Two‐objective optimization strategies using the adjoint method and game theory in inverse natural convection problems

This paper considers the problem of estimating the strengths of two time‐varying heat sources simultaneously, from measurements of the temperature inside the square domain in a porous medium, when prior knowledge of the source functions is not available. This problem is an inverse natural convection problem. In order to circumvent this problem, we define several optimization criteria (objective functionals) that measure discrepancies between model and measured data, where objective functionals depend on two heat sources and use multi‐criteria optimization to identify Nash equilibria, which are solutions to the non‐cooperative game according to game theory. Two non‐cooperative game strategies are considered: competitive (Nash) game and hierarchical (modified Stackelberg) game. The methodology that we employ relies on a combination of mixed finite element space approximations, finite difference time discretizations, adjoint equation and sensitivity equation techniques, and nonlinear conjugate gradient algorithms for the solutions of estimating two heat sources. Applying the Sobolev gradient for the noise removal is investigated. The performance of the present technique of inverse analysis is evaluated, by means of several numerical experiments, and is found to be very accurate as well as efficient. Copyright © 2012 John Wiley & Sons, Ltd.     

The emergence of leadership in coordination-intensive groups

Although group coordination was introduced to psychology in the early 1990s, it was not until the advent of nonlinear dynamical systems (NDS) that it was possible to gain an understanding of how the process of implicit learning and self-organization take place in conditions where no hierarchical (management) influences are involved. This experimental study examined how leaders might emerge from coordination-intensive task groups where verbal interaction is not possible. NDS and game-theoretical research indicate that the presence of leaders is not required to produce coordination. Thus the question remains as to whether leaders would emerge from coordination-intensive task groups in the similar manner to how they emerge from other types of groups. In the experiment, 13 4-person groups were allowed to discuss the coordination (card game) task while performing it; 13 other groups worked nonverbally. Split-plot ANOVA showed that verbalizing groups performed better than nonverbalizing groups overall and showed more acute coordination learning curves. Nonlinear regression for temporal dynamics within verbalizing and nonverbalizing groups showed asymptotic stability for initial coordination learning and transfer to a coordination rule of equal difficulty, but a chaotic function was observed when the teams switched to a more difficult coordination rule. A questionnaire measured leadership emergence at the end of the game along with other social contributions to the groups' efforts. The average level of leadership emergence for individuals did not differ between verbal and nonverbal conditions, although differences in other social contributions were observed. This experiment illustrates, furthermore, how the nonlinear science paradigm produces new hypotheses concerning verbalization that would not have been formulated otherwise. New avenues of study concerning coordination, leadership, and hierarchies are also discussed.     

Origins of group coordination: nonlinear dynamics and the role of verbalization

Coordination occurs when two or more people do the same or complimentary tasks simultaneously; its explanation game theory, nonlinear dynamics, and implicit learning theory. In the experiment, 12 four-person groups were allowed to discuss the coordination (card game) task while performing it; 12 other groups worked nonverbally. One to three group members were replaced during the game. Split-plot ANOVA showed that verbalizing groups performed better than nonverbalizing groups overall and showed more acute coordination learning curves, but verbalization did not compensate for the replacement of personnel. Groups that changed one or two players showed positive coordination transfer, but groups that changed three players did not. Nonlinear regression for temporal dynamics within verbalizing and nonverbalizing groups showed asymptotic stability for initial coordination learning and transfer to a difficult rule, a chaotic function when replacements were introduced, and asymptotic stability again when the team with replacements switched to the difficult rule.     

Nonlinear models for the adoption and diffusion of innovations for industrial energy conservation

Four different theoretical models for explaining the diffusion of innovation were compared for 13 energy-related innovations--the Theory of Planned Behavior, the S-curve for Diffusion of Innovations, the power law distribution, and the cusp catastrophe. The substantive concern was to explore the roles of facilitative and obstructive factors in diffusing industrial and commercial innovations. Participants were 102 industrial plant and facilities managers from sites that were among the top users of electrical energy and natural gas in the United States. They completed a survey that contained measurements of positive attitudes toward innovation, organizational resistance to innovation, and the extent to which they had investigated or adopted each of the target innovations. Seven of the 13 innovations exhibited strong cusp catastrophe models (via nonlinear regression, average R2 = .91) compared to linear alternative models (average R2 = .31) for those innovations; the S-curve for diffusion was regarded as a simplified version of the cusp. One innovation was characterized best by a power law distribution (R2 = .94), and the remaining five were characterized best by a linear model that was based on the Theory of Planned Behavior (R2 = .41). Different underlying dynamics for the various innovations were implied by these results.     

Phase Transformations in Electrically Conductive Ferromagnetic Shape-Memory Alloys, Their Thermodynamics and Analysis

We derive a thermodynamically consistent general continuum-mechanical model describing mutually coupled martensitic and ferro/paramagnetic phase transformations in electrically-conductive magnetostrictive materials such as NiMnGa. We use small-strain and eddy-current approximations, yet large velocities and electric current injected through the boundary are allowed. Fully nonlinear coupling of magneto-mechanical and thermal effects is considered. The existence of energy-preserving weak solutions is proved by showing convergence of time-discrete approximations constructed by a carefully designed semi-implicit regularized scheme.     

Evolving Graphs by Singular Weighted Curvature

A new notion of solutions is introduced to study degenerate nonlinear parabolic equations in one space dimension whose diffusion effect is so strong at particular slopes of the unknowns that the equation is no longer a partial differential equation. By extending the theory of viscosity solutions, a comparison principle is established. For periodic continuous initial data a unique global continuous solution (periodic in space) is constructed. The theory applies to motion of interfacial curves by crystalline energy or more generally by anisotropic interfacial energy with corners when the curves are the graphs of functions. Even if the driving force term (homogeneous in space) exists, the initial-value problem is solvable for general nonadmissible continuous (periodic) initial data. (Accepted July 5, 1996)     

Multi-objective Optimization Strategies Using Adjoint Method and Game Theory in Aerodynamics

There are currently three different game strategies originated in economics: (1) Cooperative games (Pareto front), (2) Competitive games (Nash game) and (3) Hierarchical games (Stackelberg game). Each game achieves different equilibria with different performance, and their players play different roles in the games. Here, we introduced game concept into aerodynamic design, and combined it with adjoint method to solve multi-criteria aerodynamic optimization problems. The performance distinction of the equilibria of these three game strategies was investigated by numerical experiments. We computed Pareto front, Nash and Stackelberg equilibria of the same optimization problem with two conflicting and hierarchical targets under different parameterizations by using the deterministic optimization method. The numerical results show clearly that all the equilibria solutions are inferior to the Pareto front. Non-dominated Pareto front solutions are obtained, however the CPU cost to capture a set of solutions makes the Pareto front an expensive tool to the designer.The project supported by the National Natural Science Foundation of China (10372040) and Scientific Research Foundation (SRF) for Returned Oversea’s Chinese Scholars (ROCS) (2003-091). The English text was polished by Yunming Chen.     

Second Order Abstract Neutral Functional Differential Equations

In this paper we are concerned with a class of second order abstract neutral functional differential equations with finite delay in a Banach space. We establish the existence of mild and classical solutions for the nonlinear equation, and we show that the map defined by the mild solutions of the linear equation is a strongly continuous semigroup of bounded linear operators on an appropriate space. We use this semigroup to establish a variation of constants formula to solve the inhomogeneous linear equation.     

Tracking control of uncertain switched nonlinear cascade systems: a nonlinear H ∞ sliding mode control method

This paper considers the tracking control problem for a class of uncertain switched nonlinear cascade systems via the multiple Lyapunov functions (MLFs) method. Each subsystem under consideration is composed of two cascade-connected parts: the null space dynamics part and the range space dynamics part. The two main robust control strategies, nonlinear H _∞ control (NHC) and the sliding mode control (SMC), are integrated to function in a complementary manner for tracking control tasks. Furthermore, sufficient conditions for the solvability of the tracking control problem of the switched system and design of both switching laws and controllers are presented. Finally, a simulation example is provided to demonstrate the feasibility of the developed method.     

Solitary Waves in Nonlinear Beam Equations: Stability,Fission and Fusion

We continue work by the second author and co-workers onsolitary wave solutions of nonlinear beam equations and their stabilityand interaction properties. The equations are partial differentialequations that are fourth-order in space and second-order in time.First, we highlight similarities between the intricate structure ofsolitary wave solutions for two different nonlinearities; apiecewise-linear term versus an exponential approximation to thisnonlinearity which was shown in earlier work to possess remarkablystable solitary waves. Second, we compare two different numericalmethods for solving the time dependent problem. One uses a fixed griddiscretization and the other a moving mesh method. We use these methodsto shed light on the nonlinear dynamics of the solitary waves. Earlywork has reported how even quite complex solitary waves appear stable,and that stable waves appear to interact like solitons. Here we show twofurther effects. The first effect is that large complex waves can, as aresult of roundoff error, spontaneously decompose into two simplerwaves, a process we call fission. The second is the fusion of twostable waves into another plus a small amount of radiation.     

Unsteady water wave modulations: fully nonlinear solutions and comparison with the nonlinear Schrödinger equation

The time evolution of a uniform wave train with a small modulation which grows is computed with a fully nonlinear irrotational flow solver. Many numerical runs have been performed varying the initial steepness of the wave train and the number of waves in the imposed modulation. It is observed that the energy becomes focussed into a short group of steep waves which either contains a wave which becomes too steep and therefore breaks or otherwise having reached a maximum modulation then recedes until an almost regular wave train is recovered. This latter case typically occurs over a few hundred time periods. We have also carried out some much longer computations, over several thousands of time periods in which several steep wave events occur. Several features of these modulations are consistent with analytic solutions for modulations using weakly nonlinear theory, which leads to the nonlinear Schrödinger equation. The steeper events are shorter in both space and time than the lower events. Solutions of the nonlinear Schrödinger equation can be transformed from one steepness to another by suitable scaling of the length and time variables. We use this scaling on the modulations and find excellent agreement particularly for waves that do not grow too steep. Hence the number of waves in the initial modulation becomes an almost redundant parameter and allows wider use of each computation. A potentially useful property of the nonlinear Schrödinger equation is that there are explicit solutions which correspond to the growth and decay of an isolated steep wave event. We have also investigated how changing the phase of the initial modulation effects the first steep wave event that occurs.     

Harmonic Balance Based Averaging: Approximate Realizations of an Asymptotic Technique

Averaging is a classical asymptotic technique commonly used to studyweakly nonlinear oscillations via small perturbations of the harmonicoscillator. If the unperturbed oscillator is autonomous and stronglynonlinear, but with a two-parameter family of periodic solutions, thenaveraging is allowed in principle but typically not considered feasibleunless (a) the required family of unperturbed periodic solutions can befound in closed form, and (b) the averaging integrals can be found inclosed form. Often, the foregoing requirements cannot be met. Here, itis shown how both these difficulties can be bypassed using the classicalbut heuristic approximation method of harmonic balance, to obtain approximate realizations of the asymptotic analytical technique. Theadvantages of the present approach are that (a) closed form solutions tothe unperturbed problem are not needed, and (b) the heuristic andasymptotic parts of the calculation are kept conceptually distinct, withscope for refining the former, while preserving the asymptotic nature ofthe latter. Several examples are provided, including oscillators with astrong cubic nonlinearity, velocity dependent nonlinear terms (includinga strongly nonconservative system), a nondifferentiable characteristic,and a strongly nonlinear but homogeneous function of order 1; dynamicphenomena investigated include damped oscillations, limit cycles, forcedoscillations near resonance, and subharmonic entrainment. Goodapproximations are obtained in each case.     

Social-support moderated stress: a nonlinear dynamical model and the stress-buffering hypothesis

Health psychology has studied the cross-sectional, stationary relationships linking stress, social support, and health. Levels of stress-related illness are generally modeled by including a nonlinear multiplicative or 'buffering' effect, corresponding to the interaction of stressor levels with social support from family and friends. The motivation of the present research is to extend an iterative, dynamic model of this well-investigated psychological process using a dynamical systems model expressed as a set of continuous, nonlinear differential equations similar to those of the 'Oregonator,' a model of a nonlinear dynamic chemical system. This model of the behavior of an individual is amenable to numerical investigation of its stationary-state stability properties, temporal evolution, and cause-effect relationships. The continuous variables in this new approach refer to varying states of an individual; they are Perceived stress (X), Symptoms (Y), and Social support (Z). It is expected that poor health in this model, represented by Symptoms (Y), is directly related to Perceived stress, as well as being tied in more complicated ways to Social support. A number of such models may be envisioned, some including a multiplicative, 'buffering' (- X x Z) effect of social support dependent on stress levels. We explore the behavior of this model over ranges of parameter values and initial conditions and relate these results to how an individual reacts to environmental challenges at various levels of stressors and social-support recruitment. Data generated by the model are in turn analyzed with a traditional cross-sectional statistical technique. Similarities and differences between chemical and psychological systems are discussed.     

<Emphasis Type="Italic">L</Emphasis><Superscript>1</Superscript> Stability of the Boltzmann Equation for the Hard-Sphere Model

We study the L¹ stability of classical solutions to the Boltzmann equation for a hard-sphere model, when initial datum is a small perturbation of a vacuum, and tends to zero exponentially fast at infinity in the phase space. For this, we introduce nonlinear functionals measuring potential interactions between particles with different velocities and L¹ distance between classical solutions. We use pointwise estimates for a solution and the gain term of a collision operator to control the time-evolution of nonlinear functionals.Dedicated to Marshall Slemrod on the occasion of his 60th birthday     

Computational Aeroacoustics: An Overview of Computational Challenges and Applications

The objective of this paper is to present an overview of recent advances in computational aeroacoustics (CAA). During the last decade, CAA has developed quite independent of computational fluid dynamics (CFD). There are computational issues that are unique to CAA and are, generally, not considered in CFD. In this paper, these issues are discussed and explained. In CAA, there is a great need to resolve high-frequency short waves with the minimum number of mesh points per wavelength. There is also a special need to minimize numerical dispersion and dissipation associated with wave propagation computation. All these have led to the development of large-stencil high-resolution schemes for CAA. A careful examination of dispersion and dissipation errors due to spatial and temporal discretization is provided. These errors are quantified and analyzed in wave number space through the use of Fourier-Laplace transforms. At this time, some of the original computational challenges to CAA have been resolved satisfactorily. A discussion of how some of these computational issues are resolved is presented. Several important CAA applications with interesting or unusual computational innovations are highlighted. Finally, a few of the most pressing outstanding computational challenges to CAA are elaborated.     

On the accuracy of a nonlinear finite volume method for the solution of diffusion problems using different interpolations strategies

In this paper, we consider a nonlinear finite volume method to solve the steady‐state diffusion equation in nonhomogeneous and non‐isotropic media. The method is nonlinear even if the original problem is linear. In its original form, the scheme is monotone, because the coefficient matrix is monotone under certain assumptions and, as a consequence, whenever the analytic operator demands, it preserves the positivity of numerical solutions. On the other hand, the scheme is unable to reproduce piecewise linear solutions exactly. In order to recover this interesting feature, we use two different interpolation strategies. In this case, even though we are unable to prove monotonicity, we show some numerical evidences that the combined method has an improved behavior, producing second order accurate solutions, even for nonhomogeneous and strongly anisotropic media. Copyright © 2013 John Wiley & Sons, Ltd.     

Singularities and spectral asymptotics of a random nonlinear wave in a nondispersive system

In this paper, we study the propagation of high-intensity acoustic noise in free space and in waveguide systems. A mathematical model generalizing the Burgers equation is used. It describes the nonlinear wave evolution inside tubes of variable cross-section, as well as in ray tubes, if the geometric approximation for heterogeneous media is used. The generalized equation transforms to the common Burgers equation with a dissipative parameter, known as the “Reynolds–Goldberg number”. In our model, this number depends on the distance travelled by the wave. With a zero “viscous” dissipative term, the model reduces to the Riemann (or Hopf) equation. Its solution presents the field by an implicit function. The spectral form of this solution makes it possible to derive explicit expressions for both dynamic and statistical characteristics of intense waves. The use of a spectral approach allowed us to describe the high-intensity noise in media with zero and finite viscosity. Applicability conditions of these solutions are defined. Since the phase matching is fulfilled for any triplet of interacting spectral components, there is an avalanche-like increase in the number of harmonics and the formation of shocks. The relationship between these discontinuities and other singularities and the high-frequency asymptotic of intense noise is studied. The possibility is shown to enhance nonlinear effects in waveguide systems during the evolution of noise.     

Numerical solutions to regularized long wave equation based on mixed covolume method

The mixed covolume method for the regularized long wave equation is developed and studied. By introducing a transfer operator γh , which maps the trial function space into the test function space, and combining the mixed finite element with the finite volume method, the nonlinear and linear Euler fully discrete mixed covolume schemes are constructed, and the existence and uniqueness of the solutions are proved. The optimal error estimates for these schemes are obtained. Finally, a numerical example is provided to examine the efficiency of the proposed schemes.     

基于混合行为博弈的多目标仿生设计方法

将多个设计目标视为不同的博弈方,通过计算设计变量对目标函数的影响因子和模糊聚类, 将设计变量集合分割为各博弈方拥有的策略空间. 对蜥蜴种群的繁衍生存机理进行仿生,将 3种蜥蜴的行为方式分别定义为利己主义、集体主义和投机主义,并赋予相应的博弈方, 各博弈方根据所仿生蜥蜴的行为特点,建立自身博弈得益函数与目标函数之间的映射关系. 各博弈方分别以自身博弈得益函数为目标,在各自的策略空间中进行单目标优化,获得本博 弈方对其余博弈方的最佳对策,所有博弈方的最佳对策形成一轮博弈的策略组合,并根据收 敛判别,通过多轮博弈,获得最终的博弈解. 以白鹤滩拱坝体型的三目标优化设计为例,设 计结果显示坝体体积方量减少了16.412万方,或2.38%; 最大主拉应力降低 了0.036MPa, 或0.31%; 整体应变能下降了0.167GJ 或4.47% 体现了基 于混合行为博弈方式的多目标仿生设计方法的有效性.