First steps towards more numerical reproducibility*
Questions whether numerical simulation is reproducible or not have been reported in several sensitive applications. Numerical reproducibility failure mainly comes from the finite precision of computer arithmetic. Results of floating-point computation depends on the comp...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
EDP Sciences
2014-09-01
|
| Series: | ESAIM: Proceedings and Surveys |
| Online Access: | http://dx.doi.org/10.1051/proc/201445023 |
| _version_ | 1797967743146786816 |
|---|---|
| author | Jézéquel Fabienne Langlois Philippe Revol Nathalie |
| author_facet | Jézéquel Fabienne Langlois Philippe Revol Nathalie |
| author_sort | Jézéquel Fabienne |
| collection | DOAJ |
| description | Questions whether numerical simulation is reproducible or not have been reported in
several sensitive applications. Numerical reproducibility failure mainly comes from the
finite precision of computer arithmetic. Results of floating-point computation depends on
the computer arithmetic precision and on the order of arithmetic operations. Massive
parallel HPC which merges, for instance, many-core CPU and GPU, clearly modifies these two
parameters even from run to run on a given computing platform. How to trust such computed
results? This paper presents how three classic approaches in computer arithmetic may
provide some first steps towards more numerical reproducibility. |
| first_indexed | 2024-04-11T02:35:59Z |
| format | Article |
| id | doaj.art-c15e637371d24a188be197a7f76a5896 |
| institution | Directory Open Access Journal |
| issn | 2267-3059 |
| language | English |
| last_indexed | 2024-04-11T02:35:59Z |
| publishDate | 2014-09-01 |
| publisher | EDP Sciences |
| record_format | Article |
| series | ESAIM: Proceedings and Surveys |
| spelling | doaj.art-c15e637371d24a188be197a7f76a58962023-01-02T20:17:01ZengEDP SciencesESAIM: Proceedings and Surveys2267-30592014-09-014522923810.1051/proc/201445023proc144523First steps towards more numerical reproducibility*Jézéquel Fabienne0Langlois Philippe1Revol Nathalie2Sorbonne Universités, UPMC Univ Paris 06, CNRS, UMR 7606, LIP6Université de Perpignan Via Domitia, DALI, LIRMM (UMR5506 CNRS-Université Montpellier 2)INRIA, U. de Lyon, LIP (UMR5668 CNRS-ENS de Lyon-INRIA-UCBL), ENS de LyonQuestions whether numerical simulation is reproducible or not have been reported in several sensitive applications. Numerical reproducibility failure mainly comes from the finite precision of computer arithmetic. Results of floating-point computation depends on the computer arithmetic precision and on the order of arithmetic operations. Massive parallel HPC which merges, for instance, many-core CPU and GPU, clearly modifies these two parameters even from run to run on a given computing platform. How to trust such computed results? This paper presents how three classic approaches in computer arithmetic may provide some first steps towards more numerical reproducibility.http://dx.doi.org/10.1051/proc/201445023 |
| spellingShingle | Jézéquel Fabienne Langlois Philippe Revol Nathalie First steps towards more numerical reproducibility* ESAIM: Proceedings and Surveys |
| title | First steps towards more numerical reproducibility* |
| title_full | First steps towards more numerical reproducibility* |
| title_fullStr | First steps towards more numerical reproducibility* |
| title_full_unstemmed | First steps towards more numerical reproducibility* |
| title_short | First steps towards more numerical reproducibility* |
| title_sort | first steps towards more numerical reproducibility |
| url | http://dx.doi.org/10.1051/proc/201445023 |
| work_keys_str_mv | AT jezequelfabienne firststepstowardsmorenumericalreproducibility AT langloisphilippe firststepstowardsmorenumericalreproducibility AT revolnathalie firststepstowardsmorenumericalreproducibility |