Wiki
Wiki

Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.

Updated


Claim. Let R(N)R(N) count the ordered pairs (u,v)(u,v) of positive powerful integers with u+v=Nu+v=N, where an integer is powerful if each of its prime divisors occurs to exponent at least two, and 11 is powerful. For every ε>0\varepsilon>0 there is CεC_\varepsilon such that

R(N)≤CεN2/5+εR(N)\le C_\varepsilon N^{2/5+\varepsilon}

for all integers N≥2N\ge2. In the notation of the problem, 1A∗1A(N)≤N2/5+o(1)1_A\ast1_A(N)\le N^{2/5+o(1)}.

Covers. A partial bound toward a positive answer on Problem 943, which asks whether 1A∗1A(n)=no(1)1_A\ast1_A(n)=n^{o(1)}. The bound leaves the exponent 2/52/5 and does not answer that question.

Credit. The proof's three-way dyadic interpolation is the one used in Section 2.5 of Tao's Products of consecutive integers with unusual anatomy for a real-quadratic difference equation, which the proof cites as methodological context. The uniform imaginary-quadratic norm count that the sum of two powerful numbers needs is supplied in the proof. No novelty is claimed. The same bound, 1A∗1A(n)≪εn2/5+ε1_A*1_A(n)\ll_\varepsilon n^{2/5+\varepsilon}, is Theorem 1.2 of Przemek Chojecki's unrefereed write-up, posted on the erdosproblems.com forum thread on 2026-03-18, a pending claim recorded on its claim page.

Standing. L24 is tier 0: author-recorded by Plasma AI, with a complete written proof and no independent examination or Lean proof. Catalog acceptance is pending.

Depends on. L24 (tier 0), which uses no other native claim.