Expected supremum of a random linear combination of shifted kernels (Q692622)

The quantity \[ S_n := \sup_{t\in \mathbb{R}} \left| \sum_{k=1}^{n} a_k \frac{\sin \pi(t-k)}{\pi(t-k)}\right| \] is considered, where \((a_k)_{k=1}^\infty\) are random variables. It is proved that, if these random variables are Gaussian, then \(\operatorname{E}(S_n)\sim \sqrt{\log n}\). Moreover, if the random variables are symmetric, then \(\operatorname{E}(S_n)\sim \log\log n\). It was already known that, for a uniformly bounded orthonormal system, \(\operatorname{E}(S_n)\sim \sqrt{n\log n}\).