Skip to content

Erdős Problem #120

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

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

Suppose someone chooses an infinite set of real numbers \(A\). Can we build a measurable set \(E\) of positive Lebesgue measure which contains no nontrivial affine image of the entire set \(A\)?

Formally, Erdős #120 asks whether for every infinite \(A\subseteq\mathbb R\) there is a measurable set \(E\subseteq\mathbb R\), \(m(E)>0\), such that

\[ aA+b\not\subseteq E \quad\text{for every }a\ne0,\ b\in\mathbb R. \]

The scale \(a\) may be negative. Crucially, \(E\) may depend on \(A\). This is the Erdős similarity conjecture. For a finite \(A\), Steinhaus's theorem supplies scaled and translated copies inside any positive-measure set, so the word infinite cannot be omitted.

A small example

Take \(A=\mathbb Z\). The interval \(E=[0,1]\) has positive measure but cannot contain a translate and nonzero rescaling of the entire unbounded set \(\mathbb Z\). For a bounded example, take a decreasing sequence \(A=\{1,1/2,1/4,\ldots\}\) tending to zero. A bounded interval can contain an affine copy of this sequence; excluding every possible translated, rescaled copy is substantially harder.

If \(A\) is dense in some interval, an affine copy is dense in a nondegenerate interval. A positive-measure set with no complete interval, chosen appropriately, can avoid such a copy. That observation alone does not handle rapidly convergent sequences.

What is known

The conjecture is OPEN, including the geometric sequence \(\{2^{-j}:j\ge0\}\). The primary survey by Yeonwook Jung, Chun-Kit Lai and Yuveshen Mooroogen, Fifty years of the Erdős similarity conjecture, published in a January 2025 revised edition, reviews major partial results. Eigen and Falconer show that a decreasing null sequence with successive ratios tending to one is not measure-universal. Bourgain proves the result for sumsets of three infinite subsets of \(\mathbb R\). Kolountzakis supplies other gap-structure criteria and almost-everywhere variants.

The maintained Erdős #120, last edited 23 January 2026, identifies unbounded sets and sets dense in an interval as solved classes. Those reductions leave the hard lacunary-sequence case. A theorem that almost every scale fails to embed does not prove that no scale embeds, and replacing positive measure with residual Baire category gives another distinct problem.

GCL's current state

The original authenticated assignment is #1389, with pinned status OPEN. The protected Formal Conjectures 120 statement was read: it quantifies over all infinite real sets, measurable positive-volume \(E\), and every nonzero scale. Its declaration contains a proof hole, not a Lean resolution. The complete proof of each survey result has not been independently rederived.

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 formalize the unbounded-set \(A\) case: construct \(E=[0,1]\), prove that for every \(a\ne0\) the image \(aA+b\) is unbounded, and therefore not contained in \(E\). Acceptance requires both positive measure and the correct order of quantifiers. Stop at this special case. It does not address bounded lacunary sets such as \(\{2^{-j}\}\), and must not be used to claim a proof of the full conjecture.

Sources and claim boundaries

No solution for lacunary sequences, independently kernel-checked proof, or editorial admission is claimed.

Catalogue and evidence provenance

  • Pinned catalogue source: teorth/erdosproblems at 536408716724.
  • GCL issue: #1389 — protected issue/assignment identity; its labels and results are live, not reproduced automatically here.
  • GCL triage: PROTECTED_READY_BACKLOG.
  • Formal-statement coverage: A provider statement was included in the pinned 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