STABLE AND EFFICIENT COMPUTATIONAL METHODS FOR DYNAMIC PROGRAMMING
针对动态经济分析中动态规划求解的数值方法，指出标准方法存在慢或不稳定的问题，提出采用现代数值优化和近似方法可避免不稳定，并利用并行计算架构求解大规模问题。
Dynamic programming is the foundation of dynamic economic analysis and often requires numerical solution methods. Standard methods are either slow or unstable. These instabilities are avoided when one uses modern methods from numerical optimization and approximation. Furthermore, large dynamic programming problems can be solved by using modern parallel computing architectures.