Abstract:In the 1960s Moser asked how dense a subset of $\mathbb{R}^d$ can be if no pairs of points in the subset are exactly distance 1 apart. There has been a long line of work showing upper bounds on this density. One curious feature of dense unit distance avoiding sets is that they appear to be ``clumpy,'' i.e. forbidding unit distances comes hand in hand with having more than the expected number distance $\approx 2$ pairs. In this work we rigorously establish this phenomenon in $\mathbb{R}^2$. We show that dense unit distance avoiding sets have over-represented distance $\approx 2$ pairs, and that this clustering extends to typical unit distance avoiding sets. To do so, we build off of the linear programming approach used previously to prove upper bounds on the density of unit distance avoiding sets.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is about the relationship between the density of sets that avoid unit distances in two - dimensional space (that is, the distance between any two points in the set cannot be exactly 1) and the over - representation of specific distance pairs (for example, pairs of points with a distance of approximately 2). Specifically, the authors want to prove that in two - dimensional space, in those sets that avoid unit distances and have a high density, the number of pairs of points with a distance of approximately 2 will exceed expectations. This phenomenon indicates that the restriction of avoiding unit distances not only reduces the number of pairs of points with a distance of exactly 1, but also leads to an increase in the number of pairs of points with other specific distances, which is counter - intuitive. ### Background of the paper - **Moser problem**: In the 1960s, Moser proposed a question: in a space $\mathbb{R}^d$ of a given dimension, if there is no distance exactly equal to 1 between any two points in a subset $A$, then what is the maximum density of this subset? - **Related research**: This problem is closely related to the Hadwiger - Nelson problem, which requires calculating the minimum number of colorings of $\mathbb{R}^d$ such that all pairs of points with a distance exactly equal to 1 have different colors. - **Planar case**: In the two - dimensional plane, Erdős conjectured in 1985 that the maximum density does not exceed 1/4. This conjecture was recently confirmed by Ambrus et al., who proved that the maximum density of any Lebesgue - measurable 1 - avoiding planar set is strictly less than 0.2470. ### Main contributions - **Over - representation phenomenon**: The authors proved through strict mathematical methods that in two - dimensional space, in high - density sets that avoid unit distances, the number of pairs of points with a distance of approximately 2 will exceed expectations. - **Linear programming method**: The authors use the linear programming method to study the autocorrelation function $f^\circ(r)$ of 1 - avoiding sets and prove the above phenomenon by solving linear programming problems. - **Container method**: The authors also use the graph container method to prove that typical 1 - avoiding sets also have the over - representation phenomenon of pairs of points with a distance of approximately 2. ### Specific results - **Theorem 1.3**: There exists a constant $\gamma>0$. For a periodic 1 - avoiding set $A\subset\mathbb{R}^2$, if $s(1; A)\leq\gamma$ and $\delta(A)\geq m_1(\mathbb{R}^2)-\gamma$, then $s(1.96; A)\geq1 +\gamma$. - **Theorem 1.4**: For sufficiently large $K$ and $N$, randomly select a set $A$ from all 1 - avoiding $\frac{1}{N}$ - local - constant and $K$ - periodic sets. With high probability, $s(1.96; A)\geq1+\frac{\gamma}{2}$. ### Method overview - **Linear programming**: By studying the Fourier expansion of the autocorrelation function of 1 - avoiding sets, construct linear programming problems to prove the upper bound of density. - **Graph container method**: Utilize the supersaturation property of graphs and the container lemma to prove that typical 1 - avoiding sets also have the over - representation phenomenon of pairs of points with specific distances. Through these methods, the authors not only verified the theoretical conjecture, but also provided a new perspective to understand the structural characteristics of 1 - avoiding sets.

Clustering in typical unit-distance avoiding sets

On the Advice Complexity of Online Unit Clustering

Clustering Based on Pairwise Distances When the Data is of Mixed Dimensions

Low-Distortion Clustering in Bounded Growth Graphs

A Statistical Information-Based Clustering Approach in Distance Space

Efficient Clustering with Limited Distance Information

Unit and distinct distances in typical norms

On Generalization Bounds for Projective Clustering

Density-ratio Based Clustering for Discovering Clusters with Varying Densities.

Avoiding intersections of given size in finite affine spaces AG(n,2)

Avoiding intersections of given size in finite affine spaces AG( n ,2)

On cliques in three-dimensional dense point-line arrangements

On Clustering Induced Voronoi Diagrams

Clustering Mixtures of Bounded Covariance Distributions Under Optimal Separation

DECWA : Density-Based Clustering using Wasserstein Distance

A Distance Scaling Method to Improve Density-Based Clustering.

A new model for natural groupings in high-dimensional data

A new lower bound for the density of planar Sets avoiding Unit Distances

Inapproximability of Maximum Diameter Clustering for Few Clusters

Clustering What Matters in Constrained Settings