Technical reports:
Jimmy Lam, "An Efficient Simulated
Annealing Schedule," Technical Report 8818, Ph.D. Dissertation, Yale Electrical Engineering Department,
New Haven, CT, September 1988.
Jimmy Lam and Jean-Marc Delosme, "An Efficient
Simulated Annealing Schedule: Derivation," Technical Report 8816, Yale Electrical
Engineering Department, New Haven, CT, September 1988.
Jimmy Lam and Jean-Marc Delosme, "An Efficient
Simulated Annealing Schedule: Implementation and Evaluation," Technical Report
8817, Yale Electrical Engineering Department, New Haven, CT, September 1988.
Conference papers:
Jimmy Lam and Jean-Marc Delosme,
"Performance of a New Annealing Schedule," Proc. of the 25th ACM/IEEE DAC, pp.
306-311, 1988.
Jimmy Lam and Jean-Marc
Delosme, "Simulated Annealing: A Fast Heuristic for Some Generic Layout Problems,"
Proc. IEEE ICCAD, pp. 510-513, 1988.
Simulated Annealing Applet:
Simulated Annealing Traveling Salesman Problem Java Applet
Basic control parameters:
City: the number of cities. A maximum of 10,000 is allowed.
Lambda: the quality factor. The smaller the lambda, the better the quality of the final solution, the longer the run time.
Seed: the random seed. Specify 0 to use the system clock as random seed.
Advanced control parameters:
Initial Move: the number of initial moves to randomize the system and gather statistics. Since the initial solution is already randomized, the default value of 1,000 is sufficient.
Mean Memory: the lambda mean memory product. The smaller this value, the more adaptive and less stable the mean estimator.
S.D. Memory: the lambda standard deviation memory product. The smaller this value, the more adaptive and less stable the standard deviation estimator.
Display control parameter:
Update Freq: the screen update frequency in seconds. Specify 0 to disable screen update.
Package Annealing, about 300 lines of code, contains the application independent part of the simulated annealing schedule.
Package Tour contains the annealing implementation for the traveling salesman problem. The basic movement strategy is 2-opt move with feedback control. The target acceptance ratio is 44%.
E-mail: <user_name_in_hyperlink>@comcast.net