Representation and solution of decision problems using sequential decision diagrams

Article Abstract:

A new sequential decision diagram is introduced as a tool for modeling, developing and solving sequential decision problems under uncertainty. This graph is as compact as an influence diagram, but is as effective as a decision tree in capturing the asymmetric and sequential dimensions of decision problems. Aside from being able to identify scenarios sharing common sequences of realized variables, this new graphical representation of decision problems can be employed in both discrete and continuous variable domains. It is demonstrated that, with the use in unison of a sequential diagram, an influence diagram and a common formulation table, all dimensions of a decision problem can be represented in a complete, compact and consistent manner.

author: Covaliu, Zvi, Oliver, Robert M.
Analysis, Markov processes, Decision tree, Decision trees, Sequential processing

User Contributions:

Comment about this article or add new information about this topic:

CAPTCHA

An algorithm for a class of nonconvex programming problems with nonlinear fractional objectives

Article Abstract:

When modeling decision systems appropriate for either public policy making or capital investment planning, the model maker encounters fractional criterion functions. In such situations the algorithm developed here would be helpful, since it identifies a desirable set of target values for given nonconvex programming problems. The primary and secondary equations for the algorithm developed are provided and explained. The basic concept behind the algorithmic maximization solution is the minimization of penalties occurring when fractional objectives deviate from target values.

author: Jagannathan, R.
Case studies, Capital investments, Nonlinear programming

User Contributions:

Comment about this article or add new information about this topic:

CAPTCHA

A new algorithm for computing the maximal closure of a graph

Article Abstract:

A new algorithm is suggested to address the problem of maximum closure in which it is necessary to determine a subset of nodes, all of whose successors belong to the subset, with the greatest possible sum of node weights. The new algorithm compares positively to minimal cut and maximal flow processes on classes of problems which were generated randomly.

author: Faaland, Bruce, Kim, Kiseog, Schmitt, Tom
Graphic methods

User Contributions:

Comment about this article or add new information about this topic:

CAPTCHA


subjects list: Decision-making, Models, Decision making, Research, Algorithms
This website is not affiliated with document authors or copyright owners. This page is provided for informational purposes only. Unintentional errors are possible.