Нашли опечатку? Выделите ее мышкой и нажмите Ctrl+Enter
Название: Programming the Finite Element Method
Авторы: Smith I., Griffiths D.
The theme of the successful earlier editions has been maintained. A modular programming style, now expressed in Fortran95, facilitates both the consolidation of previous programs into easier-o-use units, and the very important advance into parallel computing environments. Chapter 1 has been extended by a description of the parallelisation strategy adopted. This involves message passing using the de facto standard, MPI, although other possibilities such as OpenMP are addressed. In Chapter 3, iterative equation solution methods have been extended to cover non- symmetric systems by the use of bi-conjugate gradient (BiCG) methods. An element-by-element strategy therefore allows the solution of all large finite element problems on modestly sized computers. From Chapter 4 onwards, programs that were distinct in earlier editions have been consolidated into easier-to-use units within which, for example, problem dimensionality, element type, and so on are selected by the user from inside the same program. Output has been made more concise and readable. Although mesh generation, a topic in itself, has been avoided, programs that interface readily with mesh generation packages have been provided. For the first time results display using PostScript files has been included in most chapters. Exercises for students to attempt on their own have been extended. Therein example programs from Chapters 5 to 11 have been parallelised using element-by-element strategies and MPI. The parallelisation process is largely hidden from the user and the central theme of program conciseness and readability has been maintained. Performance statistics show that efficient use has been made of parallel hardwares ranging from supercomputers to clusters of PCs.