Digital Library of the European Council for Modelling and Simulation

Title:	Investigation Into The Behaviour Of An Asynchronous Population Based Hill Climbing Algorithm
Authors:	Lars Nolle, Jens Werner
Published in:	(2019). ECMS 2019 Proceedings Edited by: Mauro Iacono, Francesco Palmieri, Marco Gribaudo, Massimo Ficco, European Council for Modeling and Simulation. DOI: http://doi.org/10.7148/2019 ISSN: 2522-2422 (ONLINE) ISSN: 2522-2414 (PRINT) ISSN: 2522-2430 (CD-ROM) 33rd International ECMS Conference on Modelling and Simulation, Caserta, Italy, June 11^th – June 14^th, 2019
Citation format:	Lars Nolle, Jens Werner (2019). Investigation Into The Behaviour Of An Asynchronous Population Based Hill Climbing Algorithm, ECMS 2019 Proceedings Edited by: Mauro Iacono, Francesco Palmieri, Marco Gribaudo, Massimo Ficco European Council for Modeling and Simulation. doi: 10.7148/2019-0050
DOI:	https://doi.org/10.7148/2019-0050
Abstract:	Previously, a variation of Asynchronous Population Based Hill Climbing was applied to a discrete optimisation task from the field of electronic circuit design with a challenging 9 dimensional search space. The algorithm exhibited a certain behaviour, which is analysed in this new research. A problem was the asynchronous nature of the search algorithm, which did not allow the search threads to save the internal state data into log files. Instead, a novel monitoring strategy had to be developed, which accesses this internal data via shared memory in specified times intervals. It was possible to show that the algorithm under investigation is capable of switching between exploration and exploitation of the search space during the search. This prevents the algorithm from being stuck in a local optimum.
Full text: