Case study: A gem of a user
Neasan O'Neill finds how Ernesto Garcia is putting the grid to good use
How much grid does one person need? Well one man, Professor Ernesto Garcia Para, apparently needs a lot. He has used over 15 million hours of CPU on the European Grid Infrastructure and his work is still going strong. Garcia expects to use 5 million hours this year.
Based at the University of the Basque Country in Vitoria, Garcia and his team are interested in investigating what is going on at the atomic level in chemical reactions. This is a very wide field and the reactions studied have many applications. Garcia’s recent results have focused on reactions in the atmosphere including how they are affected by carbon emissions or their impact on spacecraft re-entry.
Garcia started using the grid back in 2006. “We couldn’t do our research without computers, the days of the back of an envelope or the blackboard are well gone. However, doing detailed calculations of reactions takes a long time on the standard resources available to us. To be honest when I was first shown the grid I was a little sceptical but decided that if it delivered on its promises it would be a boon to our field.”
This research combines cutting edge theoretical chemistry with computer modelling, to examine, and predict real world scenarios. The computer models take into account as many variables as possible, to ensure that they follow theory and match experimental results. This requires an enormous amount of computing resources.
He explains: “Some of the reactions we are investigating look quite simple, but saying A+B give you C+D hides a huge amount of complexity. What are the intermediate steps, what part do the local conditions play, can the reaction go backwards? This made even a 'simple' reaction that involves only three atoms extremely problematic.”
Garcia and his team are not alone. Computational chemistry was one of the first communities that saw the potential of grid computing and by 2006, when they started, there were applications already there for them to try out. The tools have matured with the infrastructure and Garcia has seen this progress first hand: “When we started the tools were useful but slightly crude. Now the guys at the University of Perugia have developed the Grid Empowered Molecular Simulator (GEMS), which makes my life a lot easier.”
One of the people at Perugia is Antonio Lagana, from the university’s Department of Chemistry, who coordinates a joint team of molecular and computer scientists. He says: “I think one of the most important steps was the creation of the COMPCHEM VO [Virtual Organisation]. This gave new, wary, users instant access to resources and applications but more importantly it gave them a support network and a way for them and developers to communicate.”
The GEMS application is built in ‘blocks’, making it modular and customisable. Each block performs a certain kind of calculation and can be used or discarded, as the user wants. The user can also determine the code/algorithm each block uses.
GEMS has also tackled another problem. “The lack of standard data formats has plagued this community for years,” says Lagana. “This has made collaboration difficult as the different codes people use just simply can’t talk to each other. During our work on GEMS we have made some progress on this. We needed the different blocks and codes to be able to exchange information so we have developed and adopted standards that will benefit even the non-grid users in our community.” •
