On the number of nonzero digits in the beta-expansions of algebraic numbers (Q509931)

Let \(\beta\) be a Pisot or Salem number. For any real number \(x\in (0;1)\) we can obtain its beta expansion \(x=\sum_{n=1}^{\infty} t_n(\beta,x)\beta^{-n}\). The coefficients \(t_n(\beta,x)\) can be found by a greedy algorithm. Put \(\nu(\beta,x,N)=\text{card}\{n\in\mathbb{N}:1\leq n\leq N, t_n(\beta,x)\neq 0 \}\). Let \(\xi\in (0;1)\) be an algebraic number with \([\mathbb{Q}(\beta,\xi):\mathbb{Q}(\beta)]=D\). Suppose that there exist infinitely many nonzero digits \(t_n(\beta,\xi)\). Then it is proved that there exist effectively computable positive constants \(C_1(\beta,\xi)\) and \(C_2(\beta,\xi)\) such that \(\nu(\beta,\xi,N)\geq C_1(\beta,x)\left( \frac{N}{\log N} \right)^{1/D}\) for \(N\geq C_2(\beta,\xi)\). An earlier similar result known has exponent \(1/(2D-1)\) instead of \(1/D\). As an application, it is proved that for any real \(z>1\) we have \([\mathbb{Q}(\psi(z,\beta^{-1})):\mathbb{Q}(\beta)]\geq \lceil z\rceil\) where \(\psi(z,\beta^{-1})=\sum_{n=1}^{\infty} \beta^{-\lfloor n^z\rfloor}\).