MSV-2035 Astronomy Document - Inside Design - FINAL - FINAL

Astronomy & Astrophysics 93 developed software skills and its aspirations in high-performance computing to fulfil this critical role. The resources needed to fully process, distribute, curate and utilise data flowing from the SKAobservatory represents an enormous challenge. To develop a plan for an international network of data centres - referred to as SKARegional Centres (SRCs) - the SKA organisation (SKAO) has formed the SRC Coordination Group with active Indian participation. The SRCs are required to provide the compute, storage and analysis/visualisation capabilities needed to handle exabyte-sized datasets. It is expected that there will be several such SRCs located across the globe. The SKAO and the network of SRCs will need to work collaboratively to shape and establish a shared, distributed data, computing and networking capability that draws on international cooperation and supports the broad spectrum of SKA science. The full SRC network is expected to be operational by 2029, with one node planned to be located in India. Over the multi- decade operation of the SKA, the SRCs will provide the computing resources needed by the international community for maximal scientific utilisation of SKAdata.Asmall scale prototype version of the SRC that can cater to the needs of the SKA precursor and pathfinder facilities, such as the upgraded GMRT, will be made operational by 2024, and the full-scale SKAPhase-1 compatible version is planned to be ready by 2028, with periodic technology refresh activities planned every subsequent 4-5 years. 5.7.3 GravitationalWaveAstronomy - DataAnalysis andNumerical Relativity Detection of gravitational waves in noisy data requires enormous amount of computing power. Several searches for GW are presently limited by computing power. For example, the searches assume the spins of the binary components are aligned with the orbital angular momentum, without any strong physical motivation. Thus, at present, the chances of detection of binaries with non-aligned spins are less. The computation power necessary to relax this assumption in the models is orders of magnitude more than currently available.Ahybrid scheme to “interpolate” numerical relativity templates in the parameter space is used even for the present simpler searches. While several algorithms are being developed to speed up computation, state-of-the-art data centres are being built for such analyses across the world by the international community. IUCAAhas built a data centre with ~ 500 TeraFlops peak theoretical computing power, which is expected to ramp up by the time LIGO-India comes online. 5.7.4 Requirements of OtherMega Projects As with SKA and LIGO-India, other mega projects such as TMT, NLST, NLOT, etc. require significant amount of computational resources for project specific requirements. These requirements include data acquisition, preliminary analysis, data centre for storing raw data and computational platforms for analysis of data. Some of the instruments, such as integral field units and fibre based spectrographs, generate large data volumes during each observation that require special platforms for analysis. The data are often processed in batches, and with the recent progress in machine learning and related techniques, it has become possible to explore data for unanticipated features as well. Each mega project also requires resources for making data available: raw data as well as data products and catalogs need to be made available to all users to ensure optimal utilisation of the data. MEGA SCIENCE VISION-2035