🔢 mathematics

Remarks on the disproof of the unit distance conjecture

This paper presents a concise, human-verified summary of an OpenAI-generated counterexample to the Erdős unit distance conjecture, supported by reflections on its reliance on mathematical concepts attributed to Ellenberg-Venkatesh, Golod-Shafarevich, and Hajir-Maire-Ramakrishna.

Original authors: Noga Alon, Thomas F. Bloom, W. T. Gowers, Daniel Litt, Will Sawin, Arul Shankar, Jacob Tsimerman, Victor Wang, Melanie Matchett Wood

Published 2026-05-21

📖 1 min read🧠 Deep dive

CC BY 4.0

Original authors: Noga Alon, Thomas F. Bloom, W. T. Gowers, Daniel Litt, Will Sawin, Arul Shankar, Jacob Tsimerman, Victor Wang, Melanie Matchett Wood

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: A Puzzle Solved by a Digital Detective

Imagine a famous, unsolved puzzle in mathematics called the Erdős Unit Distance Conjecture.

For decades, mathematicians have been trying to figure out a specific question: If you scatter $n$ dots on a piece of paper, what is the maximum number of pairs of dots that can be exactly one inch apart?

The Old Guess: Most experts believed the answer was "not very many." They thought the number of pairs would grow slowly, roughly like $n$ (the number of dots) plus a tiny, shrinking amount. They thought the dots would eventually run out of space to form new one-inch connections.
The New Discovery: This paper announces that the experts were wrong. A team of human mathematicians, working with an AI model from OpenAI, found a way to arrange the dots so that the number of one-inch pairs explodes. It grows much faster than anyone expected—specifically, it grows like $n$ raised to a power slightly bigger than 1 (like $n^{1.000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000

Technical Summary: Remarks on the Disproof of the Unit Distance Conjecture

1. The Problem
The paper addresses the Erdős unit distance conjecture, a central open problem in discrete geometry posed in 1946. The problem asks for the maximum number of unit distances, denoted $U(n)$ , that can be determined by a set of $n$ points in the Euclidean plane $\mathbb{R}^2$ .

Historical Context: Erdős established a lower bound of $n^{1+\Omega(1/\log \log n)}$ using a $\sqrt{n} \times \sqrt{n}$ integer grid. The best known upper bound is $O(n^{4/3})$ , proven by Spencer, Szemerédi, and Trotter.
The Conjecture: Erdős conjectured that $U(n) \le n^{1+o(1)}$ . This belief was widely held, with many experts assuming the conjecture was true, partly because the distinct distances problem (proven by Guth and Katz to have $\Omega(n/\log n)$ distinct distances) suggested a structural rigidity that would limit unit distances.

2. Methodology and Proof Strategy
The paper presents a human-verified, simplified, and generalized version of a counterexample generated by an internal OpenAI model. The argument constructs a sequence of point sets $P_i \subset \mathbb{R}^2$ such that $|P_i| \to \infty$ and the number of unit distances is at least $|P_i|^{1+\varepsilon}$ for some $\varepsilon > 0$ .

The proof relies on two main lemmas bridging algebraic number theory and geometry of numbers:

Lemma 2.1 (Geometry of Numbers): This lemma establishes that if a full-rank lattice $\Lambda \subset \mathbb{C}^f$ contains a large set of points $U_\Lambda$ with complex absolute value 1 (unit modulus) in every embedding, one can project a bounded window of the lattice to the plane to create a point set with many unit distances. The number of unit distances scales super-linearly with the number of points if the ratio of "unit" elements to the lattice's "skewness" is sufficiently large and the dimension $f$ grows.
Lemma 2.2 (Algebraic Number Theory): This lemma constructs the required large set of unit-modulus elements. It utilizes the pigeonhole principle on the ideal class group of a number field $K$ . Specifically, it considers an ideal $Q$ formed by products of conjugate prime ideals $(P_j \bar{P}_j)^{k_j}$ . By counting the number of ways to form principal ideals within this structure, the lemma guarantees the existence of many elements $u \in Q^{-2}$ such that $|u|=1$ in all embeddings.

Key Technical Innovations:

Infinite Class Field Towers: Unlike previous constructions that fixed the number field and varied the primes, this proof varies the number field itself. It employs an infinite tower of number fields (specifically CM fields $K_j$ ) with bounded root discriminant but growing degree $[K_j : \mathbb{Q}] \to \infty$ .
Golod-Shafarevich Theory: The existence of such towers is guaranteed by the Golod-Shafarevich theorem. The authors utilize a specific construction involving a maximal pro- $p$ extension unramified outside a finite set of primes $T$ and split completely at a set $S$ . This ensures the class number $h(K)$ and discriminant remain controlled relative to the degree, allowing the number of unit elements to grow exponentially with the degree.
Splitting Primes: The construction requires the number fields to have many split primes (specifically, primes that split completely). The paper notes that while the original AI argument required a finite number of split primes, the human-verified version simplifies this by using towers with infinitely many split primes (a known result in the literature, e.g., by Hajir, Maire, and Ramakrishna).

3. Key Contributions and Results

Disproof of the Conjecture: The primary result is Theorem 1.1, which asserts the existence of $\varepsilon > 0$ and a sequence of point sets $P_i$ where the number of unit distances is $\ge |P_i|^{1+\varepsilon}$ . This definitively disproves the conjecture that $U(n) \le n^{1+o(1)}$ .
Human-AI Collaboration: The paper serves as a "digest" of an AI-generated proof. It validates the AI's core insight—that infinite class field towers could provide the necessary asymptotic growth—while simplifying the technical details (e.g., reducing the number of required split primes) and providing rigorous verification.
Synthesis of Fields: The work demonstrates a deep connection between extremal combinatorics, algebraic number theory (specifically class field towers and unit groups), and the geometry of numbers. It highlights that "enormous degree" in algebraic realizations, previously viewed as an annoyance, can be a source of counterexamples.

4. Significance and Reflections
The authors, including prominent mathematicians in combinatorics, number theory, and geometry, reflect on the implications of this result:

Surprise Factor: The result is surprising because the community had largely focused on proving the upper bound $n^{1+o(1)}$ , assuming the conjecture was true. The counterexample required a shift in perspective: moving from fixed fields to infinite towers and prioritizing disproof over proof.
AI's Role: The paper argues that AI demonstrated a unique capability to combine disparate technical tools (Golod-Shafarevich, class field theory, geometry of numbers) and persist in exploring a path (disproof via high-degree fields) that human experts might have dismissed as unlikely or too complex.
Limitations of the Result: The authors note that while the construction is elegant, it does not introduce new geometric tools or structural results that a proof of the conjecture might have offered. It is a "negative" result (a counterexample) rather than a structural characterization.
Future Directions: The paper suggests that similar number-theoretic constructions might be relevant to other open problems in discrete geometry, such as the distinct distances problem or the unit distance problem in $\mathbb{R}^3$ , though significant obstacles remain for those cases.

In summary, the paper presents a rigorous, human-verified disproof of the Erdős unit distance conjecture, leveraging infinite class field towers to construct point sets with super-linear unit distances. It stands as a landmark example of AI-assisted mathematical discovery, where an AI generated a novel counterexample that human experts subsequently refined and validated.

The Big Picture: A Puzzle Solved by a Digital Detective

Technical Summary: Remarks on the Disproof of the Unit Distance Conjecture

More like this