In this work, we present and analyze the use of a reconfigurable job scheduling simulator called RJSSim as an aid tool for parallel processing learning. This software is a functional and performance Java-based simulat...
详细信息
This paper introduces a new mapping of geometrical transformation on the MorphoSys (M1) reconfigurable computing (RC) system. New mapping techniques for some linear algebraic functions are recalled. A new mapping for ...
详细信息
This paper introduces a new mapping of geometrical transformation on the MorphoSys (M1) reconfigurable computing (RC) system. New mapping techniques for some linear algebraic functions are recalled. A new mapping for geometrical transformation operations is introduced and their performance on the M1 system is evaluated. The translation and scaling transformation addressed in this mapping employ some vector vector and vector scalar operations [6 and 7]. A performance analysis study of the M1 RC system is also presented to evaluate the efficiency of the algorithm execution. Numerical examples were simulated to validate our results, using the MorphoSys mULATE program, which emulates M1 operations.
Convergence results for the restricted multiplicative Schwarz (RMS) method, the multiplicative version of the restricted additive Schwarz (RAS) method for the solution of linear systems of the form Ax = b, are provide...
详细信息
Convergence results for the restricted multiplicative Schwarz (RMS) method, the multiplicative version of the restricted additive Schwarz (RAS) method for the solution of linear systems of the form Ax = b, are provided. An algebraic approach is used to prove convergence results for nonsymmetric M-matrices. Several comparison theorems are also established. These theorems compare the asymptotic rate of convergence with respect to the amount of overlap, the exactness of the subdomain solver, and the number of domains. Moreover, comparison theorems are given between the RMS and RAS methods as well as between the RMS and the classical multiplicative Schwarz method.
This work addresses the issue of finding the programming procedure that results to the fastest implementation of the core calculations of the model predictive control (MPC) algorithms that are amenable to parallel pro...
详细信息
This work addresses the issue of finding the programming procedure that results to the fastest implementation of the core calculations of the model predictive control (MPC) algorithms that are amenable to parallel processing on a real-time multiprocessing system. Three concurrent programming procedures were considered for the MPC implementation, each one consisting of a number of tasks configured to a linear array, mesh and tree parallel architectures. Both theoretical analysis and measurements, taken from running these procedures on a four-processor computer platform, indicate that the procedure of the linear array architecture presents the best speed-up ratio. (C) 2003 Elsevier Science B.V. All rights reserved.
This paper presents an overview of different methods proposed in the last several decades for computing the inverse of a polynomial matrix, concentrating on Buslowicz's algorithm. A detailed description of Buslowi...
详细信息
In the data-accumulating paradigm, inputs arrive continuously in real time, and the computation terminates when all the already received data are processed before another datum arrives. Previous research states that a...
详细信息
In the data-accumulating paradigm, inputs arrive continuously in real time, and the computation terminates when all the already received data are processed before another datum arrives. Previous research states that a constant upper bound on the running time of a successful algorithm within this paradigm exists only for particular forms of the data arrival law. This contradicts our recent conjecture that those problems that are solvable in real time are included in the class of logarithmic space-bounded computations. However, we prove that such an upper bound does exist in fact in both the parallel and sequential cases and for any polynomial arrival law, thus strengthening the mentioned conjecture. Then, we analyze an example of a noncontinuous data arrival law. We find similar properties for the sorting algorithm under such a law, namely the existence of an upper bound on the running time, suggesting that such properties do not depend on the form of the arrival law. (C) 2003 Elsevier Science B.V. All rights reserved.
The dynamic lattice liquid model was implemented as a simulation algorithm for simple liquids. This model works correctly for the density factor rho=1 and becomes a base for a parallel algorithm. Such an algorithm tak...
详细信息
The dynamic lattice liquid model was implemented as a simulation algorithm for simple liquids. This model works correctly for the density factor rho=1 and becomes a base for a parallel algorithm. Such an algorithm takes into account coincidences of elementary molecular motion attempts resulting in local cooperative structural transformations. In this paper general conditions for simulating simple liquids on various lattices are presented. Interdiffusion coefficients were directly monitored as changes of composition profiles with time for both athermal and interacting binary (AB) mixtures. In the athermal case, the coefficients of self-diffusion and interdiffusion are equal. For interacting binary mixtures the interdiffusion exhibits a thermally activated character. (C) 2003 American Institute of Physics.
The problem of finding possible limit reliability functions of basic systems is well known. For instance Gnedenko [Ann. Math. 44 (1943) 432-453] has found a class of possible limit reliability functions for series and...
详细信息
The problem of finding possible limit reliability functions of basic systems is well known. For instance Gnedenko [Ann. Math. 44 (1943) 432-453] has found a class of possible limit reliability functions for series and parallel systems. Smirnov [Trudy Matem Inst. im. V;A. Stelkova, 1949] has solved the problem for "k-out-of-n" systems. In this paper these two areas are brought together and a class of limit reliability functions of homogeneous series-"k-out-of-n" systems is presented. Moreover some applications are given as well. (C) 2002 Elsevier Science Inc. All rights reserved.
We present and analyze efficient new algorithms for generating a random variate distributed according to a dynamically changing set of N weights. The base version of each algorithm generates the discrete random variat...
详细信息
We present and analyze efficient new algorithms for generating a random variate distributed according to a dynamically changing set of N weights. The base version of each algorithm generates the discrete random variate in O(log* N) expected time and updates a weight in O(2(log)* (N)) expected time in the worst case. We then show how to reduce the update time to O(log* N) amortized expected time. We finally show how to apply our techniques to a lookup-table technique in order to obtain expected constant time in the worst case for generation and update. We give parallel algorithms for parallel generation and update having optimal processor-time product. Besides the usual application in computer simulation, our method can be usedto perform constant-time prediction in prefetching applications. We also apply our techniques to obtain an efficient dynamic algorithm for maintaining an approximate heap of N elements, in which each query is required to return an element whose value is within an epsilon multiplicative factor of the maximal element value. For epsilon = 1/polylog(N), each query, insertion, or deletion takes O(log log log N) time.
In this paper, we present a quasi-Newton parallel algorithm to solve the Inverse Additive Symmetric Eigenvalue Problem. Our approach differs from other state-of-the-art algorithms that use Newton type iteration, in th...
详细信息
ISBN:
(纸本)1892512416
In this paper, we present a quasi-Newton parallel algorithm to solve the Inverse Additive Symmetric Eigenvalue Problem. Our approach differs from other state-of-the-art algorithms that use Newton type iteration, in that we carry out the computation of the Jacobian matrix once in the first iteration, thus giving better performance than Newton algorithms even in the sequential case. In addition, our algorithm has been parallelized by using public domain software like ScaLAPACK and MPI, thus guaranteeing portability. Good performances have been obtained on a cluster of PCs connected through a Fast Ethernet network.
暂无评论