On the eigenstructure of sparse matrices related to the prime number theorem (Q2332445)

Let \(R_{n}\) be the \(n\times n\) matrix whose \((i,j)\)-th entry is \(1\) if \(j=1\) or if \(i\mid j\) and is \(0\) otherwise. It is an elementary exercise to show that \(\det R_{n}=\sum_{k=1}^{n}\mu(k)\) where \(\mu(k)\) is the Möbius function (see [\textit{R. Redheffer}, ISNM, Int. Ser. Numer. Math. 36, 213--216 (1977; Zbl 0363.65062)]). On the other hand it is a classical theorem that the Mertens function \(M(n):=\sum_{k=1}^{n}\mu(k)\) satisfies \(M(n)=o(n^{1/2+\varepsilon})\) for each \(\varepsilon>0\) if and only if the Riemann hypothesis (RH) is true, so there is an intriguing relationship between \(R_{n}\) and RH. This has led to papers by various authors investigating the spectral properties of \(R_{n}\) and a related matrix \(\mathcal{R}_{n}\). The main theorem of the present paper shows that the largest and smallest eigenvalues of \(\mathcal{R}_{n}\) are asymptotically of the form \(r_{\pm}=\pm\sqrt{\pi(n)}+\frac{1}{2}\log\log n+O(1)\).