope, and in particular determine whether f(n)=2nf(n) = 2n for all nn or whether lim⁡n→∞f(n)/n<2\lim_{n \to \infty} f(n)/n < 2.

Research frontier as of 2026

As of 2026, the asymptotic ratio lim⁡n→∞f(n)/n\lim_{n \to \infty} f(n)/n remains unknown, trapped between the 1975 Hall–Jackson–Sudbery–Wild algebraic lower bound 32=1.5\frac{3}{2} = 1.5 and the trivial pigeonhole upper bound 22. Computer searches (by Flammenkamp, Prellberg, Heule, and others) have found configurations of 2n2n points with no three in line for every n≤74n \le 74, yet the number of symmetric 2n2n-point solutions decreases for larger nn, lending empirical support to the Guy–Kelly heuristic f(n)∼π3n≈1.8138nf(n) \sim \frac{\pi}{\sqrt{3}} n \approx 1.8138n.

Best known results

  • For every n≥2n \ge 2, (32−o(1))n≤f(n)≤2n(\frac{3}{2} - o(1))n \le f(n) \le 2n (Hall, Jackson, Sudbery, and Wild, 1975).
  • Exact equality f(n)=2nf(n) = 2n holds for all 2≤n≤742 \le n \le 74 by explicit computer-constructed configurations.

Tools and where they stop

ToolAchievedWhere it stops
Algebraic curves over Fp\mathbb{F}_p (Erdős, Hall–Jackson–Sudbery–Wild)Uses Bézout's theorem over Fp\mathbb{F}_p—that a line intersects an irreducible conic such as y≡x2(modp)y \equiv x^2 \pmod p or xy≡k(modp)xy \equiv k \pmod p in at most 22 points modulo pp, hence at most 22 points in Z2\mathbb{Z}^2—to build (3/2−o(1))n(3/2 - o(1))n points.Curves of degree d≥3d \ge 3 over Fp\mathbb{F}_p can intersect a line in 33 or more points, while tiling multiple shifted conics creates collinear triples across different tiles once density exceeds 3n/23n/2.

Open questions

  • Does there exist an integer n≥2n \ge 2 for which f(n)<2nf(n) < 2n?
  • Does the limit lim⁡n→∞f(n)/n\lim_{n \to \infty} f(n)/n exist, and is it strictly less than 22 (for instance π/3≈1.8138\pi/\sqrt{3} \approx 1.8138)?

References

  1. Klaus Friedrich Roth (with an appendix by Paul Erdős) (1951). On a problem of formal logic · DOI:10.1112/jlms/s1-26.3.198
  2. Richard K. Guy, Patrick A. Kelly (1968). The no-three-in-line problem · DOI:10.4153/CMB-1968-062-3
  3. R. R. Hall, T. H. Jackson, A. Sudbery, K. Wild (1975). Some advances in the no-three-in-line problem · DOI:10.1016/0097-3165(75)90043-6