Two-Level Space–Time Domain Decomposition Methods for Flow Control Problems

For time-dependent control problems, the class of sub-optimal algorithms is popular and the parallelization is usually applied in the spatial dimension only. In the paper, we develop a class of fully-optimal methods based on space–time domain decomposition methods for some boundary and distributed control of fluid flow and heat transfer problems. In the fully-optimal approach, we focus on the use of an inexact Newton solver for the necessary optimality condition arising from the implicit discretization of the optimization problem and the use of one-level and two-level space–time overlapping Schwarz preconditioners for the Jacobian system. We show that the numerical solution from the fully-optimal approach is generally better than the solution from the sub-optimal approach in terms of meeting the objective of the optimization problem. To demonstrate the robustness and parallel scalability and efficiency of the proposed algorithm, we present some numerical results obtained on a parallel computer with a few thousand processors.     

Reliability-based design optimization for elastoplastic mechanical structures

In this paper, two special formulations to carry out a reliability-based design optimization of elastoplastic mechanical structures are introduced. The first approach is based on a well-known two-level method where the first level involves the optimization for the design parameters whereas the evaluation of the probabilistic constraints is carried out in a sub-optimization level. Because the evaluation of the probabilistic constraints in a sub-optimization level causes non-convergence behavior for some problems as indicated in the literature, an alternative formulation based on one-level is developed considering the optimality conditions of the β-computation by which the probabilistic constraint appears in the first level reliability-based design optimization formulation. In both approaches, an explicit parameter optimization problem is proposed for the computation of a design point for elastoplastic structures.Three examples in this paper demonstrate that the one-level reliability-based design optimization formulation is superior in terms of convergence to an optimal design than the two-level reliability-based design optimization formulation.     

An empirical comparison of backtracking algorithms

In this paper we report the results of experimental studies of zero-level, one-level, and two-level search rearrangement backtracking. We establish upper and lower limits for the size problem for which one-level backtracking is preferred over zero-level and two-level methods, thereby showing that the zero-level method is best for very small problems. The one-level method is best for moderate size problems, and the two-level method is best for extremely large problems. Together with our theoretical asymptotic formulas, these measurements provide a useful guide for selecting the best search rearrangement method for a particular problem.     

Optimization of weakly coupled subsystems by a two-level method

A two-level method to find the optimal control of a linear system is presented. We consider that the system is composed of coupled subsystems. And we solve two decoupled first-level problems and one coordinating second-level problem iteratively. When the system can be decoupled into weakly coupled subsystems, the solution fast converges to the optimal.     

Biobjective optimization for analytical target cascading: optimality vs. achievability

Analytical target cascading (ATC), a hierarchical, multilevel multidisciplinary methodology, has proved to be an effective solution strategy for complex design problems. ATC decomposition and coordination incorporates compromise between the performance of the system and the demands of the subproblems reflected in their feasibility constraints. Optimal system performance regardless of subproblem feasibility may yield targets that are not achievable by the subproblems. Compromise is needed to accept deterioration of the optimal system performance and to increase the achievability of the targets. Biobjective optimization is used to reconcile system optimality and subproblem achievability of targets while solving the ATC-decomposed problem and generating an overall optimal solution. Three algorithms are proposed for two-level ATC-decomposed problems. The effectiveness of the algorithms is evaluated on mathematical and engineering examples. For a class of ATC problems, the performance of the proposed algorithms is superior to the performance of the ATC methods currently considered to be state of the art.     

A hierarchical game under uncertainty conditions with using player risk function and guarantied strategy estimate

A game-theoretic model of optimal interaction of players in a two-level hierarchic system under uncertainty conditions is proposed. The players keep in mind the outcome with the worst value of the uncertain factor, so that the upper-level player also applies guaranteed approach to the lower-level player. The convergence of the type of penalty function method we use has been proved, the problem of passing from iterated limits by the penalty parameters to the conventional limit by means of their match has been solved. The optimality conditions were obtained.     

Analysis of optimality criteria and gradient projection methods for optimal structural design

A general mathematical model for optimal design of structural and mechanical systems is defined. The finite element techniques for analysis of these systems are incorporated into the model. Optimality conditions for the model are derived. It is shown that an analytical solution of the optimality conditions is impossible except for trivial design problems. Numerical methods for solving these optimality conditions are presented. A direct method based on the gradient projection concept is derived. Numerical aspects for the method are discussed. These include step size selection, calculation of Langrange multipliers, identification of dependent design sensitivity vectors of constraint functions, and convergence criterion. Both direct and indirect approaches require total derivatives of cost and various constraint functions with respect to design variables. This is accomplished by integrating into both the approaches the state space design sensitivity analysis that has proven to be very general and efficient. It is shown that other major calculations of the two approaches are essentially the same. Thus there is potential for continuous transition between various algorithms for structural optimization. Optimal solutions for two design examples are obtained by hybrid methods to show the potential for further development of these methods.     

Methods of search for stable solutions of some optimization problems in the theory of recognition by precedents

In constructing some models of recognition algorithms, there arise a number of optimization problems. The search for the optimal consistent subsystem of a given system of linear inequalities plays an important role in the process of data analysis in the theory of recognition by precedent. The optimality of a required subsystem is defined by a number of the conditions imposed on it. Earlier the author developed several algorithms for solving the problems of this type. These algorithms are based on exhaustive search for nodal subsystems of a given system of linear inequalities. In the search for optimal consistent subsystem, these algorithms find boundary decisions. However, in practical application often it is necessary to find a stable solution. So, when looking for logical regularities of a special type, it is required to find a set of non-degenerate polyhedra describing a certain class of objects in a space of features. Therefore, linear inequalities systems corresponding to these polyhedra must be stable. In this paper, we propose a method for modifying the previously developed algorithms to select the stable consistent subsystem of highest possible power and find its stable solution.     

Mixed coordination method for long-horizon optimal control problems

We present a new approach to solving long-horizon, discrete-time optimal control problems using the mixed coordination method. The idea is to decompose a long-horizon problem into subproblems along the time axis. The requirement that the initial state of a subproblem equal the terminal state of the preceding subproblem is relaxed by using Lagrange multipliers. The Lagrange multipliers and initial state of each subproblem are then selected as high-level variables. The equivalence of the two-level formulation and the original problem is proved for both convex and non-convex cases. The low-level subproblems are solved in parallel using extended differential dynamic programming (DDP). An efficient way to find the gradient and hessian of a low-level objective function with respect to high-level variables is developed. The high-level problem is solved using the modified Newton method. An effective procedure is developed to select initial values of multipliers based on the initial trajectory. The method can convexify the high-level problem while maintaining the separability of an originally non-convex problem. The method performs better and faster than one-level DDP for both convex and non-convex test problems.     

基于分级网络的车牌字符识别算法*

提出了一种基于分级径向基神经网络的车牌字符识别算法,采用两级径向基神经网络结构,由一级网络识别后,通过对识别结果和置信度进行分析,确定了12个二级子网;采用动态误差分割算法进行网络训练,通过大量实验研究,车牌整体识别率达到了86.58%,平均识别时间为181ms,对比性研究验证了该方法的有效性。     

Numerical solutions of a nonlinear reaction-diffusion system

This paper is concerned with finite difference solutions of a coupled system of nonlinear reaction-diffusion equations. The investigation is devoted to the finite difference system for both the time-dependent problem and its corresponding steady-state problem. The existence and uniqueness of a non-negative finite difference solution and three monotone iterative algorithms for the computation of the solutions are given. It is shown that the time-dependent problem has a unique non-negative solution, whereas the steady-state problem may have multiple non-negative solutions depending on the parameters in the problem. The different non-negative steady-state solutions can be computed from the monotone iterative algorithms by choosing different initial iterations. Also discussed is the asymptotic behaviour of the time-dependent solution in relation to the steady-state solutions. The asymptotic behaviour result gives some conditions ensuring the convergence of the time-dependent solution to a positive or semitrivial non-negative steady-state solution. Numerical results are given to demonstrate the theoretical analysis results.     

Nested multigrid methods for time-periodic,parabolic optimal control problems

We present a nested multigrid method to optimize time-periodic, parabolic, partial differential equations (PDE). We consider a quadratic tracking objective with a linear parabolic PDE constraint. The first order optimality conditions, given by a coupled system of boundary value problems can be rewritten as an Fredholm integral equation of the second kind, which is solved by a multigrid of the second kind. The evaluation of the integral operator consists of solving sequentially a boundary value problem for respectively the state and the adjoints. Both problems are solved efficiently by a time-periodic space-time multigrid method.     

One-shot methods in function space for PDE-constrained optimal control problems

We state and analyse one-shot methods in function space for the optimal control of nonlinear partial differential equations (PDEs) that can be formulated in terms of a fixed-point operator. A general convergence theorem is proved by generalizing the previously obtained results in finite dimensions. As application examples we consider two nonlinear elliptic model problems: the stationary solid fuel ignition model and the stationary viscous Burgers equation. For these problems we present a more detailed convergence analysis of the method. The resulting algorithms are computationally implemented in combination with an adaptive mesh refinement strategy, which leads to an improvement in the performance of the one-shot approach.     

求解线性规划问题的光滑型牛顿算法

对线性规划的最优性条件,给出一个扩展系统,设计一个连续化的光滑型算法求解该系统。所设计的算法的全局收敛性不需要添加任何假设条件。在每一个迭代点处,只需要解一个线性方程组和做一次线性搜索,比现有求解线性规划问题的连续化方法具有更好的收敛性质。     

A new approach to differential dynamic programming for discrete time systems

This paper proposes a new differential dynamic programming algorithm for solving discrete time optimal control problems with equality and inequality constraints on both control and state variables and proves its convergence. The present algorithm is different from differential dynamic programming algorithms developed in [10]-[15], which can hardly solve optimal control problems with inequality constraints on state variables and whose convergence has not been proved. Composed of iterative methods for solving systems of nonlinear equations, it is based upon Kuhn-Tucker conditions for recurrence relations of dynamic programming. Numerical examples show file efficiency of the present algorithm.     

Convergence analysis of the Neumann–Neumann waveform relaxation method for time-fractional RC circuits

The classical waveform relaxation (WR) methods rely on decoupling the large-scale ODEs system into small-scale subsystems and then solving these subsystems in a Jacobi or Gauss–Seidel pattern. However, in general it is hard to find a clever partition and for strongly coupled systems the classical WR methods usually converge slowly and non-uniformly. On the contrary, the WR methods of longitudinal type, such as the Robin-WR method and the Neumann–Neumann waveform relaxation (NN-WR) method, possess the advantages of simple partitioning procedure and uniform convergence rate. The Robin-WR method has been extensively studied in the past few years, while the NN-WR method is just proposed very recently and does not get much attention. It was shown in our previous work that the NN-WR method converges much faster than the Robin-WR method, provided the involved parameter, namely β, is chosen properly. In this paper, we perform a convergence analysis of the NN-WR method for time-fractional RC circuits, with special attention to the optimization of the parameter β. For time-fractional PDEs, this work corresponds to the study of the NN-WR method at the semi-discrete level. We present a detailed numerical test of this method, with respect to convergence rate, CPU time and asymptotic dependence on the problem/discretization parameters, in the case of two- and multi-subcircuits.     

A robust finite element solver for a multiharmonic parabolic optimal control problem

This paper presents the analysis of a distributed parabolic optimal control problem in a multiharmonic setting. In particular, the desired state is assumed to be multiharmonic. After eliminating the control from the optimality system, we arrive at the reduced optimality system for the state and the co-state that is nothing but a coupled system of a forward and a backward parabolic partial differential equation. Due to the linearity, the state and the co-state are multiharmonic as well. We discretize the Fourier coefficients by the finite element method. This leads to a large system of algebraic equations, which fortunately decouples into smaller systems each of them defining the cosine and sine Fourier coefficients for the state and co-state with respect to a single frequency. For these smaller systems, we construct preconditioners resulting in a fast converging minimal residual solver with a parameter-independent convergence rate. All these systems can be solved totally in parallel.     

Continuous-time Wiener system identification

A continuous-time Wiener system is identified. The system consists of a linear dynamic subsystem and a memoryless nonlinear one connected in a cascade. The input signal is a stationary white Gaussian random process. The system is disturbed by stationary white random Gaussian noise. Both subsystems are identified from input-output observations taken at the input and output of the whole system. The a priori information is very small and, therefore, resulting identification problems are nonparametric. The impulse impulse of the linear part is recovered by a correlation method, while the nonlinear characteristic is estimated with the help of the nonparametric kernel regression method. The authors prove convergence of the proposed identification algorithms and examine their convergence rates     

一类带有非匹配不确定性的非线性组合系统的按指数镇定

本文讨论一类带有非匹配不确定性的非线性大系统的按指数镇定，在标称系统是按指数稳定且非匹配不确定性为等效匹配的假设条件下提出了两种鲁棒控制算法，分散控制仅利用每个子系统的状态作为反馈信息，而双层控制使用每个子系统及相关子系统的状态，最后用一个说明性的例子来说明本文所提出方法的可用性。     

A Recursive Method for Solving a Climate–Economy Model: Value Function Iterations with Logarithmic Approximations

A recursive method for solving an integrated assessment model of climate and the economy is developed in this paper. The method approximates a value function with a logarithmic basis function and searches for solutions on a set satisfying optimality conditions. These features make the method suitable for a nonlinear model with many state variables and various constraints. The method produces exact solutions to a simple economic growth model and is useful for solving more demanding models such as the well-known DICE model (dynamic integrated model of climate and the economy).