Julien Peloton: when computer science meets astronomy and combines with open science

“Science is beautiful when it is open and accessible to everyone,” begins Julien Peloton, research engineer at IJCLab since 2017. Holder of a doctorate degree in theoretical astrophysics with a particular focus on analysis and exploitation of data coming from cosmological phenomena, his principal activity is to provide research teams with technical solutions adapted to the analysis of large datasets (up to a few petabytes, or 10¹⁵ bytes). In this context, he launched the FINK community project in 2019, which he has been coordinating since then. FINK is a software capable of managing, in real time, the flow of alerts per night, that are associated with transient phenomena in the sky and generated by the LSST (Large Survey of Space and Time) telescope of the Vera C. Rubin Observatory, in Chile. In 2023, FINK was awarded with the open science prize for free software by the Ministry of Higher Education and Research in the jury’s “favourite” category.

A growing appetite for software development

“Before starting my PhD, I was not particularly interested in software development,” recalls Julien Peloton. “From the very first day, I was only programming, but fortunately, I benefited from a very rich exchange with the interdisciplinary team in which I was. Not only physicists, but also engineers, mathematicians, and statisticians worked together there. At that moment, I realised that computing was not only a tool but also a science in itself. While it was a technical need at the start of my thesis, I gradually began to develop an in-depth interest in it. This subsequently allowed me to combine informatics with the knowledge of physics to become a research engineer.” Julien Peloton notes, however, that the codes are often closed. “I told myself that maybe we were doing something that had probably already been done elsewhere, and it was also to change this way of working that I very quickly began to freely distribute my codes.”

The daily life of a research engineer

At the IJCLab, Julien Peloton's daily life consists of several “back-and-forths” with the research teams, in order to understand their technical and scientific needs, but also to provide training. As the question of software quality is rarely addressed by the students during their studies, it is, for a research engineer, not only the issue of writing code, but also of making it readable using existing standards, to test and document it, and to be able to deploy and maintain it over the long term. Ultimately, this turns out to be more time-consuming than writing lines of code itself. “Software quality and support are rarely addressed in school programs,” emphasises Julien Peloton.
As a research engineer, he considers himself at the crossroads, between technical and scientific expertise. “Astronomy is a very computerised scientific field, and it is with our technical skills and expertise that we cross important steps. This requires strong collaboration between the teams to advance research, as no one alone has all the keys.” He regularly ensures that no user is left in the dark when it comes to FINK.

The contributions of FINK software

FINK provides an interface between the scientists and massive data coming from astronomical surveys. The cases studied are varied - from the constituents of the solar system to the stars of our galaxy and even beyond - and the evolution of the transient phenomena involved ranges from a few minutes to several years. While observatories regularly send data in the form of alert streams, FINK analyses it in real time to understand the nature of the underlying physical processes. “The users access a probability function as an output, allowing them to select the objects that interest them from among the millions present at the input,” explains Julien Peloton. The analysis part of the software is based on the contributions from about sixty people involved in a dozen of teams around the world, and each providing a brick of the software that benefits everyone. “It's exciting from a scientific point of view to bring together so many different communities, but it's also challenging from a sociological point of view.” The project ultimately creates a bridge between scientific communities who necessarily do not work with each other otherwise and whose daily practices are different. For Julien Peloton, the great strength of free software is being able to run, test, modify, reuse and distribute this support.

Automated and hosted on a cloud, FINK makes it possible to continuously extract scientific information, automatically made public. “I often receive very positive feedback from the users on the fact that we have made their work easier,” rejoices Julien Peloton about the impact of the software, while urging us to keep an eye on scientific treatments. Because even if the use of artificial intelligence is great in processing chains, nothing replaces scientific teams when it comes to making complex decisions.

Open, but not just any way

“Informatics has become almost essential for scientists. They are increasingly asked to produce codes or software for their work, without prior training,” testifies the research engineer at the IJCLab. However, codes poorly described by some often become black boxes for others. Using open science is thus, a way to help overcome difficulties, especially with the advent of artificial intelligence. For Julien Peloton, “the opening of codes is absolutely necessary, but not sufficient.” And it should not be done anyhow.

Opening is not just about making the source code accessible to everyone: it must also be easy to interpret and use. You must succeed in deciphering what the software contains, using its documentation and the tools around it (API, test suite, etc.). “On the other hand, doing this takes a lot of time and represents a considerable investment. This is not always easy to do, especially at the start of your career.” The difficulty often lies in the gap between the expectations of the scientific teams (a scientific result) and the means to achieve it (for example, the production of a software). Appropriate training, timely technical support, or even development management by a technical team can help alleviate this problem.

Motivating young people to do research

“Our mission is also to show the younger generation that research is interesting,” comments Julien Peloton. “You have to give them a taste for it, without being too technical from the beginning. Of course, we do not go through technical and scientific revolutions every day, but we modestly make our contribution.” How can we, in this case, avoid the gap between the public's perception of research work and what scientists carry out on a daily basis? This involves in particular, science outreach actions, “unfortunately rarely taken into consideration while evaluating our career,” regrets Julien Peloton. “For example, how can we highlight the impact of a scientist speaking on the radio about the contribution of open science?”

The next steps?

With a new observatory under construction, whose operation will begin in 2025 and continue until 2035, the FINK project is already anticipating new challenges. To respond to this, he plans to deploy the software on a larger computing infrastructure, at the computing centre of the CNRS National Institute of Nuclear and Particle Physics (CC-IN2P3). The archiving of the code on Software Heritage, an open archive for software source codes, is also foreseen, in order to ensure its long-term preservation.