An empirical study of the stochastic evolution algorithm for the VLSI cell placement problem
MetadataShow full item record
The Stochastic Evolution (SE) algorithm is a relatively new heuristic method that is used for combinatorial optimization that exploits an analogy between biological evolution and combinatorial optimization. The SE algorithm begins with a random initial solution or with a previously found good solution to the problem and simulates the evolution process by eliminating the bad characteristics of the older generation resulting in an improved newer generation solution. The SE algorithm achieves this using functions and operations which test the suitability of characteristics for the existing environment. Each characteristic of a species in the current generation has to prove its suitability under the existing environmental conditions in order to remain unchanged in the next generation. This process is repeated until a certain number of iterations is completed or until no significant improvement is noticed and the solution to the problem is obtained. In this project, the SE algorithm is studied and implemented to solve the very large scale integration cell placement problem, and the quality of the solutions and the running times of the algorithm are compared with those generated by the Simulated Annealing (SA) algorithm. The SE algorithm after experiments shows that it produces results that are comparable to the results that were generated by the SA algorithm. The SE algorithm seems to be suitable in cases where the size of the input is considerably large. The SE algorithm starts consuming more time than the SA algorithm as the size of the input increases. The feature in the SE algorithm which increases the number of trials if the newer generation is better than the older could increase the running time of the SE algorithm considerably.