This paper offers a new way of thinking about dynamic resource management problems, a problem class which we have named dynamic resource transformation problems. We have tried to offer as much generality as possible, subject to the constraint that the properties of the problem be sufficiently well defined that the problem can be easily represented as a well defined set of data and software. An important contribution of our representation is that it exposes many of the prominent dimensions of dynamic problems that are often lost or ignored in specific problems. A software library built around this representation has been developed in Java. Elements of the architecture of this system are described in Shapiro and Powell . We have intentionally left out any algorithmic details, since these are viewed as being problem specific. This decision, of course, raises the common question in operations research, why model a problem that cannot be solved? A central claim of our representation is that any problem can be decomposed into suitably small subproblems and then solved. In Shapiro and Powell  we propose an algorithmic metastrategy that describes how DRTP's can be decomposed and solved using a class of approximation strategies drawn from dynamic programming. The mathematical foundation of this metastrategy is based on dynamic programming techniques developed especially for solving resource management problems. Readers are referred to Powell et al.  for a description of these techniques.
All Science Journal Classification (ASJC) codes
- Decision Sciences(all)
- Management Science and Operations Research