Erdős Problem #324
Editorial depth: Reviewed explanatory article — evidence qualified.
Snapshot classification: open (not automatically the live result).
Subject areas: number theory, powers, sidon sets.
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 324 asks whether there is one integer-coefficient polynomial whose values behave like a particularly collision-resistant list of integers. More precisely, seek a polynomial \(f\in\mathbb{Z}[x]\) such that whenever \(a<b\) and \(c<d\) are nonnegative integers, the equality
forces \((a,b)=(c,d)\). Thus every sum of two different entries is unique: one can recover the two inputs from their sum. A set with this property is called a Sidon set. The question is existential and global: it concerns every pair of nonnegative inputs, not merely a large finite range or a tail of the sequence.
The maintained Erdős Problems page paraphrases the object as the set of values for \(n\ge 1\). The catalogue statement and the pinned formal statement use nonnegative integers, including \(0\). That indexing difference should not be silently erased when comparing results; a tail theorem and the full problem are different assertions.
A small example
The candidate quartic family discussed in the live forum is \(f_t(x)=x^4+t^3x\). It has an exact collision for every positive integer \(t\):
For a fully concrete check, take \(t=1\). Then \(f_1(23)=279864\), \(f_1(27)=531468\), \(f_1(6)=1302\), and \(f_1(30)=810030\). Hence both sums equal \(811332\), while \((23,27)\ne(6,30)\) and both pairs have increasing, nonnegative entries. This is a reproducible exact-arithmetic disproof of the proposed family, not a disproof of every quartic polynomial. It also illustrates the standard obstruction: one nontrivial equal-sum quadruple is enough to destroy the Sidon property.
What is known
The original source is Erdős and Graham's Old and New Problems and Results in Combinatorial Number Theory, p. 53, as identified by the maintained entry. It calls the problem difficult and suggests \(x^5\) as a likely candidate. Ruzsa's primary paper, “An almost polynomial Sidon sequence,” proves a nearby result: for some real \(\xi\in[0,1]\) and a suitable \(n_0\), the tail set
is Sidon. This is strong evidence for the shape of a construction, but it is not an integer polynomial and does not cover the initial values.
The strongest directly relevant obstruction located in this audit is Dubickas and Novikas, “No cubic integer polynomial generates a Sidon sequence,” Mathematische Nachrichten 294 (2021), 1859–1865. The publisher's primary-paper abstract states that no integer-coefficient polynomial of degree at most three generates a Sidon tail, and that the proof constructs infinitely many solutions of \(f(m)+f(n)=f(r)+f(s)\) in pairwise distinct positive integers. The publisher record dates first publication to 3 October 2021. The live entry also reports that linear and quadratic cases are elementary and that \(x^4\) fails by a classical result of Swinnerton-Dyer, “\(A^4+B^4=C^4+D^4\) Revisited,” Journal of the London Mathematical Society 43 (1968), 149–151.
These results leave the central construction question unresolved. In particular, the cited sources do not establish that \(x^5\) works, nor do they settle whether some quartic polynomial works. The live forum's geometric discussion is explicitly speculative, and the finite quartic-family collision above is a check of a particular family rather than a published classification.
GCL's current state
The frozen catalogue record at teorth/erdosproblems commit 536408716724a2402bdfa7c1d764bf83d64a8c89 records Problem 324 as open, last updated 31 August 2025, with formal status unformalized and statement-formalized metadata yes. Its exact Git blob is 4e60263a80af6207aca494d2e1388e44fa372889; the Programme registry class is PROTECTED_READY_BACKLOG.
The pinned FormalConjectures file at commit 85f863718beeec7b58a3a1926ee92e3472bc2020 has blob df78bf69d0ce3828d73a67fb872136f4c65e5318. Its main theorem quantifies over \(f:\mathbb{Z}[X]\) and injectivity on pairs of naturals with \(a<b\), matching the core quantifiers up to the usual coercion for evaluation. It also contains a separate quintic variant. Both theorem bodies end in sorry: presence of a formal statement is not a proof, replay, or certification.
The source-audit issue is MATH-PROGRAMME #1594, assignment ERDOS-CATALOGUE-0324-S01. It had no comments or existing RESULT/1 at audit time. The current issue body hashes to b0eaf7ff95c907fee875c7cde41dcce75523a15778e5bfe54f950bcefa747c01, while the protected registry records a2a496accbf403e0ffa196c54f2beec2691e5585f0a9070ccfb75a3807de67c1; that documentary drift is recorded, not treated as mathematical evidence.
What remains
The next useful bounded question is narrower than the full problem: decide a specified candidate quartic or quintic polynomial by producing either a symbolic collision family or a proof that its equal-sum equation has only the trivial pair symmetries. A computational search may find a collision and can falsify a candidate, but a finite collision-free range cannot prove an infinite Sidon property. Any proposed successor should fix the domain, distinguish ordered from unordered pairs, and state whether it targets the full sequence or only a tail.
For \(x^5\), the missing object is an all-input proof that \(a^5+b^5=c^5+d^5\) with \(a<b\) and \(c<d\) implies equality of pairs. For general quartics, the missing object is either one explicit polynomial with a proof of that implication or a theorem ruling out all integer quartics. Neither is supplied here.
Sources and claim boundaries
The maintained entry is erdosproblems.com/324, last edited 11 April 2026 when read. Primary references are Erdős–Graham (1980), Ruzsa (2001), Dubickas–Novikas (2021), and Swinnerton-Dyer (1968). The Dubickas–Novikas publisher page and abstract were readable, but the article's full text was not available in this audit environment; proof details beyond the abstract are therefore not independently checked here.
The original issue is the inquiry anchor. Original SOURCE RESULT/1 citation: authenticated original RESULT/1. This article is documentary source reconciliation only. It supplies no GCL mathematical certification, and accepted coverage is decided by the V2 ledger rather than by this manuscript alone.
Catalogue and evidence provenance
- Pinned catalogue source: teorth/erdosproblems at
536408716724. - GCL issue: #1594 — 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.