A parallel FDTD algorithm for the analysis of optical data storage systems: Master's thesis

The need for information storage is increasing at an explosive rate, fueled by global interconnection networks, miniaturized mobile devices, and multimedia requirements for text, images, video and audio. To meet these needs, new storage methods in magnetic and optical data storage are extensively studied; our attention is confined to optical data storage. This thesis presents a brief survey of Finite Difference Time Domain (FDTD) method, which is widely used to estimate the viability of new optical data storage methods. When the light interacts with structural elements comparable in size to the wavelength of the incident light, as in optical data storage, it is not permissible to invoke assumptions of scalar diffraction theories. The solution has to be sought via numerical methods based on the Maxwell's equations. The FDTD method approximates continuous time and space derivatives of Maxwell's equations in a spatial grid by finite difference operators. This leads to an algorithm, which provides reliable solutions of field distributions and is applicable for a wide range of problems of computational electrodynamics. The disadvantages of the FDTD method are its relatively high memory requirements and long computational times. The available memory limits the maximum size of a computational domain that can be simulated, while the long computational times restrict the use of the FDTD method for 'what-if' simulations. These limitations can be significantly reduced by using parallel computing. As a part of this thesis, a three-dimensional parallel FDTD algorithm based on the MPI (Message Passing Interface) library is introduced for Beowulf cluster systems, where many PCs are connected to each other by a network. The applicability of the developed FDTD algorithm for purposes of optical data storage is proved by investigating the interaction of a Gaussian laser beam and a DVD-RAM disk. The electric field distribution in the vicinity of the data layer is observed to be clearly polarization dependent.