Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.
Updated
Ducz 2026 unit distance graph plane independence ratio
conjecture_1: The authors conjecture that f(n)/n = m_1(R^2) + o(1), and in particular that m_1(R^2) equals Croft's density 0.22936...; the paper proves nothing toward it.
corollary_1: The fractional chromatic number of the plane is strictly greater than 4, which falsifies Conjecture 1 of Matolcsi, Ruzsa, Varga and Zsámboki.
corollary_2: The chromatic number of the plane is at least 5, which the paper presents as recovering de Grey's lower bound by a different route.
corollary_3: The supremum m_1 of the upper densities of measurable planar sets with no two points at distance 1 is strictly less than 1/4, a Fourier-free proof of a conjecture of Erdős first proved by Ambrus, Csiszárik, Matolcsi, Varga and Zsámboki.
lemma_1: The unit-distance graph G_29, the 27-vertex configuration of Matolcsi, Ruzsa, Varga and Zsámboki with two added points, has geometric fractional chromatic number greater than 4.0007, certified by a rational dual solution of a linear program.
theorem_1: Some finite unit-distance graph in the plane has independence number less than a quarter of its number of vertices, which answers in the negative Erdős's question whether f(n) >= n/4 for all n; the graph is shown to exist but is not exhibited.
Ákos Dúcz, Dániel Varga, A unit-distance graph in the plane with independence ratio below 1/4. arXiv preprint (2026). arXiv:2606.28157. The copy read for this card is arXiv:2606.28157v1, dated 26 June 2026.
Dúcz and Varga prove Theorem 1: some finite planar unit-distance graph G has alpha(G)/|V(G)| < 1/4, answering negatively Erdős's question of whether f(n) >= n/4 for all n. They work in the geometric fractional chromatic number framework of Matolcsi, Ruzsa, Varga and Zsámboki, adding two carefully chosen points p and q to the 27-vertex configuration G_27 to form a 29-vertex 'snail' graph; Lemma 1 gives chi_gf(G_29) > 4.0007, certified by a rational feasible solution of the dual linear program, which the authors do not claim is optimal (that dual program has 16860 variables and 498168 constraints, and its exact solution was beyond their computational resources), and following the arguments in the proofs of MRVZ's Theorems 1 and 2 then gives finite 'blow-ups' of G_29 with independence ratio below 1/4. This falsifies MRVZ Conjecture 1 and yields Corollary 1, chi_f(R^2) > 4, Corollary 2, chi(R^2) >= 5 (recovering de Grey), and Corollary 3, m_1(R^2) < 1/4 (a simpler proof of a result of Ambrus et al.). The search for p and q rests on a characterization of all optimal geometric fractional colorings of G_27: they are exactly the convex combinations of 23 extremal colorings and span a set of affine dimension 11, each extremal coloring being realizable as a k-fold coloring with k <= 22. The authors also conjecture (Conjecture 1, p. 3) that f(n)/n = m_1(R^2) + o(1), and in particular that m_1(R^2) equals Croft's density 0.22936.... For problem 1070 the paper settles the particular n/4 question negatively while leaving the estimation of f(n) open, for problem 508 it gives chi_f(R^2) > 4 and a new proof of chi(R^2) >= 5 but no new bound on chi(R^2), and for problem 232 it gives a new proof of m_1(R^2) < 1/4 without a numerical bound; the graph is only shown to exist, and the paper notes that an explicit construction, though possible in principle, is astronomically large.
Source: https://arxiv.org/abs/2606.28157. The arXiv record names arXiv's non-exclusive distribution license (arXiv:2606.28157), every other right reserved.
Bears on. #1070: Theorem 1 gives an N-point planar set with f(N) < N/4, the negative answer to the particular question whether f(n) >= n/4 for all n, as the paper states (p. 1); it does not estimate f(n), and Conjecture 1 is the authors' conjectured asymptotic, unproved. #508: Corollary 1 gives chi_f(R^2) > 4 and Corollary 2 a second proof of the known bound chi(R^2) >= 5 (p. 1); no new bound on chi(R^2). #232: Corollary 3 gives a Fourier-free proof of m_1(R^2) < 1/4, first proved by Ambrus, Csiszárik, Matolcsi, Varga and Zsámboki (pp. 1, 3), with no numerical bound on m_1; Conjecture 1 conjectures its value.
Results. Labels and pages are those of v1.
- Theorem 1 (p. 1, restated p. 6): a finite planar unit-distance graph with independence ratio below 1/4.
- Corollary 1 (p. 1): chi_f(R^2) > 4.
- Corollary 2 (p. 1): chi(R^2) >= 5.
- Corollary 3 (p. 1): m_1(R^2) < 1/4.
- Conjecture 1 (p. 3): f(n)/n = m_1(R^2) + o(1).
- Lemma 1 (p. 6): chi_gf(G_29) > 4.0007.
The Section 5 characterization (pp. 6--8) of the optimal geometric fractional colorings of G_27, as the convex hull of 23 extremal colorings spanning an affine set of dimension 11, each a k-fold coloring with k <= 22, is an unnumbered computational result and has no page of its own.
No file of this source is held: no license on record permits its redistribution, and the card cites the edition it names above.