Skip to content

Erdős Problem #12

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

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 #12 concerns infinite sets of positive integers with a restriction on a smaller element dividing the sum of two larger elements. Its historical Property P forbids a configuration

\[ a\mid b+c,\qquad a<b\le c,\qquad a,b,c\in A. \]

The pinned formal provider instead only forbids such triples when the two larger summands are different: it allows \(b=c\). This is a real mathematical difference, not a harmless change of wording.

The problem has several parts about how numerous such sets can be. Its central remaining question is whether every eligible set has a convergent sum of reciprocals. Published results settling other parts do not settle this one.

A small example

For the finite set \(A=\{3,6\}\), the smaller number 3 divides \(6+6=12\). The historical property rejects this repeated-summand configuration. The frozen formal predicate allows it because its conclusion when divisibility holds is \(b=c\).

The GCL audit even constructs an infinite example that satisfies the frozen predicate but not the historical condition:

\[ A_*=\{3,6\}\cup\{p^2:p>3\text{ prime},\ p\equiv3\pmod4\}. \]

The argument uses residue restrictions and the fact that \(-1\) is not a quadratic residue modulo primes congruent to 3 modulo 4. The reciprocal series for this particular set converges, so it is not a counterexample to the outstanding reciprocal-sum conjecture.

What is known

The three-part formal record distinguishes two density-related statements with known mathematical answers from the unresolved question of convergence of the reciprocal series. Sparse natural density alone does not guarantee reciprocal summability.

A useful exact reduction is to group positive elements into dyadic intervals. Put \(D_j=\#(A\cap[2^j,2^{j+1}))\). Then

\[ \frac12\sum_{j\ge0}\frac{D_j}{2^j} \le \sum_{\substack{a\in A\\a\ge1}}\frac1a \le\sum_{j\ge0}\frac{D_j}{2^j}. \]

Consequently convergence of the reciprocal sum is equivalent to convergence of the dyadic weighted counting series. This elementary criterion does not itself prove convergence for every eligible set.

GCL's current state

Source audit #1281 checks the original-versus-frozen predicate mismatch, gives the infinite countermodel and a finite exact regression, and identifies the unresolved reciprocal-summability boundary. The formal source contains unfilled proof bodies, so it is not a local compiled certification.

What remains

The central task is to establish or refute reciprocal-sum convergence for the precisely stated admissible class. Before importing any historical theorem, the repeated-summand distinction must be resolved: a result for the stronger historical Property P cannot automatically be applied to every set allowed by the weaker frozen predicate.

Sources and claim boundaries

  • Erdős Problems #12: original family of questions and bibliography.
  • GCL source audit #1281: exact countermodel, dyadic criterion and source-semantics caveat; ordinary mathematical analysis, not kernel certification.

Catalogue and evidence provenance

  • Pinned catalogue source: teorth/erdosproblems at 536408716724.
  • GCL issue: #1281 — 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