Panos Pardalos will make a keynote talk on the conference PPSN-2012 in Taormina, Italy

Panos Pardalos will make a keynote talk "Global equilibrium search algorithm for combinatorial optimization problems" on the 12th International Conference on Parallel Problem Solving From Nature which takes place in Taormina, Italy 1-5 Sep 2012.

Panos Pardalos will make a keynote talk "Global equilibrium search algorithm for combinatorial optimization problems" on 12th International Conference on Parallel Problem Solving From Nature which takes place in Taormina, Italy on 1-5 September, 2012.

Global equilibrium search algorithm for combinatorial optimization problems

Panos M. Pardalos, University of Florida, USA

 

Abstract

Global Equilibrium Search (GES) is a meta-heuristic search method that shares similar ideas with simulated annealing method. GES accumulates a compact set of information about search space of an optimization problem that is used to generate promising initial solutions for local search techniques. This method has been successfully applied to classic discrete optimization problems, such as the unconstrained quadratic programming problem, the maximum satisfiability problem, the max-cut problem, the multidimensional knapsack problem and the job-shop scheduling problem. On all these domains, GES provides state-of-the-art performance compared to the current best known algorithms when used for large scale problems. In this talk, we provide an overview of Global Equilibrium Search, and discuss some successful applications. We explain counter-intuitive empirical observations of super linear speedup in parallel implementations and reveal how parallel acceleration is linked to restart properties of underlying serial algorithms.

This is joint work with Dmytro Korenkevych and Oleg Shylo.