Search

This paper concerns a global optimality principle for fully coupled mean-field control systems. Both the first-order and the second-order variational equations are fully coupled mean-field linear FBSDEs. A new linear relation is introduced, with which we successfully decouple the fully coupled first-order variational equations. We give a new second-order expansion of Yε that can work well in mean-field framework. Based on this result, the stochastic maximum principle is proved. The comparison with the stochastic maximum principle for controlled mean-field stochastic differential equations is supplied. [ABSTRACT FROM AUTHOR]

This paper derives necessary optimality conditions for a certain class of optimal control problems without linearity or convexity assumptions. The optimal control problem has a general objective function of integral type and a finite number of isoperimetric type constraints. For proving the main result derived in this paper, the Lagrange function and the control Hamiltonian are introduced and an adjoint differential equation is stated. In addition, we formulate some examples where the derived necessary optimality conditions are applied. [ABSTRACT FROM AUTHOR]

The connection between numerical methods for solving differential equations and machine learning has been revealed recently. Differential equations have been proposed as continuous analogues of deep neural networks, and then used in handling certain tasks, such as image recognition, where the training of a model includes learning the parameters of systems of ODEs from certain points along their trajectories. Treating this inverse problem of determining the parameters of a dynamical system that minimize the difference between data and trajectory by a gradient-based optimization method presents the solution of the adjoint equation as the continuous analogue of backpropagation that yields the appropriate gradients. The paper explores an abstract approach that can be used to construct a family of loss functions with the aim of fitting the solution of an initial value problem to a set of discrete or continuous measurements. It is shown, that an extension of the adjoint equation can be used to derive the gradient of the loss function as a continuous analogue of backpropagation in machine learning. Numerical evidence is presented that under reasonably controlled circumstances the gradients obtained this way can be used in a gradient descent to fit the solution of an initial value problem to a set of continuous noisy measurements, and a set of discrete noisy measurements that are recorded at uncertain times. [ABSTRACT FROM AUTHOR]

This paper is concerned with a class of mean-field type stochastic optimal control systems, which are governed by fully coupled mean-field forward-backward stochastic differential equations with Teugels martingales associated to Lévy processes. In these systems, the coefficients contain not only the state processes but also their marginal distribution, and the cost function is of mean-field type as well. The necessary and sufficient conditions for such optimal problems are obtained. Furthermore, the applications to the linear quadratic stochastic optimization control problem are investigated. [ABSTRACT FROM AUTHOR]

This paper studies the risk-sensitive optimal control problem for a backward stochastic system. More precisely, we set up a necessary stochastic maximum principle for a risk-sensitive optimal control of this kind of equations. The control domain is assumed to be convex and the generator coefficient of such system is allowed to be depend on the control variable. As a preliminary step, we study the risk-neutral problem for which an optimal solution exists. This is an extension of initial control system to this type of problem, where the set of admissible controls is convex. An example to carried out to illustrate our main result of risk-sensitive control problem under linear stochastic dynamics with exponential quadratic cost function. [ABSTRACT FROM AUTHOR]

This paper considers the numerical solution of optimal control problems based on ODEs. We assume that an explicit Runge-Kutta method is applied to integrate the state equation in the context of a recursive discretization approach. To compute the gradient of the cost function, one may employ Automatic Differentiation (AD). This paper presents the integration schemes that are automatically generated when differentiating the discretization of the state equation using AD. We show that they can be seen as discretization methods for the sensitivity and adjoint differential equation of the underlying control problem. Furthermore, we prove that the convergence rate of the scheme automatically derived for the sensitivity equation coincides with the convergence rate of the integration scheme for the state equation. Under mild additional assumptions on the coefficients of the integration scheme for the state equation, we show a similar result for the scheme automatically derived for the adjoint equation. Numerical results illustrate the presented theoretical results. [ABSTRACT FROM AUTHOR]

This paper is mainly concerned with the solutions to both forward and backward mean-field stochastic partial differential equation and the corresponding optimal control problem for mean-field stochastic partial differential equation. The authors first prove the continuous dependence theorems of forward and backward mean-field stochastic partial differential equations and show the existence and uniqueness of solutions to them. Then they establish necessary and sufficient optimality conditions of the control problem in the form of Pontryagin's maximum principles. To illustrate the theoretical results, the authors apply stochastic maximum principles to study the infinite-dimensional linear-quadratic control problem of mean-field type. Further, an application to a Cauchy problem for a controlled stochastic linear PDE of mean-field type is studied. [ABSTRACT FROM AUTHOR]

In the present paper, the Ulam stability of the first order linear dynamic equation and its adjoint equation on a time scale is established by using the integrating factor method. Meantime, two examples are provided to illustrate our main results. [ABSTRACT FROM AUTHOR]

This paper studies an optimal control problem for partially observed forward-backward stochastic control system with a convex control domain and the forward diffusion term containing control variable. A maximum principle is proved for this kind of partially observable optimal control problems and the corresponding adjoint processes are characterized by the solutions of certain forward-backward stochastic differential equations in finite-dimensional spaces. One partially observed recursive linear-quadratic (LQ) optimal control example is also given to show the application of the obtained maximum principle. An explicit observable optimal control is obtained in this example. [ABSTRACT FROM AUTHOR]

The paper is concerned with a stochastic optimal control problem where the controlled systems are driven by Teugel’s martingales and an independent multi-dimensional Brownian motion. Necessary and sufficient conditions for an optimal control of the control problem with the control domain being convex are proved by the classical method of convex variation, and the coefficients appearing in the systems are allowed to depend on the control variables. As an application, the linear quadratic stochastic optimal control problem is studied. [ABSTRACT FROM AUTHOR]

The optimum planning of industrial pollutant sources, which optimizes the economic object without violating environmental constraints, is an important and hard task to be conquered. In this paper, an adjoint method is developed to solve the problem. The penalty function is introduced to deal with the environmental inequality constraints, and Lagrange function is constructed to derive the adjoint equation and the gradient of the object function. In this means, the gradient of the object function can be calculated by solving the adjoint equation, and the information from the gradient is used to make the object function descend and approach to an optimal solution after some iterations. A two-dimensional, simplified model is used for numerical experiments. The theoretical derivations are verified by the results of the experiments. Furthermore, the adjoint method is shown to be of excellent convergence and efficiency, which is adaptive to the fast development of air quality numerical models and super computers. [ABSTRACT FROM AUTHOR]

This paper deals with the numerical solution of optimal control problems for ODEs. The methods considered here rely on some standard optimization code to solve a discretized version of the control problem under consideration. We aim to make available to the optimization software not only the discrete objective functional, but also its gradient. The objective gradient can be computed either from forward (sensitivity) information or backward (adjoint) information. The purpose of this paper is to discuss various ways of adjoint computation. It will be shown both theoretically and numerically that methods based on the continuous adjoint equation require a careful choice of both the integrator and gradient assembly formulas in order to obtain a gradient consistent with the discretized control problem. Particular attention is given to automatic differentiation techniques which generate automatically a suitable integrator. [ABSTRACT FROM AUTHOR]

This paper is devoted to the analysis of adjoint pairs of regular differential-algebraic equations with arbitrarily high tractability index. We consider both standard form DAEs and DAEs with properly involved derivative. We introduce the notion of factorization-adjoint pairs and show their common structure including index and characteristic values. We precisely describe the relations between the so-called inherent explicit regular ODE (IERODE) and the essential underlying ODEs (EUODEs) of a regular DAE. We prove that among the EUODEs of an adjoint pair of regular DAEs there are always those which are adjoint to each other. Moreover, we extend the Lyapunov exponent theory to DAEs with arbitrarily high index and establish the general class of DAEs being regular in Lyapunov's sense. The Perron identity which is well known in the ODE theory does not hold in general for adjoint pairs of Lyapunov regular DAEs. We establish criteria for the Perron identity to be valid. Examples are also given for illustrating the new results. [ABSTRACT FROM AUTHOR]

In this paper, we consider optimal control problems governed by linear unsteady fluid-structure interaction problems. Based on a novel symmetric monolithic formulation, we derive optimality systems and provide regularity results for optimal solutions. The proposed formulation allows for natural application of gradient-based optimization algorithms and for space-time finite element discretizations. [ABSTRACT FROM AUTHOR]

In this paper, we deal with near optimality for stochastic control problems where the controlled system is described by a linear fully coupled forward–backward stochastic differential equation.We assume that the forward diffusion coefficient depends explicitly on the control variable and the control domain is not necessarily convex.We establish necessary as well as sufficient conditions for near optimality, satisfied by all near optimal controls. The proof of the main result is based on Ekeland’s variational principle and some estimates on the state and the adjoint processes with respect to the control variable. Finally an example which illustrate our results is presented. [ABSTRACT FROM AUTHOR]

Adjoint methods have recently gained considerable importance in the finance sector, because they allow to quickly compute option sensitivities with respect to a large number of model parameters. In this paper we investigate how the efficiency of adjoint methods can be exploited to speed up the Monte Carlo-based calibration of financial market models. After analyzing the calibration problem both theoretically and numerically, we derive the associated adjoint equation and propose its application in combination with a multi-layer method, for which we prove convergence to a stationary point of the underlying optimization problem. Detailed numerical examples illustrate the performance of the method. In particular, the proposed algorithm reduces the calibration time for a typical equity market model with time-dependent model parameters from over three hours to less than ten minutes on a usual desktop PC. [ABSTRACT FROM AUTHOR]

We consider a stochastic control problem in the case where the set of control domain is convex, the system is governed by a nonlinear forward-backward doubly stochastic differential equation with given terminal condition. The criteria to be minimized is in the general form, with initial and terminal costs. We derive a maximum principle of optimality. [ABSTRACT FROM AUTHOR]

We consider a stochastic control problem where the system is governed by a non linear stochastic differential equation with jumps. The control is allowed to enter into both diffusion and jump terms. By only using the first order expansion and the associated adjoint equation, we establish necessary as well as sufficient optimality conditions of controls for relaxed controls, who are a measure-valued processes. [ABSTRACT FROM AUTHOR]

Sufficient conditions are given for the asymptotic constancy of the solutions of a linear system of difference equations with delays. Moreover, it is shown that the limits of the solutions, as t→∞, can be computed in terms of the initial function and a special matrix solution of the corresponding adjoint equation. [ABSTRACT FROM AUTHOR]

In this article, we obtain the adjoint difference equation for the Nikiforov–Uvarov–Suslov difference equation of hypergeometric type on non-uniform lattices, and prove it to be a difference equation of hypergeometric type on non-uniform lattices as well. The particular solutions of the adjoint difference equation are then obtained. As an application of these particular solutions, we use them to obtain the particular solutions for the original difference equation of hypergeometric type on non-uniform lattices. In addition, we give another kind of fundamental theorems for the Nikiforov–Uvarov–Suslov difference equation of hypergeometric type, which are essentially new results and their expressions are different from the Suslov Theorem. Finally, we give an example to illustrate the application of the new fundamental theorems. [ABSTRACT FROM AUTHOR]

In this article, we obtain a new fundamental theorems for Nikiforov–Uvarov–Suslov complex difference equation of hypergeometric type by the method of Euler integral transformation, its expression is different from Suslov's Theorem. We also establish the adjoint equation for Nikiforov–Uvarov–Suslov difference equation of hypergeometric type on non-uniform lattices, and prove it to be a difference equation of hypergeometric type on non-uniform lattices as well. The particular solutions of the adjoint equation are then obtained. As an appliction of these particular solutions, we use them to obtain the particular solutions for the original difference equation of hypergeometric type on non-uniform lattices and other important results. [ABSTRACT FROM AUTHOR]

We consider the free endpoint Mayer problem for a controlled Moreau process, the control acting as a perturbation of the dynamics driven by the normal cone, and derive necessary optimality conditions of Pontryagin’s Maximum Principle type. The results are also discussed through an example. We combine techniques from Sene and Thibault (Journal of Nonlinear and Convex Analysis 15, 647-663, 2014) and from Brokate and Krejčí (Discrete and Continuous Dynamical Systems Series B 18, 331-348, 2013), which in particular deals with a different but related control problem. Our assumptions include the smoothness of the boundary of the moving set C(t), but, differently from Brokate and Krejčí, do not require strict convexity and time independence of C(t). Rather, a kind of inward/outward pointing condition is assumed on the reference optimal trajectory at the times where the boundary of C(t) is touched. The state space is finite dimensional. [ABSTRACT FROM AUTHOR]

This study develops a direct optimal growth algorithm for three-dimensional transient growth analysis of perturbations in channel flows which are globally stable but locally unstable. Different from traditional non-modal methods based on the Orr-Sommerfeld and Squire (OSS) equations that assume simple base flows, this algorithm can be applied to arbitrarily complex base flows. In the proposed algorithm, a re-orthogonalization Arnoldi method is used to improve orthogonality of the orthogonal basis of the Krylov subspace generated by solving the linearized forward and adjoint Navier-Stokes (N-S) equations. The linearized adjoint N-S equations with the specific boundary conditions for the channel are derived, and a new convergence criterion is pro-posed. The algorithm is then applied to a one-dimensional base flow (the plane Poiseuille flow) and a two-dimensional base flow (the plane Poiseuille flow with a low-speed streak) in a channel. For one-dimensional cases, the effects of the spanwise width of the chan-nel and the Reynolds number on the transient growth of perturbations are studied. For two-dimensional cases, the effect of strength of initial low-speed streak is discussed. The presence of the streak in the plane Poiseuille flow leads to a larger and quicker growth of the perturbations than that in the one-dimensional case. For both cases, the results show that an optimal flow field leading to the largest growth of perturbations is character-ized by high-and low-speed streaks and the corresponding streamwise vortical structures. The lift-up mechanism that induces the transient growth of perturbations is discussed. The performance of the re-orthogonalization Arnoldi technique in the algorithm for both one-and two-dimensional base flows is demonstrated, and the algorithm is validated by comparing the results with those obtained from the OSS equations method and the cross-check method. [ABSTRACT FROM AUTHOR]

Cardiac muscle tissue during relaxation is commonly modeled as a hyperelastic material with strongly nonlinear and anisotropic stress response. Adapting the behavior of such a model to experimental or patient data gives rise to a parameter estimation problem which involves a significant number of parameters. Gradient-based optimization algorithms provide a way to solve such nonlinear parameter estimation problems with relatively few iterations, but require the gradient of the objective functional with respect to the model parameters. This gradient has traditionally been obtained using finite differences, the calculation of which scales linearly with the number of model parameters, and introduces a differencing error. By using an automatically derived adjoint equation, we are able to calculate this gradient more efficiently, and with minimal implementation effort. We test this adjoint framework on a least squares fitting problem involving data from simple shear tests on cardiac tissue samples. A second challenge which arises in gradient-based optimization is the dependency of the algorithm on a suitable initial guess. We show how a multi-start procedure can alleviate this dependency. Finally, we provide estimates for the material parameters of the Holzapfel and Ogden strain energy law using finite element models together with experimental shear data. [ABSTRACT FROM AUTHOR]

We consider a nonlinear stochastic optimal control problem associated with a stochastic evolution equation. This equation is driven by a continuous martingale in a separable Hilbert space and an unbounded time-dependent linear operator. We derive a stochastic maximum principle for this optimal control problem. Our results are achieved by using the adjoint backward stochastic partial differential equation. [ABSTRACT FROM AUTHOR]

We present a spectrally accurate numerical method for finding nontrivial time-periodic solutions of nonlinear partial differential equations. The method is based on minimizing a functional (of the initial condition and the period) that is positive unless the solution is periodic, in which case it is zero. We solve an adjoint PDE to compute the gradient of this functional with respect to the initial condition. We include additional terms in the functional to specify the free parameters, which in the case of the Benjamin–Ono equation, are the mean, a spatial phase, a temporal phase, and the real part of one of the Fourier modes at t=0. We use our method to study global paths of nontrivial time-periodic solutions connecting stationary and traveling waves of the Benjamin–Ono equation. As a starting guess for each path, we compute periodic solutions of the linearized problem by solving an infinite dimensional eigenvalue problem in closed form. We then use our numerical method to continue these solutions beyond the realm of linear theory until another traveling wave is reached. By experimentation with data fitting, we identify the analytical form of the solutions on the path connecting the one-hump stationary solution to the two-hump traveling wave. We then derive exact formulas for these solutions by explicitly solving the system of ODEs governing the evolution of solitons using the ansatz suggested by the numerical simulations. [ABSTRACT FROM AUTHOR]

In this work, a deterministic and a probabilistic method for the delineation of well head protection areas are applied and compared. The deterministic method was implemented using the automatic backward particle tracking algorithm (APA, Tosco et al., Water Resour Res, 44(7):W07419, 2008). The backward probability model rests upon the backward adjoint-based model developed by Neupauer and Wilson, and allows the inclusion of dispersion in the definition of capture zones. The two methods are evaluated comparing the “advective front” of the probability protection area and the perimeter given by the particle tracking method. Furthermore, a semi-quantitative study was performed over probability protection areas, in order to evaluate the influence of dispersivity on the extent and growth rate of capture zones identified by fixed probability isolines. [ABSTRACT FROM AUTHOR]

Consider the third order differential operator L given by $$L\left(\cdot\right) \equiv \frac{1}{{a_3 (t)}}\frac{d}{{dt}}\frac{1}{{a_2 (t)}}\frac{d}{{dt}}\frac{1}{{a_1 (t)}}\frac{d}{{d(t)}}\left(\cdot\right)$$ and the related linear differential equation L( x)( t) + x( t) = 0. We study the relations between L, its adjoint operator, the canonical representation of L, the operator obtained by a cyclic permutation of coefficients a i, i = 1,2,3, in L and the relations between the corresponding equations. We give the commutative diagrams for such equations and show some applications (oscillation, property A). [ABSTRACT FROM AUTHOR]