Sharp interface tracking using the phase-field equation

A general interface tracking method based on the phase-field equation is presented. The zero phase-field contour is used to implicitly track the sharp interface on a fixed grid. The phase-field propagation equation is derived from an interface advection equation by expressing the interface normal and curvature in terms of a hyperbolic tangent phase-field profile across the interface. In addition to normal interface motion driven by a given interface speed or by interface curvature, interface advection by an arbitrary external velocity field is also considered. In the absence of curvature-driven interface motion, a previously developed counter term is used in the phase-field equation to cancel out such motion. Various modifications of the phase-field equation, including nonlinear preconditioning, are also investigated. The accuracy of the present method is demonstrated in several numerical examples for a variety of interface motions and shapes that include singularities, such as sharp corners and topology changes. Good convergence with respect to the grid spacing is obtained. Mass conservation is achieved without the use of separate re-initialization schemes or Lagrangian marker particles. Similarities with and differences to other interface tracking approaches are emphasized.     

Generalized Pseudospectral Method for Solving the Time-Dependent Schrödinger Equation Involving the Coulomb Potential

We present an accurate and efficient generalized pseudospectral method for solving the time-dependent Schrödinger equation for atomic systems interacting with intense laser fields. In this method, the time propagation of the wave function is calculated using the well-known second-order split-operator method implemented by the numerically exact, fast transform between the grid and spectral representations. In the grid representation, the radial coordinate is discretized using the Coulomb wave discrete variable representation (CWDVR), and the angular dependence of the wave function is expanded in the Gauss-Legendre-Fourier grid. In the spectral representation, the wave function is expanded in terms of the eigenfunctions of the field-free zero-order Hamiltonian. Calculations on the high order harmonic generation and ionization dynamics of hydrogen atom in strong laser pulses are presented to demonstrate the accuracy and efficiency of the present method. This new algorithm will be found more computationally attractive than the close-coupled wave packet method using CWDVR and/or methods based on evenly spaced grids.     

High order matched interface and boundary methods for the Helmholtz equation in media with arbitrarily curved interfaces

This work overcomes the difficulty of the previous matched interface and boundary (MIB) method in dealing with interfaces with non-constant curvatures for optical waveguide analysis. This difficulty is essentially bypassed by avoiding the use of local cylindrical coordinates in the improved MIB method. Instead, novel jump conditions are derived along global Cartesian directions for the transverse magnetic field components. Effective interface treatments are proposed to rigorously impose jump conditions across arbitrarily curved interfaces based on a simple Cartesian grid. Even though each field component satisfies the scalar Helmholtz equation, the enforcement of jump conditions couples two transverse magnetic field components, so that the resulting MIB method is a full-vectorial approach for the modal analysis of optical waveguides. The numerical performance of the proposed MIB method is investigated by considering interface problems with both constant and general curvatures. The MIB method is shown to be able to deliver a fourth order of accuracy in all cases, even when a high frequency solution is involved.     

Numerical modeling of transient two-dimensional viscoelastic waves

This paper deals with the numerical modeling of transient mechanical waves in linear viscoelastic solids. Dissipation mechanisms are described using the generalized Zener model. No time convolutions are required thanks to the introduction of memory variables that satisfy local-in-time differential equations. By appropriately choosing the relaxation parameters, it is possible to accurately describe a large range of materials, such as solids with constant quality factors. The evolution equations satisfied by the velocity, the stress, and the memory variables are written in the form of a first-order system of PDEs with a source term. This system is solved by splitting it into two parts: the propagative part is discretized explicitly, using a fourth-order ADER scheme on a Cartesian grid, and the diffusive part is then solved exactly. Jump conditions along the interfaces are discretized by applying an immersed interface method. Numerical experiments of wave propagation in viscoelastic and fluid media show the efficiency of this numerical modeling for dealing with challenging problems, such as multiple scattering configurations.     

Full multi grid method for electric field computation in point-to-plane streamer discharge in air at atmospheric pressure

Streamers dynamics are characterized by the fast propagation of ionized shock waves at the nanosecond scale under very sharp space charge variations. The streamer dynamics modelling needs the solution of charged particle transport equations coupled to the elliptic Poisson’s equation. The latter has to be solved at each time step of the streamers evolution in order to follow the propagation of the resulting space charge electric field. In the present paper, a full multi grid (FMG) and a multi grid (MG) methods have been adapted to solve Poisson’s equation for streamer discharge simulations between asymmetric electrodes. The validity of the FMG method for the computation of the potential field is first shown by performing direct comparisons with analytic solution of the Laplacian potential in the case of a point-to-plane geometry. The efficiency of the method is also compared with the classical successive over relaxation method (SOR) and MUltifrontal massively parallel solver (MUMPS). MG method is then applied in the case of the simulation of positive streamer propagation and its efficiency is evaluated from comparisons to SOR and MUMPS methods in the chosen point-to-plane configuration. Very good agreements are obtained between the three methods for all electro-hydrodynamics characteristics of the streamer during its propagation in the inter-electrode gap. However in the case of MG method, the computational time to solve the Poisson’s equation is at least 2 times faster in our simulation conditions.     

数值求解三维含时Schrodinger方程及其应用

发展了一套高精度、高效率的伪谱方法,以非微扰的方式求解真实原子三维含时Schrodinger方程.该方法选用二阶劈裂算符作为时间演化算子,分别选择能谱表象和坐标表象作为含时波函数演化的两个表象.在坐标表象下波函数的径向部分使用库仑波函数离散变量表象来离散;角向波函数展开在两维的Gauss-Legendre-Fourier格点上.以H原子的光激发和光电离过程为例,进行了数值计算并和解析解进行了比对.结果表明二者符合很好.该方法很好地处理了库仑奇点问题.还计算了强激光辐照H原子的多光子电离过程,并和其他的数值方案进行了比较.结果表明,在计算收敛的前提下本方法计算效率更高.

Abstract：

We present an accurate and effective pseudospectral method for solving the three-dimensional time-dependent Schrodinger equation involving the Coulomb potential. In this method, the Hamiltonian is evaluated by exploiting the two representations of the wave function. One is a grid representation, in which the angular dependence of the wave function is expanded in a two-dimensional Gauss-Legendre-Fourier grid in the coordinate space of polar and azimuthal angles. The radial coordinate is discretized using a discrete variable representation constructed from the Coulomb wave function (CWDVR) . The other is a spectral representation, in which the wave function is expanded in a set of square integrable functions chosen as the eigenfunctions of a zero-order Hamiltonian. The time of propagation of the wave function is calculated using the well-known second-order split-operator method implemented through the transform between the grid and spectral representations. Calculations on the photo-absorption strength of hydrogen atom are presented to demonstrate the accuracy of present method in low energy limit by the time-dependent wave-packet propagation method. As another example, the present method is applied to multiphoton ionization of H atom. For a wide range of field parameters, ionization rates calculated using the present method are in excellent agreement with those from other accurate numerical calculations. The new algorithm will be found more efficient than the close coupled wave packet method using CWDVR and/or methods based on evenly spaced grids.     

一种模拟大密度比多相流的混合算法

在扩散界面法(diffuse interface method,DIM)的基础上提出一种能够处理大密度比(large density ratio)的多相流混合算法.流场信息通过格子波尔兹曼方法(lattice Boltzmann method,LBM)获得;相界面通过直接求解Cahn-Hilliard(C-H)方程确定.为保证在大密度比情况下求解界面方程的稳定性,采用二阶迎风格式来离散方程的对流项.通过对Rayleigh-Taylor(R-T)不稳定、液体中的气泡上升及液滴撞击干燥壁面的数值模拟,验证了方法的可行性.     

数值求解三维含时Schr?dinger方程及其应用

发展了一套高精度、高效率的伪谱方法,以非微扰的方式求解真实原子三维含时Schrdinger方程.该方法选用二阶劈裂算符作为时间演化算子,分别选择能谱表象和坐标表象作为含时波函数演化的两个表象.在坐标表象下波函数的径向部分使用库仑波函数离散变量表象来离散;角向波函数展开在两维的Gauss-Legendre-Fourier格点上.以H原子的光激发和光电离过程为例,进行了数值计算并和解析解进行了比对.结果表明二者符合很好.该方法很好地处理了库仑奇点问题.还计算了强激光辐照H原子的多光子电离过程,并和其他的数值方案进行了比较.结果表明,在计算收敛的前提下本方法计算效率更高. 关键词： 三维含时Schrdinger方程 库仑奇点 强场 含时波包传播     

Plasmonic wave propagation along a magnetized plasmon–dielectric boundary

The emphasis of this paper is on investigation of the TM/TE depolarization along a magnetized surface. In this regard, initially, the wave equation with appropriate boundary conditions is solved for the propagation constant of the wave along the interface. Next, in order to investigate the electromagnetic field TM/TE depolarization along a linear magnetoplasmon–dielectric interface, coupled partial differential equations are derived for the amplitudes of the TE and TM polarized fields along the surface. These equations are then solved using the Fourier transform and 4th order Runge–Kutta methods and the envelopes of the electric field are obtained. The dependency of the propagation constants of the slow and fast modes on the frequency and magnitude of DC magnetic bias field is investigated. A discussion of TE mode deflection is also provided.     

Quasi-steady-state modeling of dendritic growth

A new method of the moving-boundary problem analysis is developed. After proposed coordinate transformation, the quasi-steady-state differential equation was reduced to equivalent Laplace equation. Transformed boundary conditions of a rescaled field distribution reflect the presence of rate-dependent sources on interface. Taking into account the local rate-dependence of the Neumann's boundary conditions, non-equilibrium pattern formation was considered. The problem of dendrite fractal growth was reduced to one of interaction of conformal to tips charged particles. Conditions of the quasi-steady growth and the recurrence formula for k-order dendrite spacing were derived. Theoretically obtained scaling laws for interface shape, dendrite spacing, critical sidebranch distance are confirmed by available experimental data.     

颗粒圆孔射流碰撞并行直接数值模拟算法研究

本文对空间模式发展的颗粒圆孔射流碰撞进行了并行直接数值模拟算法研究。气相采用可压缩的N-S方程直接求解。颗粒相采用Lagrangian方法跟踪实际的颗粒运动。利用并行求解算法,实现了颗粒穿越边界面的模拟以及高效颗粒碰撞算法。考虑了颗粒和流体的双相耦合以及颗粒之间的碰撞。在本文的计算条件下,颗粒的直径远小于网格的间距,平均的Kolmogorov尺度和网格的间距在一个量级。气相和颗粒相的应力与实验的对比研究表明,本文的颗粒并行程序是可信的。     

Two-component description of dynamical systems that can be approximated by solitons: The case of the ion acoustic wave equations of plasma physics

A new approach to the perturbative analysis of dynamical systems, which can be described approximately by soliton solutions of integrable non-linear wave equations, is employed in the case of small-amplitude solutions of the ion acoustic wave equations of plasma physics. Instead of pursuing the traditional derivation of a perturbed KdV equation, the ion velocity is written as a sum of two components: elastic and inelastic. In the single-soliton case, the elastic component is the full solution. In the multiple-soliton case, it is complemented by the inelastic component. The original system is transformed into two evolution equations: An asymptotically integrable Normal Form for ordinary KdV solitons, and an equation for the inelastic component. The zero-order term of the elastic component is a single-soliton or multiple-soliton solution of the Normal Form. The inelastic component asymptotes into a linear combination of single-soliton solutions of the Normal Form, with amplitudes determined by soliton interactions, plus a second-order decaying dispersive wave. Satisfaction of a conservation law by the inelastic component and of mass conservation by the disturbance to the ion density is determined solely by the initial data and/or boundary conditions imposed on the inelastic component. The electrostatic potential is a first-order quantity. It is affected by the inelastic component only in second order. The charge density displays a triple-layer structure. The analysis is carried out through the third order.     

Das kritische Magnetfeld von dünnen supraleitenden Doppelschichten

Thin superposed films of two superconductors or of a superconductor and a normal metal in a surface parallel magnetic field are considered, assuming that the mean free path of electrons is very short (dirty limit) and that the films are very thin (Cooper limit). From the linearized integral equation for the order parameter and the diffusion equation for the kernel with appropriate boundary conditions at the surfaces and at the interface, an equation for the critical magnetic field is derived. For small magnetic fields and for superposed films consisting of metals which are not too different in their physical properties, our results are in agreement with those obtained earlier byKlein andFischer. For high magnetic fields, however, there are essential differences.     

A real decoupled method and free interface component mode synthesis methods for generally damped systems

This paper reports on the development of a new transformation method. In contrast to most existing mode transformation methods in which the first-order state-space equation of the damped vibration system is transformed into a decoupled form with complex coefficient matrices, using the decoupled method presented in this paper, the equation of the damped system can be decomposed into a decoupled equation with real coefficient matrices. Two new free interface component mode synthesis methods are also presented. The equivalent full-mode matrix of the damped structure is used to capture the effects of the higher-order modes. Additionally, this work modifies the compatibility conditions at the junctions that are employed in most of the previous component mode synthesis methods for generally damped systems. The first component mode synthesis method is performed in complex space, whereas the second method can be applied in real space. Because the coefficient matrices of the coupled equation constructed by the second component mode synthesis method are all real-valued, the solution of the eigenproblem for this coupled equation can be performed in real space as well. Additionally, numerical examples demonstrate the accuracy and validity of these two component mode synthesis methods.     

非线性左右手媒质界面的Goos-Hänchen位移

 由均匀平面电磁波在左右手媒质界面满足的切向边界条件出发,推导了电磁波由线性传统媒质入射到非线性左手媒质时波的传播特性。利用时间延迟的方法,给出全反射情况下媒质界面非线性Goos-Hänchen位移表达式。分析了非线性左手媒质界面的侧向位移随入射角及入射波电场强度的变化关系,发现入射波场强对传输特性起决定作用:当入射波电场小于临界场强时,调节入射场强可以控制相应的侧向位移;当入射波电场大于临界场强时,不再满足全反射条件,部分入射波透射到非线性介质中。波导中加入非线性介质不仅可以调节侧向位移的大小,且可以实现对入射波场强的控制。     

Modeling and computation of two phase geometric biomembranes using surface finite elements

Biomembranes consisting of multiple lipids may involve phase separation phenomena leading to coexisting domains of different lipid compositions. The modeling of such biomembranes involves an elastic or bending energy together with a line energy associated with the phase interfaces. This leads to a free boundary problem for the phase interface on the unknown equilibrium surface which minimizes an energy functional subject to volume and area constraints. In this paper we propose a new computational tool for computing equilibria based on an L² relaxation flow for the total energy in which the line energy is approximated by a surface Ginzburg–Landau phase field functional. The relaxation dynamics couple a nonlinear fourth order geometric evolution equation of Willmore flow type for the membrane with a surface Allen–Cahn equation describing the lateral decomposition. A novel system is derived involving second order elliptic operators where the field variables are the positions of material points of the surface, the mean curvature vector and the surface phase field function. The resulting variational formulation uses H¹ spaces, and we employ triangulated surfaces and H¹ conforming quadratic surface finite elements for approximating solutions. Together with a semi-implicit time discretization of the evolution equations an iterative scheme is obtained essentially requiring linear solvers only. Numerical experiments are presented which exhibit convergence and the power of this new method for two component geometric biomembranes by computing equilibria such as dumbbells, discocytes and starfishes with lateral phase separation.     

抛物方程的多重非均匀网格模型及其应用

提出了抛物方程的多重非均匀网格模型,以准确求解三维空间存在多辐射源的电波传播问题。通过对不同辐射源建立不同的坐标系,并对其仿真空间采用不同的非均匀网格划分,构建了抛物方程的多重非均匀网格模型。在此基础上,实现了三维多辐射源问题的并行计算。实例仿真了空间存在四个辐射源的电波传播特性。结果表明,抛物方程的多重非均匀网格模型能够准确求解多源的空间电磁场分布特性,且在该算例中,并行技术使得抛物方程的计算速度提升了2.41倍,极大地提高了抛物方程对三维多源问题的求解效率。     

边界条件对对流扩散方程数值稳定性的影响

本文利用数值计算方法对采用均分网格的一维线性无源的对流－扩散方程在各种边界条件下的稳定性进行了分析,燕求出了不同边界条件下一维问题的中心差分和ＱＵＩＣＫ格式的临界网格Ｐｅｃｌｅｔ数。指出按现有方法得出的临界网格Ｐｅｃｌｅｔ数是判别差分格式对流数值稳定性的最苛刻的要求。对中心差分和ＱＵＩＣＫ格式,除两点边值问题以外的其它边界条件下的稳定性范围均不小于或远远大于两点边值问题的稳定性范围。通过计算还得出了格式的数值稳定性主要取决于计算区域下游侧的边界条件类型而与计算区域上游侧的边界条件类型无关的结论。     

3维楔形阴极场致发射模型

分析了场致发射的物理机理,在3维Yee网格模型的基础上导出并实现了场致发射的发射条件,并引入Pade网格算法加密尖端网格。以一个楔形阴极场致发射模型为例得到了场致发射的伏安特性曲线,并且在高电压时得到了3/2次方的空间电荷限制流。同时为了解决粒子数与网格的巨大引起的计算量大的问题,引入了消息传递机制并行算法,得到3.0以上的并行加速比。     

一种基于波动方程的新高阶FDTD方法

 提出了一种基于二阶波动方程的（2M,4）高阶时域有限差分(FDTD)方法,通过使用辛积分传播子(SIP)在时域上获得4阶精度,使用离散奇异卷积(DSC)方法在空域上达到2M阶精度。与已有的(2M,4) 阶FDTD方法相比,虽然两者都采用SIP和DSC方法,但是此二者的不同点在于：第一,新方法基于二阶波动方程;第二,在离散计算空间时使用单一网格而不是传统的Yee网格;第三,单独计算某一场分量从而节约内存并减少计算量。数值计算结果表明,与传统高阶算法相比,基于波动方程的高阶FDTD方法耗费的机时只有它的50%,内存消耗下降10%, 而两者的计算结果之间相对误差小于5‰。