To avoid the complexity and inefficiency for specific applications of the current software architecture, a novel approach using partial evaluation is proposed to improve the running performance of components. The gene...To avoid the complexity and inefficiency for specific applications of the current software architecture, a novel approach using partial evaluation is proposed to improve the running performance of components. The generic program was specialized into domain-specific realization for the known knowledge and environments. The syntax and semantic(adj.) were analyzed based on byte code instruction sequences, and partial evaluation rules depicted how to perform the specialization. The partial evaluation for object-oriented programs was implemented. The experimental results show that partial evaluation is effective to speed up the running efficiency. The more generality and scalability can be obtained by the integration of partial evaluation with the favorable design mechanisms and compiler optimization technology.展开更多
In order to solve the hybrid and dependent task scheduling and critical source allocation problems, a task scheduling algorithm has been developed by first presenting the tasks, and then describing the hybrid and depe...In order to solve the hybrid and dependent task scheduling and critical source allocation problems, a task scheduling algorithm has been developed by first presenting the tasks, and then describing the hybrid and dependent scheduling algorithm and deriving the predictable schedulability condition. The performance of this agorithm was evaluated through simulation, and it is concluded from the evaluation results that the hybrid task scheduling subalgorithm based on the comparison factor can be used to solve the problem of aperiodic task being blocked by periodic task in the traditional operating system for a very long time, which results in poor scheduling predictability; and the resource allocation subalgorithm based on schedulability analysis can be used to solve the problems of critical section conflict, ceiling blocking and priority inversion; and the scheduling algorithm is nearest optimal when the abortable critical section is 0.6.展开更多
EEG inverse problem has great significance and importance for both cli nical and research applications. It discusses EEG dipole source localization pro blems solved by nonlinear local optimization methods, such as Lev...EEG inverse problem has great significance and importance for both cli nical and research applications. It discusses EEG dipole source localization pro blems solved by nonlinear local optimization methods, such as Levenberg-Marquar t b. This paper presents the relation between location errors and noise level on c ondition that the source number is known; if the source number is not known, the selected number in model may not equal to the actual one, and a computation is carried out and a corresponding discrimination criteria is proposed. Computer si mulation demonstrates that Levenberg-Marquardt algorithm is better than global methods if the source number is small.展开更多
Chebfun is a Matlab-based software system that overloads Matlab's discrete operations for vectors and matrices to analogous continuous operations for functions and operators.We begin by describing Chebfun's fa...Chebfun is a Matlab-based software system that overloads Matlab's discrete operations for vectors and matrices to analogous continuous operations for functions and operators.We begin by describing Chebfun's fast capabilities for Clenshaw-Curtis and also Gauss-Legendre,-Jacobi,-Hermite,and-Laguerre quadrature,based on algorithms of Waldvogel and Glaser,Liu and Rokhlin.Then we consider how such methods can be applied to quadrature problems including 2D integrals over rectangles,fractional derivatives and integrals,functions defined on unbounded intervals,and the fast computation of weights for barycentric interpolation.展开更多
基金Sponsored by the National High-Tech Research and Development Program of China (Grant No 2001AA113160,2004AA104280,and 2007AA010302)the National Natural Science Foundation of China(Grant No90718004)
文摘To avoid the complexity and inefficiency for specific applications of the current software architecture, a novel approach using partial evaluation is proposed to improve the running performance of components. The generic program was specialized into domain-specific realization for the known knowledge and environments. The syntax and semantic(adj.) were analyzed based on byte code instruction sequences, and partial evaluation rules depicted how to perform the specialization. The partial evaluation for object-oriented programs was implemented. The experimental results show that partial evaluation is effective to speed up the running efficiency. The more generality and scalability can be obtained by the integration of partial evaluation with the favorable design mechanisms and compiler optimization technology.
文摘In order to solve the hybrid and dependent task scheduling and critical source allocation problems, a task scheduling algorithm has been developed by first presenting the tasks, and then describing the hybrid and dependent scheduling algorithm and deriving the predictable schedulability condition. The performance of this agorithm was evaluated through simulation, and it is concluded from the evaluation results that the hybrid task scheduling subalgorithm based on the comparison factor can be used to solve the problem of aperiodic task being blocked by periodic task in the traditional operating system for a very long time, which results in poor scheduling predictability; and the resource allocation subalgorithm based on schedulability analysis can be used to solve the problems of critical section conflict, ceiling blocking and priority inversion; and the scheduling algorithm is nearest optimal when the abortable critical section is 0.6.
文摘EEG inverse problem has great significance and importance for both cli nical and research applications. It discusses EEG dipole source localization pro blems solved by nonlinear local optimization methods, such as Levenberg-Marquar t b. This paper presents the relation between location errors and noise level on c ondition that the source number is known; if the source number is not known, the selected number in model may not equal to the actual one, and a computation is carried out and a corresponding discrimination criteria is proposed. Computer si mulation demonstrates that Levenberg-Marquardt algorithm is better than global methods if the source number is small.
基金supported by the MathWorks,Inc.,King Abdullah University of Science and Technology (KAUST) (Award No. KUK-C1-013-04)the European Research Council under the European Union’s Seventh Framework Programme (FP7/2007-2013)/ERC (Grant Agreement No. 291068)
文摘Chebfun is a Matlab-based software system that overloads Matlab's discrete operations for vectors and matrices to analogous continuous operations for functions and operators.We begin by describing Chebfun's fast capabilities for Clenshaw-Curtis and also Gauss-Legendre,-Jacobi,-Hermite,and-Laguerre quadrature,based on algorithms of Waldvogel and Glaser,Liu and Rokhlin.Then we consider how such methods can be applied to quadrature problems including 2D integrals over rectangles,fractional derivatives and integrals,functions defined on unbounded intervals,and the fast computation of weights for barycentric interpolation.