A Dynamic Programming Approach to Network Problems : A Model for On-Line Computer Systems.
Derivation, documentation, and listing of a dynamic programming model on JOSS for finding optimal solutions (minimum-cost paths) to network problems. The network represents the possible connections between decisions and expresses costs of going from one to another. The model encompasses a value-iterative algorithm which successively converges to the solution. The principle of optimality, which permits finding the minimum-cost (shortest) path as a function of the maximum number of arcs allowed, is: regardless of how a particular state was arrived at, the remaining decisions as to what path to take en route to the terminal must themselves constitute an optimal solution. The costs of traversing arcs may be input, recalled from other files, or calculated by JOSS from input functions. Model flexibility allows for testing the sensitivity of a decision process to changes in the terminal point.