Berlin 2012 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
DY: Fachverband Dynamik und Statistische Physik
DY 22: Posters I
DY 22.5: Poster
Wednesday, March 28, 2012, 17:00–19:00, Poster A
Influence of rounding errors on the quality of heuristic optimization algorithms — •Martin Ransberger1, Ingo Morgenstern1, and Johannes Josef Schneider2 — 1Faculty of Physics, University of Regensburg, 93040 Regensburg, Germany — 2Department of Physics, Mathematics, and Computer Science, Johannes Gutenberg University of Mainz, Staudinger Weg 7, 55099 Mainz, Germany
Simulated annealing and search space smoothing are both widely used optimization algorithms. While simulated annealing overcomes barriers in the energy landscape at finite temperatures, search space smoothing intends to remove these barriers, such that a greedy algorithm is sufficient to find the global minimum. Rounding errors in the calculation of the energy landscape can affect the quality of the results and even lead to a new physical behavior.
In this presentation, we thoroughly investigate the effect of rounding errors on the energy landscape and their influence on optimization processes with simulated annealing and search space smoothing. We present computational results for the traveling salesman problem.