Tabu Search for Managing Groundwater
Contribution to Book
Hydrology, Hydraulics and Water Resources Management
The family of tabu search (TS) heuristic global search optimizer types can solve problem types ranging from linear programming (LP) through mixed integer nonlinear programming (MINLP). TS is most commonly used for problems that are not readily solved by classical linear programming and gradient search nonlinear programming algorithms (e.g .. problems having nondifferentiable objective functions). TS uses its memory of tested solutions to guide development of new solutions. For example. a TS algorithm maintains a tabu list of inferior solutions, and will avoid expending effort on solutions similar to the poor ones. User input designates the minimum acceptable difference between a new strategy and a previous poor strategy. On the other hand. the TS will try to develop new solutions that are in the neighborhood of known good solutions. User input determines acceptable and desirable proximity. Other rules for discarding and favoring candidate solutions he! p make TS very adaptable to many situations. As it moves discrete distances thru the solution space. TS can avoid being trapped in local optima, making it a global optimizer. TS is especially useful for discrete problems (having a finite number of solutions). Especially when it directs other optimization algorithms. TS is referred to as a metaheuristic optimizer. Because of the ease with which TS can couple with other optimizers. and incorporate problem-specific rules. TS applications in hydrologic science arc increasing. This chapter discusses TS concepts. features, and coupling with other heuristic optimizers (HO) such as genetic algorithm. The resulting hybrid HO effectively and efficiently sol\cs complex nonlinear groundwater optimization problems.
Peralta, Richard C. and Kalwij, Ineke M., "Tabu Search for Managing Groundwater" (2012). Civil and Environmental Engineering Faculty Publications. Paper 1172.