Talbot effect for the cubic non-linear Schröedinger equation on the torus (Q2511485)

In this paper the authors investigate the Talbot effect for cubic nonlinear Schrödinger (NLS) equation \[ iu_t+u_{xx}+| u|^2u=0, \quad t\in\mathbb R, \quad x\in\mathbb T\equiv\mathbb R/2\pi \mathbb Z \] with periodic boundary condition \(u(x,0)=g(x)\), where \(g\) is of bounded variation. The main idea here is to generalize the known results of \textit{K. I. Oskolkov} [in: Progress in approximation theory. An international perspective. Proceedings of the international conference on approximation theory, Tampa, South Florida, USA, March 19-22, 1990. New York: Springer-Verlag. 353--402 (1992; Zbl 0815.42003)] and \textit{I. Rodnianski} [Contemp. Math. 255, 181--187 (2000; Zbl 1054.81018)] to be generalized. \textit{H. F. Talbot} [``Facts related to optical science'', Philos. Mag., Ser. 9, No. 56, 401--407 (1836; \url{doi:10.1080/14786443608649032})] shows by an optical experiment the existence of white light passing through a diffraction grating. The images are produced by magnifying lens. It is noticed a sharp focused pattern with a certain periodicity depending on the distance. Denote the Talbot distance \(d = \alpha^2/\lambda \) [\textit{L. Rayleigh}, ``On copying diffraction-gratings, and on some phenomena connected therewith'', Philos. Mag., Ser. 11, No. 67, 196--205 (1881; 10.1080/14786448108626995)], where \(\alpha \) is the spacing of the grating, and \(\lambda \) is the wavelength of the incoming light. \textit{M. V. Berry} and \textit{S. Klein} [J. Mod. Opt. 43, No. 10, 2139--2164 (1996; Zbl 0941.78524)] propose an evolutionary model of the Talbot effect using for this purpose the linear Schrödinger (LS) equation. They show that ``at rational multiples of the Talbot distance overlapping copies of the grating pattern reappear, while at irrational times the images have a fractal nowhere differentiable profile''. \textit{M. V. Berry} [J. Phys. A, Math. Gen. 29, No. 20, 6617--6629 (1996; Zbl 0910.60087)] conjectured that ``for the \(n\)-dimensional linear Schrödinger equation confined in a box the imaginary part \(\Im u(x, t)\), the real part \(\Re u(x, t)\) and the density \(| u(x, t)| ^2\) of the solution is a fractal set with dimension \(D=n+1/2\) for most irrational times.'' Note that in the one-dimensional case one may think that there are ``space slices whose time fractal dimension is \(7/4\) and there are diagonal slices with dimension \(5/4\).'' The main result here is that if \(t/2\pi\) is an irrational number, and for a large class of initial data then the solution \(u(x,t)\) of the above stated NLS equation, is a continuous w.r.t. \(x\). If assumed that \(t/2\pi \) is rational, then the solution is a bounded function with at most countably many discontinuities. In the case when the initial function \(g\) is continuous, then the solution \(u\) is continuous w.r.t. both \(x\) and \(t\). If assumed that \(g\) does not belong to the class \(\cup_{\epsilon >0}H^{1/2+\epsilon }\), then for almost all values of \(t\) either the real part or the imaginary part of the graph of \(u(\cdot ,t)\) has the upper Minkowski dimension \(3/2\). We note that the proof of the main result is too sophisticated. Therefore, the reader may try to prove it in detail.