An approximate block Newton method for coupled iterations of nonlinear solvers: Theory and conjugate heat transfer applications

A new, approximate block Newton (ABN) method is derived and tested for the coupled solution of nonlinear models, each of which is treated as a modular, black box. Such an approach is motivated by a desire to maintain software flexibility without sacrificing solution efficiency or robustness. Though block Newton methods of similar type have been proposed and studied, we present a unique derivation and use it to sort out some of the more confusing points in the literature. In particular, we show that our ABN method behaves like a Newton iteration preconditioned by an inexact Newton solver derived from subproblem Jacobians. The method is demonstrated on several conjugate heat transfer problems modeled after melt crystal growth processes. These problems are represented by partitioned spatial regions, each modeled by independent heat transfer codes and linked by temperature and flux matching conditions at the boundaries common to the partitions. Whereas a typical block Gauss–Seidel iteration fails about half the time for the model problem, quadratic convergence is achieved by the ABN method under all conditions studied here. Additional performance advantages over existing methods are demonstrated and discussed.     

Numerical analysis of the 2D acoustic field with epistemic uncertainty

Aiming at the problem that the epistemic uncertain parameters exist in an acoustic field,an evidence theory-based finite element method(ETFEM) is proposed by introducing the evidence theory,in which the focal element and basic probability assignment(BPA) are used to describe the epistemic uncertainty.In order to reduce the computational cost,the interval analysis technique based on perturbation method is adopted to acquire the approximate sound pressure response bounds for each focal element.The corresponding formulations of intervals of expectation and standard deviation of the sound pressure response with epistemic uncertainty are deduced.The sound pressure response of a 2D acoustic tube and a 2D car acoustic cavity with epistemic uncertain parameters are analyzed by the proposed method.The proposed method is verified through the comparison of the analysis results of random acoustic field with that of epistemic uncertain acoustic field.Numerical analysis results show that the proposed method can analyze the 2D acoustic field with epistemic uncertainty effectively,and has good prospect of engineering application.     

认知不确定二维声场的响应特性数值分析

针对声学参数存在认知不确定性的问题,为实现认知不确定声场声压响应的预测。提出了解决二维认知不确定声场的有限元法(Evidence Theory-based Finite Element Method,ETFEM),引入证据理论,采用焦元和基本可信度的概念来描述认知不确定参数,基于摄动法的区间分析技术,推导了认知不确定声场声压响应的标准差、期望求解公式。为验证本文方法的可行性。以认知不确定参数下的二维管道声场模型和某轿车二维声腔模型为例进行了数值计算,对比离散随机变量得到认知不确定参数的声场分析结果和相应的随机声场所得分析结果,研究表明:该方法能够有效的处理认知不确定参数下的二维声场,为工程问题中噪声预测提供可靠的分析模型。      

On the Convergence of Kikuchi’s Natural Iteration Method

In this article we investigate on the convergence of the natural iteration method, a numerical procedure widely employed in the statistical mechanics of lattice systems, to minimize Kikuchi’s cluster variational free energies. We discuss a sufficient condition for the convergence, based on the coefficients of the cluster entropy expansion, depending on the lattice geometry. We also show that such a condition is satisfied for many lattices usually studied in applications. Finally, we consider a recently proposed class of methods for the minimization of Kikuchi functionals, showing that the natural iteration method turns out as a particular instance of that class.     

An efficient and general numerical method to compute steady uniform vortices

Steady uniform vortices are widely used to represent high Reynolds number flows, yet their efficient computation still presents some challenges. Existing Newton iteration methods become inefficient as the vortices develop fine-scale features; in addition, these methods cannot, in general, find solutions with specified Casimir invariants. On the other hand, available relaxation approaches are computationally inexpensive, but can fail to converge to a solution. In this paper, we overcome these limitations by introducing a new discretization, based on an inverse-velocity map, which radically increases the efficiency of Newton iteration methods. In addition, we introduce a procedure to prescribe Casimirs and remove the degeneracies in the steady vorticity equation, thus ensuring convergence for general vortex configurations. We illustrate our methodology by considering several unbounded flows involving one or two vortices. Our method enables the computation, for the first time, of steady vortices that do not exhibit any geometric symmetry. In addition, we discover that, as the limiting vortex state for each flow is approached, each family of solutions traces a clockwise spiral in a bifurcation plot consisting of a velocity-impulse diagram. By the recently introduced “IVI diagram” stability approach [Phys. Rev. Lett. 104 (2010) 044504], each turn of this spiral is associated with a loss of stability for the steady flows. Such spiral structure is suggested to be a universal feature of steady, uniform-vorticity flows.     

Potential-based reduced Newton algorithm for nonlinear multiphase flow in porous media

We present a phase-based potential ordering that is an extension of the Cascade ordering introduced by Appleyard and Cheshire [John R. Appleyard, Ian M. Cheshire, The cascade method for accelerated convergence in implicit simulators, in: European Petroleum Conference, 1982, pp. 113–122]. The proposed ordering is valid for both two-phase and three-phase flow, and it can handle countercurrent flow due to gravity and/or capillarity. We show how this ordering can be used to reduce the nonlinear algebraic system that arises from the fully-implicit method (FIM) into one with only pressure dependence. The potential-based reduced Newton algorithm is then obtained by applying Newton’s method to this reduced-order system. Numerical evidence shows that our potential-based reduced Newton solver is able to converge for time steps that are much larger than what the standard Newton’s method can handle. In addition, whenever standard Newton converges, so does the reduced Newton algorithm, and the number of global nonlinear iterations required for convergence is significantly reduced compared with the standard Newton’s method.     

The residue harmonic balance for fractional order van der Pol like oscillators

When seeking a solution in series form, the number of terms needed to satisfy some preset requirements is unknown in the beginning. An iterative formulation is proposed so that when an approximation is available, the number of effective terms can be doubled in one iteration by solving a set of linear equations. This is a new extension of the Newton iteration in solving nonlinear algebraic equations to solving nonlinear differential equations by series. When Fourier series is employed, the method is called the residue harmonic balance. In this paper, the fractional order van der Pol oscillator with fractional restoring and damping forces is considered. The residue harmonic balance method is used for generating the higher-order approximations to the angular frequency and the period solutions of above mentioned fractional oscillator. The highly accurate solutions to angular frequency and limit cycle of the fractional order van der Pol equations are obtained analytically. The results that are obtained reveal that the proposed method is very effective for obtaining asymptotic solutions of autonomous nonlinear oscillation systems containing fractional derivatives. The influence of the fractional order on the geometry of the limit cycle is investigated for the first time.     

An analysis on convergence and convergence rate estimate of elitist genetic algorithms in noisy environments

Random noise perturbs objective functions in many practical problems, and genetic algorithms (GAs) have been widely proposed as an effective optimization tool for dealing with noisy objective functions. However, little papers for convergence and convergence speed of genetic algorithms in noisy environments (GA-NE) have been published. In this paper, a Markov chain that models elitist genetic algorithms in noisy environments (EGA-NE) was constructed under the circumstance that objective function is perturbed only by additive random noise, and it was proved to be an absorbing state Markov chain. The convergence of EGA-NE was proved on the basis of the character of the absorbing state Markov chain, its convergence rate was analyzed, and its upper and lower bounds for the iteration number expectation were derived when EGA-NE first gets a globally optimal solution.     

水声通信加权反馈双向Turb均衡器

为了提升自适应双向Turbo均衡器的收敛速度及降低误比特率,提出了采用加权反馈的双向Turbo均衡算法。首先在单个均衡器反馈输入中采用后验均值与先验均值混合的反馈方案,有效提升一轮迭代中均衡器输出的准确性;其次通过后验均值与先验均值的加权合并作为另一均衡器反馈的非因果项输入,在提升反馈输入准确性的同时提升了数据的利用率;最后在权值迭代中采用优化的比例归一化最小均方算法,提升训练阶段均衡器收敛速度。千岛湖试验中,在同样3.75 kbps通信速率的情况下,该方法误比特率仅为传统双向Turbo均衡器的1/3。仿真和试验数据表明,均衡器要达到同样的误比特率,本方法所需迭代轮数更少,在时变信道中系统稳定性更好,误比特率更低,提升了水声通信效率。      

一类离散时间广义系统的迭代学习控制

针对一类离散时间广义系统,提出了一种离散迭代学习控制算法.首先,通过非奇异变换将离散时间广义系统分解为正常离散状态方程和代数方程的形式.然后,利用上一次迭代学习获得的前一时刻误差和当前时刻误差来修正上一次的控制量,从而获得下一次迭代学习的新控制量,并对算法的收敛性进行了理论证明,给出了算法收敛的充分条件.研究结果表明,所提算法能够在有限时间区间内实现系统状态对期望状态的完全跟踪.最后,通过仿真算例进一步验证了所提算法的有效性.     

非定常Navier-Stokes方程基于两重网格离散的有限元并行算法

基于两重网格离散和区域分解技巧,提出三种求解非定常Navier-Stokes方程的有限元并行算法.算法的基本思想是在每一时间迭代步,在粗网格上采用Oseen迭代法求解非线性问题,在细网格上分别并行求解Oseen、Newton、Stokes线性问题以校正粗网格解.对于空间变量采用有限元离散,时间变量采用向后Euler格式离散.数值实验验证了算法的有效性.     

The Convergence Estimation of the Parallel Algorithm of the Linear Cauchy Problem Solution for Large Systems of First-Order Ordinary Differential Equations Using the Solution as Expansion over Orthogonal Polynomials

This paper is devoted to the algorithm of the linear Cauchy problem solution for large systems of first-order ordinary differential equations using parallel calculations. The proof of the convergence of the iteration process using the solution as expansion over orthogonal polynomials for the interval [0,1] is presented. The features of this algorithm are its simplicity, the opportunity to get a solution by parallel calculations, and also the possibility to get a solution for nonlinear problems by changing the operator using the solution from the iteration process.     

Efficient full Newton–Raphson technique for the solution of molecular integral equations – example of the SPC/E water-like system

It is shown how a full Newton–Raphson technique speeds up in impressive proportions the iterative resolution of molecular integral equations and makes it possible to reach quadratically complete convergence down to machine precision in a very few cycles. The technique generalises what has been originally proposed by Zerah and extensively used since then with great success for various fluids and mixtures of spherical objects. At each main iteration, the linearised cycle obtained by differentiating the Ornstein–Zernike and the integral equations is itself solved iteratively in terms of Δg^mnl_μν(r) projections. Its solution is reached very rapidly thanks to the powerful biconjugate gradient method and to the absence of any Euler angle manipulation. The virial equation is written in a shape formally different from the standard one, which allows a much higher numerical precision for the pressure without extra numerical work. The complete scheme is illustrated on the popular SPC/E water model.     

An implicit immersed boundary method for three-dimensional fluid–membrane interactions

We present an implicit immersed boundary method for the incompressible Navier–Stokes equations capable of handling three-dimensional membrane–fluid flow interactions. The goal of our approach is to greatly improve the time step by using the Jacobian-free Newton–Krylov method (JFNK) to advance the location of the elastic membrane implicitly. The most attractive feature of this Jacobian-free approach is Newton-like nonlinear convergence without the cost of forming and storing the true Jacobian. The Generalized Minimal Residual method (GMRES), which is a widely used Krylov-subspace iterative method, is used to update the search direction required for each Newton iteration. Each GMRES iteration only requires the action of the Jacobian in the form of matrix–vector products and therefore avoids the need of forming and storing the Jacobian matrix explicitly. Once the location of the boundary is obtained, the elastic forces acting at the discrete nodes of the membrane are computed using a finite element model. We then use the immersed boundary method to calculate the hydrodynamic effects and fluid–structure interaction effects such as membrane deformation. The present scheme has been validated by several examples including an oscillatory membrane initially placed in a still fluid, capsule membranes in shear flows and large deformation of red blood cells subjected to stretching force.     

Pair approximation for polarization interaction and adiabatic nuclear and electronic sampling method for fluids with dipole polarizability

The adiabatic nuclear and electronic sampling method (ANES), originally formulated as an efficient Monte Carlo algorithm for systems with fluctuating charges, is applied to the simulation of a polarizable water model with induced dipole moments. Structural, thermodynamic and dipolar properties obtained by ANES and a newer algorithm, the pair approximation for polarization interaction (PAPI), are compared with full iteration. With the best parameters, the inaccuracy of both approximate methods was found to be comparable with the uncertainty of the full iteration. The PAPI method with iteration radius equal to the second minimum of the oxygen—oxygen correlation function is, depending on the convergence tolerance, 10–15 times faster than the full iteration for 256 molecules, and yields very accurate structure and thermodynamics with deviation about 0.3%. When the iteration radius is increased to the cutoff distance, exact results are recovered at the cost of decreased efficiency. The ANES method with small nuclear displacements proved to inefficiently sample the configurational space. Simulations at low electronic temperatures with large nuclear displacements are inaccurate for up to 100 electronic moves, and increasing this number would make the simulations as slow as the full iteration. The most accurate and efficient adiabatic ANES simulations are those with infinite electronic temperature, large nuclear displacements and 1–10 electronic moves. The extra freedom of induced dipoles in the ANES method at high electronic temperatures modifies the observed dipolar properties; however, the question of whether the dielectric constant is also modified needs further consideration.     

A framework for iterative analysis of non-classically damped dynamical systems

In this paper, we propose a general iterative framework to solve the dynamic problem for linear systems with non-classical viscous damping. A systematic approach is used to derive families of stationary iterative schemes that, as an instance of particular interest, decouple the equations of motion for numerical study of the system response. For such schemes, we present a detailed convergence analysis and propose several solution strategies suitable for a broad class of systems. These techniques are based on spectral analysis of particular iteration matrices arising in the derivation and aim at optimizing the convergence performance of the method. We demonstrate that the proposed systematic framework, based on a novel application of the homotopy analysis method, generalizes iterative schemes previously reported in the literature and, importantly, provides a unified perspective for the study of iterative solutions of dynamic problems. Further, we establish a connection between our results and the theory of iterative schemes for algebraic linear systems, thus providing insights on convergence results and applicability of the method. Numerical examples illustrate the effectiveness of the approach and indicate future research directions.     

Method for fitting crystal field parameters and the energy level fitting for Yb3+ in crystal Sc2O3

<正>A method to compute the numerical derivative of eigenvalues of parameterized crystal field Hamiltonian matrix is given,based on the numerical derivatives the general iteration methods such as Levenberg-Marquardt,Newton method, and so on,can be used to solve crystal field parameters by fitting to experimental energy levels.With the numerical eigenvalue derivative,a detailed iteration algorithm to compute crystal field parameters by fitting experimental energy levels has also been described.This method is used to compute the crystal parameters of Yb~(3+) in Sc_2O_3 crystal, which is prepared by a co-precipitation method and whose structure was refined by Rietveld method.By fitting on the parameters of a simple overlap model of crystal field,the results show that the new method can fit the crystal field energy splitting with fast convergence and good stability.     

一种新的稳健波束形成算法及其一维搜索策略

存在条件失配时自适应波束形成器的性能急剧下降,凸优化技术的引入使稳健波束形成器的设计更加灵活,但同时带来了计算复杂度的增加和工程实现上的困难.针对上述问题,提出了一种基于最小二乘估计的稳健波束形成算法,并推导得到一种基于一维搜索的求解方法.首先利用广义旁瓣对消器的结构将标准Capon波束形成器转化为稳健最小二乘问题,并将该问题转化为二阶锥规划的形式.为了减少计算量,利用二阶锥规划问题的原始问题和对偶问题的关系,将求解过程转化为一维搜索,并利用牛顿迭代法获得最优解,从而获得与标准Capon波束形成相近的计算复杂度.仿真分析表明,该算法具有良好的抗导向矢量失配和快拍数不足的稳健性.     

On Entropic Uncertainty Relations for Measurements of Energy and Its “Complement”

Heisenberg's uncertainty principle in application to energy and time is a powerful heuristics. This statement plays an important role in foundations of quantum theory and statistical physics. If some state exists for a finite interval of time, then it cannot have a completely definite value of energy. It is well known that the case of energy and time principally differs from more familiar examples of two non‐commuting observables. Since quantum theory was originated, many approaches to energy–time uncertainties have been proposed. Entropic way to formulate the uncertainty principle is currently the subject of active researches. Using the Pegg concept of complementarity of the Hamiltonian, uncertainty relations of the “energy–time” type are obtained in terms of Rényi and Tsallis entropies. Although this concept is somehow restricted in scope, derived relations can be applied to systems typically used in quantum information processing. Both the state‐dependent and state‐independent formulations are of interest. Some of the derived state‐independent bounds are similar to the results obtained within a more general approach on the basis of sandwiched relative entropies. The developed method allows us to address the case of detection inefficiencies.     

Two-Level Newton Iteration Methods for Navier-Stokes Type Variational Inequality Problem

This paper deals with the two-level Newton iteration method based on the pressure projection stabilized finite element approximation to solve the numerical solution of the Navier-Stokes type variational inequality problem. We solve a small Navier-Stokes problem on the coarse mesh with mesh size $H$ and solve a large linearized Navier-Stokes problem on the fine mesh with mesh size $h$. The error estimates derived show that if we choose $h=\mathcal{O}(|\log h|^{1/2}H^3)$, then the two-level method we provide has the same $H^1$ and $L^2$ convergence orders of the velocity and the pressure as the one-level stabilized method. However, the $L^2$ convergence order of the velocity is not consistent with that of one-level stabilized method. Finally, we give the numerical results to support the theoretical analysis.