The importance of large scale data analysis increases year by year in not only a scientific domain such as high energy physics, astronomy, biology but also an enterprise domain such as web and mail search engines. Moreover, collaborative e-Science activity requires efficient data sharing and data analysis within a virtual organization that may consist of several institutes located distantly.

The international data analysis challenge --- Finding Supernovae --- facilitates such a trend and encourages efficient data analysis efforts in distributed environment. In this challenge, huge amount of real scientific data taken by the Subaru telescope for seven years is used to find new and unknown supernova candidates using several cluster systems in Japan that consist of thousands of CPU cores, which includes the InTrigger info-plosion platform and Tokyo Tech TSUBAME.

We are inviting research teams or individuals to participate in the challenge. Researchers and students working on parallel and distributed computing and large scale data analysis as well as astronomical scientists are welcome. Selected teams will be invited to present their efforts at the IEEE Cluster/Grid conference. The winner will receive a student travel award to help with travel to the conference.

• Subaru Telescope

  Subaru is an 8.2 meter optical-infrared telescope at the summit of Mauna Kea, Hawaii, operated by the National Astronomical Observatory of Japan (NAOJ), National Institutes of Natural Science. For details, see http://www.subarutelescope.org/

• InTrigger info-plosion platform

  InTrigger is a distributed computing platform spread around Japan. As of the competition, it consists of about 850 CPU cores across eleven clusters.

Real scientific data taken by the Subaru telescope is dispersed across several clusters in the Challenge platform. The challengers find all supernova candidates using the Challenge platform. To find all supernova candidates, it is necessary to compare every two shots taken at an interval of one month. The data analysis program is given, while it is allowed to optimize it as long as the result is not changed. There are two challenge categories BS and FT.

** Category BS: basic category

The challengers compete in the total time to find all supernova candidates. There is no restriction on programming language and distributed computing middleware. Any failure during the execution is not assumed in this category.

** Category FT: fault tolerant category

The challengers compete in the total time to find all supernova candidates under a scenario of artificial node failure. Some processes will be killed according to the failure scenario. Although a trial scenario is prepared, a scenario for the challenge is not provided until the final measurement. There is no restriction on programming language and distributed computing middleware as the Category BS.