General-purpose graphics processing units (GPGPUs) could bring huge performance improvements in scientific and numerical fields. We presented two approaches utilizing hybrid CPU/GPU system in Cholesky factorization. F...
详细信息
ISBN:
(纸本)9781467392952
General-purpose graphics processing units (GPGPUs) could bring huge performance improvements in scientific and numerical fields. We presented two approaches utilizing hybrid CPU/GPU system in Cholesky factorization. First, we analyzed the implementation of Cholesky factorization in MAGMA and identified the bottleneck of the current implementation, which is the use of fixed block size without considering any factors in the computing environment. So, we designed an algorithm, which could determine the optimal block size of Cholesky factorization based on multiple factors (input matrix size, CPU/GPU performance, and CPU/GPU bandwidth, etc.). Then, we presented a new improvement on MAGMA's implementation utilize the algorithm. Test results showed that our approach is more efficient than MAGMA's fixed block size implementation under some circumstance. After combining our implementation with MAGMA's implementation, the new hybrid implementation could outperform the current MAGMA implementation. Second, we identified that all the implementations of Cholesky factorization, to our best knowledge, that utilized the GPU do not fully utilized the multicore CPU. So, after studied other researchers approaches, we designed a new algorithm that could utilize multicore CPU and GPU simultaneously in Cholesky factorization. Our approach could keep the block size and workload distribution between CPU and GPU dynamically. Testing results showed the optimal data distribution ratio for our current implementation.
This paper is concerned with L-2-gain optimal control approach for rotary electro-hydraulic servo-system. The electro-hydraulic dynamics with respect to hydraulic motor velocity, with input voltage to the servo valve ...
详细信息
ISBN:
(纸本)9781479959013
This paper is concerned with L-2-gain optimal control approach for rotary electro-hydraulic servo-system. The electro-hydraulic dynamics with respect to hydraulic motor velocity, with input voltage to the servo valve as control input and load torque as disturbance input, is formulated. The mathematical model results in input-affine nonlinear system. A numerical algorithm based on Newton method to solve a finite-horizon minimax problem for L-2-gain minimisation of electro-hydraulic system is presented. The feedback control and disturbance variables are formulated as linear combination of approximation functions. The proposed algorithm, which has recursive matrix structure, directly finds approximations of the feedback control and the "worst case" disturbance variables. Developed controller has been tested experimentally in the laboratory model of an electro-hydraulic servo system.
Non-obstructive particle damping (NOPD) is a new composite damping technology based on traditional partical damping and impacting damping, and it has good vibration-reduced results. Through studying thermodynamics con...
详细信息
ISBN:
(纸本)9783037850770
Non-obstructive particle damping (NOPD) is a new composite damping technology based on traditional partical damping and impacting damping, and it has good vibration-reduced results. Through studying thermodynamics constraint condition for non-inverse variation of inner tissue structure of the material, the endochronic theory obtained variation regularity of inner variable, further gave out evolution path of specific non-inverse thermodynamics variable under specific condition, and thus the constitutive relationship of the material was obtained. Based on endochronic theory, an incremental form of the endochronic constitutive relationship of discrete particle was derived, and then penalty element was introduced to solve the connection problem between structure and discrete particle, lastly finite element dynamics equation of NOPD composite structure was constructed. Based on the model of NOPD composite structure, simulation and experiment for response of NOPD free beam were made. It indicated that applying endochronic theory to analyze the response of NOPD is rational, which supplied an effective method for engineering application of NOPD.
Let S be an operator in a Banach space H and S i (u) (i = 0, 1, ..., u H) be the evolutionary process specified by S. The following problem is considered: for a given point z 0 and a given initial condition a 0, find ...
详细信息
In this paper, we propose an efficient numerical algorithm for solving linear two-point boundary value problem for delay differential equations with loadings. The Dzhumabaev parametrization method is applied to constr...
详细信息
In this paper, we propose an efficient numerical algorithm for solving linear two-point boundary value problem for delay differential equations with loadings. The Dzhumabaev parametrization method is applied to construct the numerical solutions. The application of the this method leads the considering problem to a system of algebraic equations and Cauchy problems for differential equation without delay argument that are easy to solve. Several illustrative examples are given to demonstrate the effectiveness of the proposed algorithm.
The paper is devoted to the solution of the energy minimization problem for a moving train. The train movement is governed by the system of the first order ordinary differential equations where the train speed and the...
详细信息
ISBN:
(纸本)9783319771793;9783319771786
The paper is devoted to the solution of the energy minimization problem for a moving train. The train movement is governed by the system of the first order ordinary differential equations where the train speed and the distance along the track are the state variables. The provided locomotive power depends on the control function. The generated traction force is assumed to depend on the velocity of the train and on the control function. Each non-negative value of the control function determines a traction force control while negative values determine a braking force control. The cost functional is defined as the train energy. It is dependent on traction force, speed and control functions. The speed, distance and control functions are assumed bounded. Using the maximum principle and Lagrangian multipliers the system of equations constituting the necessary optimality conditions is formulated. Based on the analysis of the train movement the optimal trajectories in terms of train speed and associated optimal control functions are calculated. A new simplified method is used to calculate the set of the switching times implementing the optimal control function. numerical examples are provided and discussed.
In this paper, a new numerical algorithm for high-accurate computation of exact Carson formulas is developed. Carson formulas are used for computing of perunit-length (pul) self and mutual impedances of infinitely lon...
详细信息
ISBN:
(纸本)9789532900712
In this paper, a new numerical algorithm for high-accurate computation of exact Carson formulas is developed. Carson formulas are used for computing of perunit-length (pul) self and mutual impedances of infinitely long parallel conductors. The proposed algorithm is based on piecewise linear approximation of kernel function and analytical integrations of piecewise-linearized kernel function multiplied by the rest of two integrands. Using proposed numerical algorithm, high-accurate results with desired n-digit accuracy can be easily obtained. Results computed by two approximation methods are compared with high-accurate results computed by proposed numerical algorithm for power frequency.
This essay describes seismic damage mechanism of concrete material and a general method induced from a mixed Lagrange approach based on irreversible *** kind of method permits nonlinear effects of geometric,material a...
详细信息
This essay describes seismic damage mechanism of concrete material and a general method induced from a mixed Lagrange approach based on irreversible *** kind of method permits nonlinear effects of geometric,material and damage during the catastrophic *** simulate the damage behaviors in concrete materials,the authors make usage of a damage model deduced from a generalized standard material framework. According to the generalized Hamilton principle,the Euler-Lagrange equations of the material and component can be derived from the integral action of generalized standard model based upon Gateaux differentiable form. A numerical implementation technology is exploited to calculate the collapse course of concrete structures.
The Local Taylor Polynomial (LTP) method treats fields and sources as local analytic polynomials for rapid numerical solution of PDE's. Knowledge of the source polynomial would allow sources to be treated analytic...
详细信息
ISBN:
(纸本)9781467312257;9781467312226
The Local Taylor Polynomial (LTP) method treats fields and sources as local analytic polynomials for rapid numerical solution of PDE's. Knowledge of the source polynomial would allow sources to be treated analytically in LTP formulas, since the derivatives follow directly for polynomial forms. Aside from idealized cases, such information is seldom available;however, use of macro-particles with finite polynomial shape functions for charge and current deposition can make analytic treatment possible for general cases. The macro-particle polynomial coefficients can be used directly in LTP solution formulas. This scheme allows one to choose a representation of particles that enables high fidelity with fewer particles compared to traditional particle methods, since the polynomials can include information such as transverse profile information from a beam, for example. This paper presents a simple polynomial form for representing macro-particles in numerical simulations. The particle shape must be continuous and differentiable to a specified order, bounded in size and magnitude, zero in magnitude and derivatives at the edges, symmetric, positive definite, with analytic coefficients. Two key properties for use in LTP will be demonstrated: zero derivatives at the particle boundary and analytically computable coefficients.
This paper presents a ready-to-use procedure for detecting atypical (rarely occurring) elements, in one-and multidimensional spaces. The issue is considered through a conditional approach. The application of nonparame...
详细信息
ISBN:
(纸本)9783319546094;9783319546087
This paper presents a ready-to-use procedure for detecting atypical (rarely occurring) elements, in one-and multidimensional spaces. The issue is considered through a conditional approach. The application of nonparametric concepts frees the investigated procedure from distributions of describing and conditioning variables. Ease of interpretation and completeness of the presented material lend themselves to the use of the worked out method in a wide range of tasks in various applications of data analysis in science and practice, engineering, economy and management, environmental and social issues, biomedicine, and related fields.
暂无评论