SAMURAI: A general and efficient simulated-annealing schedule with fully adaptive annealing parameters

Integration ◽  
1988 ◽  
Vol 6 (2) ◽  
pp. 147-178 ◽  
Author(s):  
Francky Catthoor ◽  
Hugo de Man ◽  
Joos Vandewalle
Keyword(s):  
Simulated Annealing ◽  
Annealing Schedule ◽  
Fully Adaptive

Convergence and finite-time behavior of simulated annealing

1986 ◽  
Vol 18 (03) ◽  
pp. 747-771 ◽  
Author(s):  
Debasis Mitra ◽  
Fabio Romeo ◽  
Alberto Sangiovanni-Vincentelli
Keyword(s):  
Markov Chain ◽  
Simulated Annealing ◽  
Finite Time ◽  
Randomized Algorithm ◽  
Global Optimum ◽  
Time Behavior ◽  
Annealing Schedule ◽  
Complete Problems ◽  
Inhomogeneous Markov Chain ◽  
Np Complete

Simulated annealing is a randomized algorithm which has been proposed for finding globally optimum least-cost configurations in large NP-complete problems with cost functions which may have many local minima. A theoretical analysis of simulated annealing based on its precise model, a time-inhomogeneous Markov chain, is presented. An annealing schedule is given for which the Markov chain is strongly ergodic and the algorithm converges to a global optimum. The finite-time behavior of simulated annealing is also analyzed and a bound obtained on the departure of the probability distribution of the state at finite time from the optimum. This bound gives an estimate of the rate of convergence and insights into the conditions on the annealing schedule which gives optimum performance.

Convergence and finite-time behavior of simulated annealing

1986 ◽  
Vol 18 (3) ◽  
pp. 747-771 ◽  
Author(s):  
Debasis Mitra ◽  
Fabio Romeo ◽  
Alberto Sangiovanni-Vincentelli
Keyword(s):  
Markov Chain ◽  
Simulated Annealing ◽  
Finite Time ◽  
Randomized Algorithm ◽  
Global Optimum ◽  
Time Behavior ◽  
Annealing Schedule ◽  
Complete Problems ◽  
Inhomogeneous Markov Chain ◽  
Np Complete

Simulated annealing is a randomized algorithm which has been proposed for finding globally optimum least-cost configurations in large NP-complete problems with cost functions which may have many local minima. A theoretical analysis of simulated annealing based on its precise model, a time-inhomogeneous Markov chain, is presented. An annealing schedule is given for which the Markov chain is strongly ergodic and the algorithm converges to a global optimum. The finite-time behavior of simulated annealing is also analyzed and a bound obtained on the departure of the probability distribution of the state at finite time from the optimum. This bound gives an estimate of the rate of convergence and insights into the conditions on the annealing schedule which gives optimum performance.

THE THERMODYNAMICS OF DYSLEXIC LEARNING

1992 ◽  
Vol 03 (04) ◽  
pp. 351-360
Author(s):  
H. MAKSE ◽  
R.P.J. PERAZZO
Keyword(s):  
Simulated Annealing ◽  
Phase Transitions ◽  
Numerical Analysis ◽  
Boolean Functions ◽  
Schematic Model ◽  
Annealing Schedule ◽  
Nonzero Temperature ◽  
Symmetry Classes ◽  
Cooling Schedule ◽  
Partial Learning

The dyslexic behaviour of a layered network is interpreted as arising from its incomplete training using a cost function that is sensitive to the grouping of Boolean functions into symmetry classes. The training is envisaged as a simulated annealing and a partial learning as the interruption of the cooling schedule at a finite nonzero temperature. We present the thermodynamics of the process. The onset of dyslexic and normal behaviours arise from phase transitions that take place during the annealing schedule. We exemplify the theory with the numerical analysis of a schematic model.

Improving the Efficiency of Simulated Annealing Optimization Through Detection of Productive Search

1997 ◽  
Author(s):  
Simon Szykman ◽  
Linda C. Schmidt ◽  
Harshawardhan Shetty
Keyword(s):  
Simulated Annealing ◽  
Traveling Salesman Problem ◽  
Traveling Salesman ◽  
Annealing Schedule ◽  
Statistical Process ◽  
New Techniques ◽  
Algorithm Performance ◽  
Traditional Schedules ◽  
Improve Algorithm ◽  
Potential Impact

Abstract The popularity of simulated annealing for engineering design applications has grown in recent years, increasing the need for new techniques that improve algorithm performance. Simulated annealing is a time-consuming, iteration-intensive algorithm. One area of algorithm enhancement with high potential impact is the development of methods for improving the algorithm by reducing the amount of wasted or non-productive search. This paper presents an approach to detection of productive search based on statistical process control (SPC) concepts. The proposed Detection of Productive Search (DPS) annealing schedule is compared to three other viable schedules using a 100-city traveling salesman problem. The DPS schedule produces results on par with the best from the more traditional schedules but does so with significantly fewer iterations.

Sign in / Sign up

Export Citation Format

Share Document