Centres of Excellence

La commission européenne va lancer un appel en 2014 portant sur des "centres d'excellence". Cet appel, doté d'environs 45M€, concerne en principe les communautés de mathématiques appliquées et de calcul scientifique, ainsi que des champs disciplinaires tels que: la dynamique des fluides, la combustion, la météorologie, la climatologie, l'océanographie, etc.

Selon l'appel (version préliminaire) :

EINFRA 5-2015 – Centres of Excellence for computing applications

Specific challenge: Establishing a limited number of Centres of Excellence (CoE) is necessary to ensure EU competitiveness in the application of HPC for addressing scientific, industrial or societal challenges. CoEs will develop a culture of excellence, both scientific and industrial, placing computational science and the harnessing of 'big data' at the centre of scientific discovery and industrial competitiveness. CoEs may be 'thematic', addressing specific application domains such as medicine, life science or energy; 'transversal' on computational science (e.g. algorithms, analytics, numerical methods etc.); or 'challenge-driven', addressing societal or industrial challenges (e.g. ageing, climate change, clean transport etc.); or a combination of these types. 

Scope: The CoE's are expected to be: 

(1) integrated: encompassing not only HPC software but also relevant aspects of hardware, data management and storage, connectivity, security, etc.; 

(2) multidisciplinary: with domain expertise co-located alongside HPC system, software and algorithm expertise; 

(3) user-driven, with the application users and owners playing a decisive role in governance; and

(4) distributed with a possible central hub, federating capabilities around Europe, exploiting available competences, and ensuring synergies with national/local programmes;

Proposals for CoEs will address: 

• Provision of services such as: developing, optimising (including if needed re-design) and scaling HPC application codes towards peta and exascale computing; testing, validating and maintaining codes and managing the associated data; quality assurance; co-design of hardware, software and codes; consultancy to industry and SMEs; research in HPC applications; and addressing the skills gap in computational science.

• Working in synergy with the European Technology Platform for HPC (www.etp4hpc.eu) and with the pan-European HPC infrastructure (www.prace-ri.eu), including by identifying suitable applications for co-design activities relevant to the development of HPC technologies towards exa-scale. 

• Sustainability embracing a wide range of service models and funding from a mixture of sources, including through sponsorship by industry or hybrid public-private models. Clear business plans need to be presented in the proposal.

• Creating communities around specific codes that impact the target sectors, involving ISVs (independent software vendors) where appropriate, and exchange of best practices in particular for SMEs.

• A governance structure driven by the needs of the users. Commercial management expertise will be needed along with technical expertise to manage industry clients and supply chains.

8-10 CoEs are expected to be funded in this topic in order to test the concept. A follow up Call is expected in the future that will build on the results and lessons learnt from the present Call. 

International co-operation is encouraged where there are clear mutual benefits and the partners have the relevant HPC capacity. 

Expected impact:

• Improved access to computing applications and expertise that enables researchers and industry to be more productive, leading to scientific excellence;
• Improved competitiveness for companies and SMEs through access to CoE expertise and services;
• European leadership in applications that address societal challenges or are important for industrial applications through better code performance and better code maintenance and availability;
• More scientists and engineers trained in the use of computational methods and optimisation of applications.

Type of action: Research & Innovation Action

Cet appel est, en partie, inspiré du rapport EESI (European Exascale Software Initiative) "Deliverable D5.6 - Final report on roadmap and recommendations development", Janvier 2012,  (téléchargeable ici) dont voici quelques extraits.

Pour un CoE dédié à "exascale Climate and Weather Science (ECWS)", ils préconisent :

Mathematical models (e.g. for new grids), numerical algorithms (e.g. for improved solvers), and scientific software codes must be reformulated to take advantage of these emerging computational platforms. Achieving these improvements within the framework, and applying the methodology, of the proposed co-design centre will require the combined expertise of, and the cooperation between, domain scientists, application developers, applied mathematicians, computer architects, system software developers, and vendors, working together to make informed decisions about features and tradeoffs in the design of the hardware, software and underlying algorithms.

A well balanced interdisciplinary team would consequently include:

• Climate and weather science experts, applied mathematicians, computer scientists, hardware architects and software engineers;

• Software developers who would be able to evaluate and implement multiple alternatives on an aggressive schedule to support the architecture development timeline;

• Experience with application codes, i.e. an existing body of work with one or more codes that are under active development, targeting at the exascale design points;

• Experience with scientific kernels and algorithms design, and optimization, uncertainty quantification, verification, and validation techniques.