Skip to content

Erdős Problem #421

Editorial depth: Reviewed explanatory article — evidence qualified.
Snapshot classification: solved (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

Choose a strictly increasing sequence of positive integers \(1\leq d_1<d_2<\cdots\). For any indices \(u\leq v\), multiply the consecutive entries from \(d_u\) through \(d_v\). Erdős Problem 421 asks whether the sequence can be chosen so that every such consecutive-block product is different, while its selected integers have natural density one. If \(D\) is the selected set, the second requirement is

\[ \lim_{x\to\infty}\frac{|D\cap[1,x]|}{x}=1. \]

This permits deleting an asymptotically negligible proportion of the positive integers but not a fixed positive fraction. The two demands are in tension: removing many numbers makes products easier to distinguish, whereas density one allows very few omissions. The requirement concerns all pairs of nonempty index intervals, not merely consecutive intervals of a given length. A finite prefix cannot establish density one because the latter is an asymptotic property.

A small example

The sequence \((2,3,6)\) fails: the product of its first two terms equals \(2\cdot3=6\), the product of its final one-term block. Thus index ranges \([1,2]\) and \([3,3]\) collide despite having different lengths. This is exactly the kind of equality the conjectured construction would have to avoid.

By contrast, \((2,3,5,7,11)\) has fifteen nonempty consecutive blocks and all fifteen products are different. This was verified by exact multiplication of every interval. Indeed, if we take all primes as our infinite sequence, unique prime factorization makes the consecutive-block products distinct: each product determines the precise prime factors that occurred and hence the block. But prime numbers have natural density zero, so this does not answer the actual question.

If every selected entry is at least \(2\), a strict extension of one block multiplies its product by at least \(2\) and cannot preserve it. In that setting two blocks that are properly nested cannot collide; the interesting finite collisions must involve nonnested intervals. This elementary reduction helps with exhaustive searches but is much weaker than a proof that dense infinite sequences exist.

What is known

The original statement appears in P. Erdős and R. L. Graham, Old and New Problems and Results in Combinatorial Number Theory (1980), printed page 84 of the scanned book. The original source asks the exact density-one consecutive-product question and gives no full construction there.

The maintained Erdős Problems #421 page, observed in the source inquiry on 13 October 2026, labels the problem OPEN. It reports Selfridge's construction of sequences of density greater than \(1/e-\varepsilon\) for any prescribed positive \(\varepsilon\). This is meaningful positive-density progress, but its numerical threshold is strictly below one. It does not provide the required density-one set. Related Problem #786 discusses the construction and adjacent ideas.

The maintained page also indicates that a solution has been claimed in user comments. Comments are not automatically adjudicated mathematical proofs. The page retains OPEN status, and this bounded source inquiry has not located an independently verified published density-one construction. This is a material evidence boundary: a reported solution claim should trigger critical examination, not automatic promotion to a theorem.

GCL's current state

The protected original source issue #1691 pins the catalogue at teorth/erdosproblems commit 536408716724a2402bdfa7c1d764bf83d64a8c89, file data/problems.yaml, Git blob 4e60263a80af6207aca494d2e1388e44fa372889. That frozen entry describes the problem as solved as of August 2025, but the currently maintained mathematical page describes it as open. This discrepancy is a substantive source-status conflict. It would be incorrect either to erase the frozen record or to treat its label as proof of an unstated theorem.

The original issue also identifies the pinned FormalConjectures 421.lean, Git blob e904247c666facad9b828a645dce343e09c966c8. It formulates an existence assertion for a strictly monotone natural-number sequence whose range has density one and whose products over all inclusive index intervals are injective. The definition is useful for checking quantifiers: two distinct intervals must never have the same product. The theorem is tagged research open, its answer parameter is not a proved determination, and its proof body is literally by sorry. No new Lean compilation or mathematical certification is claimed here.

The original authenticated RESULT/1 reference is authenticated original RESULT/1. This source-linked article is separate from the V2 editorial admission ledger and does not confer accepted coverage.

What remains

The useful next task is a focused source-fidelity audit of the purported resolution. Inspect all relevant maintained-page comments, identify any complete underlying proof or paper, and demand a verifiable count showing that the proposed sequence has density exactly one. A construction with density \(1/e\) or even a fixed density \(0.99\) would not answer the question. Distinguish a user assertion, a correct finite pattern and an independently reviewable universal construction.

A separate bounded arithmetic fixture should enumerate every interval product for increasing finite test sequences. Begin with \((2,3,6)\) and \((2,3,5,7,11)\); return exact pairs of colliding blocks when they exist, and an exact count of unique products otherwise. Also verify the observation about strict interval containment when all terms exceed one. Acceptance is an exact replayable collision test and a dated source-concordance report. Falsification includes missed collisions, claiming density from a finite prefix, or promoting a commented solution without proof of the density-one bound. Stop at that source and arithmetic interface if the full mathematical claim cannot be established.

Sources and claim boundaries

Historical primary question: Erdős–Graham 1980, printed p.84. Maintained status and reported Selfridge bound: Problem #421, accessed 2026-10-13. Related source: Problem #786. Frozen original assignment: GCL issue #1691. Catalogue: teorth archive. Formal statement: pinned 421.lean.

The explicit finite products are independently verified. The Selfridge lower-density construction is attributed to the maintained source and requires primary-proof comparison for full fidelity. The conflict between frozen solved metadata and live open status remains unresolved. No novel construction, accepted proof, mathematical certification or editorial admission is asserted.

Catalogue and evidence provenance

  • Pinned catalogue source: teorth/erdosproblems at 536408716724.
  • GCL issue: #1691 — protected issue/assignment identity; its labels and results are live, not reproduced automatically here.
  • GCL triage: OUTSIDE_STRICT_OPEN_TRIAGE.
  • 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