Controller synthesis for infinite-state systems / synthese de controleurs pour des systemes a espace d'états infini

``软件必须运行闭环系统。‘’Robert Laddaga的这一说法源于对软件代理从传统程序演变的思考。软件代理需要更多的监视和控制，因为它们具有更大程度的自主性、适应性和移动性。动态系统控制是上世纪初诞生的一门学科。许多基于强大的数学理论的解决方案已经被提出用于连续系统，最近又被用于作为程序或软件代理的离散事件系统。在最后一种情况下，实际影响没有那么显著，尽管已经做出了几项努力来阐述一种名为监督控制理论的实质性理论。在程序执行期间控制程序意味着解决与以数学形式表示程序有关的基本问题。表示程序的一种自然方式是使用自动机，自动机具有无限大小的堆栈以及象征无界值的参数。它是由这种建模的下推系统和具有无限多个可达状态的参数化系统产生的。因此，无限状态系统的控制提出了新的重要问题，这些问题过去在监督控制理论的背景下很少被研究，并构成了本研究的主要主题。

``Software must run closed loop.'' This statement from Robert Laddaga emerges from thinking about software agent evolution from conventional programs. Software agents require more monitoring and control because of their greater degree of autonomy, adaptation and mobility. The control of dynamic systems is a discipline that is born at the beginning of last century. Many solutions founded on powerful mathematical theories have been proposed for continuous systems and more recently for discrete event systems as programs or software agents. In the last case, the real impact is less significant, although several efforts have been made to elaborate a substantial theory called Supervisory Control Theory. Controlling programs during their execution implies, amongst other, solving fundamental problems related to their representation in a mathematical form. A natural way to represent a program is to use an automaton with a stack of unlimited size as well parameters that symbolize unbounded values. It results from this modelling pushdown systems and parameterized systems with infinitely many reachable states. Therefore, the control of infinite-state systems raises new important issues, which have been study very little in the past in the context of the Supervisory Control Theory, and constitutes the main subject of the present research program.