Title :
Probabilistic rejectionless anti-voter optimization algorithm
Author :
Patkonjak, M. ; Robaey, J.
Author_Institution :
NEC, Princeton, NJ
Abstract :
A novel probabilistic algorithm, called the probabilistic rejectionless anti-voter (PRAV) algorithm, is introduced. Due to its property of concentrating computational efforts on places where there are the greatest chances of the improvement of an objective function and due to the rejectionless mechanism used, a significant speed-up over other probabilistic methods is achieved. The extensive superior experimental results with respect to simulated annealing on both generic NP-complete problems and high level synthesis tasks are presented and statistically analyzed. The problems discussed include a novel application of commutativity for design improvement. In addition, the proof of convergence is presented
Keywords :
computational complexity; graph colouring; graph theory; optimisation; probability; simulated annealing; commutativity for design; generic NP-complete problems; graph colouring; graph partitioning; high level synthesis; probabilistic algorithm; probabilistic rejectionless antivoter optimization algorithm; proof of convergence; simulated annealing; Analytical models; Computational modeling; Convergence; High level synthesis; Laboratories; Mechanical factors; National electric code; Neural networks; Partitioning algorithms; Simulated annealing;
Conference_Titel :
Circuits and Systems, 1992. ISCAS '92. Proceedings., 1992 IEEE International Symposium on
Conference_Location :
San Diego, CA
Print_ISBN :
0-7803-0593-0
DOI :
10.1109/ISCAS.1992.230228
