Skip to content

Erdős Problem #106

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

Pack \(n\) Euclidean squares inside a square of side one, requiring that no two squares share an interior point. The smaller squares may be rotated. Let \(f(n)\) be the supremum of the sum of their side lengths. Erdős #106 conjectured that, for every positive integer \(k\),

\[ f(k^2+1)=k. \]

The condition permits boundary contacts and rotations. Replacing it by a restriction that all squares have sides parallel to the container changes the problem.

A small example

Divide the unit square into a \(k\times k\) array of squares of side \(1/k\). Their total side length is exactly \(k\). Replacing one square by two smaller squares of side \(1/(2k)\) preserves that sum while increasing the number of pieces from \(k^2\) to \(k^2+1\). Hence \(f(k^2+1)\ge k\).

For the axis-parallel variant, Baek, Koizumi and Ueoro proved equality \(g(k^2+1)=k\) in A note on the Erdős conjecture about square packing. Their restriction is substantive, not merely a coordinate choice.

What is known

DISPROVED for unrestricted orientations, by an explicit 2026 rational construction. The maintained Erdős #106, edited March 2026, still labels the unrestricted problem FALSIFIABLE/OPEN. It predates the public August 2026 counterexamples and must not control the mathematical conclusion.

In Sprite143's primary construction, 17 squares, 16 axis-parallel and one rotated through an angle with cosine \(255/257\) and sine \(32/257\), fit inside the unit square with pairwise disjoint interiors and total side length

\[ \frac{2190452873}{547596200} =4+\frac{68073}{547596200}>4. \]

The corresponding conjecture predicts \(f(17)=4\) at \(k=4\). This explicit strict inequality refutes its universal assertion. The repository's accompanying exact rational checkers are not the only evidence: GCL independently reconstructed each square's four vertices from the publicly specified closed-form configuration, then tested all 68 corners for containment and all 136 unordered pairs using a separate exact-rational separating-axis implementation. The results were 100 strictly separated pairs, 36 contacting pairs and zero interior overlaps. No floating-point tolerance was used.

The tested source revision is Sprite143/erdos-106-counterexample@edbfb534e78d4f1f9b1a5a7298b9b022285b4962. The independent machine receipt is stored at /home/jim/GCL/EC-BULK-002/2026-10-14-next-104-111/ERDOS_106_EXACT_GEOMETRY_QA.json. A separate Seamons 2026 Zenodo preprint gives another exact rational 26-square counterexample. These are construction certificates; external peer review and independent replay of the reported third-party Lean proof are different and remain unclaimed.

GCL's current state

The original source issue #1375 carries a pinned DISPROVED status, whereas the maintained public page shows an earlier OPEN status. The discrepancy is now reconciled at the mathematical level by the exact witness above. The protected original source metadata and independent axis-parallel theorem are both preserved. No prize award or protected formal proof certification follows from this editorial audit.

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 successor is to preserve a minimal, separately implemented exact-rational verifier as a reproducible project fixture. Its acceptance test must reconstruct all corners, verify containment, enumerate all 136 pairs, permit boundary contact, and independently sum 17 positive side lengths to the rational above. A future referee can replay the certificate without trusting a geometric optimization package. Stop short of claiming the optimal value of \(f(17)\) or any result about packings of exactly \(k^2\) squares.

Sources and claim boundaries

The unrestricted conjecture's counterexample is independently verified as finite rational geometry, not peer-reviewed mathematical attribution, protected Lean acceptance or editorial admission.

Catalogue and evidence provenance

  • Pinned catalogue source: teorth/erdosproblems at 536408716724.
  • GCL issue: #1375 — 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