A comparison between gradient descent and stochastic approaches for parameter optimization of a sea ice model
Two types of optimization methods were applied to a parameter optimization problem in a coupled ocean–sea ice model of the Arctic, and applicability and efficiency of the respective methods were examined. One optimization utilizes a finite difference (FD) method based on a traditional gradient desce...
Published in: | Ocean Science |
---|---|
Main Authors: | , , , , |
Format: | Other/Unknown Material |
Language: | English |
Published: |
2018
|
Subjects: | |
Online Access: | https://doi.org/10.5194/os-9-609-2013 https://os.copernicus.org/articles/9/609/2013/ |
id |
ftcopernicus:oai:publications.copernicus.org:os17741 |
---|---|
record_format |
openpolar |
spelling |
ftcopernicus:oai:publications.copernicus.org:os17741 2023-05-15T15:07:33+02:00 A comparison between gradient descent and stochastic approaches for parameter optimization of a sea ice model Sumata, H. Kauker, F. Gerdes, R. Köberle, C. Karcher, M. 2018-01-15 info:eu-repo/semantics/application/pdf https://doi.org/10.5194/os-9-609-2013 https://os.copernicus.org/articles/9/609/2013/ eng eng info:eu-repo/grantAgreement/EC/FP7/265863 doi:10.5194/os-9-609-2013 https://os.copernicus.org/articles/9/609/2013/ info:eu-repo/semantics/openAccess eISSN: 1812-0792 info:eu-repo/semantics/Text 2018 ftcopernicus https://doi.org/10.5194/os-9-609-2013 2020-07-20T16:25:25Z Two types of optimization methods were applied to a parameter optimization problem in a coupled ocean–sea ice model of the Arctic, and applicability and efficiency of the respective methods were examined. One optimization utilizes a finite difference (FD) method based on a traditional gradient descent approach, while the other adopts a micro-genetic algorithm (μGA) as an example of a stochastic approach. The optimizations were performed by minimizing a cost function composed of model–data misfit of ice concentration, ice drift velocity and ice thickness. A series of optimizations were conducted that differ in the model formulation ("smoothed code" versus standard code) with respect to the FD method and in the population size and number of possibilities with respect to the μGA method. The FD method fails to estimate optimal parameters due to the ill-shaped nature of the cost function caused by the strong non-linearity of the system, whereas the genetic algorithms can effectively estimate near optimal parameters. The results of the study indicate that the sophisticated stochastic approach (μGA) is of practical use for parameter optimization of a coupled ocean–sea ice model with a medium-sized horizontal resolution of 50 km × 50 km as used in this study. Other/Unknown Material Arctic Sea ice Copernicus Publications: E-Journals Arctic Ocean Science 9 4 609 630 |
institution |
Open Polar |
collection |
Copernicus Publications: E-Journals |
op_collection_id |
ftcopernicus |
language |
English |
description |
Two types of optimization methods were applied to a parameter optimization problem in a coupled ocean–sea ice model of the Arctic, and applicability and efficiency of the respective methods were examined. One optimization utilizes a finite difference (FD) method based on a traditional gradient descent approach, while the other adopts a micro-genetic algorithm (μGA) as an example of a stochastic approach. The optimizations were performed by minimizing a cost function composed of model–data misfit of ice concentration, ice drift velocity and ice thickness. A series of optimizations were conducted that differ in the model formulation ("smoothed code" versus standard code) with respect to the FD method and in the population size and number of possibilities with respect to the μGA method. The FD method fails to estimate optimal parameters due to the ill-shaped nature of the cost function caused by the strong non-linearity of the system, whereas the genetic algorithms can effectively estimate near optimal parameters. The results of the study indicate that the sophisticated stochastic approach (μGA) is of practical use for parameter optimization of a coupled ocean–sea ice model with a medium-sized horizontal resolution of 50 km × 50 km as used in this study. |
format |
Other/Unknown Material |
author |
Sumata, H. Kauker, F. Gerdes, R. Köberle, C. Karcher, M. |
spellingShingle |
Sumata, H. Kauker, F. Gerdes, R. Köberle, C. Karcher, M. A comparison between gradient descent and stochastic approaches for parameter optimization of a sea ice model |
author_facet |
Sumata, H. Kauker, F. Gerdes, R. Köberle, C. Karcher, M. |
author_sort |
Sumata, H. |
title |
A comparison between gradient descent and stochastic approaches for parameter optimization of a sea ice model |
title_short |
A comparison between gradient descent and stochastic approaches for parameter optimization of a sea ice model |
title_full |
A comparison between gradient descent and stochastic approaches for parameter optimization of a sea ice model |
title_fullStr |
A comparison between gradient descent and stochastic approaches for parameter optimization of a sea ice model |
title_full_unstemmed |
A comparison between gradient descent and stochastic approaches for parameter optimization of a sea ice model |
title_sort |
comparison between gradient descent and stochastic approaches for parameter optimization of a sea ice model |
publishDate |
2018 |
url |
https://doi.org/10.5194/os-9-609-2013 https://os.copernicus.org/articles/9/609/2013/ |
geographic |
Arctic |
geographic_facet |
Arctic |
genre |
Arctic Sea ice |
genre_facet |
Arctic Sea ice |
op_source |
eISSN: 1812-0792 |
op_relation |
info:eu-repo/grantAgreement/EC/FP7/265863 doi:10.5194/os-9-609-2013 https://os.copernicus.org/articles/9/609/2013/ |
op_rights |
info:eu-repo/semantics/openAccess |
op_doi |
https://doi.org/10.5194/os-9-609-2013 |
container_title |
Ocean Science |
container_volume |
9 |
container_issue |
4 |
container_start_page |
609 |
op_container_end_page |
630 |
_version_ |
1766339034518388736 |