A sneak peek at an early version of the challenge’s data cube, showing galaxies up to a redshift of 0.5 – that’s a distance of around 4 billion light years. Credit: Dr Philippa Hartley
SKA SCIENCE TEAM GEARING UP FOR DATA CHALLENGE #2 BY SKAO Preparations are accelerating for SKA’s Science Data Challenge 2, which will be launched later this year. The aim of the Data Challenges is to prepare the scientific community for the reality of working with SKA data. Run by the SKAO Science team, the new challenge will be an HI galaxy simulation in which teams are asked to locate galaxies and discern their properties using mock SKA data. They’ll do so by running their own software analysis tools on a 3D “data cube” – a series of stacked radio images, each reflecting a different frequency. Unlike a 2D image, the data cube will allow researchers to determine things like how fast the galaxy rotates, and how much mass it must contain.
This time instead of downloading the data to their own computers, participants will access and process these data using major computing facilities, just as SKA users will eventually have to do via the SKA Regional Centres (SRC). This allows future users to familiarise themselves with the kind of computing environments the SKA will provide. “Using computing clusters allows us to test that SRC model, and also means we can create a much bigger data set for the challenge, around one terabyte in this case, which otherwise could take weeks to download,” says SKA Project Scientist Dr Anna Bonaldi, who is co-leading the work.
The Shanghai SKA Regional Centre Prototype and ENGAGESKA Portugal cluster are lined up to store and process the data, and others may join them. As well as helping the SKA Science and Operations teams to test the model of transferring data back and forth to dispersed computing clusters, the challenge will also start a dialogue between users and the facilities involved to identify any training or additional tools that would be useful. The SKA HI Science Working Group will soon begin testing and debugging a subset of the data as part of the preparations.
USING COMPUTING CLUSTERS ALLOWS US TO TEST THAT SRC MODEL, AND ALSO MEANS WE CAN CREATE A MUCH BIGGER DATA SET FOR THE CHALLENGE. Dr Anna Bonaldi, SKA Project Scientist
5
