Grid Computing for the Large Hadron Collider

Wednesday 8th July 2009, 6:30 pm.

Speaker: Professor Tony Doyle, University of Glasgow.

University of Edinburgh Informatics Forum, 10 Crichton Street, Edinburgh, EH8 9AB - map (click on Informatics Forum in the list of buildings).

This talk is free of charge. Refreshments available from 6:00 pm.

Synopsis

Imagine you want to explore the atto-scale universe to determine what drives the earliest nanosecond quantum fluctuations at the birth of our Universe. To do so you'll need to build a computing system of 100,000 CPUs with access to PetaBytes of data that scale with Moore's law and Kryder's law. Network capacity will increase with Butter's law such that a Global system can be considered. You'll have to build this up within ten years and then make sure its stable and secure for thousands of users yet scales in compute power for up to twenty years. These considerations led to the adoption of the Grid for particle physics ten years ago. As the associated machine is now ready to start we highlight the challenges that have been overcome and those that lie ahead in Grid computing for the LHC.

The Large Hadron Collider (LHC) will enable us to explore the smallest distance scales and reach back to the earliest moments after the Big Bang. The complexity of the collider and detector systems is matched with a computing system that enables thousands of collaborators to share data worldwide. The Grid that has been developed is now providing compute and data resources for particle physics and a wide range of other disciplines. The talk will describe the steps taken to build the current Grid system, highlight the current issues, and explore what might be achievable using distributed computing systems of the future.

About the speaker

  Tony Doyle is a professor of physics and group leader of the experimental particle physics group at the University of Glasgow. He is currently working on particle physics experiments based at CERN (Geneva) and DESY (Hamburg). He has worked on various data analysis issues, starting with online trigger systems, progressing to Monte Carlo simulation studies and through to physics data analysis. As an Alexander von Humboldt Fellow, DESY Visiting Scientist and member of the ZEUS Collaboration, he worked on the analysis of Structure Functions and Hadronic Final States in Deep Inelastic Scattering.

He is GridPP Technical Director, working with a collaboration of Particle Physicists and Computing Scientists from the UK and CERN, who are building a Grid for Particle Physics. He co-directs the ScotGrid project, collaborates within the EU EGEE project, and is a member of various UK e-Science Boards and Committees.

Most recently, as an STFC Senior Research Fellow and member of the ATLAS Collaboration, he has been working on analysis methods to search for the Higgs boson at the LHC. This task requires petabyte-scale data resources and led to an interest in Grid Data Management, including efficient navigation mechanisms between datasets using tags. Secondly, despite the large computational resources available via the Grid, the task necessitates improved simulation methods. Thirdly, it requires enhanced analysis methods, including multivariate analyses and advanced statistical methods to extract the Higgs signal from the significant backgrounds.

He is a Fellow of the Institute of Physics and the Royal Society of Edinburgh.