Skip to content

Erdős Problem #656

Editorial depth: Reviewed explanatory article — evidence qualified.
Snapshot classification: proved (not automatically the live result).
Subject areas: number theory, additive 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 \(A\) is a set of positive whole numbers that occupies a non-negligible fraction of the integers infinitely often. Can we find infinitely many members of \(A\) whose pairwise sums, after adding one fixed shift, all return to \(A\)?

The precise Erdős Problem #656 asks for an infinite \(B\subseteq A\) and an integer \(t\) such that

\[ \{b_1+b_2+t:b_1,b_2\in B,\ b_1\ne b_2\}\subseteq A. \]

Two subtleties matter: the elements of \(B\) must themselves lie in \(A\), and the sums use two distinct elements. We may translate all sums by one integer \(t\), but we do not choose a different shift for each pair.

This statement is proved by a published 2024 theorem of Bryna Kra, Joel Moreira, Florian K. Richter and Donald Robertson. It is not an open mathematical conjecture.

A small example

Let \(A\) consist of all positive integers leaving remainder \(1\) on division by \(3\):

\[ A=\{1,4,7,10,13,\ldots\}. \]

This has density \(1/3\). Choose \(B=A\) and \(t=2\). Every pair \(b_1,b_2\in B\) has

\[ b_1+b_2\equiv 2\pmod 3, \quad\text{so}\quad b_1+b_2+2\equiv1\pmod3. \]

For instance, the distinct members \(1\) and \(4\) have \(1+4+2=7\in A\); \(4\) and \(7\) give \(13\in A\). Indeed the property holds for all pairs. Conversely, choosing an infinite \(B\subseteq A\) cannot make unshifted pairwise sums belong to \(A\) in this example: they always have remainder \(2\), not \(1\). So the permission to shift is mathematically significant.

This example is a direct elementary check, not a proof for every positive-density set.

What is known

The upper asymptotic density of \(A\subseteq\mathbb N\) is

\[ \overline d(A)=\limsup_{N\to\infty} \frac{|A\cap\{1,\ldots,N\}|}{N}. \]

The registered problem assumes \(\overline d(A)>0\). A related, broader notion is upper Banach density: one is allowed to find long intervals anywhere on the number line, rather than only initial intervals, that have a positive fraction of members in \(A\).

Kra, Moreira, Richter and Robertson proved a result for the broader class. Their published A proof of Erdős's \(B+B+t\) conjecture, Communications of the American Mathematical Society 4 (2024), 480–494, states in Theorem 1.2(i), page 481:

For every \(A\subseteq\mathbb N\) of positive upper Banach density, there are an infinite \(B\subseteq A\) and \(t\in\mathbb N\) for which the restricted pairwise-sum set of \(B\) is contained in \(A-t\).

Writing \(S\subseteq A-t\) means that \(s+t\in A\) for every \(s\in S\). Thus the theorem's shift has the same sign as the original registered requirement. The authors also emphasize that the diagonal exclusion and the translation cannot simply be dropped.

Why does this settle the upper-asymptotic-density version? If \(\overline d(A)>0\), choose a sequence \(N_k\to\infty\) along which the initial-interval densities converge to some positive number. The intervals \([1,N_k]\) form a Følner sequence because translating them by a fixed \(h\) changes only \(O(h)\) positions out of \(N_k\). Thus \(A\) has positive upper Banach density, so the published theorem applies with precisely \(B\subseteq A\) as required.

The theorem is not merely an assertion about two unrelated infinite sets \(B,C\) with \(B+C\subseteq A\). Nor is it the later density finite sums theorem, which permits \(B\subseteq\mathbb N\) and asks simultaneously for several subset-sum lengths. The \(B\subseteq A\) condition and fixed two-summand constraint identify this particular problem.

At a high level the 2024 paper transports an arithmetic density condition into an ergodic-dynamical system and then extracts an infinite combinatorial pattern. The precise bridge passes through its Theorem 1.4. This is a published proof, not an independently reconstructed or Lean-replayed GCL proof.

GCL's current state

The authenticated original RESULT/1 source return retains its original BOUNDED_TASK_DESIGNED disposition and mathematical qualifications. This source-grounded explanation is editorially reviewed through the exact-hash EC-BULK-005-R8 protected transaction, effective on protected merge. It is not a fresh proof of the attributed mathematical theorem or an internally replayed Lean certificate.

Critical source-scope distinction: Kra–Moreira–Richter–Robertson B+B+t positive upper density result, B subset A, distinct pairs, not B+B including diagonal.

What remains

The bounded verification task is to put the exact theorem and the registered conjecture side by side and check every logical component: universal quantifier on \(A\), positive density, infinite \(B\), \(B\subseteq A\), distinct indices, and single fixed positive-direction shift. The density implication via Følner intervals should be written as a short self-contained lemma. The modular example should check the shift's necessity without claiming necessity for every set.

The original references [Er75b] and [Er80, p.105] are documented by the modern authors and maintained catalogue, but their historical passages were not independently inspected during this cohort audit. They remain a documentary source-fidelity check, not an objection to the published proof's exact modern theorem statement.

Before editorial acceptance, an accountable source-critical review must confirm the references and example, followed by consolidated protected publication, a real-browser MathJax readback, and a distinct exact-hash V2 editorial admission. No additional Solve campaign is warranted merely because the manuscript is not yet editorially admitted.

Sources and claim boundaries

Authority boundary: an exact match to a published theorem, a checked reduction from upper asymptotic to upper Banach density, and a finite congruence example. The original full proof and external formal kernel certification have not been independently replayed.

Editorial provenance: protected R8 source-critical review. Original issue-level source and research-proof holds remain. Status: REVIEWED explanatory article; full historical theorem proofs, unresolved conjectures and external Lean proofs are not newly certified.

Catalogue and evidence provenance

  • Pinned catalogue source: teorth/erdosproblems at 536408716724.
  • GCL issue: #1927 — 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