Cloud Computing Busy Week
In February, a five-day busy week was conducted at the University of Arizona in order for PIRE members to work on large-scale synthetic data generation for the Event Horizon Telescope (EHT) project. The effort was coordinated with senior team members from CyVerse, Tyson Swetnam and Nirav Merchant, and included visitors Freek Roelofs and Michael Janssen (Radboud University), Angelo Ricarte (Harvard University), and David Lee (University of Illinois) along with University Arizona PIRE members Chi-kwan Chan, Feryal Özel, Dimitrios Psaltis, Kaushik Satapathy, Tyler Trent, and Sean Dougall. The initial effort entailed organizing all available simulations of accretion flows around black holes in a consistent format on the CyVerse environment. SYMBA software was used to create synthetic data sets and launch online jobs on the cloud environment.
SYMBA simulates the entire EHT signal path — how radiation passes through the Earth’s atmosphere, how this radiation is captured by the EHT telescope array, and how this raw synthetic data is further processed for scientific analysis (similar to observational data). Therefore, SYMBA enables accurate predictions of how the EHT ‘sees’ a particular source model. During the busy week, a dockerized (and therefore fully reproducible) workflow was created using the Open Science Grid to run SYMBA and storing the synthetic datasets on CyVerse. First test runs have been successful and we are now moving towards large-scale production.
The resulting images and synthetic data will be used in the ongoing EHT efforts to study M87 and Sgr A*. New tools were also developed during the busy week to manage large simulation libraries as well as scaling data- and computation-intensive jobs to the cloud and the grid to further support the EHT’s science activities.
View the mini-gallery here.