Comparison of a random search algorithm and mixed integer programming for solving area-based forest plans

Abstract

An area-based forest plan is formulated and solved by mixed integer programming and a random search algorithm. This is a computationally difficult problem because operational and environmental constraints require that harvest units and road projects be defined as strict binary variables. It was found that the random search algorithm could easily identify several solutions with objective function values within 10% of the true optimum. The best solution found was within 3% of the optimum. The random search algorithm is simple and can be readily implemented on the microcomputer. It is concluded that the random search algorithm is an effective technique for generating acceptable alternatives to complex area-based planning problems.