An Improved Genetic Algorithm for Cell Placement

Genetic algorithm, an effective methodology for solving combinatorial optimization problems, is a very computationally expensive algorithm and, as such, numerous researchers have undertaken efforts to improve it. In this paper, we presented the partial mapped crossover and cell move or cells exchange mutation operators in the genetic algorithm when applied to cell placement problem. Traditional initially placement method may cause overlaps between two or more cells, so a heuristic initial placement approach and method of timely updating the coordinates of cells involved were used in order to eliminate overlaps between cells, meanwhile, considering the characters of different circuits to be placed, the punishment item in objective function was simplified. This algorithm was applied to test a set of benchmark circuits, and experiments reveal its advantages in placement results and time performance when compared with the traditional simulated annealing algorithm.