Preliminary Computational Experiments with a Genetic Algorithm for the Open Stacks Problem

The Minimization of the maximum number of Opens Stacks Problem (MOSP) can be described as follows. We consider a cutting stock environment, where saw machines cut large sheets (standard size panels) into smaller pieces (items) of different sizes. Each panel is cut according to a determined cutting pattern, specifying a set of item types. All the panels to be cut according to the same pattern are processed consecutively. When the first panel containing a given item type is processed, a stack is reserved for that item type. The stack is closed as soon as the last panel containing that item type is processed. Given the set of item types and the set of cutting patterns, MOSP consists in determining a sequence of patterns that minimizes the maximum number of simultaneously open stacks. MOSP can be used to model problems in different application contexts: production planning, Very Large Scale Integration (VLSI) circuit design etc. [4]. In the following, we refer the the Adaptive Genetic Algorithm (AGA) presented in [2] and we give the results of a preliminary computational campaign on a variety of literature benchmarks, and on a set of real instances from woodcutting industry. AGA parameter settings are reported in [2]. Results are organized in five groups: results on aggregated instances from [8], results on individual instances from [8], results on random instances from [1], results on real instances from VLSI industry, results on real instances from woodcutting companies. AGA has been implemented in C++ and all the results have