A Dynamic Programming Approach to Network Problems: A Model for On-Line Computer Systems
This report has yet to be scanned by Contrails staff
Report Number: RM-6338-ARPA
Author(s): Pipes, L. J.
Corporate Author(s): The RAND Coporation
Date of Publication: 1970-08
Contract: DAHC15 67 C 0142
DoD Task:
Identifier: AD0711811
Abstract:
This Memorandum contains the derivation of a dynamic-programming model for finding optimal solutions to problems involving certain multi-stage decision processes. The model has been implemented on JOSS, and instructions on its use are also included. The dynamic-programming approach analyzes an optimization problem with various constraints and variables by decomposing the problem into a sequence of stages at which lower-order optimization takes place. The model presented here encompasses a value-iterative method which is less restrictive and which offers desirable advantages over other currently used techniques One such advantage is a reduction in the actual amount of computer storage required to reach an optimal solution. The model is flexible and allows for testing the sensitivity of a decision process to changes in the terminal point and thus in the associated costs. Additional information may be gained from examining the buildup of an optimal solution, which can also be printed out if conditions permit.
Author(s): Pipes, L. J.
Corporate Author(s): The RAND Coporation
Date of Publication: 1970-08
Contract: DAHC15 67 C 0142
DoD Task:
Identifier: AD0711811
Abstract:
This Memorandum contains the derivation of a dynamic-programming model for finding optimal solutions to problems involving certain multi-stage decision processes. The model has been implemented on JOSS, and instructions on its use are also included. The dynamic-programming approach analyzes an optimization problem with various constraints and variables by decomposing the problem into a sequence of stages at which lower-order optimization takes place. The model presented here encompasses a value-iterative method which is less restrictive and which offers desirable advantages over other currently used techniques One such advantage is a reduction in the actual amount of computer storage required to reach an optimal solution. The model is flexible and allows for testing the sensitivity of a decision process to changes in the terminal point and thus in the associated costs. Additional information may be gained from examining the buildup of an optimal solution, which can also be printed out if conditions permit.