文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > Finite Time--Horizon Risk Sensitive Control and the Robust Limit under a Quadratic Growth A
侵权投诉

Finite Time--Horizon Risk Sensitive Control and the Robust Limit under a Quadratic Growth A

The finite time--horizon risk sensitive limit problem for continuous, nonlinear systems is considered. Previous results are extended to cover more typical examples. In particular, the cost may grow quadratically, and the diffusion coefficient may depend on

Finite Time{Horizon Risk Sensitive Control and the Robust Limit under a Quadratic Growth Assumption Francesca Da Lio William M. McEneaneyy October 21, 1998 Abstract The nite time{horizon risk sensitive limit problem for continuous, nonlinear systems is considered. Previous results are extended to cover more typical examples. In particular, the cost may grow quadratically, and the di usion coe cient may depend on the state. It is shown that the risk sensitive value function is the solution of the corresponding dynamic programming equation. It is also shown that this value converges to the value of the Robust control problem as the cost becomes in ntely risk averse with corresponding scaling of the di usion coe cient. Key Words: risk-sensitive control, robust, H1, viscosity solutions, nonlinear HJB equations, nonlinear Isaacs equations AMS subject classi cations: 35B37, 49L25, 90D25, 93B36, 93C10, 93E05, 93E20

1 Introduction The nonlinear, nite time{horizon risk sensitive limit problem is considered. It is, by now, well known that the value functions of risk{sensitive stochastic control problems tend to converge to the value functions of the corresponding Robust/H1 control problems as one approaches in nite risk aversion. This was addressed rst in the LEQG (linear{exponential{quadratic{gaussian) case where, in fact, one does not need to take the risk averse limit 23], 42], 5], 9]. In the nonlinear, case, results were rst developed for nite time{horizon control problems 43], 24], 14], 32], 33]. Further studies considered nonlinear, in nite time{horizon problems (in which case one gets the H1 limit) 15], 33], 16], 13], 36], 18], 38], 28], 34], 4] and nonlinear escape problems 27], 7]. Other studies have involved discrete systems 39], 10], 12], and the partial observations case 6], 25]. The results for both the nite time{horizon problem and the in nite time{horizon problem have mainly been obtained under assumptions which preclude quadratic cost criteria. In other words, the LEQG results were not subsumed by the nonlinear results. Since control systems for nonlinear systems are often designed by analogy with the linear{quadratic theory, it would be most desirable for the nonlinear theory to subsume the LEQG theory. For the in nite time-horizon problem, this is addressed in a preliminary fashion in 22] and 28] (with some results taken from 29], 30] in this latter case). The in nite time{ horizon case presents certain technical di culties which are not present in the nite time horizon case. In particular, there is a certain lack of uniqueness of solutions to the dynamic programming equation Dept. of Mathematics, University of Padova, via Belzoni, 7 35131 Padova, ITALY, dalio@math.unipd.it of Mathematics, North Carolina State University, Raleigh, NC 27695{8205, USA, wmm@math.ncsu.edu. Research partially supported by AFOSR grant F49620-95-1-0296. y Dept.

第1页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email