Abstract
This paper presents fundamental contents of a general-purpose genetic algorithm (GA) Workbench under development. We consider for the workbench to be more versatile by adopting superior GA operations peculiar local rules different to each problem. The local rules work effectively to accelerate convergence of GA operations. The bench is developing on the viewpoints of 1) being available for many kinds of GA operations, 2) being a platform-independent and 3) making the bench so as to be black box by adopting object-oriented language C++. Consequently the bench will become possible to apply to various kinds of optimization problems.
