Asymptotic expansions for interior penalty solutions of control constrained linear-quadratic problems
نویسندگان
چکیده
We consider a quadratic optimal control problem governed by a nonautonomous affine differential equation subject to nonnegativity control constraints. For a general class of interior penalty functions, we show how to compute the principal term of the pointwise expansion of the state and the adjoint state. Our main argument relies on the following fact: If the control of the initial problem satisfies strict complementarity conditions for its Hamiltonian except for a finite number of times, the estimates for the penalized optimal control problem can be derived from the estimations of a related stationary problem. Our results provide several types of efficiency measures of the penalization technique: error estimations of the control for L norms (s in [1,+∞]), error estimations of the state and the adjoint state in Sobolev spaces W 1,s (s in [1,+∞)) and also error estimates for the value function. For the L norm and the logarithmic penalty, the optimal results are given. In this case we indeed establish that the penalized control and the value function errors are of order O(ε| log ε|). Key-words: Optimal control, interior-point algorithms, sensitivity, expansion of value function and solutions. ∗ Department of Mathematical Engineering, Universidad de Chile and Centre for Mathematical Modelling, Universidad de Chile-CNRS. Casilla 170-3 Santiago 3, Chile. This author was partially supported by grants from FONDECYT 1050706 and the Millennium Scientific Institute on Complex Engineering Systems funded by MIDEPLAN-Chile ([email protected]) † Université Paris 06, Equipe Combinatoire et Optimisation, UMR 7090, Case 189, 75252 Paris, France and INRIA-Saclay and CMAP, École Polytechnique, 91128 Palaiseau, France ([email protected]) ‡ INRIA-Saclay and CMAP, École Polytechnique, 91128 Palaiseau, France ([email protected]) § INRIA-Saclay and CMAP, École Polytechnique, 91128 Palaiseau, France ([email protected]) Développements asymptotiques pour les solutions intérieures de pénalité d’un problème linéaire quadratique avec contraintes sur la commande Résumé : On considère un problème de commande optimale dont la fonction coût est quadratique et la dynamique est régie par une équation différentielle ordinaire. Pour une classe générale de fonctions de pénalité intérieure, on montre comment calculer le terme principal du développement ponctuelle de l’état et de l’état adjoint. Notre argument principal se base sur le fait suivant: Si la commande optimale pour le problème initial satisfait les conditions de complémentarité stricte pour son Hamiltonien sauf en un nombre fini d’instants, les estimations pour le problème de commande optimale pénalisé peuvent être obtenues à partir des estimations pour un problème stationnaire associé. Nos résultats fournissent plusieurs types de mesures d’efficacité pour la technique de pénalisation: estimations des erreurs de la commande pour les normes L (s dans [1,+∞]), estimations des erreurs pour l’état et l’état adjoint dans les espaces de Sobolev W 1,s (s dans [1,+∞)) et aussi des estimations des erreurs pour la fonction valeur. Pour la norme L et la pénalisation logarithmique, on montre que les erreurs pour la commande optimale du problème pénalise et pour la fonction valeur sont de l’ordre O(ε| log ε|). Mots-clés : Commande optimale, algorithmes de point intérieur, sensibilité, développement de la valeur et des solutions. Asymptotic expansions for interior solutions of linear-quadratic problems 3
منابع مشابه
Superlinearly convergent exact penalty projected structured Hessian updating schemes for constrained nonlinear least squares: asymptotic analysis
We present a structured algorithm for solving constrained nonlinear least squares problems, and establish its local two-step Q-superlinear convergence. The approach is based on an adaptive structured scheme due to Mahdavi-Amiri and Bartels of the exact penalty method of Coleman and Conn for nonlinearly constrained optimization problems. The structured adaptation also makes use of the ideas of N...
متن کاملA constructive interior penalty method for optimal control problems
This paper exposes a methodology which allows us to address constrained optimal control of non linear systems by interior penalty methods. A constructive choice for the penalty functions that are introduced to account for the constraints is established in the article. It is shown that it allows us to approach the solution of the non linear optimal control problem using a sequence of unconstrain...
متن کاملA Fully Unconstrained Interior Point Algorithm for Multivariable State and Input Constrained Optimal Control Problems
This paper exposes a methodology to solve constrained optimal control problems for non linear systems using interior penalty methods. A constructive choice for the penalty functions that are introduced to account for the constraints is established in the article. It is shown that this choice allows one to approach a solution of the non linear optimal control problem using a sequence of unconstr...
متن کاملHigh order Newton-penalty algorithms ?
Recent efforts in differentiable non-linear programming have been focused on interior point methods, akin to penalty and barrier algorithms. In this paper, we address the classical equality constrained program solved using the simple quadratic loss penalty function/algorithm. The suggestion to use extrapolations to track the differentiable trajectory associated with penalized subproblems goes b...
متن کاملFast interior point solution of quadratic programming problems arising from PDE-constrained optimization
Interior point methods provide an attractive class of approaches for solving linear, quadratic and nonlinear programming problems, due to their excellent efficiency and wide applicability. In this paper, we consider PDE-constrained optimization problems with bound constraints on the state and control variables, and their representation on the discrete level as quadratic programming problems. To...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید
ثبت ناماگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید
ورودعنوان ژورنال:
- Math. Program.
دوره 135 شماره
صفحات -
تاریخ انتشار 2012