arXiv++ Combinatorics

Several algorithms for similarity search employ seeding techniques to quickly discard very dissimilar regions. In this paper, we study theoretical properties of lossless seeds, i.e., spaced seeds having full sensitivity. We prove that lossless seeds coincide with languages of certain sofic subshifts, hence they can be recognized by finite automata. Moreover, we show that these subshifts are fully given by the number of allowed errors k and the seed margin l. We also show that for a fixed k, optimal seeds must asymptotically satisfy l ~ m^(k/(k+1)).

The $m$-bonacci word is a generalization of the Fibonacci word to the $m$-letter alphabet $\mathcal{A} = {0,...,m-1}$. It is the unique fixed point of the Pisot--type substitution $ \varphi_m: 0\to 01, 1\to 02, ..., (m-2)\to0(m-1), and (m-1)\to0$. A result of Adamczewski implies the existence of constants $c^{(m)}$ such that the $m$-bonacci word is $c^{(m)}$-balanced, i.e., numbers of letter $a$ occurring in two factors of the same length differ at most by $c^{(m)}$ for any letter $a\in \mathcal{A}$. The constants $c^{(m)}$ have been already determined for $m=2$ and $m=3$. In this paper we study the bounds $c^{(m)}$ for a general $m\geq2$. We show that the $m$-bonacci word is $(\lfloor κm \rfloor +12)$-balanced, where $κ\approx 0.58$. For $m\leq 12$, we improve the constant $c^{(m)}$ by a computer numerical calculation to the value $\lceil\frac{m+1}{2}\rceil$.

We derive an explicit formula for the Abelian complexity of infinite words associated with quadratic Parry numbers.