Summary
We prove that the optimal convergence speed exponent for parallel annealing based on periodically interacting multiple searches with time periodr is always worse than for independent multiple searches whenever the cost function has only one global minimum. Our proofs will be based on large deviation estimates coming from the theory of generalized simulated annealing (G.S.A.).
Article PDF
Similar content being viewed by others
References
E. H. L. Aarts, F. M. J. de Bond, J. H. A. Habers, and P. J. M. van Laarhoven. Parallel implementations of statistical cooling algorithms.Integration VLSI J., 4:209–238, 1986.
E. H. L. Aarts and P. J. M. van Laarhoven.Simulated Annealing: Theory and applications. D. Reidel, Dordrect, 1987.
R. Azencott et al., editors.Simulated annealing: Parallelization techniques. Willey and Sons, 1992.
R. Azencott and C. Graffigne. Parallel annealing by periodically interacting multiple searches: acceleration rates. In R. Azencott et al., editors,Simulated annealing: Parallelization techniques, chapter 6, pages 81–90. Willey and Sons, 1992.
O. Catoni. Multi-temperature parallelization of simulated annealing. Unpublished paper.
O. Catoni. Rough large deviation estimates for simulated annealing. Application to exponential schedules.Ann. Probab., 20:1109–1146, 1992.
T.-S. Chiang and Y. Chow. A limit theorem for a class of inhomogeneous Markov processes.Ann. Probab., 1989.
J.-D. Deuschel and C. Mazza.L 2 convergence of time non-homogeneous Markov processes: I. spectral estimates. Université de Fribourg, Institut de Mathématiques, preprint (to appear in the Ann. of Appl. Prob.), 1992.
M. I. Freidlin and A. D. Wentzell.Random Perturbations of Dynamical Systems, volume 260. Springer-Verlag, 1984.
C. Graffigne. Parallel annealing by periodically interacting multiple searches: experimental study. In R. Azencott et al., editors,Simulated anealing: Parallelization techniques. Willey and Sons, 1992.
C.-R. Hwang and S.-J., Sheu. Singular perturbed Markov chains and exact behaviours of simulated annealing process.J. Theoret. Probab., 5(2):223–249, 1992.
L. Miclo,Evolution de l'énergie libre. Applications à l'étude de la convergence des algorithmes du recuit simulé. PhD thesis, Université d'Orsay, February 1991.
A. Trouvé. Optimal convergence rate for generalized simulated annealing.C. R. Acad. Sci. Paris, Serie I., 315:1197–1202, 1992.
A. Trouvé.Parallélisation massive du recuit simulé. PhD thesis, Université d'Orsay, Jan. 1993.
A. Trouvé. Rough large deviation estimates for the optimal convergence speed exponent of generalized simulated annealing algorithms. Rapport de recherche du LMENS (to appear in the Annales de l'I.H.P.), 1994.
Author information
Authors and Affiliations
Additional information
This article was processed by the author using the Latex style filepljourlm from Springer-Verlag.
Rights and permissions
About this article
Cite this article
Trouvé, A. Asymptotical behaviour of several interacting annealing processes. Probab. Theory Relat. Fields 102, 123–143 (1995). https://doi.org/10.1007/BF01295225
Received:
Revised:
Published:
Issue Date:
DOI: https://doi.org/10.1007/BF01295225