Skip to content

Erdős problem 291

More generally, if the leading digit of nn in base pp is p−1p-1 then p∣(an,Ln)p\mid (a_n,L_n). There is in fact a necessary and sufficient condition: a prime p≤np\leq n divides (an,Ln)(a_n,L_n) if and only if pp divides the numerator of 1+⋯+1k1+\cdots+\frac{1}{k}, where kk is the leading digit of nn in base pp. This can be seen by writing an=Ln1+⋯+Lnna_n = \frac{L_n}{1}+\cdots+\frac{L_n}{n} and observing that the right-hand side is congruent to 1+⋯+1/k1+\cdots+1/k modulo pp. (The previous claim about p−1p-1 follows immediately from Wolstenholme's theorem.)

Sources

The source's own material, retained exactly.

8 retained statements36c6d0fef71a

Open selected source

FormalConjectures/ErdosProblems/

291.lean

Retained formal statement4 of 8

This leads to a heuristic prediction (see for example a preprint of Shiu [Sh16]) of ≍xlog⁡x\asymp\frac{x}{\log x} for the number of n∈[1,x]n\in [1,x] such that (an,Ln)=1(a_n,L_n)=1.

FormalConjectures/ErdosProblems/291.leanErdos291.erdos_291.variants.shiu_heuristic_asymptotic1 lineExact file
(fun x => ↑{n ∈ Finset.Icc 1 x | (Erdos291.a n).gcd (Erdos291.L n) = 1}.card) =Θ[Filter.atTop] fun x => ↑x / Real.log ↑x
OpenStatement only, no proof