State-constrained Optimal Control Problems Research Articles

Combination of Direct Methods and Homotopy in Numerical Optimal Control: Application to the Optimization of Chemotherapy in Cancer

We consider a state-constrained optimal control problem of a system of two non-local partial-differential equations, which is an extension of the one introduced in a previous work in mathematical oncology. The aim is to minimize the tumor size through chemotherapy while avoiding the emergence of resistance to the drugs. The numerical approach to solve the problem was the combination of direct methods and continuation on discretization parameters, which happen to be insufficient for the more complicated model, where diffusion is added to account for mutations. In the present paper, we propose an approach relying on changing the problem so that it can theoretically be solved thanks to a Pontryagin Maximum Principle in infinite dimension. This provides an excellent starting point for a much more reliable and efficient algorithm combining direct methods and continuations. The global idea is new and can be thought of as an alternative to other numerical optimal control techniques.

Structure Exploitation in an Interior-Point Method for Fully Discretized, State Constrained Optimal Control Problems

We discuss a direct discretization method for state-constrained optimal control problems and an interior-point method, which is used to solve the resulting large-scale and sparse nonlinear optimization problems. The main focus of the paper is on the investigation of an efficient method to solve the occurring linear equations with saddle-point structure. To this end, we exploit the particular structure that arises from the optimal control problem and the discretization scheme and use a tailored linear algebra solver alglin in combination with a re-ordering of the saddle-point matrices. Numerical experiments for a simple optimal control problem show a significant speed-up compared to state-of-the-art sparse LU decomposition methods like MA57 or MUMPS in combination with Ipopt.

Primal-Dual Extragradient Methods for Nonlinear Nonsmooth PDE-Constrained Optimization

We study the extension of the Chambolle--Pock primal-dual algorithm to nonsmooth optimization problems involving nonlinear operators between function spaces. Local convergence is shown under technical conditions including metric regularity of the corresponding primal-dual optimality conditions. We also show convergence for a Nesterov-type accelerated variant provided one part of the functional is strongly convex. We show the applicability of the accelerated algorithm to examples of inverse problems with $L^1$- and $L^\infty$-fitting terms as well as of state-constrained optimal control problems, where convergence can be guaranteed after introducing an (arbitrary small, still nonsmooth) Moreau--Yosida regularization. This is verified in numerical examples.

Semi-definite relaxations for optimal control problems with oscillation and concentration effects

Converging hierarchies of finite-dimensional semi-definite relaxations have been proposed for state-constrained optimal control problems featuring oscillation phenomena, by relaxing controls as Young measures. These semi-definite relaxations were later on extended to optimal control problems depending linearly on the control input and typically featuring concentration phenomena, interpreting the control as a measure of time with a discrete singular component modeling discontinuities or jumps of the state trajectories. In this contribution, we use measures introduced originally by DiPerna and Majda in the partial differential equations literature to model simultaneously, and in a unified framework, possible oscillation and concentration effects of the optimal control policy. We show that hierarchies of semi-definite relaxations can also be constructed to deal numerically with nonconvex optimal control problems with polynomial vector field and semialgebraic state constraints.

Line search globalization of a semismooth Newton method for operator equations in Hilbert spaces with applications in optimal control

We consider the numerical solution of nonlinear and nonsmooth operator equations in Hilbert spaces. A semismooth Newton method is used for search direction generation. The operator equation is solved by a globalized semismooth Newton method that is equipped with an Armijo linesearch using a semismooth merit function. We prove that an accumulation point of the globalized algorithm is a solution and transition to fast local convergence under a directional Hadamard-like continuity assumption on the Newton matrix. In particular, no auxiliary descent directions or smoothing steps are required. Finally, we apply this method to a control-constrained and also to a regularized state-constrained optimal control problem subject to partial differential equations.

State-constrained optimal control problems governed by coupled nonlinear wave equations with memory

This work considers the optimal control problems governed by coupled nonlinear wave equations with memory, where the state constraint is a type of integral. First, we investigate the existence of the optimal solution for the optimal control problem, and then we deduce the maximum principle for the optimal solution with state constraint. Moreover, we give a concrete example which is covered by the main result.

State constrained $L^\infty$ optimal control problems interpreted as differential games

We consider state constrained optimal control problems in which the cost to minimize comprises an $L^\infty$ functional, i.e. the maximum of a running cost along the trajectories. In absence of state constraints, a new approach has been suggested by a recent paper [9]. The main purpose of the present paper is to extend this approach and the related results to state constrained $L^\infty$ optimal control problems. More precisely, using the $(L^\infty, L^1)$-duality, the reference optimal control problem can be seen as a <em> static differential game</em>, in which an extra variable is introduced and plays the role of an opponent player who wants to <em> maximize</em> the cost. Under appropriate assumptions and employing suitable Filippov's type results, this static game turns out to be equivalent to the corresponding <em> dynamic differential game</em>, whose (upper) value function is the unique viscosity solution to a <em> constrained boundary value problem</em>, which involves a Hamilton-Jacobi equation with a <em> continuous</em> Hamiltonian.

Optimal Control Problem for Cahn–Hilliard Equations with State Constraint

This paper is concerned with the state-constrained optimal control problem governed by Cahn---Hilliard equations. After showing the relationship between the control problem and its approximation, we derive the optimality conditions for an optimal control of our original problem by using the one of the approximate problems.

The Constants in A Posteriori Error Indicator for State-Constrained Optimal Control Problems with Spectral Methods

We employ Legendre-Galerkin spectral methods to solve state-constrained optimal control problems. The constraint on the state variable is an integration form. We choose one-dimensional case to illustrate the techniques. Meanwhile, we investigate the explicit formulae of constants within a posteriori error indicator.

State-Constrained Optimal Control Problems of Impulsive Differential Equations

The present paper studies an optimal control problem governed by measure driven differential systems and in presence of state constraints. The first result shows that using the graph completion of the measure, the optimal solutions can be obtained by solving a reparametrized control problem of absolutely continuous trajectories but with time-dependent state-constraints. The second result shows that it is possible to characterize the epigraph of the reparametrized value function by a Hamilton-Jacobi equation without assuming any controllability assumption.

Preconditioners for state-constrained optimal control problems with Moreau-Yosida penalty function

SUMMARY Optimal control problems with partial differential equations as constraints play an important role in many applications. The inclusion of bound constraints for the state variable poses a significant challenge for optimization methods. Our focus here is on the incorporation of the constraints via the Moreau–Yosida regularization technique. This method has been studied recently and has proven to be advantageous compared with other approaches. In this paper, we develop robust preconditioners for the efficient solution of the Newton steps associated with the fast solution of the Moreau–Yosida regularized problem. Numerical results illustrate the efficiency of our approach. Copyright © 2012 John Wiley & Sons, Ltd.

Preface

This issue of Numerical Algebra, Control and Optimization is dedicated to Professor Helmut Maurer on the occasion of his 65th birthday. Professor Maurer has made many outstanding contributions to the field of optimal control, particularly in terms of sufficient optimality conditions, sensitivity analysis, and computational methods for singular and state-constrained optimal control problems. Professor Maurer also has a keen interest in solving practical optimal control models arising in applications such as robotics, economics, chemistry and biomedical engineering. <br><br> For more information please click the “Full Text” above.

Uniqueness Criteria for the Adjoint Equation in State-Constrained Elliptic Optimal Control

The article considers linear elliptic equations with regular Borel measures as inhomogeneity. Such equations frequently appear in state-constrained optimal control problems. By a counter example of Serrin [18], it is known that, in the presence of non-smooth data, a standard weak formulation does not ensure uniqueness for such equations. Therefore several notions of solution have been developed that guarantee uniqueness. In this note, we compare different definitions of solutions, namely the ones of Stampacchia [19] and Boccardo-Galouët [4] and the two notions of solutions of [2, 7], and show that they are equivalent. As side results, we reformulate the solution in the sense of [19], and prove the existence of solutions in the sense of [2, 4, 7] in case of mixed boundary conditions.

Pontryagin’s principle for state-constrained evolutionary differential complementarity system

This paper is devoted to the state-constrained optimal control problem of evolutionary variational inequality. In this paper, the control domain is not necessarily convex. Moreover, since our state constraint is quite general and, in many cases, it requires pointwise behavior of the state, the framework of the partial differential equation (instead of the abstract framework) is used. Some optimality conditions (in the form of Pontryagin’s principle) for optimal controls are established.

Deterministic state-constrained optimal control problems without controllability assumptions

In the present paper, we consider nonlinear optimal control problems with constraints on the state of the system. We are interested in the characterization of the value function without any controllability assumption. In the unconstrained case, it is possible to derive a characterization of the value function by means of a Hamilton-Jacobi-Bellman (HJB) equation. This equation expresses the behavior of the value function along the trajectories arriving or starting from any position $x$. In the constrained case, when no controllability assumption is made, the HJB equation may have several solutions. Our first result aims to give the precise information that should be added to the HJB equation in order to obtain a characterization of the value function. This result is very general and holds even when the dynamics is not continuous and the state constraints set is not smooth. On the other hand we study also some stability results for relaxed or penalized control problems.

The smoothed-penalty algorithm for state constrained optimal control problems for partial differential equations

We present an algorithm for the solution of general inequality constrained optimization problems. The algorithm is based upon an exact penalty function that is approximated by a family of smooth functions. We present convergence results. As numerical examples we treat state constrained optimal control problems for elliptic partial differential equations. We compare the results with existing methods.

On the Use of Outer Approximations as an External Active Set Strategy

Outer approximations are a well known technique for solving semiinfinite optimization problems. We show that a straightforward adaptation of this technique results in a new, external, active-set strategy that can easily be added to existing software packages for solving nonlinear programming problems with a large number of inequality constraints. Our external active-set strategy is very easy to implement, and, as our numerical results show, it is particularly effective when applied to discretized semiinfinite optimization or state-constrained optimal control problems. Its effects can be spectacular, with reductions in computing time that become progressively more pronounced as the number of inequalities is increased.

A Priori Error Analysis for Linear Quadratic Elliptic Neumann Boundary Control Problems with Control and State Constraints

In this paper we consider a state-constrained optimal control problem with boundary control, where the state constraints are imposed only in an interior subdomain. Our goal is to derive a priori error estimates for a finite element discretization with and without additional regularization. We will show that the separation of the set where the control acts and the set where the state constraints are given improves the approximation rates significantly. The theoretical results are illustrated by numerical computations.

Finite Element Approximations of an Optimal Control Problem with Integral State Constraint

An integral state-constrained optimal control problem governed by an elliptic partial differential equation and its finite element approximation are considered. The finite element approximation is constructed on multimeshes. An $L^2$-norm a priori error estimate of the finite element approximation is obtained. Further, some superconvergence results are proved. Based on these superconvergence results, almost optimal $L^\infty$-norm error estimates are derived. Some recovery algorithms are then proposed to produce a posteriori error estimators of gradient type. To solve the finite element system, a simple and yet efficient iterative gradient projection algorithm is proposed and its convergence rate is proved. Some numerical examples are performed to confirm theoretical analysis.

A sensitivity-based extrapolation technique for the numerical solution of state-constrained optimal control problems

Sensitivity analysis (with respect to the regularization parameter) of the solution of a class of regularized state constrained optimal control problems is performed. The theoretical results are then used to establish an extrapolation-based numerical scheme for solving the regularized problem for vanishing regularization parameter. In this context, the extrapolation technique provides excellent initializations along the sequence of reducing regularization parameters. Finally, the favorable numerical behavior of the new method is demonstrated and a comparison to classical continuation methods is provided.