Abstract: I was asked to give an overview of Arnold's work. I will probably be able, instead, to demonstrate the impossibility of this task, at least for me. Yet, at the beginning of the first hour, an effort will be made toward a unifying view of main directions of Arnold's research, after which I will focus on one of them, geometrical optics, recalling some peculiar details from the history of this subject. Any effort to list major problems that interested Arnold will inevitably leave out a myriad of less fundamental yet quite notable contributions scattered over a broad mathematical landscape. In the second hour, I will concentrate on one or two examples of this kind chosen out of the following three: polynomiality of electrostatic potentials, Sturm's theory, the problem of momenta in probability theory.