S.G. Akl and S.D. Bruda, Improving a solution's quality through parallel processing, The Journal of Supercomputing , Vol. 19, No. 2, 2001, pp. 219--231.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....superlinear in n. Evidently, what constitutes a better solution depends on the problem under consideration. Thus, for example, better means closer to optimal for optimization problems [3, 6, 18, 27, 28] more accurate for numerical problems [7] and more secure for cryptographic problems [8]. Different models of computation were used to develop the parallel solutions described in the cited references. They include the linear array [21] the reconfigurable multiple bus model [32] and the parallel random access machine [1] Furthermore, and partly as a consequence of the particular ....

....If the time available for solving the problem, or for making each decision, is restricted, the importance of making the best possible decision at each step is paramount. Examples of computations with these characteristics and algorithms for solving them in real time using SDA are provided in [1] [8], 11] 16] and [22] 28] 5 Computer models In order to study the effect of parallelism on real time state space traversal, we define two distinct models of computers for solving the problem at hand. 5.1 Sequential model The first computer model is sequential: It consists of one processor ....

S.G. Akl and S.D. Bruda, Improving a solution's quality through parallel processing, The Journal of Supercomputing , Vol. 19, No. 2, 2001, pp. 219--231.

....superlinear in n. Evidently, what constitutes a better solution depends on the problem under consideration. Thus, for example, better means closer to optimal for optimization problems [2, 8, 15, 20, 21] more accurate for numerical problems [7] and more secure for cryptographic problems [6]. 2 Different models of computation were used to develop the parallel solutions described in the cited references. They include the linear array, the reconfigurable multiple bus model, and the parallel random access machine. Furthermore, and partly as a consequence of the particular choice of a ....

S.G. Akl and S.D. Bruda, Improving a solution's quality through parallel processing, The Journal of Supercomputing, Vol. 19, No. 2, 2001, pp. 219--231.

....these cases is possible during one time interval. From another perspective, the paper confirms that the classic approach in measuring the performance of a parallel algorithm (namely, by analyzing the speedup achieved) is not appropriate for real time paradigms of the type defined herein (see also [4, 2, 3, 5, 6, 7, 8]) The true superiority of a parallel algorithm over its sequential counterpart, in the real time area, can be fully exposed only if alternative performance measures are employed. These include the generated error and the quality up, as defined in Sections 3 and 4, respectively. 15 It is also ....

S. G. Akl and S. D. Bruda. Improving a solution's quality through parallel processing. The Journal of Supercomputing, 19:219--231, 2001.

....time constraints. The input is not all available at the beginning and arrives during the computation. Output has to be produced before a deadline. Depending on the input data arrival law and the output time constraints, different sub paradigms have been defined and studied for different problems [1, 2, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 25]. The real time maximum flow problem, as defined in this paper, is a variant of the data accumulating paradigm [8] and the data correcting paradigm [12] The initial data undergo small adjustments during the computation. The initial flow network N = G = V; E) c; s; t) together with the already ....

S. G. Akl and S. D. Bruda. Improving a solution's quality through parallel processing. The Journal of Supercomputing, 2001. To appear.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC