Nonlinear systems occur in diverse applications, e.g., in the steady state analysis of chemical processes. If safety concerns require the results to be provably correct then result-verifying algorithms relying on inte...
详细信息
Certain numerically intensive applications executed within a grid computing environment crucially depend on the properties of floating-point arithmetic implemented on the respective platform. Differences in these prop...
详细信息
ISBN:
(纸本)076952821X
Certain numerically intensive applications executed within a grid computing environment crucially depend on the properties of floating-point arithmetic implemented on the respective platform. Differences in these properties may have drastic effects. This paper identifies the central problems related to this situation. We propose an approach which gives the user valuable information on the various platforms available in a grid computing environment in order to assess the numerical quality of an algorithm run on each of these platforms. In this manner, the user will at least have very strong hints whether a program will perform reliably in a grid before actually executing it. Our approach extends the existing IeeeCC754 test suite by two "grid-enabled" modes: The first mode calculates a "numerical checksum" on a specific grid host and executes the job only if the checksum is identical to a locally generated one. The second mode provides the user with information on the reliability and IEEE 754-conformity of the underlying floating-point implementation of various platforms. Furthermore, it can help to find a set of compiler options to optimize the application's performance while retaining numerical stability.
Fuzzy Connectedness is an important image segmentation routine for image processing of medical images. It is often used in preparation for surgery and sometimes during surgery. It is important to have an algorithm whi...
详细信息
In recent years, it has become obvious that mRNA expression does not always correlate with protein expression. It seems that a full understanding of the complexity of life can only be obtained by examining abundances ...
详细信息
作者:
El-Owny, H.Bergische Universität Wuppertal
Faculty C - Department of Mathematics and Computer Science Scientific Computing / Software Engineering Gaußstraße 20 42097 Wuppertal Germany
This paper addresses the problem of solving parametric linear systems of equations whose coefficients are, in the general case, nonlinear functions of interval parameters. Such systems, are encountered in many practic...
详细信息
ISBN:
(纸本)076952821X
This paper addresses the problem of solving parametric linear systems of equations whose coefficients are, in the general case, nonlinear functions of interval parameters. Such systems, are encountered in many practical problems, e.g in electrical engineering and mechanical systems. A C-XSC[8] implementation of a parametric fixed-point iteration method for computing an outer enclosure for the solution set is proposed in this paper. Numerical examples illustrating the applicability of the proposed method are solved, and compared with other methods.
A stationary convection-diffusion problem with a small parameter multiplying the highest derivative is considered. The problem is discretized on a uniform rectangular grid by the central-difference scheme. A new class...
详细信息
Application performance on high-performance shared-memory systems is often limited by sharing patterns resulting in cache-coherence bottlenecks. Current approaches to identify coherence bottlenecks incur considerable ...
详细信息
Certain numerically intensive applications executed within a grid computing environment crucially depend on the properties of floating-point arithmetic implemented on the respective platform. Differences in these prop...
详细信息
ISBN:
(纸本)9781424431410;9780769528212;076952821X
Certain numerically intensive applications executed within a grid computing environment crucially depend on the properties of floating-point arithmetic implemented on the respective platform. Differences in these properties may have drastic effects. This paper identifies the central problems related to this situation. We propose an approach which gives the user valuable information on the various platforms available in a grid computing environment in order to assess the numerical quality of an algorithm run on each of these platforms. In this manner, the user will at least have very strong hints whether a program will perform reliably in a grid before actually executing it. Our approach extends the existing IeeeCC754 test suite by two "grid-enabled" modes: The first mode calculates a "numerical checksum" on a specific grid host and executes the job only if the checksum is identical to a locally generated one. The second mode provides the user with information on the reliability and IEEE 754-conformity of the underlying floating-point implementation of various platforms. Furthermore, it can help to find a set of compiler options to optimize the application's performance while retaining numerical stability.
We investigate the use of the multistep successive preconditioning strategies (MSP) to construct a class of parallel multilevel sparse approximate inverse (SAI) preconditioners. We do not use independent set ordering,...
详细信息
We propose an approach which gives the user valuable information on the various platforms avail able in a grid in order to assess the numerical quality of an algorithm run on each of these platforms. In this manner, t...
详细信息
We propose an approach which gives the user valuable information on the various platforms avail able in a grid in order to assess the numerical quality of an algorithm run on each of these platforms. In this manner, the user is provided with at least very strong hints whether a program performs reliably in a grid before actually executing it. Our approach extends IeeeCC754 by two "grid-enabled" modes: The first mode calculates a "numerical checksum" on a specific grid host and executes the job only if the check sum is identical to a locally generated one. The second mode provides the user with information on the reliability and IEEE 754-conformity of the underlying floating-point implementation of various platforms. In addition, it can help to find a set of compiler options to optimize the application's performance while retaining numerical stability
暂无评论