Skip to content

Erdős Problem #662

Editorial depth: Reviewed explanatory article — evidence qualified.
Snapshot classification: open (not automatically the live result).
Subject areas: geometry, distances.

Read this status correctly

The status above is reproduced from GCL's immutable external-catalogue snapshot. It is not a new claim of current mathematical resolution, a source-semantic review, an independent proof check, or GCL certification. A published proof and a GCL replay are separate events.

The question in plain language

Erdős Problem #662, attributed to Erdős, Lovász and Vesztergombi, compares well-separated planar point sets with the triangular lattice, the familiar arrangement in which every point has six closest neighbours.

The catalogue asks about how many distances can occur below a given threshold \(t\), under the constraint that every pair of distinct points is at least one unit apart. Unfortunately, the maintained question is internally inconsistent as written. It uses the word distances for quantities that appear to alternate between numbers of neighbouring points and numbers of distinct distance values. Some of its numerical examples do not agree with either interpretation.

The source itself warns that there must be at least one typo. The most responsible editorial treatment is therefore a source-fidelity investigation, not an assertion that the ill-posed literal statement is a new open theorem.

A small example

Let the unit triangular lattice be

\[ \Lambda=\left\{\left(a+\frac b2,\frac{\sqrt3}{2}b\right): a,b\in\mathbb Z\right\}. \]

The squared distance from its origin to the lattice point indexed by \((a,b)\) is

\[ Q(a,b)=a^2+ab+b^2. \]

There are six lattice points with \(Q=1\), six with \(Q=3\), six with \(Q=4\), twelve with \(Q=7\), and six with \(Q=9\). These numbers were independently verified by exact integer enumeration.

Thus the numbers of other lattice points within a disk centred at the origin are:

Radius \(t\) Count of lattice neighbours with distance \(\leq t\) Distinct positive distance values
\(1\) 6 1
\(\sqrt3\) 12 2
\(2\) 18 3
\(3\) 36 5

This gives a precise test for the source's displayed examples. The maintained values \(f(1)=6\) and \(f(\sqrt3)=12\) match a neighbour count, not a count of different lengths. But its \(f(3)=18\) does not match the same definition; 18 is the neighbour count at radius \(2\).

That discrepancy is mathematical evidence of a transcription or definition problem, not permission to silently replace the historical question with our preferred correction.

What is known

The recorded version asks whether a set \(X=\{x_1,\ldots,x_n\}\) satisfying

\[ \lVert x_i-x_j\rVert\geq1 \quad(i\ne j) \]

has at most \(f(t)\) distances of size at most \(t\), once \(n\) becomes large. It also asks whether the number of distances below \(\sqrt3-\varepsilon\) is less than one.

Taken literally as the number of distinct distance values anywhere in the configuration, the proposed upper bound is immediately false for every fixed \(t>1\).

To see why, choose any integer \(K\) and \(K\) distinct lengths

\[ 1<r_1<\cdots<r_K<t. \]

For each \(i\), create a pair of points at separation \(r_i\), and put the pairs far apart, with gaps larger than \(3t\). Every pair of distinct points in the full set is at least one unit apart. Every cross-pair distance between different chosen pairs is larger than \(t\), but each of the \(K\) chosen pairs contributes its own different short distance \(r_i\). We therefore have an arbitrarily large number \(K\) of distinct distances at most \(t\).

This directly refutes any universal constant upper bound \(f(t)\) on the global number of short distance values. It remains a counterexample even when the number of points is arbitrarily large.

The second literal assertion is also suspect: if \(0<\varepsilon<\sqrt3-1\), then \(t=\sqrt3-\varepsilon>1\), and any configuration containing a unit-distance pair has at least one distance below \(t\). Its global distance count cannot be less than one.

What this does not settle. An intended question might instead concern the maximum number of neighbours of one point within a disk of radius \(t\), subject to unit separation, or the number of short edges incident to a chosen vertex. That is a genuine local packing problem, quite different from the number of distinct lengths anywhere among all point pairs. The catalogue's examples at radii \(1\) and \(\sqrt3\) suggest a local interpretation, but the original intended quantifiers remain unverified.

A further source inconsistency occurs in a displayed sequence of successive triangular-lattice lengths reported as \(1,\sqrt3,3,5,\ldots\). The exact quadratic form above shows that the first positive lattice lengths include

\[ 1,\quad\sqrt3,\quad2,\quad\sqrt7,\quad3,\ldots. \]

The displayed sequence therefore cannot literally be the increasing sequence of all positive triangular-lattice distances.

GCL's current state

The original authenticated RESULT/1 source return has historical disposition STATUS_CONFLICT. This status and the original research exceptions are preserved, not silently converted into a theorem verdict. The maintained Erdős source and source-specific mathematical definitions were reviewed, and this article is editorially reviewed as a source-exception-annotated explanation under EC-BULK-005-R10, effective only after the exact-hash protected admission merges.

Source interpretation and unresolved mathematical qualification: The maintained lattice shell tally conflicts with exact radius-squared counting, and arbitrarily separated short-distance pairs invalidate the literal global f(t) bound.

The explanation is accepted for publication of established background, examples, and explicitly attributed current claims. It does not establish or formally certify the disputed original theorem, external model-assisted proof, historical interpretation, or any unreplayed Lean kernel file. Original source-fidelity and proof-review research holds remain.

What remains

The first obligation is an exact primary-source reconstruction. Inspect Erdős's original Some of my favourite unsolved problems (1997), page 532, including surrounding explanatory text. Determine whether \(f(t)\) counts neighbours, edges, length values, or some other object; whether \(t\) is an ordinary distance, a squared-distance threshold or a shell index; and whether the inequality counts distances globally or from a fixed centre.

Then check candidate corrected readings against the lattice-shell table and the separated-pairs construction. A proposed formulation that fails either an elementary finite fixture or the basic large-separation example should be rejected rather than forwarded to an unrestricted mathematical worker.

Only after such a source lock can a bounded mathematical successor be defined. For example, a repaired local packing version could ask how many unit-separated points fit inside an annulus or disk about a fixed centre, with exact thresholds and extremizers specified. But that would be a proposed reconstruction, not necessarily Erdős's question.

The article should remain on source-fidelity hold until the historical formulation is independently checked. Any later publication would still need accountable critical review, protected merge, public MathJax browser readback and a separate V2 editorial admission.

Sources and claim boundaries

  • T. F. Bloom, Erdős Problem #662: maintained transcription, OPEN label and explicit warning that at least one typo must be present.
  • Erdős #662 revision history: preserved wording and the questionable triangular-lattice sequence.
  • GCL authenticated #662 source return: independent finite lattice enumeration, counterexample and exact source-reconstruction task.
  • P. Erdős, Some of my favourite unsolved problems, Math. Japon. (1997), p.532 [Er97e]: cited historical authority, original page not independently inspected.
  • Erdős–Lovász–Vesztergombi attribution: maintained-source record, not independently recovered original manuscript.

Authority boundary: elementary checking that exposes a statement defect, not a resolution of a reliably reconstructed open problem. No original-paper source lock, mathematical proof certification, protected production merge, deployed MathJax rendering or editorial admission.

Qualified editorial critical review: R10 source-exception dossier. The source-semantic exception is visible and its original research hold remains. Publication classification: REVIEWED EXPLANATORY ARTICLE WITH UNRESOLVED SOURCE/PROOF QUALIFICATIONS. Independent theorem proof or Lean certification: NOT CLAIMED.

Catalogue and evidence provenance

  • Pinned catalogue source: teorth/erdosproblems at 536408716724.
  • GCL issue: #1933 — protected issue/assignment identity; its labels and results are live, not reproduced automatically here.
  • GCL triage: STATEMENT_CAPTURE_REQUIRED.
  • Formal-statement coverage: No statement was confirmed in GCL's pinned formal-provider snapshot. Presence does not mean the statement or proof has been checked.
  • External mathematical status: consult current problem page and its dated bibliography.

Back to catalogue · Compendium introduction