What is BLASST

The Brazilian Lab for Astronomical Survey Studies (BLASST) is a pioneering initiative designed to significantly amplify the scientific return from existing as well as future astronomical surveys, such as the Vera Rubin Legacy Survey of Space and Time (LSST). The primary motivation behind BLASST stems from the unprecedented data deluge expected from these facilities, which demands sophisticated, cutting-edge approaches to data processing, analysis, and interpretation. While raw data distribution is crucial, BLASST recognizes the critical need for creating "value-added" data products and developing novel methodologies that go beyond basic data access. By leveraging state-of-the-art software, including advanced Machine Learning (ML) models and intelligent agents, BLASST aims to transform raw observational data into readily usable scientific catalogs, and from value-added catalogs to the final scientific exploitation of the data, thereby multiplying the scientific impact for the broader astronomical community in Brazil and globally.

BLASST is an inclusive group that is constantly adding new members. Presently the team is made of 40 experienced researchers spanning 14 Brazilian institutions: CBPF, IAG-USP, IF-USP, IFT-UNESP, IME-USP, ON, INPE, UESC, UFRJ, UFES, UFRGS, UNIVAP, UFS, and UFSC. This diverse expertise covers various areas of astronomy, astrophysics and cosmology, with significant expertise in computation science and Machine Learning. Functionally, BLASST will complement existing data distribution systems by focusing on the creation of these high-level, value-added catalogs from various surveys, enabling real-time LSST image subtraction and advanced photometry techniques, with a particular emphasis on LSST data. Its scope goes well beyond creating and curating catalogs, and includes developing new analytical methods, fine-tuning Large-Language Models (LLMs) and agentes, integrating artificial intelligence into the analysis of astronomical survey data, in particular that from LSST. This comprehensive approach ensures that BLASST will not only process data but also innovate in how astronomical information is extracted and utilized to produce high-impact science

5-year plan

The implementation of BLASST is envisioned through a strategic five-year plan. In the initial two years, the focus will be on establishing the core computational infrastructure, developing foundational data pipelines, and creating initial versions of value-added catalog generation tools. This phase will involve recruiting the first cohort of graduate students (MSc and PhD) to develop key methodologies and undergraduate students for data curation and testing. Several post-doctoral researchers will be brought on board to lead specific research areas, such as advanced ML applications or time-domain astronomy. Essential technical support, including system administrators, software engineers and project management, will also be secured to ensure robust system operation. The first steps towards greater community engagement will start with training, user-friendly web-based capabilities as well as workshops.

Years three and four will see an expansion of catalog generation capabilities and the integration of more sophisticated ML and AI techniques. Student involvement will increase, and additional post-doctoral researchers will be recruited to specialize in advanced data analytics. The technical team will also expand to provide comprehensive system maintenance and user support.

By the fifth year, BLASST will concentrate on refining its catalogs, publishing significant scientific results, and expanding its community engagement. At this point some post-doctoral researchers may transition into permanent research positions, while student training and technical support will continue to ensure the project's long-term sustainability.

Hardware requirements will include high-performance computing clusters equipped with both CPUs and GPUs for intensive data processing and machine learning tasks, petabyte-scale data storage solutions, and high-bandwidth network connectivity to facilitate data transfer and access.