An incomplete attack on the upper bound of the unit distance problem

Steven Senger · 2026

Read An incomplete attack on the upper bound of the unit distance problem by Steven Senger (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-26145.txt.

Introduction

Senger offers an incomplete attempt to show that the standard upper bound on unit distances is not sharp. The bound caps unit distances from a big finite point set in the plane at order n to the four-thirds. Senger's methods aim to break that bound's sharpness, but the attempt stays incomplete, as the paper states. The same methods also speak to point-line sets attaining the sharp Szemerédi-Trotter incidence bound. Readers should treat the work as a partial attack, not a proof.

Catalog record

CREATORSteven Senger
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE29874
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.26145

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

An incomplete attack on the upper bound of the unit distance problem Steven Senger Abstract This is an incomplete attempt to show that the upper bound of ≲ n 4 3 \lesssim n^{\frac{4}{3}} on the number unit distances determined by a large finite set of n n points in the plane is not sharp. The methods also say something about sets of n n points and n n lines that attain the sharp bound of the Szemerédi-Trotter point-line incidence bound. 1 Introduction The unit distance problem asks for upper bounds on how often the most popular distance can occur in a large finite set of n points in the plane. In 1946, in [ 2 ] , Erdős conjectured that this could occur no more than n^{1+\epsilon} times for any \epsilon>0. This was recently disproved by artificial intelligence (see [ 5 ] and [ 1 ] ), causing an upsurge in interest in the problem. Following this proof, Sawin gave an explicit improvement exhibiting Cn^{1.014114} unit distances for some absolute constant C>0, in [ 6 ] . There is still a gap between the best known upper bound of n^{\frac{4}{3}} due to Spencer, Szemerédi, and Trotter in [ 7 ] , and the recent results. Because of the recent interest, I have decided to make this document publicly available. This work was mostly done between 2015–2021, but I got stuck. The basic idea is to suppose that n^{\frac{4}{3}} is tight, then derive a contradiction by following the crossing number lemma backwards. This type of approach worked in the case of distinct dot products, and lead to a concrete improvement for that problem (see [ 3 ] ), so I was hopeful it could work here as well. However, when I saw that in [ 4 ] , Katz and Silier produced results that appear to be significantly stronger that what I proved, I essentially abandoned the attack. Still, here is a description of the approach in case anyone finds it useful. 1.1 Preliminaries By way of contradiction, we suppose that n^{\frac{4}{3}} is a sharp upper bound for the unit distance problem. We derive necessary facts about any point set attaining this bound. In what follows, we use the notation X\lesssim Y to denote that X=O(Y), X\approx Y to denote that X=\Theta(Y). We also use the fairly standard notation X=o(Y) or Y=\omega(X). Moreover, we use c to denote an undetermined constant, but if it has a subscript, c_{j} , we will keep track of it. Also, we assume that n is a large finite number. We begin with the celebrated result of Spencer, Szemerédi, and Trotter, from [ 7 ] . Let u(n) denote the maximum number of pairs of points separated by a unit distance in any set of n points in the plane. Theorem 1.1 . u(n)\lesssim n^{\frac{4}{3}}. We sketch a of the proof of Theorem 1.1 due to Székely, from [ 8 ] . It relies on the following lemma about the crossing number of a graph, which is the smallest number of times edges can cross under any redrawing of the graph in the plane. We we record a statement given in [ 8 ] , though various statements and proofs exist elsewhere in the literature. Lemma 1.2 . Given a graph G(V,E) with |E|>4|V|, we have that the crossing number of G is bounded below by cr(G)\geq\frac{|E|^{3}}{100|V|^{2}}. Proof. To prove Theorem 1.1 , we construct a graph out of any set of n points in the plane. Let the points be the vertices of the graph, so |V|=n. Draw unit circles centered at each point. The arcs of the circles connecting consecutive points will be edges. Notice that by construction, |E| is exactly two times the number of unit distances. So either |E|\leq 4|V|, which implies that the number of unit distances is no more than 4n, or we can employ Lemma 1.2 . In the latter case, notice that since the edges are arcs of circles, we can have no more than two crossings from any pair of circles. There are exactly n(n-1)/2 pairs of circles, so there can be no more than n(n-1) crossings. We plug in the various quantities and get \frac{|E|^{3}}{100|V|^{2}}\leq cr(G)\leq n(n-1), (1) which gives us that |E|\leq(100n^{3}(n-1))^{\frac{1}{3}}\lesssim n^{\frac{4}{3}}. ∎ 1.2 Average crossings per edge Suppose that there exists a set P , of n points in the plane with c_{1}n^{\frac{4}{3}} pairs of points separated by a unit distance. We now claim that in the associated graph for such a set, almost every edge must have about n^{\frac{2}{3}} crossings. Proposition 1.3 . In the graph associated to the point set, there are no more than o\left(n^{\frac{4}{3}}\right) edges with o\left(n^{\frac{2}{3}}\right) crossings or with \omega\left(n^{\frac{2}{3}}\right) crossings. Proof. Both claims will follow by way of contradiction. For the first, suppose that there is a subset E^{\prime}\subset E of cn^{\frac{4}{3}} edges with m=o\left(n^{\frac{2}{3}}\right) crossings. Then we could construct a subgraph of G called G^{\prime}, with the same vertex set, but this smaller edge set, E^{\prime}. We count the total crossings of G^{\prime} by simply adding up the number of crossings per edge, giving us cr(G^{\prime})=o\left(|E^{\prime}|n^{\frac{2}{3}}\right)=o\left(n^{2}\right). Notice that we still have |E^{\prime}|>4|V|, so Lemma 1.2 tells us that cr(G^{\prime})\gtrsim\frac{|E^{\prime}|^{3}}{|V|^{2}}\approx n^{2}, which contradicts the upper bound on cr(G^{\prime}). The second claim follows by a similar argument. Suppose there is a subset, E^{\prime\prime}\subset E of cn^{\frac{4}{3}} edges with m=\omega\left(n^{\frac{2}{3}}\right) crossings. We again count the total number of crossings and get that cr(G)=\omega\left(|E^{\prime\prime}|n^{\frac{2}{3}}\right)=\omega\left(n^{2}\right), but this violates the upper bound of n(n-1) as given in ( 2 ). ∎ Proposition 1.4 . Any set of n points in the plane with c_{1}n^{\frac{4}{3}} unit distances must consist of at least c_{2}n points each at a unit distance to at least c_{3}n^{\frac{1}{3}} other points in the set. Proof. We begin by claiming that there are \gtrsim n circles that each cross \gtrsim n other circles. If this were not the case, we would need to have o(n) circles with \gtrsim n crossings. But then by referring to ( 2 ), we could reduce the upper bound on the crossing number, and get that there are fewer than c_{1}n^{\frac{4}{3}} unit distances. Proposition 4.2 has guaranteed that \gtrsim n^{\frac{4}{3}} of the edges in our graph have \approx n^{\frac{2}{3}} crossings. So when we count the \approx n total crossings for any of these \approx n circles, they must break down into \approx n^{\frac{1}{3}} edges with \approx n^{\frac{2}{3}} crossings each. ∎ We denote the unit circle centered at the point p by C(p). When two points q and r are less than a unit distance apart, we define L(q,r) to be the union of the two crescent-shaped regions bounded by the C(q) and C(r). We call these regions lunes . We use the notation |L(q,r)\cap P| to denote the number of points from our set within a given lune. We say that a point p from our set is typical if there are at least c_{4}n^{\frac{1}{3}} pairs of points q and r that are consecutive on C(p) , satisfying c_{5}n^{\frac{2}{3}}\leq|L(q,r)\cap P|\leq c_{5}^{\prime}n^{\frac{2}{3}}. As a consequence of this definition, a point p is typical if and only if there are \gtrsim n^{\frac{1}{3}} circles passing through p that define lunes with \approx n^{\frac{2}{3}} points from P properly contained

Read the complete text