Skip to content

Erdős Problem #105

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

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

Let \(A\) and \(B\) be disjoint finite sets of points in the Euclidean plane, with \(|A|=n\) and \(|B|=n-3\). Assume \(A\) is not collinear. Erdős and Purdy asked whether some straight line determined by two distinct points of \(A\) contains no point of \(B\).

This is not merely a question about the number of lines spanned by \(A\). The same line may contain three or more points of \(A\); a single point of \(B\) on that line makes it forbidden. The two colour classes must have the exact specified sizes.

A small example

With \(n=3\), the second set has size zero. Any three noncollinear points form a triangle, and all three of its side lines avoid \(B\) vacuously. To disprove the universal statement one needs a noncollinear \(A\) and a smaller set \(B\) meeting every line determined by a pair of \(A\) points.

A counterexample to the stronger historical version with \(|B|=n-2\) is not sufficient: removing one point from \(B\) might expose an unblocked line. The exact three-point difference is therefore essential.

What is known

DISPROVED. The maintained Erdős #105 credits Wu Xichuan with several explicit finite counterexamples. The original mathematical source is linked as forum post #1430. An independent third-party Lean project provides a formalized counterexample in Erdos105.lean, including explicit integer-coordinate point sets; its file explains that a projective construction was translated into finite affine coordinates.

The maintained source reports the third-party proof as Lean-verified, but GCL has not run the declared Lean toolchain or audited all dependencies. A theorem displayed online with a proof is evidence of formalization, not a protected GCL kernel-replay receipt. The result does not settle whether the assertion becomes true with \(|B|=n-4\), or with \(|B|=n-O(1)\).

The original Erdős–Purdy 1995 paper and the 1983 Beck/Szemerédi–Trotter bounds are separate historical results, including a weaker affirmative statement when \(|B|\) is only a sufficiently small positive fraction of \(n\).

GCL's current state

The pinned and maintained catalogues mark #105 DISPROVED; the original authenticated issue is #1374. The counterexample statement and the third-party formal-source headers were inspected, but its rational geometry and formal proof have not been independently replayed here. The disproof is correctly attributed to its source author, not to GCL.

The original issue contains an authenticated RESULT/1 source report. This is unadjudicated source evidence only, not independent mathematical certification or editorial admission.

What remains

A bounded independent replay is to extract the exact two finite coordinate sets from the cited Lean witness and check them with integer determinants. Acceptance: \(|A|=|B|+3\), disjointness, at least one noncollinear triple in \(A\), and for every pair of distinct \(A\) points, some \(B\) point collinear with them. Check a second implementation's pair-incidence table. Stop at these finite facts without claiming an audit of the entire Lean kernel proof.

Sources and claim boundaries

No independently certified source-programme proof, new counterexample, or editorial admission is claimed.

Catalogue and evidence provenance

  • Pinned catalogue source: teorth/erdosproblems at 536408716724.
  • GCL issue: #1374 — protected issue/assignment identity; its labels and results are live, not reproduced automatically here.
  • GCL triage: OUTSIDE_STRICT_OPEN_TRIAGE.
  • 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