A hybrid genetic algorithm for the floorplan optimization problem

The floorplan optimization problem is an important problem in VLSI design. In a floorplan optimization problem, a set of rectangular blocks is given and the objective is to find a rectangle with minimum area that encloses all blocks. In this paper, a genetic algorithm combined with a local search is proposed for this problem. The chromosome is represented by the sequence pair. The genetic algorithm, acting as a global search method, tries to search large areas while the local search tries to search small areas. As revealed by the experimental results on MCNC benchmark circuits, the proposed method is highly competitive.