Periodic aspects of sequences generated by two special mappings (Q2472556)

The authors study \(\beta\)-transformation \(T(x)=\beta x\bmod1\) with rational \(\beta=\frac{q}{p}\), \(2\leq q<p\), \(\gcd(p,q)=1\). Expressing \(n\)th iterates \(T^{(n)}(1)\) as \(p\)-adic expansion \(T^{(n)}(1)=\frac{t(n)_1}{p}+\dots+\frac{t(n)_n}{p^n}\), \(0\leq t(n)_i<p\), they proved that, for every \(k=0,1,2,\dots\), the sequence \(t(n+k)_n\), \(n=1,2,\dots\), is purely periodic, and the length of the period, marked \(m_k\), divides \(m_k|e_k\) -- the multiplicative order of \(q\) modulo \(p^{k+1}\). On the other hand the authors found some integers \(\mu_k\) such that either \(\frac{e_k}{\mu_k}|m_k\) or \(\frac{e_k}{2\mu_k}|m_k\). This implies that if \(k>0\) and every prime factor of \(p\) divides \(\frac{e_k}{e_{k-1}}\), then \(m_k=e_k\). Finally, the authors claim, without proof, that the same periodicity properties there has the sequence \(s(n+k)_n\), \(n=1,2,\dots\), where \(\beta^n\bmod1=\frac{s(n)_1}{p}+\dots+\frac{s(n)_n}{p^n}\). All proofs are elementary.