News
Workshop Program onlineImportant Dates
Papers due:May 25, 2016 (final) Notification: June 17, 2016 Workshop Date: Aug 22, 2016
Links



Workshop Proceedings


Workshop Program
Workshop Date: 22 August 2015Place: TBA
Time | Talk |
---|---|
9:00 - 9:15 |
Workshop Introduction Session chair: TBA |
9:15 - 10:00 | Tutorial: Reproducibility of Computational Experiments: How prova! Can Help? Danilo Guerrera (presenter), Antonio Maffia, Helmar Burkhart, and Florina Ciorba, University of Basel, Switzerland Abstract: see here |
10:00 - 10:30 | Verification of Simulation via Reproducibility Franziska Hoffeins, Florina M. Ciorba, Ioana Banicescu, and Wolfgang E. Nagel (Contributed Talk) Abstract: see PDF |
10:30 - 11:00 | Coffee break |
11:00 - 11:45 | Keynote: Reproducible research in image processing: the case of IPOL Enric Meinhardt-Llopis, ENS Cachan Abstract: see here |
11:45 - 12:15 |
Paper session 1 Session chair: TBA
|
12:30 - 14:00 | Lunch Break |
14:00 - 15:00 |
Paper session 2 Session chair: TBA
|
15:00 - 15:15 | Discussion and Workshop Closing |
Tutorial: Reproducibility of Computational Experiments: How prova! Can Help?
Danilo Guerrera (presenter), Antonio Maffia, Helmar Burkhart, and Florina Ciorba, University of Basel, SwitzerlandAbstract
We will start by introducing the Basel taxonomy of computational reproducibility, which subdivides into different complexity levels: repetition, replication, and re-experimentation. Our tool prova! (try, prove, convince) is based on the experiment description triple problem/system/method and supports these reproducibility demands. We introduce both the functional requirements, the prototype architecture as well the basic user interface of prova!. Next, use cases of different complexity levels are demonstrated. We further sketch how prova! has been used in a recent graduate course on high-performance computing. Attendants are invited to visit our website https://prova.io and download the software before or after the tutorial for their own experiments. Keywords: Reproducibility, performance engineering, stencil applications, science gateway, EasyBuild, LMod.
Enric Meinhardt-Llopis, ENS Cachan
Title: Reproducible research in image processing: the case of IPOLAbstract
IPOL -Image Processing On Line- is a journal that publishes articles on mathematical image processing. It is an extreme case of reproducible research: not only each article comes with its peer-reviewed source code, but there is also a website associated to each article where the readers can try the algorithm with their own images. Moreover, there is a public archive of all experiences run by previous readers. This archive turns out to be, unexpectedly, one of the most interesting features. In this talk we explain the rationale behind IPOL and what are the main advantages and disadvantages of such an open system.