Targeting at SN algorithm for the neutron transport equation in the two-dimensional spherical coordinate system, we propose a directed graph model based on a (cell, direction) two-tuple, and design a multi-level parallel SN algorithm with controllable granularity on the basis of the existing parallel pipeline algorithm based on directed graph. Among them, a combination of domain decomposition and parallel pipeline is used to mine parallelism in the space-angle direction, and an energy group pipeline parallel method is proposed. Furthermore, by setting appropriate pipeline granularity, the overhead of scheduling, communication and idle waiting are well balanced. Experimental results show that the algorithm can effectively solve the neutron transport equation in the two-dimensional spherical coordinate system. For a typical neutron transport problem with 960 000 grids, 60 directions, 24 energy groups, and billions of degrees of freedom, the parallel program achieved 71% parallel efficiency on 1920 cores of a domestic parallel machine.

Supersonic compression ramp flow is studied. Based on Rytov approximation, a generalized convolution-fast Fourier transform (GCV-FFT) method is used to solve directly Helmholtz equation in inhomogeneous medium. Scattering field of the laser beam and complex amplitude distribution of diffracted field are obtained. Aero-optical parameters including light intensity, optical path difference, Strehl ratio, relative light intensity, beam centroid position and effective beam width are given. It can be seen that the laser beam is transmitted in the turbulent flow field, and the light intensity distribution appears gradually undulates and deviates. As passing through flow field, the beam produces a large degree of distortion, deviation and breakage due to the effect of diffraction. Effect of flow field in compressed ramp region on beam quality is significantly greater than that in fully developed turbulent region. From distribution of relative light intensity, the beam breakage seems to be related to small-scale coherent structure of flow field. The error of Strehl ratio obtained by traditional OPD mean square error or weighted mean square error estimation is greater. More accurate method should be evaluated with PSF method.

Effects of thickness and density of plate on impact load and cavitation evolution in non-contact underwater explosions are investigated.A modified ghost fluid method developed recently is extended to treat nonlinear interaction between compressible flow and elastic plate.With assumption that fracture of a plate does not occur, it discloses that relatively small thickness or density of plate weakens impact of blast wave on structure.Impact load tends to result of rigid wall boundary nonlinearly with increase of thickness or density.Furthermore, smaller thickness or density causes cavitation to occur earlier and creates a larger cavitation region.It delays and weakens secondary impact on structure induced by cavitation collapse as well.

By analyzing ghost fluid method for compressible multi-medium flows, fundamentals for defining ghost fluid state with a general equation of state are described. All possible definitions of ghost fluid states are derived according to wave patterns and free variables in ghost fluid region. Following these treatments, it reveals that solution of multi-medium Riemann problem can be obtained exactly in theory. Several simple but effective definitions independent with wave patterns in ghost fluid region, are further summarized. One of these ways is similar to reflective boundary condition. An essential prerequisite is that interfacial velocity must be predicted exactly. Numerical results show that this methodology indeed works reasonably for multi-medium flows.

In order to improve efficiency and resolution of inversion and computation in concrete tomography, we propose a probability weighted conjugate gradient algorithm, in which weight is added on imaging unit instead of equation. IART algorithm is adopt to determine weight factor and initial value. Simulation examples and a real physical experiment show that the algorithm is of fast convergence, feasibility and validity.

A weighting approach for ray matrix in ART algorithm for ultrasonic computerized tomogrnphy of concrete is developed. Firstly。a low-resolution image is generated by BPT algorithm and which is used asinitial value of iteration.Then,information of lowresolution image is used as weighting coefficients for ray matrix of ART algorithm.Image information and numerical information are used fully.High speed and low speed abnormal units can be weighted.As data error is less than 0.5%.examples show that the accuracy and iteration speed of improved method are better than that of traditional ART algorithm.

Electronic structure of BaMgF₄ crystal containing F color centers is studied within the framework of fully relativistic self-consistent Dirac-Slater theory,using a numerically discrete variational(DV-X_α) method.2p state of H atom is added into the program.It is concluded that energy levels of F color center are located in forbidden band of the crystal.Optical transition energy from ground state to excited state of the F color center is 5.12 eV,which corresponds to 242 nm absorption bands.They are well consisted with experiments.Absorption bands resulted from an F color center in BaMgF₄ crystal are located in 236-274 nm.

Smoothed particle hydrodynamics(SPH) is a free-Lagrangian computational method.Simulating shock tube problems is a common measure to test SPH methods and SPH codes.The fundamentals of SPH are presented and the implementation of SPH for 1-D shock tube problem simulation is discussed.Some numerical results and explanations are provided.

Two types of preconditionings coupled with the GMRES algorithm for the two-dimensional Euler equations are presented.An approach is used for reducing the memory requirement of GMRES algorithm and preserving its efficiency.With the calculations of two inviscous flows,the GMRES algorithm and the associated two preconditionings are analyzed,and its comparison with the DDADI method is also presented.The calculation results of the NACA0012 airfoil transonic flow and the supersonic GAMM flow show that the GMRES algorithm coupled with the two preconditionings has superior advantage on convergence rate compared to DDADI method.Furthermore,from the results it can be seen that the LUSGS preconditioning is slightly better than the ILU preconditioning.

It provides an overview of the parallel computing for 2-dimensional plasma simulations with particle clouds in cells methods under both shared-memory and distrbuted-memory parallel computing environments.The parallel algorithm and its implementation are desigened on the basis of the serial algorithm and the executing characteristics of the serial code.The practical parallel computations are performed under Challenge,DAWN 1000+, Power PC Cluster and XMZY. Many important conclusions are given and quite useful for the parallel computations of both types of particle simulation methods and other applicational codes.

A two-dimensional relativistic electromagnetic particle simulation code (2DCIC) is presented. The charge and current densities are computed self-consistently using Maxwell's equations, relativistic motion equations of electrons and Newton's equations of ions by tracking 10⁴-10⁶ simulated particles. some details of wave wave and wave particle interaction are also studied together with the evolution of instability as time. Laser light impinges obliquely (or normally) on plasma, which has a density gradient (or homogeneous density). To save computational time, the parallel computational method is developed. Using this code, the ultrashort intense pulse laser plasma interaction and other plasma problems can be also studied by modifying parameters of physical model. After a lot of test computation, plasma equilibrium and propagation of ultrashort intense pulse laser in plasma are studied. The computed physical picture is reasonable.

Based on finite element and artificial transmitting boundary method, the cylindrical wave scattering by infinite circular cylinder is computed and compared with analytical solution. Then other problems on scattering induced by irregular objects are also solved on use of the same method, whose superiority and efficiency have been indicated by numerical examples.

In this paper we present a double-time-scale cloud-in-cell (CIC) algorithm for simulation of plasmas which reduce by half-quarter computational cost of standard CIC methods. The algorithm is integrating electrons by small time steps and integrating ions by large steps. The algorithm have been used in one and a half-dimensional CIC scheme. The results of theoretic analysis and simulation support the stability and correctness of the algorithm.

In this paper we study the laser plasma turbulent heating due to parametric instabilities by the numerical simulation method based on one and a half-dimensional cloud-in-cell scheme.The computer simulation results show that linear growth rate of the parametric instabilities is in reasonable agreement withthe theoretical predic tion. We also give the computer simulation results on no-nlinear heating rate,the heated electron distribution function,the ripple surfaceof the critical density and the caviton, These results are discussed.

We have studied parametric heating caused by electric field of strong laser with frequence which is a little higher than plasma frequence(w₀=1.1W_p)by use one and half dimensional CIC scheme. Results shown that both the ion-acoustic instabilities and the oscillation two-stream instabilities are stimulated together, nonlinear saturation show some fine structures, plasma is heated strongly and tur-blent spectra after saturetion fllows a typical power law.

The noise-restraining mechanism due to introducing periodical smoothing in phase space in the particle simulation is analyzed. We introduce a concept of filtrate effect of the smoothing and give a analytical expression for filtrate wave coefficent that is dependent on wavelength.The results obtained show that the smoothing restrain is mainly for short wave noises and no energy transfer happens in wavenumber space.