Using randomization to break the curse of dimensionality

被引:186
作者
Rust, J
机构
关键词
dynamic programming; curse of dimensionality; Bellman operator; random Bellman operator; computational complexity; maximal inequalities; empirical processes;
D O I
10.2307/2171751
中图分类号
F [经济];
学科分类号
02 ;
摘要
This paper introduces random versions of successive approximations and multigrid algorithms for computing approximate solutions to a class of finite and infinite horizon Markovian decision problems (MDPs). We prove that these algorithms succeed in breaking the ''curse of dimensionality'' for a subclass of MDPs known as discrete decision processes (DDFs).
引用
收藏
页码:487 / 516
页数:30
相关论文
共 42 条
[1]  
ANDERSON E, 1996, HDB COMPUTATIONAL EC, P171
[2]  
[Anonymous], 2010, Dynamic programming
[3]  
[Anonymous], ANN MATH STAT, DOI DOI 10.1214/AOMS/1177700285
[4]   A UNIFIED FRAMEWORK FOR THE DISCRETIZATION OF NON-LINEAR OPERATOR-EQUATIONS [J].
ANSELONE, PM ;
ANSORGE, R .
NUMERICAL FUNCTIONAL ANALYSIS AND OPTIMIZATION, 1981, 4 (01) :61-99
[5]  
ANSELONE PM, 1971, PRENTICE HALL SERIES
[6]  
BAKHVALOV NS, 1964, NUMERICAL METHODS SO, P5
[7]   LEARNING TO ACT USING REAL-TIME DYNAMIC-PROGRAMMING [J].
BARTO, AG ;
BRADTKE, SJ ;
SINGH, SP .
ARTIFICIAL INTELLIGENCE, 1995, 72 (1-2) :81-138
[8]  
Bellman R., 1963, Mathe- matics of Computation, V17, P155
[9]  
Bellman RE., 1962, Applied dynamic programming
[10]   CONVERGENCE OF DISCRETIZATION PROCEDURES IN DYNAMIC-PROGRAMMING [J].
BERTSEKAS, DP .
IEEE TRANSACTIONS ON AUTOMATIC CONTROL, 1975, AC20 (03) :415-419