Random multiplicative functions: the Selberg-Delange class (Q2064869): Difference between revisions

Let \(\beta\in[1/2,1)\), \(p\) be a generic prime number and \(f_\beta\) be a random multiplicative function supported on the squarefree integers such that \((f_\beta(p))_p\) is an i.i.d. sequence of random variables with distribution \(\mathbb{P}(f(p)=-1)=\beta=1-\mathbb{P}(f(p)=+1)\). In the paper under review, the author is motivated by the behaviour of partial sums of Möbius function and its relation with the Riemann hypothesis, more precisely Wintner's model in this area. The author considers the following questions: Question 1. What can be said about the partial sums \(\sum_{n\leq x}f_\beta(n)\) for \(\beta\in(1/2,1)\)? Do they have square root cancellation as in Wintner's model and as we expect for the Möbius function under the Riemann hypothesis? Question 2. If the partial sums \(\sum_{n\leq x}f_\beta(n)\) are \(O(x^{1-\delta})\) for some \(\delta>0\), almost surely, then can we say something about the partial sums of the Möbius function? The author gives a negative answer for Question 1 by showing that for any integer \(n\geq 1\) and \(\beta=1-1/2^{n+1}\), for any \(\delta>0\), \(\sum_{n\leq x}f_\beta(n)\) is not \(O(x^{1-\delta})\) almost surely. Regarding to Question 2, letting \(\omega(n)\) to be be the number of distinct primes that divide \(n\), he proves that the Riemann hypothesis is equivalent to the following statement: \[ \sum_{n\leq x}(2\beta-1)^{-\omega(n)}f_\beta(n)=O(x^{1/2+\varepsilon}), \] for all \(\varepsilon>0\) and \(x\) sufficiently large with respect to \(\varepsilon\), almost surely, for each \(\beta\in(1/2+1/(2\sqrt{2}),1)\). Hence, the Riemann hypothesis is equivalent to the square root cancellation of the above weighted partial sums of \(f_\beta\).