Document Type


Date of Original Version



We compare theoretical, experimental, and computational approaches to random rough surfaces. The aim is to produce rough surfaces with desirable correlations and to analyze the correlation functions extracted from the surface profiles. Physical applications include ultracold neutrons in a rough waveguide, lateral electronic transport, and scattering of longwave particles and waves. Results provide guidance on how to deal with experimental and computational data on rough surfaces. A supplemental goal is to optimize the neutron waveguide for GRANIT experiments. The measured correlators are identified by fitting functions or by direct spectral analysis. The results are used to compare the calculated observables with theoretical values. Because of fluctuations, the fitting procedures lead to inaccurate physical results even if the quality of the fit is very good unless one guesses the right shape of the fitting function. Reliable extraction of the correlation function from the measured surface profile seems virtually impossible without independent information on the structure of the correlation function. Direct spectral analysis of raw data rarely works better than the use of a "wrong" fitting function. Analysis of surfaces with a large correlation radius is hindered by the presence of domains and interdomain correlations.