lu bang sui ji dong tai gui hua zhong de xin xi song chi dui ou xing

With the rapid development of social economy, economics research has become increasingly complex, and traditional quantitative mathematical models have been difficult to simulate economics. Stochastic dynamic programming has begun to be widely used in economics. For example, the calculation of hidden volatility of securities, the estimation and accounting of environmental costs, the financing scale of supply chain finance and the calculation of guarantee funds, etc. These all require the use of stochastic dynamic programming models. However, due to the unknown nature of the data, the traditional stochastic dynamic programming model also has limitations in its application.Stochastic dynamic programming (SDP) is one of the most mathematical technique for modelling and solving problems of decision making under uncertainty. Many problems can use stochastic dynamic programming to model and solve it. Such as asset pricing, gambling game and so on.While in traditional Stochastic dynamic programming (SDP) we only consider the random noise has the specific distribution, such as normal distribution. However, in most economic studies, due to insufficient data, it is difficult to know the true distribution of interference items. At this time, traditional stochastic dynamic programming is difficult to accurately simulate the model. Therefore, scholars have introduced the concept of robustness and established a robust stochastic dynamic programming model. However, robust Stochastic dynamic programming often cannot be solved, and because of its complexity, computational simulation also cannot get its approximate solution.In this paper, we provide an algorithm and suppose to use this algorithm to get the solution of robust stochastic dynamic programming. By introducing duality and penalty, we can obtain the upper bound and the lower bound of the solution. And then by value iteration, we can gradually reduce the gap between upper and lower solution. We prove that the gap between upper and lower solution can be reduce to zero theoretically. In this way, we can get the solution of the robust stochastic dynamic programming. The algorithm designed in this paper can efficiently solve the problem of solving the robust stochastic dynamic model, and it has high use value in the models of economics and finance.隨著社會經濟的飛速發展，經濟學研究變得日益複雜，傳統定量的數學模型已經很難完善的對經濟學進行類比，隨機動態規劃開始大量運用於經濟學之中。例如有價證券隱藏波動率的計算、環境成本的估計與核算、供應鏈金融的融資規模與擔保金的計算等等，都需要用到隨機動態規劃模型。但由於資料的未知性，目前傳統的隨機動態規劃模型也出現了應用的局限性。在傳統的隨機動態規劃模型中，我們只考慮其中的干擾項擁有一個特定的分佈，如正態分佈。但是在大多研究中，由于据量不足，我很知道干扰的真分布。的机划很准确的模型行模，所以者引入了棒性的概念，建立了棒机划模型。然而在大多數顯示情況中，我們往往不知道干擾項的真正的分佈，因此我們引入了魯邦隨機動態規劃模型，然而魯邦隨機動態規劃模型常常很難被準確計算得出解，因此其複雜程度也很難用計算機模擬計算得出近似解。本文中，我們提供一種算法，希望能夠通過這個算法來得到魯邦隨機動態規劃模型的近似解，我們引入了對偶性得到了魯邦隨機動態規劃模型的上解和下解，並且通過迭代的方式來不斷縮小上解和下解之間的差距，通過理論證明得出上解和下解的差距可以縮小為零，從而得到了魯邦隨機童泰規劃模型的準確解。本文的算法能高效率的解棒机模型的求解，在和金融的模型中有高的使用价值。Wang, Manyan."October 2020."M.Phil. Chinese University of Hong Kong 2021.Includes bibliographical references (leaves 39-41).Abstracts also in Chinese.Title from PDF title page (viewed on October 25, 2022).Wang, Manyan