M2 student or Engineer student internship : Machine Learning for Photometric redshift estimation of LSST galaxies
Supervisor: Prof. Simona Mei, APC ( https://www.linkedin.com/in/simona-mei-1721bb3a/
Astroparticule et Cosmologie, Physics Department, Université Paris Cité
Machine Learning techniques have revolutionized artificial intelligence. Their application to astrophysics and cosmology permits us to analyze the large quantity of data obtained with current surveys and expected from future surveys with the aim of improving our understanding of the cosmological model.
The internship is in the context of the Vera Rubin Observatory (>https://www.lsst.org/about) LLST (Legacy Survey of Space and Time), in particular in the context of the Dark Energy (DESC) and Galaxies Rubin Science Collaborations (https://rubinobservatory.org/for-scientists/science-collaborations), and of the Euclid space mission (https://sci.esa.int/web/euclid.
We will optimize and use two kinds of deep machine learning networks to optimally extract distances, as photometric redshifts, of galaxies detected in these large-scale surveys: an inception deep convolutional network and transformer neural networks.
The Vera Rubin Observatory’s mission is to build a well-understood system that will produce an unprecedented astronomical data set for studies of the deep and dynamic universe, make the data widely accessible to a diverse community of scientists[1] with the goal to address some of the most pressing questions about the structure and evolution of the universe and the objects in it. The Rubin Observatory will conduct a deep survey over a very large of over ten years (LSST) to achieve astronomical catalogs thousands of times larger than have ever previously been compiled. Only the development of efficient Machine learning techniques will permit us to analyze this large amount of data.
At present, photometric redshift estimates are based on the knowledge of galaxy stellar populations, which determine the galaxy SED (e.g. Newman & Gruen 2022). The major limitations are precision and redshift distribution. The first problem is that we still do not in a precise way the galaxy stellar populations, and we can calibrate them with limited spectroscopical samples, and this affect photometric redshift precision. The second problem is due the difficulty to recover the redshift distributions of a large sample of galaxies in the presence of uncertainty on individual redshifts.
The APC Rubin/LSST and Euclid team has installed a deep neural network (Pasquet et al. 2019; https://ui.adsabs.harvard.edu/abs/2019A%26A...621A..26P/abstract) in the DESC photometric redshift pipeline infrastructure and tested 40 variations of the network to optimize performance and running time. The goal of the internship is to validate it on galaxy simulations already available in the collaboration and compare its performance to transformer neural networks.
This work is in collaboration with Stanford University and the Rubin team at the SLAC National Accelerator Laboratory, in Stanford, USA. We have obtained a grant for the intern salary from 3 to 6 months with lodging and meals for 1-2 months in Stanford in March and April 2026. The intern will work with the Rubin and Euclid teams at APC, which include 3 Ph.D. students, a postdoc and a CDD engineer, in the larger context of the APC (https://apc.u-paris.fr/) Cosmology team (https://apc.u-paris.fr/fr/cosmologie).
The internship is open to M2 students in astrophysics or computer science and engineer students for 3 to 6 months (also ideal for engineer césures). If you are interested in this internship, please contact Simona Mei as soon as possible. The internship could lead to a PhD thesis, and if you are an international student, we will need to apply for international grants as soon as possible.