Wednesday, November 24, 2010

User tales from the NGS Innovation Forum, STFC Rutherford Appleton Laboratory

I have spent the last two days at the STFC Rutherford Appleton Laboratory, a place close to my heart as I worked here for nearly 7 years. As well as catching up with some ex-colleagues, and trying to rebuild my mental maps of the randomly numbered RAL buildings, I was here to find out more about UK grid services through the NGS Innovation Forum. We heard from several users of NGS resources, and also some interesting stories about innovation in the infrastructure. David Wallom of Oxford University told us about an EPSRC funded project to set up cloud systems, of a type known as Infrastructure-as-a-service (IaaS). The National eScience Centre at Edinburgh and the Oxford eResearch Centre are providing virtual machines on request that can be set up however a user wants. They are using Eucalyptus, which is an open source implementation of Amazon’s EC2 service. Once launched, they were surprised to get 20 users applying in the first week – they now support 70 users with applications ranging from bioscience to geospatial analysis, civil energy and smart electricity grids.

We also heard from several other users of the more traditional NGS services, including work on mRNA analysis in crop pests, modelling complex engineering materials and investigating microclimates inside combat equipment. Luke Rendell of the University of St Andrews gave us an insight into a social learning computer tournament called ‘Cultadaptation’. Social learning is the flow of information between individuals, whether human or animal. We are surrounded by this from the cradle to the grave, but what are the best strategies for learning in an unfamiliar environment? Luke compared this to the surviving in the jungle challenge – there are many ways to find food, from eating grubs to picking fruit, but which one gives you the best payoff (the current crop of “celebrities” in I’m a Celebrity Get Me Out of Here could probably take note). Cultadaption set up a virtual tournament to find out – competitors could submit code or pseudo code describing behaviours, and survival of the fittest then ensued. Once launched, the tournament attracted ten times the number of entries expected. Luke said that he thought ‘their goose was cooked’ when they realised they would need to run 100,000 simulations. Fortunately, the NGS resources rode to the rescue and the tournament was a great success. You can find out more about the winners and the best strategies at but it seems that one of the most successful ideas is to copy others rather than innovate yourself. But clearly this isn’t the general message that the NGS Innovation Forum is hoping we’ll take home!

No comments: