Short and long averages of a bounded multiplicative function are effectively related

You need to know: Notation {\mathbb N} for the set of positive integers, coprime integers.

Background: A function f:{\mathbb N} \to [-1,1] is called multiplicative if f(1)=1 and f(ab)=f(a)f(b) holds for all coprime a,b\in {\mathbb N}.

The Theorem: On 19th January 2015, Kaisa Matomäki and Maksym Radziwiłł submitted to arxiv a paper in which they proved the existence of absolute constants B,C (one can take B=20000) such that for any multiplicative function f:{\mathbb N} \to [-1,1], for any 2\leq h \leq X, and any \delta>0, inequality \left|\frac{1}{h}\sum\limits_{x\leq n \leq x+h}f(n) - \frac{1}{X}\sum\limits_{X\leq n \leq 2X}f(n) \right|\leq \delta + B\frac{\log\log h}{\log h} folds for all but at most CX\left(\frac{(\log h)^{1/3}}{\delta^2h^{\delta/25}} + \frac{1}{\delta^2(\log X)^{1/50}}\right) integers x\in[X,2X].

Short context: Many functions of central importance in number theory (for example, the Möbius function and the Liouville function) are multiplicative. For many applications, it is important to estimate average value \frac{1}{h}\sum\limits_{x\leq n \leq x+h}f(n) of such function in short intervals of length h. The Theorem states that this is approximately equal to the average \frac{1}{X}\sum\limits_{X\leq n \leq 2X}f(n) over “long” interval [X,2X], which is much easier to estimate. This has a lot of applications, some of them are derived in the same paper.

Links: Free arxiv version of the original paper is here, journal version is here.

Go to the list of all theorems

Leave a comment