The refined Arnoldi method proposed by Jia is used for computing some eigen-pairs of large matrices. In contrast to the Arnoldi method, the fundamental dif-ference is that the refined method seeks certain refined Ritz...The refined Arnoldi method proposed by Jia is used for computing some eigen-pairs of large matrices. In contrast to the Arnoldi method, the fundamental dif-ference is that the refined method seeks certain refined Ritz vectors, which aredifferent from the Ritz vectors obtained by the Arnoldi method, from a projection space with minimal residuals to approximate the desired eigenvectors. In com-parison with the Ritz vectors, the refined Ritz vectors are guaranteed to converge theoretically and can converge much faster numerically. In this paper we propose to replace the Ritz values, obtained by the Arnoldi method with respect to a Krylovsubspace, by the ones obtained with respect to the subspace spanned by the refined Ritz vectors. We discuss how to compute these new approximations cheaply and reliably. Theoretical error bounds between the original Ritz values and the new Ritz values are established. Finally, we present a variant of the refined Arnoldi al-gorithm for an augmented Krylov subspace and discuss restarting issue. Numerical results confirm efficiency of the new algorithm.展开更多
In this paper , numerically simulational calculations are made on aerodynamic characteristics for power car of 200km/h EMUs, using Computational Fluid Dy-namics software CFX. After calculation, surface pressure, veloc...In this paper , numerically simulational calculations are made on aerodynamic characteristics for power car of 200km/h EMUs, using Computational Fluid Dy-namics software CFX. After calculation, surface pressure, velocity field distribution as well as aerodynamic resistance are obtained. The results of calculations are com-pared with those of wind tunnel test. Comparison shows good agreement.展开更多
In order to improve generalization capability of neural networks, a model structure of Case-Based neural networks has been presented. The model blended Case-Based Reasoning method into neural networks and has the abil...In order to improve generalization capability of neural networks, a model structure of Case-Based neural networks has been presented. The model blended Case-Based Reasoning method into neural networks and has the ability of incrementally learning. The results demonstrated that the model could observably improve the generalization capability of supervised neural networks. Firstly, paper summarized the advancing front of researching on generalization capability of neural networks.Secondly, the structure of CBNN and its process of working were introduced. Finally, the results of experiments were compared and discussed.展开更多
In this paper, we present a simple and practical method to compute Fermi-Dirac integrals and modified Fermi-Dirac integrals. Moreover, we apply this method to compute the parameters A⊥^α and A⊥^β in [5] and we get...In this paper, we present a simple and practical method to compute Fermi-Dirac integrals and modified Fermi-Dirac integrals. Moreover, we apply this method to compute the parameters A⊥^α and A⊥^β in [5] and we get the desired result. The parameters are relevant to the dense plasma transport and are very important to the magnet confined plasma dynamics.展开更多
A method is presented for tracking interfaces, which is MOCL (marker on cell line) employed in two-dimensional Eulerian code. To test it, five kinds of objects with different shapes being uniform motion are numericall...A method is presented for tracking interfaces, which is MOCL (marker on cell line) employed in two-dimensional Eulerian code. To test it, five kinds of objects with different shapes being uniform motion are numerically simulated in a two- dimensional Eulerian hydrodynamics code that uses the MOCL technique to track interfaces. Results show that the method is simple and feasible.展开更多
In the processing of measured data, the number of operations of the algorithm for picking out outlier data in batches is very large. A large number of linear or nonlinear equations based on the parameter model built a...In the processing of measured data, the number of operations of the algorithm for picking out outlier data in batches is very large. A large number of linear or nonlinear equations based on the parameter model built according to the characteristics of measuring equipment and measured object are to be sovied. This paper presents the criterion of picking out outlier data point by point and a parallel algorithm for picking out outlier data in batches, with regard to large-scale linear regression model. The scalability for the parallel algorithm is analyzed, and the results for the algorithm on a group of computers are given. High speed-up is obtained.展开更多
文摘The refined Arnoldi method proposed by Jia is used for computing some eigen-pairs of large matrices. In contrast to the Arnoldi method, the fundamental dif-ference is that the refined method seeks certain refined Ritz vectors, which aredifferent from the Ritz vectors obtained by the Arnoldi method, from a projection space with minimal residuals to approximate the desired eigenvectors. In com-parison with the Ritz vectors, the refined Ritz vectors are guaranteed to converge theoretically and can converge much faster numerically. In this paper we propose to replace the Ritz values, obtained by the Arnoldi method with respect to a Krylovsubspace, by the ones obtained with respect to the subspace spanned by the refined Ritz vectors. We discuss how to compute these new approximations cheaply and reliably. Theoretical error bounds between the original Ritz values and the new Ritz values are established. Finally, we present a variant of the refined Arnoldi al-gorithm for an augmented Krylov subspace and discuss restarting issue. Numerical results confirm efficiency of the new algorithm.
文摘In this paper , numerically simulational calculations are made on aerodynamic characteristics for power car of 200km/h EMUs, using Computational Fluid Dy-namics software CFX. After calculation, surface pressure, velocity field distribution as well as aerodynamic resistance are obtained. The results of calculations are com-pared with those of wind tunnel test. Comparison shows good agreement.
文摘In order to improve generalization capability of neural networks, a model structure of Case-Based neural networks has been presented. The model blended Case-Based Reasoning method into neural networks and has the ability of incrementally learning. The results demonstrated that the model could observably improve the generalization capability of supervised neural networks. Firstly, paper summarized the advancing front of researching on generalization capability of neural networks.Secondly, the structure of CBNN and its process of working were introduced. Finally, the results of experiments were compared and discussed.
文摘In this paper, we present a simple and practical method to compute Fermi-Dirac integrals and modified Fermi-Dirac integrals. Moreover, we apply this method to compute the parameters A⊥^α and A⊥^β in [5] and we get the desired result. The parameters are relevant to the dense plasma transport and are very important to the magnet confined plasma dynamics.
文摘A method is presented for tracking interfaces, which is MOCL (marker on cell line) employed in two-dimensional Eulerian code. To test it, five kinds of objects with different shapes being uniform motion are numerically simulated in a two- dimensional Eulerian hydrodynamics code that uses the MOCL technique to track interfaces. Results show that the method is simple and feasible.
文摘In the processing of measured data, the number of operations of the algorithm for picking out outlier data in batches is very large. A large number of linear or nonlinear equations based on the parameter model built according to the characteristics of measuring equipment and measured object are to be sovied. This paper presents the criterion of picking out outlier data point by point and a parallel algorithm for picking out outlier data in batches, with regard to large-scale linear regression model. The scalability for the parallel algorithm is analyzed, and the results for the algorithm on a group of computers are given. High speed-up is obtained.
