TRANSPORTATION NETWORK INVESTMENT PROBLEM--A SYNTHESIS OF TREE-SEARCH ALGORITHMS

In this paper tree-search algorithms that are particularly adept at solving network-design problems in transportation planning are surveyed and synthesized. A unified view of the underlying principles of these tree-search algorithms is presented. Two methodologies--branch-and-bound and branch-and-backtrack--have been identified as promising techniques for solving typically nonlinear and ill-behaved network-design problems, particularly when they are coordinated with the postoptimality procedures of link lengthening and link shortening in minimum-path computation. The two algorithms are then compared, and a third algorithm--based on double bounding--is synthesized to solve transportation network-design problems more efficiently.