With the installation of three supercomputers, the University will soon be at the forefront of High-Performance Computing. The largest of these computers will be the fastest university-owned computer in the UK and among the top 100 in the world. But these new facilities will not just benefit the University, they will also help the developing world.
SCAT, which stands for Scientific Computing Advanced Training, is founded on the idea that co-operation in higher education is a means of fostering economic and social progress and, in the end, improving lives. The specific aim is to improve the conditions of training of highly qualified individuals in a skill that will undoubtedly play a dominant role in the future of scientific discovery and engineering design. The SCAT project is funded by EuropeAid and it will provide 20 high-value mobility grants for postgraduate students to spend a period of study and research in a partner institution. The European partners are in France, Spain and the UK, while the Latin American partners are in Brazil, Chile and Mexico.
The University will soon be at the forefront of high-performance computing
In addition to the mobility grants, the SCAT project organises international scientific workshops and it is under-taking both e-learning initiatives as well as distance collaboration assisted by technology. These actions, under-pinned by the goal of improving the expertise of young scientists in scientific computing, have a beneficial effect in all partners. For the University of Bristol, leading such a project increases its international standing and influence. It will also benefit from long-term professional relationships, increased collaboration, and the enriching nature of the visitor programme.
Dr Lorena Barba is the instigator and current co-ordinator of the SCAT project. She grew up in Chile during the dictatorship and obtained her first degree in mechanical engineering. During that time she became interested in fluid mechanics and in research, but her options were limited, so Dr Barba is not unfamiliar with the difficulties faced by young people aspiring to be scientists in developing countries. After finishing her degree it took seven years of hard work to pay back her university loans before she could start her postgraduate studies at the California Institute of Technology, where she obtained a PhD in aeronautics in 2004. Now resident in Bristol, she maintains ties with her first mentors in Chile and through them has extended her network of contacts to Brazil, as well as to France and Spain. Her proposal became one of only six successful projects in this final round of funding.
After a year of negotiations with the EC, the project finally got started and following the launch meeting in Barcelona (February 2006), the SCAT project held its first international meeting in the Daresbury Laboratory last June. There the international team considered the first batch of applicants for the mobility grants and chose to award three. Two of the successful applicants will come to Bristol to work with Dr Barba, while the third goes to Marseille.
Clever algorithms can achieve far more than fast computers
Felipe Cruz is currently an MSc student in the field of Informatics and Computer Science, in Valparaíso, Chile. His SCAT grant will enable him to work on the implementation of fast particle methods where the calculations involved are usually too great, and the computational resources too small, to complete them in a sensible time frame. Moore’s Law suggests that computer processors double their speed every 18 months, so in theory if we wait long enough the computer will eventually become fast enough to solve any problem. But in fact clever algorithms can achieve far more than fast computers, and in a much shorter time-scale. For example, if a problem has a million unknowns and there is a solution algorithm which requires N-squared operations, it would take 16 generations of Moore’s law (ie 24 years) before this solution method is comparable to a ‘fast’ algorithm only requiring N operations. Felipe will be working on devising such intelligent algorithms that will speed up the calculation of large numbers of interactions by making small approximations. When simulating the interactions of millions of celestial bodies under gravitational forces, for example, those that are far away from a point can be bundled together to calculate their approximate influence, thereby greatly speeding up the results.
Helmut Wahanik, from the Instituto de Matemática Pura e Aplicada, Brazil, will come to Bristol to study objects that appear in fluid flows where a number of eddies form a tight constellation and rotate together. They are called vortex multipoles and they behave in fascinating ways. The tripole, for example, is an arrangement of three eddies, where the central one rotates in the opposite sense to the two outer vortices. The whole arrangement rotates around the centre and can survive for a long time. These types of vortices are studied in relation to the oceans where the main source of variability in temperature, salinity and other properties is caused by eddies, so it is important to understand their behaviour.
For all of these young aspiring scientists, the opportunity to join a research group in Europe, have the mentoring and guidance of established international leaders in their fields, and have the chance to do hands-on computational science in world-class facilities, will be life-changing. Apart from what they learn during their stay, they will go back to their countries to become part of the academic community, having developed international collaborations that will help their career for years to come.
Lorena Barba/Department of Mathematics