Kalufya, Ali Ayub M2016-01-082016-01-0819971997http://hdl.handle.net/11693/17913Ankara : Department of Computer Engineering and Information Science and the Institute of Engineering and Science of Bilkent Univ., 1997.Thesis (Master's) -- Bilkent University, 1997.Includes bibliographical references (leaves 82-84).Cataloged from PDF version of article.The solution to the electromagnetic scattering problem may be modelled as an N-body problem. Using this model this work develops a solution that is based on a specific variant of the Fast Multipole algorithm that was proposed by V. Rokhlin[17] and modified furttier by Anderson[3], that is the Fast Multipole Method without multipoles. Because an iterative scheme is used, we also develop an preconditioning algorithm that is especially tailored for the solution of problems that may be modelled using N-body concept. Moreover, in this work parallel computing is employed to improve the solution even further by developing a program that will utilize the above mentioned fast multipole method concept in parallel so as to be able to solve even larger and more interesting real-life problems in a reasonable amount of time and using minimum possible memory space. A parallel version of the fast multipole method is developed and implemented on the Parystec Coignitive Computer 24 node multicomputer using the single program multiple data paradigm for solving the electromagnetic scattering problem in 2 dimensions.xiii, 84 leaves : illustrations, charts ; 30 cm.Englishinfo:eu-repo/semantics/openAccessN-body conceptFast multipole methodBlockwise sparse preconditioningThesisBILKUTUPB039396