Wiki
Wiki

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

Updated


Source. Definition 5.2 and Theorem 5, printed p. 1747, physical PDF p. 9; the proof runs to printed p. 1748.

Convention

Definition 5.2, which the paper takes from Brunner, Caldwell, Krywaruczenko and Lownsdale: for a positive integer bb, an integer k>1k>1 is a bb-Sierpiński number when gcd⁡(k+1,b−1)=1\gcd(k+1,b-1)=1, kk is not a power of bb, and k⋅bn+1k\cdot b^n+1 is composite for every positive integer nn. The paper records that those authors show there are infinitely many bb-Sierpiński numbers for every base b>1b>1.

Statement

"Let bb be a positive integer such that bb is not a Mersenne number (b+1b+1 is not a power of 2). There exist infinitely many bb-Sierpiński numbers kk such that k⋅bn+1k\cdot b^n+1 has at least three distinct prime divisors for all positive integers nn." (p. 1747)

The hypothesis excludes b=1b=1 and every bb of the form 2j−12^j-1; the paper describes the result as known for b=2b=2 and proves the case b>2b>2.

Proof pointer. For b>2b>2 the proof of Theorem 2 makes the Section 4 covering a (b,1)(b,1)-primitive 33-covering. Each modulus mim_i receives a primitive prime divisor pip_i of bmi−1b^{m_i}-1, and the Chinese Remainder Theorem gives infinitely many kk with $k\cdot b^{r_i}+1\equiv0 \pmod{p_i}$ for every ii, k≡0(modb−1)k\equiv0\pmod{b-1} and k≡1(modb)k\equiv1\pmod b; the last two conditions give the coprimality and non-power clauses (pp. 1747--1748). The argument was followed but not independently checked here.