ISIP 2012 Overview: Infrastructure & Software Integration Project

1. 10/12/2012 ISIP 2012 overviewF. Imbeaux, G. Manduchi, H. Klingshirn and all ISIP contributorsTF Leaders : G. Falchetto, R. Coelho, D. Coster EFDA CSU Contact Person: D. Kalupinhttps://www.efda-itm.eu/ITM/html/isip_public.html

2. Infrastructure & Software Integration Project in 2012 ISIP : develop concepts and software infrastructure for Integrated Modelling ISIP 2012 contributors L. Alegretti (CEA), T. Aniel (CEA), Y. Buravand (CEA), P. Garcia-Mueller (CIEMAT) B. Guillerminet (CEA), P. Huynh (CEA), F. Imbeaux (CEA), H. Klingshirn (IPP), L. Kos (MHST), S. Kulovec (MHST), G. Manduchi (ENEA/RFX), V. Pais (MEDC), N. Ravenel (CEA), J. Signoret (CEA), J. Storrs (CCFE) Core Programming Team: M. Plociennik, M. Owsiak, T. Zok, B. Palak (IPPLM) Continuation of the three ISIP missions, while the ITM Infrastructure is now stable since 2009 Support the users of the ITM platform (Hotline, documentation, tutorials) – the essential part of the activity Maintain and update the existing platform tools – a significant activity Identify missing functionalities in the platform and develop the required new functionalities – low activity, low pressure from users

3. Mission I: Support and Training • New Core Programming Team: M. Plociennik, M. Owsiak, T. Zok, B. Palak (IPPLM) • Answer Trackers on the General Support tracker (Gforge) • Support users during Code Camps • Provide training at 1st code camp of the year • More than 90 Tracker items posted and solved this year • This Core Programming Team will continue its activity next year ! 3

4. Mission II: 2012 platform release • 4.10a (released in June): many changes in the datastructure and transparent move to Kepler 2.3 • Several CPOs revised to use the new « arrays of structure » feature (hopefully the data structures are converging and will not undergo such deep changes every year) • Memory cache for UAL data transfer (has solved a key performance bottleneck w.r.t. last year) • Integrated Simulation Editor now handles arrays of structure • New automated testing framework for the UAL • Web Service actor generator (under testing) • Calendar for next release not decided yet • 1 major release per year is a maximum. The drive for a new major release needs to come from users/IMPs 4

5. Mission II: Improvement of the ITM Visualization Toolbox • Standardized ITM data structure allows construction of standard visualizations • Currently possible : simple 1d/2d plots with VisIt (ITM plugin) and ISE • New project : ITMVIS framework • Allows users to easily define complex visualizations (1d-3d, including task-specific data processing) • Makes these plots available through various backends (matplotlib, VisIt, ...) • Enables re-use, combination & embedding of visualizations Core data Edge data Limiter geometry Example: combined 2d visualization of electron density profiles computed by a coupled core-edge simulation

6. Mission III: Design of a new simulation catalogue • Objective: have a tool to read/write information describing the results of simulations • Database containing a summary of the simulation results • Users request the population of the catalogue from a simulation (e.g. via a Kepler actor at the end of the simulation), or can comment an existing simulation • Web-based querying interface • Design document has been finalised • First prototypes of the Catalogue DB and the system API have been implemented • Implementation to occur in 2013 6

7. Mission III: Design of advanced parallel workflows • Objective: have a standard method to execute workflows in parallel • Processing of independent time slices. Can be extended to parametric scans • One main workflow manages the batch execution of multiple Kepler instances • On the Gateway cluster now, can be extended to remote cluster / HPC / GRID • Each of this Kepler instance executes a physics workflow on a selected part of the input data • Prototype example has been impleted and tested, based on equilibrium reconstruction with EQUAL on a long Tore Supra shot (~3000 time slices calculated) 7

8. Mission III: Prepare portable version of the ITM Infrastructure • Objective: prepare local deployment of the ITM Infrastructure in the Associations, in view of wider distribution / usage of the ITM tools • Create synergies with local experiment (e.g. post-pulse processing with ITM tools) • Much easier access to the ITM technologies for people that are not in the ITM « club » • A portable version of UAL, Data Structure, FC2K, Kepler and ISE has been prepared • Can be installed directly on a local cluster • Or via a Virtual Machine environment • Legal discussions to occur at TFL, then EFDA and European Commission level • In 2013, ISIP will fund local contact persons that will try to use this local deployment facility 8

9. Summary ISIP is now essentially in a support / maintenance phase New tools can be developed but need push from the IMPs (and specifications of the needs) Feature Request tracker on the General Support page Use the IM Project Leader to formulate your requests / specifications The Core Programming Team importance has increased, with respect to the contribution of Associations It is good to have strong Core Programming team ISIP would like to see more contributors from Associations for the new developments Main lines of the 2013 ISIP programme Implement simulation catalogue Local deployment of the Infrastructure Monitoring of long/batch simulations Support IMPs in using the existing tools for GRID/HPC in workflows  need physics use cases !