A barycentric Lagrange interpolation collocation method is proposed to solve the three-dimensional and four-dimensional wave equations. Firstly, the barycentric Lagrange interpolation method is introduced and the matrix format of the collocation method is given. Secondly, the solution function and initial boundary conditions of the wave equation are approximated by Lagrange interpolation. The discrete equation is obtained by collocation method, and the matrix expression of the wave equation is obtained. Finally, the initial and boundary conditions of the wave equation are imposed by the addition method and the replacement method respectively. Numerical examples show that the barycentric Lagrange interpolation collocation method has high computational accuracy and efficiency.

A v'²-f model,a k-ε model and an indoor zero equation model are adopted respectively to simulate mixed convections which are typical in ventilation and air conditioning systems.It indicates that there is a separated vortex near the ventilation inlet,and the accuracy for predicting this vortex is a critical factor in the simulation.The separated vortex dimension obtained by experiments is between the values calculated by the v'²-f model and the indoor zero equation model,while the dimension predicted by the k-ε model is a bit smaller.The results predicted by the indoor zero equation fit experimental results best.However,it is not in general from the view of model structure.For calculating temperature field the v'²-f model is better than the indoor zero equation model,and the k-ε model is the worst among the three.

A self-heating model is presented to predict the high-frequency performance of AlGaAs/GaAs heterojunction bipolar transistors (HBT's) including high current and thermal effects. A base pushout effect is discussed at high current region. And in this case, the lattice temperature, the cut-off frequency and the maximum frequency versus the collector current density are achieved.

Based on a brief description of Corrected Smoothed Particle Hydrodynamics (CSPH) method, the discretization scheme of conservation equations in axisymmetric coordinate systems is developed and a method to determine the number of particles in the computational domain is proposed. Numerical examples for hypervelocity impact are carried out and the results clearly show that CSPH is not only a suitable method for the simulation of large deformation problems in impact dynamics, but also a new method with the advantages that can not be replaced by finite element/ finite difference method.

For the system in which the cylindrical mirrors of a hole-coupling plane waveguide free electron laser (FEL) oscillator are replaced by spherical mirrors, analytic estimation and numerical simulation are made respectively. It is shown that for the problem at hand, the influence on FEL by the mirror replacement is quite little.

Since the presence of a out coupling hole alters the transverse structure of the intra cavity light in a FEL oscillator,the finite element method is used to the optical field equations and a good proximation is made for the out-coupling process in the mirror,and then some numerical results are given.It is shown that the features of the transverse structure of the intra-cavity field are well described and the numerical results are significantly improved.

The entropic increase of a plate drived by explosive has been researched in one dimension system. Usually. The high explosive is contact with a plate immediately. There is the increase entropy of in the plate resulting from the interaction of a detonation with the plate, while increasing its kinetic energy.The rather idealized problem of detonation products initially at CJ state that expands in one dimension into a void and then accelerates a plate has been analyzed in mathmatical detail. It is believed that the technique described in this report can decrease the enlropic increase and increase some kinetic energy of the plate by choosing the proper length of void and thickness of the plate.

In this paper, for a class of nonlinear system of equations #br#F₁x₁,x₂=0#br#F₂x₁,x₂=0#br# we formulate a partial Broyden method. The basic idea is that the relation of x₁ to x₂ is determined by the first subsystem of equations F₁=0, then the Broyden method is applied to the second one F₂=0, thus forming a iterative procedure.The application of partical Broyden method to discretized fpT equations is described and numerical results is given.