New hub gears up for algorithmic exchange

Share this on social media:

Topic tags: 

Cross-fertilisation is the name of the game for Warwick University's Centre for Scientific Computing. Vanessa Spedding explores the motives behind its creation

Warwick University in the UK is on the up and up. Sometimes considered a typical 1960s, middle-of-the-road 'redbrick' institution - not known for their distinction - the 2001 UK Research Assessment Exercise (RAE) shows its research to be the fifth most highly-rated in the country, with outstanding standards in the sciences.

This impressive performance has rightly given Warwick a certain amount of muscle, which it is flexing rather effectively, aided by a snappy approach to making things happen that leaves some older institutions standing. The result is a brand new Centre for Scientific Computing (CSC), launched within a couple of years of its initial conception.

The dream was that of mathematics professor Andrew Stewart, who had been inspired by a multidisciplinary approach to research that he had seen at a similar centre at Stanford University in the USA.

'Deciding to invest the money was fairly straightforward for the university. Warwick is proud of its flat management structure - it has less bureaucracy and fewer committees, so decisions are made relatively quickly,' explained the centre's newly-appointed director, Professor Mike Allen (pictured above).

The university will invest approximately 2m in the centre through a commitment to hire 10 new staff - which means that recruitment is Allen's first priority. He is off to a running start. When Allen started, Warwick had already made moves to bag Professor Peter Taylor, deputy director of the San Diego Supercomputing Center (SDSC) in the USA. This was through the Royal Society-Wolfson Research Merit Award scheme, a UK initiative aimed at reversing the 'brain-drain' to the USA. Winners are tempted back to the UK largely by means of cash (an additional 40,000 in salaries and research expenses).

Warwick has succeeded in its bid to lure Taylor back. He'll be a jewel in the crown, not just for his experience within a worldclass supercomputing centre, but also for his expertise in quantum chemistry.

'I'm delighted at Peter Taylor's appointment,' said Allen. 'It certainly is a catch - he's a big name. SDSC is one of the major centres for the Grid and e-science in the US and I'm hoping that these will be important aspects of the centre.'

Allen himself has something of a pedigree, with a distinguished career as a theoretical physicist at Bristol University behind him and particular expertise in statistical mechanics and thermodynamics methods for the molecular modelling of liquids and liquid crystals.

The goal of maximising communication across research disciplines will be a powerful influence on how the Warwick CSC is set up.

'I expect to see enormous cross-fertilisation between disciplines,' enthused Allen. For example, the area of modelling encompasses all fields of science. Take Monte Carlo models. They are used in statistics, maths, physics, chemistry, biology and even some areas of engineering.

'Expertise in implementing algorithms in one discipline should ideally be shared across the others. It's been my experience that this sort of cross-over doesn't happen much in the conventional university setup.'

But how exactly do you facilitate this sort of sharing? 'In reality, it will be something of a compromise. Face-to-face contact is always nice and we will encourage this by having a physical location for the centre, where people can work in the same environment, and also by arranging interdisciplinary seminars, workshops and self-help groups.

'But we will be treading something of a tightrope. We don't want to suck the lifeblood out of the departments. Making them teaching-only and tying research to the research centres is not good either. The idea is to strike the right balance.'

Peter Taylor shares Allen's enthusiasm for the project. 'This is a terrific opportunity at a terrific institution. At the SDSC we put a lot of effort into cross-fertilisation.

'At Warwick we will have the opportunity to broaden the model. We expect not only to be bringing together people from different disciplines with people from computer science, but also with mathematicians. This will give us the chance to go back to fundamentals, to the algorithms behind the codes, and find new ways of tackling problems in the different subject areas.'

Taylor's own research, which he will continue, has focused on intensive computing methods for the accurate prediction of molecular properties for small molecules. He is also keen to bring new research communities into the HPC arena.

'There are communities who have not participated as much in HPC or whose needs have not been as well met.

'For example, particle physicists, people working on biological macromolecules. They are characterised by generating large amounts of data, and need rapid access to that data if they are to make meaningful comparisons and analyses. This is something we'd like to do - explore data-rich computing so that people can do more with their results. We may even extend it to the social sciences - economics and so on.'

On top of these ambitious ideas the Warwick CSC will be focusing on exactly the areas you'd expect such as climate modelling, astrophysics, bioinformatics and computational chemistry.

It will also manage a training programme to make sure that Warwick's graduates and post-doctorates are fully up to date on scientific computing techniques and software, and is putting together plans for extensive collaborations with other academic institutions and industry.

It's a tall order, and the process can't begin until more people join in - but if the plans for the centre actually come off as intended, the project could put Warwick firmly in the middle of the scientific computing map.