Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
Multi-objective parallelization of efficient global optimization
Dissertation (MSc)--University of Pretoria, 2016.
Saved in:
| Other Authors: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
University of Pretoria
2016
|
| Subjects: | |
| Tags: |
No Tags, Be the first to tag this record!
|
| _version_ | 1867613579410145280 |
|---|---|
| access_status_str | Open Access |
| author2 | Kok, Schalk |
| author_browse | Kok, Schalk |
| author_facet | Kok, Schalk |
| collection | Thesis |
| dc_rights_str_mv | © 2016 University of Pretoria. All rights reserved. The copyright in this work vests in the University of Pretoria. No part of this work may be reproduced or transmitted in any form or by any means, without the prior written permission of the University of Pretoria. |
| description | Dissertation (MSc)--University of Pretoria, 2016. |
| format | Thesis |
| id | oai:repository.up.ac.za:2263/57480 |
| institution | University of Pretoria (South Africa) |
| language | English |
| last_indexed | 2026-06-10T12:38:23.596Z |
| license_str | Other — see source repository |
| provenance_str_mv | Harvested via OAI-PMH from UPSpace — University of Pretoria Institutional Repository |
| publishDate | 2016 |
| publishDateRange | 2016 |
| publishDateSort | 2016 |
| publisher | University of Pretoria |
| publisherStr | University of Pretoria |
| record_format | dspace |
| source_str | UPSpace — University of Pretoria Institutional Repository |
| spelling | oai:repository.up.ac.za:2263/57480 Multi-objective parallelization of efficient global optimization Kok, Schalk u10077562@tuks.co.za Wilke, Daniel Nicolas Grobler, Carla UCTD Multi-objective Optimization Global Optimization Parallelization Algorithm Development Efficient Optimization Engineering, built environment and information technology theses SDG-09 SDG-09: Industry, innovation and infrastructure Engineering, built environment and information technology theses SDG-12 SDG-12: Responsible consumption and production Dissertation (MSc)--University of Pretoria, 2016. Design optimization is a subject field where mathematical algorithms are used to improve designs. Analyses of designs using computational techniques often require significant computing resources, and for these problems, an efficient optimization method is needed. Efficient Global Optimization (EGO), first proposed by Jones et al. [25] is an optimization method which aims to use few function evaluations when optimizing a design problem. In this study, we use a multi-objective strategy to parallelize EGO. EGO is part of a set of algorithms called surrogate optimization methods. A set of initial designs are analyzed and then a response surface is fitted to the evaluated designs. In each iteration, EGO selects the set of design variables for which the next analysis will be performed. It makes this decision based on two opposing criteria. EGO will either decide to sample where the predicted objective function value is low, an exploitation approach, or where there is high uncertainty, an exploration approach. In each iteration, the classical EGO only selects one design per iteration. This selected design vector is either a result of exploitation or exploration based on a measure referred to as maximum Expected Improvement (EI). However, the modern day computing environment is capable of running multiple different analyses in parallel. Thus, it would be advantageous if EGO would be able to select multiple designs to evaluate in each iteration. In this research, we treat EGO?s inherent selection criteria to either exploit or explore as a multi-objective optimization problem, since each criterion can be defined by a separate objective function. In general multi-objective optimization problems don?t only have one solution, but a set of solutions called a Pareto optimal set. In our proposed strategy multiple designs from this Pareto optimal set are selected by EGO to be analyzed in the subsequent iteration. This proposed strategy is referred to as Simple Intuitive Multiobjective ParalLElization of Efficient Global Optimization (SIMPLE-EGO). We start our study by investigating the behaviour of classical EGO. During each iteration of EGO, a new design is selected to be evaluated. This is performed by finding the maximum of the Expected Improvement (EI) function. Maximizing this function initially proved challenging. However, by exploiting information regarding the nature of the EI function, the maximization problem is simplified significantly, and the robustness of finding the maximum is enhanced. More importantly, solving this maximization problem robustly, dramatically improves the convergence behaviour once a local basin has been found. We compare our SIMPLE-EGO method to a multi-objective optimization algorithm (EGO-MO) published by Feng et al. [16]. We first investigate the behaviour of EGO, EGO-MO, and SIMPLE-EGO. Thereafter the convergence performance of these methods is quantified. As expected the parallelization of both SIMPLE-EGO and EGO-MO lead to faster convergence on a range of test functions compared to classical EGO, which only sampled one point per iteration. The convergence characteristics of SIMPLE-EGO and EGOMO are also markedly different. We conclude with a discussion on the advantages and disadvantages of the investigated methods. tm2016 mi2025 Mechanical and Aeronautical Engineering MSc Unrestricted SDG-09: Industry, innovation and infrastructure SDG-12: Responsible consumption and production 2016-10-27T07:28:29Z 2016-10-27T07:28:29Z 2016-09-01 2016 Dissertation Grobler, C 2016, Multi-objective parallelization of efficient global optimization, MSc Dissertation, University of Pretoria, Pretoria, viewed yymmdd <http://hdl.handle.net/2263/57480> S2016 http://hdl.handle.net/2263/57480 en © 2016 University of Pretoria. All rights reserved. The copyright in this work vests in the University of Pretoria. No part of this work may be reproduced or transmitted in any form or by any means, without the prior written permission of the University of Pretoria. application/pdf University of Pretoria |
| spellingShingle | UCTD Multi-objective Optimization Global Optimization Parallelization Algorithm Development Efficient Optimization Engineering, built environment and information technology theses SDG-09 SDG-09: Industry, innovation and infrastructure Engineering, built environment and information technology theses SDG-12 SDG-12: Responsible consumption and production Multi-objective parallelization of efficient global optimization |
| title | Multi-objective parallelization of efficient global optimization |
| title_full | Multi-objective parallelization of efficient global optimization |
| title_fullStr | Multi-objective parallelization of efficient global optimization |
| title_full_unstemmed | Multi-objective parallelization of efficient global optimization |
| title_short | Multi-objective parallelization of efficient global optimization |
| title_sort | multi objective parallelization of efficient global optimization |
| topic | UCTD Multi-objective Optimization Global Optimization Parallelization Algorithm Development Efficient Optimization Engineering, built environment and information technology theses SDG-09 SDG-09: Industry, innovation and infrastructure Engineering, built environment and information technology theses SDG-12 SDG-12: Responsible consumption and production |
| url | http://hdl.handle.net/2263/57480 |