NASA Funds CAII for Deep Learning Innovations in Euclid Space Mission to Study Dark Matter and Dark Energy

The Center for Artificial Intelligence Innovation (CAII) at the National Center for Supercomputing Applications has received $1 million in funding from NASA to support the Euclid space mission, which aims to study dark matter and dark energy. CAII will develop an open-source deep learning framework to address challenges in analyzing images captured by Euclid, particularly identifying blended galaxies or overlapping sources that complicate data interpretation. The project, led by Principal Investigator Xin Liu and a team of researchers, will utilize the DeepDISC tool to improve accuracy and efficiency in galaxy detection and analysis, with applications extending to other space exploration initiatives like the Vera C. Rubin Observatory.

The Center for Artificial Intelligence Innovation (CAII) has received a $1 million grant from NASA to support the Euclid space mission, which investigates dark matter and dark energy. CAII will develop an open-source deep learning framework using DeepDISC to process images from Euclid, enhancing the identification of blended galaxies that complicate data analysis.

A key challenge in Euclid’s data analysis is the presence of overlapping sources, leading to biased measurements in critical areas such as photometry and weak lensing. Xin Liu, the principal investigator, explains that DeepDISC will quantify uncertainty in predictions, improving the mission’s accuracy.

DeepDISC’s adaptability extends beyond Euclid to other missions like the Vera C. Rubin Observatory, enhancing the analysis of both ground-based and space images. This versatility underscores its potential impact across various astronomical projects.

The project benefits from a multidisciplinary team led by co-principal investigators Vlad Kindratenko, Yue Shen, and Yuxiong Wang. Their expertise in computational infrastructure, data analysis, and machine learning ensures robust methodologies for the mission’s success.

CAII serves as a pivotal hub for AI research at the National Center for Supercomputing Applications (NCSA), leveraging advanced technology to foster collaboration across disciplines. This initiative exemplifies CAII’s commitment to advancing AI applications in astrophysics and beyond, aligning with NCSA’s goal of accelerating innovation through cutting-edge research.

Interdisciplinary Collaboration for Space Exploration

The interdisciplinary collaboration behind the CAII-NASA partnership underscores the importance of integrating expertise from diverse fields to address complex challenges in space exploration. The Euclid mission, which seeks to unravel the mysteries of dark matter and dark energy, benefits from the combined skills of astrophysicists, computer scientists, and engineers. This collaborative approach ensures that innovative solutions like DeepDISC can be developed and applied effectively to real-world problems in astronomy.

DeepDISC, an artificial intelligence tool designed to detect and segment individual galaxies within blended images, exemplifies how advanced computational methods can enhance astrophysical research. By leveraging deep learning algorithms, the tool enables researchers to separate overlapping light sources, reducing biases in measurements such as photometry and weak gravitational lensing. This capability is critical for improving the accuracy of data analysis in large-scale surveys like Euclid.

The adaptability of DeepDISC extends beyond the Euclid mission, making it a valuable resource for other projects such as the Vera C. Rubin Observatory. Its ability to handle complex image processing tasks across different datasets highlights the potential for broader applications in astrophysics. This versatility not only advances specific missions but also contributes to the development of more robust tools for the wider scientific community.

The success of DeepDISC relies on the interdisciplinary expertise of its developers, who have integrated machine learning techniques with astrophysical knowledge. By quantifying uncertainties in predictions, the tool provides researchers with more reliable results, enhancing the overall quality of data interpretation. This approach demonstrates how collaboration across disciplines can lead to practical solutions that advance scientific understanding.

In summary, the development and application of DeepDISC reflect the value of interdisciplinary collaboration in addressing challenges in astrophysics. By combining expertise from multiple fields, the project not only supports the Euclid mission but also contributes to the broader advancement of computational tools for space exploration.