Next Generation Computational Modelling Summer School

Author: Adrian Jackson
Posted: 15 Jul 2015 | 15:06

Discussions on computing

Myself and Fiona ReidNGCM - Next Generation Computational Modelling recently presented a 2-day course on porting and optimising for the Xeon Phi at the NGCM (Next Generation Computational Modelling) summer academy in Southampton. 

This one-week academy is designed to give PhD students some of the skills they need to undertake the range of computational simulations and data analysis tasks that their work requires.

 As well as our course on the Xeon Phi, there were courses on CUDA, iPythonPandasVTKFEniCS, software development skills, and other things. There was quite a strong Python flavour to a lot of the courses, showing the impact it is having on computation and data analysis for scientific research.

I was also on a panel (along with Kenji Takeda who was teaching a course on Azure, Marie Rognes who was presenting the FEniCS course, and Skipper Seabold and Christopher Fonnesbeck who were teaching the Pandas course) designed to give the students the chance to quiz the instructors on their views of any aspect of PhD work and computational simulation/programming/data analytics. 

There were some lively discussions between the panel members and with the audience, but I think I upset some on the panel by suggesting that it's a shame that scientific researchers have to learn coding to work. I should be clear, and probably wasn't as clear in the panel session as I could have been (hence the upset): I'm not saying that current researchers, schoolchildren, university students, etc... shouldn't learn to code. Indeed, everyone should be coming into academia and the workforce with the skills to be able to effectively manipulate the software and data they need to work, and to interact with the internet and public through software (be it websites or apps or other things) in the best way. 

The point I was trying to make is that computers and software should just be tools for researchers and industry. Ideally we'd be in a position where the software infrastructure – and hardware it's deployed on – is functional enough that researchers don't need to write their own programs and can just input some equations to run a simulation, or process some data with the click of a few buttons to get meaningful statistics out.

NGCM Panel Discussion Panel Discussion at NGCM in Southampton

However, we're not quite there, so researchers do need the skills to manipulate their data, or write their own simulations (whether it be with C or FORTRAN or something higher-level). Also, even if we do get to a situation where the software landscape is rich enough to support people working without having to program, we will still need some specialists who can interact with the deeper levels of computing software and hardware, just like we still need people who can build and maintain our scientific instruments and tools. 

The point is, if simulation, modelling, and data analysis can become just another set of tools that are routine to manipulate, then researchers can focus on the important work of pursuing their research.