Erdős Problem #673
Editorial depth: Reviewed explanatory article — evidence qualified.
Snapshot classification: proved (not automatically the live result).
Subject areas: number theory, divisors.
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
Take an integer \(n\), write all its positive divisors in increasing order, and compare each divisor with the next one. Do these successive ratios typically add up to something large?
Erdős Problem #673 concerns
The original questions ask whether \(G(n)\) tends to infinity for almost all integers \(n\), and about the asymptotic behavior of its summatory function. The maintained catalogue lists the density-one assertion as PROVED. These questions must not be collapsed: proving a typical-growth statement does not produce a sharp mean-value formula.
A small example
For \(n=12\) the divisors are \(1,2,3,4,6,12\). Their successive ratios sum to
For a prime \(p\), however, there are only two divisors, \(1,p\), so \(G(p)=1/p\). Therefore \(G(n)\) does not tend to infinity along every sequence of integers. The important phrase is almost all, meaning the exceptional integers form a set of natural density zero.
What is known
There is a useful exact inequality. Choose any divisor \(m>1\) of \(n\). For every \(t\mid n/m\), both \(t\) and \(mt\) divide \(n\), and \(t<mt\). If \(t^+\) denotes the next divisor of \(n\) after \(t\), then
Different divisors \(t\mid n/m\) correspond to different summands, so
The upper bound follows because every successive ratio is strictly less than one. For even \(n\), taking \(m=2\) and noting \(\tau(n/2)\geq\tau(n)/2\) gives
A source-precision correction. The maintained catalogue attributes the lower inequality to Terence Tao and prints it for 'any divisor \(m\mid n\)'. The restriction \(m>1\) is essential: at \(m=1\) and \(n=5\), the displayed inequality would falsely claim \(\tau(5)=2\leq G(5)=1/5\). The GCL audit flags the literal discrepancy and gives the valid elementary argument. This is a qualification to an auxiliary statement, not evidence that the main proved density-one result is false.
Why the sum usually grows. If a prime \(p\) divides \(n\), the same argument gives
For any fixed large cutoff \(M\), the proportion of integers divisible by at least one prime \(p\leq M\) is
which approaches one as \(M\) increases. Classical number theory also shows that \(\tau(n)\) tends to infinity in natural density one. Thus, among the overwhelming majority of integers, one can choose a small prime factor \(p\leq M\) and infer a large lower bound for \(G(n)\).
More formally, for every fixed threshold \(T\), the upper density of integers with \(G(n)\leq T\) tends to zero. This establishes the almost-all conclusion, without claiming any convergence for every single integer.
The maintained record also attributes to Erdős and Tenenbaum the existence of a continuous distribution function for the normalized ratio \(G(n)/\tau(n)\). That assertion is historical attribution pending full original-paper verification; it should not be rephrased as a sharp formula for \(\sum_{n\leq X}G(n)\) without a separate theorem.
GCL's current state
The authenticated original issue #1944 source return independently proves the corrected divisor-ratio lemma and supplies the prime and \(n=12\) finite checks. Its disposition is BOUNDED_TASK_DESIGNED. The original [Er79e], [Er82e] and the Erdős–Tenenbaum distribution paper remain to be directly inspected.
This is an unaccepted draft of an established result, not an assertion of a novel density theorem or a verified sharp mean-value asymptotic.
What remains
Verify the historical source's exact quantifiers and identify where the auxiliary \(m=1\) omission arose. Recover the primary Erdős–Tenenbaum statement, its normalization and its limiting distribution theorem. Then separately determine what published results, if any, yield a precise leading term or error for the summatory function.
Any finite divisor enumeration checks only arithmetic fixtures. It cannot replace the density-one sieve/normal-order argument. Editorially, the article still needs accountable source-critical review, protected merge, deployed MathJax verification and separate V2 admission.
Sources and claim boundaries
-
Maintained Erdős Problem #673, PROVED density-one status and historical discussion.
-
Authenticated GCL #673 original issue report, explicit m>1 correction and elementary derivation.
-
Classical distribution of small prime factors and normal-order theory for \(\tau(n)\): invoked, not re-proved.
-
Erdős [Er79e], [Er82e] and Erdős–Tenenbaum: original pages not independently inspected.
Boundary: corrected elementary lemma and sourced established density-one conclusion, no sharp summatory formula, independent complete historical proof check, Lean certification, protected publication, deployed mathematical rendering or editorial admission.
Catalogue and evidence provenance
- Pinned catalogue source: teorth/erdosproblems at
536408716724. - GCL issue: #1944 — 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.