MathDB
canonical prime factorisation and iterated function

Source: Vietnam TST 1991 for the 32nd IMO, problem 5

June 25, 2005
functioninductionnumber theory unsolvednumber theory

Problem Statement

For every natural number nn we define f(n)f(n) by the following rule: f(1)=1f(1) = 1 and for n>1n>1 then f(n)=1+a1p1++akpkf(n) = 1 + a_1 \cdot p_1 + \ldots + a_k \cdot p_k, where n=p1a1pkakn = p_1^{a_1} \cdots p_k^{a_k} is the canonical prime factorisation of nn (p1,,pkp_1, \ldots, p_k are distinct primes and a1,,aka_1, \ldots, a_k are positive integers). For every positive integer ss, let fs(n)=f(f(f(n)))f_s(n) = f(f(\ldots f(n))\ldots), where on the right hand side there are exactly ss symbols ff. Show that for every given natural number aa, there is a natural number s0s_0 such that for all s>s0s > s_0, the sum fs(a)+fs1(a)f_s(a) + f_{s-1}(a) does not depend on ss.