The symplectic multistep method for special differential equation of the secondorder y" = f(x,y) was constructed in this paper. It is equals to the symmetriclinear multistep method which was presented by Lambert ...
详细信息
The symplectic multistep method for special differential equation of the secondorder y" = f(x,y) was constructed in this paper. It is equals to the symmetriclinear multistep method which was presented by Lambert & Watson, suiting themotion equations of the celestial mechanics. Obviously, we prove the Lambert’smethod possesses symplectic property. Numerical experiments are performed tothe method, and the results show that the method can preserve many inherentcharacters of the system.
In this paper, for the colltact problem in elasticity, we proposed a new mixed variational formulation, which is the base for the dual mixed finite element method of the contact problem.
In this paper, for the colltact problem in elasticity, we proposed a new mixed variational formulation, which is the base for the dual mixed finite element method of the contact problem.
In this paper, an optimal V-cycle muligrid method is presented for Q1 nonconforming element. The uniform convergence rate independent of mesh size and level is established.
In this paper, an optimal V-cycle muligrid method is presented for Q1 nonconforming element. The uniform convergence rate independent of mesh size and level is established.
In this paper, we apply the natural boundary element method to solve initialboundary value problem of parabolic equation. By Fourier expansion, we obtainthe natural integral equation of the problem and its Poisson int...
详细信息
In this paper, we apply the natural boundary element method to solve initialboundary value problem of parabolic equation. By Fourier expansion, we obtainthe natural integral equation of the problem and its Poisson integral formula overexterior circular domain. Meanwhile, the numerical implementation of the natural integral equation is given. At last some numerical examples are presented toillustrate feasibility and efficiency of our method.
In this paper, we use the flux ENO scheme for the multiresolution scheme and simplify the original scheme based on the cell average ENO. As it is well known that the present scheme may will not be conservative at disc...
详细信息
In this paper, we use the flux ENO scheme for the multiresolution scheme and simplify the original scheme based on the cell average ENO. As it is well known that the present scheme may will not be conservative at discontinuities, but numerical solutions are acceptable in practice.
In this paper, we consider the acoustic impedance inversion problem of one dimensional wave equations. from the difference scheme of one diemsional wave equations, we derive the positive property of impulsive response...
详细信息
In this paper, we consider the acoustic impedance inversion problem of one dimensional wave equations. from the difference scheme of one diemsional wave equations, we derive the positive property of impulsive response in frequency domain,and prove that the positivity is a sufficient condition for the stability of impedance inversion.
We propose an approach to construct a C1 function on a fat surface domain, asurface that has thickness, by piecewise rational functions defined on a collection ofirregular triangular prisms. We show that the interpola...
详细信息
We propose an approach to construct a C1 function on a fat surface domain, asurface that has thickness, by piecewise rational functions defined on a collection ofirregular triangular prisms. We show that the interpolation function has algebraic precision 2. Examples that show the effectiveness of the scheme is presented.
In this paper parallel implementation issues concerning computation of a set ofrecurrences on distributed memory parallel systems are discussed. It is a commonproblem when solving partial differential equations on dis...
详细信息
In this paper parallel implementation issues concerning computation of a set ofrecurrences on distributed memory parallel systems are discussed. It is a commonproblem when solving partial differential equations on distributed memory systems,especially in computational fluid dynamics. For example, solution of a set of tri-diagonalsystems of equations or Guauss-Seidel relaxations on finite difference systems all leadto this kind of computation. We emphasize on the idea of doing the computations in apipelined fashion and we show through analysis and numerical examples that by usingproperly chosen parameters, pipelined implementation often yields much better parallelefficiency with respect to other commonly used methods.
暂无评论