Summary
As is well known (cf. Derman (1970) and references cited there), dynamic programming problems with finite state and action spaces can be solved by linear programming techniques. In the present paper it will be shown that this statement can be generalized to the case of general state and action spaces.
By this approach, the underlying sequential structure is completely neglected. Instead, vectore space structures and linear programming results (such as existence and duality theorems, complementary slackness) are used to obtain optimality statements.
Article PDF
Similar content being viewed by others
Avoid common mistakes on your manuscript.
References
Blackwell, D.: Discounted dynamic programming. Ann. Math. Statist. 36, 226–235 (1965)
Derman, C.: Finite State Markovian Decision Processes. New York: Academic Press 1970
Dunford, N., Schwartz, J.T.: Linear Operators, Part I. New York: Interscience Publishers 1957
Hinderer, K.: Foundations of Non-stationary Dynamic Programming with Discrete Time Parameter. Berlin-Heidelberg-New York: Springer 1970
Horn, A., Tarski, A.: Measures in Boolean Algebras. Trans. Amer. Math. Soc. 64, 467–497 (1948)
Krabs, W.: Optimierung und Approximation. Stuttgart: Teubner 1975
Krafft, O., Witting, H.: Optimale Tests und ungünstigste Verteilungen. Z. Wahrscheinlichkeitstheorie und verw. Gebiete 7, 289–302 (1967)
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Heilmann, W.R. Solving a general discounted dynamic program by linear programming. Z. Wahrscheinlichkeitstheorie verw Gebiete 48, 339–346 (1979). https://doi.org/10.1007/BF00537529
Received:
Issue Date:
DOI: https://doi.org/10.1007/BF00537529