Brussels, 14 Apr 2006
The European Science Foundation has launched a 'Forward Look' aimed at keeping Europe at the forefront of computational science. The scheme will enable researchers to develop codes and programs collaboratively under one infrastructure, so that rather than learn and then re-learn new computing tools, scientists will be able to build new systems and tools in tandem with developments in hardware.
At the moment, there are several High Performance Computing Centres (HPCs) which provide excellent support for scientific software development. However, there is no EU-wide support, and there is little coordination. Simply, this means that resources are wasted through the duplication of effort and skill.
The Forward Look aims to put software developers in collaborative contact and to minimise the time spent on developing new types of software, so that researchers can concentrate on what they do best - research.
The initiative is the brainchild of Professor Berend Smit from the Centre Européen de Calcul Atomique Moléculaire (CECAM) in Lyon, France, who spoke to CORDIS news. 'If you ask for research, a PhD student may use 50 per cent of his time working on code for the research, and so not concentrate on the research itself.'
Professor Smit compared his vision to the synchrotron facility in Grenoble, where researchers can arrive, conduct their research and then leave - the device and the infrastructure supports all researchers, and benefits all. Professor Smit believes his vision is essential to the development of research and computational science in the next ten to 20 years.
'We have to make sure that programs are transferable between, for example, computers and supercomputers. Within a mechanism, we can make sure that when there is a new release, we can ensure it will work on all the relevant machines, and not just one. To ensure we can do this, we need a good infrastructure and good training,' he said.
'We do not need to have a building - we need a cyber-infrastructure, but the important focus of this is related to software, which is more important than hardware. The idea is for computer simulators, who have levels of specialisation can share their data and their program infrastructures.
Professor Smit is now looking at the next stage. 'We have to raise the issue, and make funding agencies realise that there is a gap. We need to approach researchers and find out exactly what the infrastructure should look like and what kinds of service they would use.'
'It is vital that the development of this infrastructure is done at the European level, so that all of the support systems will be effective throughout Europe to enable all scientists to work more effectively and to make collaborative work possible,' he said.
'If this infrastructure is successfully developed, we should see future scientific advances, which would not be possible in the current climate. Scientists will take the initiative with this,' said Professor Smit. 'We will develop a case that can be addressed by establishing the necessary policy to keep European computational science moving in the right direction.'