Erdős Problem #9
Editorial depth: Reviewed explanatory article — evidence qualified.
Snapshot classification: open (not automatically the live result).
Subject areas: number theory, additive basis, primes.
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 #9 concerns odd integers that cannot be expressed as a prime plus two powers of two. More precisely, let
The exponents may be zero or equal. The question is whether
The denominator counts all integers, not just odd ones. Positive upper natural density means that some fixed positive proportion of integers are exceptions along arbitrarily large cutoffs. It is stronger than saying the set is infinite. The question remains open in the October 2026 source record.
A small example
Exponent zero matters:
The numbers 1 and 3 are exceptions because the smallest possible sum is \(2+1+1=4\). The GCL exact arithmetic audit enumerated the odd integers through 200,000 and found exactly these two exceptions. The observed exceptional fraction among all 200,000 positive integers is \(2/200000=10^{-5}\). This is a finite sample, not an asymptotic density estimate.
There are three parity possibilities for a representation of an odd \(n\): (1) an odd prime \(p=n-2\) with \(k=\ell=0\); (2) an odd prime and both exponents positive; or (3) \(p=2\), exactly one zero exponent, giving \(n=3+2^j\) with \(j\ge1\). Ignoring zero exponents would falsely classify 5 as exceptional.
What is known
Roger Crocker's On the sum of a prime and of two powers of two, Pacific Journal of Mathematics 36 (1971), 103–107, proves an infinitude statement with positive exponents. It cannot be transferred without qualification to the present definition that permits zero exponents.
Hao Pan, On the integers not of the form \(p+2^a+2^b\), Acta Arithmetica 148 (2011), 55–61, Theorem 1.1 (printed p. 56), gives a stronger quantitative result. His proof (printed p. 59) permits \(0\le a\le b\) and yields, for some \(C>0\),
where \(\log_j\) denotes an iterated logarithm. The source's bound concerns a subset excluding prime-power summands, hence also lower-bounds \(A\). In particular the exception count grows at least as fast as \(X^{1-\varepsilon}\) (up to an epsilon-dependent multiplicative constant) for each fixed \(\varepsilon>0\) and sufficiently large \(X\).
This does not prove a positive density: the lower-bound expression divided by \(X\) tends to zero. Nor does the small finite sample refute Pan's theorem.
GCL's current state
GCL source audit #1278 records the exact zero-inclusive pinned formal statement, checks its parity decomposition, and independently verifies 99,998 finite representation witnesses for odd integers from 5 to 199,999. The source packet provides the canonical witness digest and primary-source locations. These checks are bounded arithmetic, not proof certification of an asymptotic claim.
What remains
To establish positive upper density one must find \(c>0\) and an unbounded sequence \(X_j\) with \(\#(A\cap[1,X_j])\ge cX_j\). To refute it one must show the upper density is zero. Neither Pan's nearly linear lower bound nor bounded enumeration decides the alternatives.
This explanatory article is not a GCL mathematical certification or a resolution of Erdős #9. Publication acceptance is governed separately from mathematical research status.
Sources and claim boundaries
- Erdős Problems #9, history: public statement and dated status, not a proof.
- Crocker (1971), publisher facsimile, Theorem I, p. 103: positive-exponent infinitude theorem.
- Pan (2011), publisher PDF, Theorem 1.1, p. 56, and zero-inclusive proof, p. 59: quantitative theorem; the deep analytic proof was not reproduced here.
- GCL #9 source and numerical audit: statement normalization and finite checks, not independent theorem certification.
Catalogue and evidence provenance
- Pinned catalogue source: teorth/erdosproblems at
536408716724. - GCL issue: #1278 — 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.