On the behavior of arithmetic functions on intervals of fixed length (Q1324881): Difference between revisions

The author defines certain probability measures \(\mu_ n\), which are weakly convergent to a probability measure \(\mu\), as \(n\to\infty\), and which can be applied in order to prove results on the local value distribution of certain arithmetical functions. These functions are those satisfying \(\sum_{q\in Q_ f} 1/q<\infty\), where \(Q_ f=\{ p^ \beta\): \(\exists m\), \(\text{gcd} (m,p^ \beta)=1\), such that \(f(mp^ \beta)\neq f(m)\}\). Then the limit \[ \lim_{N\to\infty} {\textstyle {1\over N}} \sum _{\substack{ m\leq N\\ \#\{n;\;m\leq n<m+k,\;f(n)=y\}=\ell }} 1 \] exists. The limit is computed, for example, for \(f=\omega_ 2\), where \(\omega_ 2(m)= \# \{p^ 2\); \(p^ 2| m\}\), yielding (if \(1\leq s\leq 3\), \(y\geq 0\)) \[ \begin{multlined} \lim_{N\to\infty} {\textstyle {1\over N}} \#\{m\leq N;\;\omega_ 2(m)= \omega_ 2(m+1)= \dots= \omega_ 2 (m+s-1)=y\}\\= \prod (1- {\textstyle {s\over p^ 2}} \sum_{\substack{ m\\ 1\leq\omega(m) \leq y\cdot s}} \mu^ 2(m) \sum_{\substack{ m=d_ 1\dots d_ s\\ \omega(d_ i)=y}} \prod^ s_{i=1} \Bigl( \prod_{p\mid d_ i} {\textstyle {1\over {p^ 2-s}}} \Bigr).\end{multlined} \] {}.