Job sequencing with quadratic penalties: An A*-based graph search approach

A.K. Sen, A. Bagchi
Proceedings of 9th IEEE Conference on Artificial Intelligence for Applications  
In minimum-penalty job sequencing on one machine, the representation of the search graph as a directed acyclic graph instead of as a tree can sometimes s p d up execution times by a foctor of over 25. This demonstrates the existence of Operclrons Research problem that can be solved faster by Algorithm A* than by traditional branchand-bound methods. Consider the best-first branch-andbound procedure &ked by Townsend in 1978 for optimal job sequencing on one machine with quadratic penalties, which
more » ... ic penalties, which still remains, with minor improvements, the only efficient way to solve that particular problem. A graph-basedA* formulation that uses Townsend's lower boumh at nodes as heuristic estimates runs many times faster, showing that the practical applicability ofA * is not limited to search graphs that are trees. It i s surprising that such high speedups can be achieved even when the heuristic estimare fiurction is consistent.
doi:10.1109/caia.1993.366611 fatcat:zidpp5fw7najhozqrl2lxjdtri