- If stars are like diamonds, why don’t they fall down?
As told by Rodrigo Ibata

– The universe is asking to be understood
I grew up in constant motion travelling backwards and forwards between Britain and Bolivia, and perhaps due to those huge contrasts I was drawn from an early age to questions that seemed almost impossibly large. The clear skies of the Bolivian highlands come to life at night with a breathtaking show of natural beauty. It is hard not to feel, beneath such a sky, that the universe is asking to be understood. My mother recalls me asking, at the age of three, “if stars are like diamonds, why don’t they fall down?” That pull toward the fundamental led me to study physics at the University of Bristol, where I graduated in 1989. I then moved to Cambridge to take the Mathematical Tripos, an experience that sharpened my instinct for rigorous, analytic thinking, a habit of mind that has shaped everything I have done since.

I was drawn from an early age to questions that seemed almost impossibly large.

Something entirely unexpected
My doctoral work at the Institute of Astronomy in Cambridge, under the supervision of Gerry Gilmore, set the course for my career. I embarked upon studying the stellar populations toward the Galactic Bulge, the dense region of stars at the center of the Milky Way. In the course of that work, in 1994, I stumbled upon something entirely unexpected: I found a previously unknown galaxy, the Sagittarius dwarf, being slowly torn apart and absorbed by our Galaxy. It was a discovery that arrived almost by accident, as the best ones sometimes do, but its implications were profound. Here was direct, tangible evidence that our Galaxy had grown by cannibalising smaller systems, a "smoking gun" for the hierarchical theory of galaxy formation. That single discovery planted a seed that has grown to define my entire research program: the idea that the faint stellar debris scattered across the Galactic halo is not noise to be cleaned away, but a very interesting signal: a fossil record of the Milky Way's violent past, and a precision instrument for measuring its invisible dark matter.

It was a discovery that arrived almost by accident, as the best ones sometimes do, but its implications were profound.

After completing my PhD I moved through a series of postdoctoral positions, at the University of British Columbia in Vancouver, the European Southern Observatory in Munich, and the Max-Planck Institut für Astronomie in Heidelberg. I finally settled down at the Observatoire de Strasbourg in 2000, where I have been based ever since as a researcher with the CNRS. These years of moving between institutions and continents were formative. I cannot thank enough my amazing mentors during those years: Mike Irwin, Harvey Richer and Hans-Walter Rix who taught me to see the grand canvas and showed me how to hone my tools to paint it with rigour and precision.

Research philosophy
My research philosophy has always been to steer a course between observations, data analysis and theory, refusing to specialise exclusively in any one. I believe that a deep understanding of galaxy formation only comes from engaging seriously with all three: knowing the limitations of the data, the strengths and weaknesses of the models, and the assumptions embedded in the underlying theory. This approach has sometimes meant working on problems before the community was fully ready for them. In the 1990s and early 2000s, studying nearby galaxies in forensic detail was relatively unfashionable, as most colleagues were focused on statistical studies of distant systems. But I was convinced that the extraordinary richness of information available in nearby objects more than compensated for their small number, and I think in retrospect that conviction proved well-founded.

Turning point
The years following the Sagittarius discovery brought a succession of results that gradually built up a new picture of how galaxies like our own have assembled. My team and I showed that the tidal stream of the Sagittarius dwarf wraps entirely around the Milky Way, and that the surprising lack of precession in that stream places strong constraints on the shape of our Galaxy's dark matter halo. We uncovered the Monoceros Ring, an immense stellar structure also encircling the Galaxy, but at the outer edge of its disk.

I have always believed that the development of rigorous, optimal analysis tools is as important to scientific progress as the telescopes themselves.

A turning point came with the Pan-Andromeda Archaeological Survey, a large imaging survey of the Andromeda galaxy using the Canada-France-Hawaii Telescope that Alan McConnachie and I co-led. The survey gave us an unprecedented panoramic view of a giant galaxy almost identical to our own, which we studied precisely to have a global perspective and to be able to put the findings in our Galaxy into a wider cosmological context. In Andromeda we discovered a very similar analog to the Sagittarius stream, the "Giant Stellar Stream" which is the largest identifiable accretion event in that galaxy. We showed that the halo of Andromeda is criss-crossed by a rich web of ghostly stellar streams, the remnants of galaxies long since consumed. One of our most striking results was the discovery that roughly half of Andromeda's dwarf satellite galaxies are arranged in an immense, rotating planar structure, a finding that poses a challenge to standard models of galaxy formation and continues to stimulate debate. Each of these discoveries was made possible not only by new observations, but by new algorithms; I have always believed that the development of rigorous, optimal analysis tools is as important to scientific progress as the telescopes themselves. This work was only made possible thanks to my amazing collaborators, especially Geraint Lewis, Nicolas Martin, Scott Chapman, Annette Ferguson, Michele Bellazzini, Benoit Famaey and Guillaume Thomas.

– The most exciting years of my career
The past five years have brought what I regard as the most exciting phase of my career, made possible by the transformative data from ESA's Gaia mission. Gaia has delivered precise positions, distances and motions for over a billion stars, enabling, for the first time, full six-dimensional maps of stellar populations throughout the Milky Way's halo. With Khyati Malhan I developed the STREAMFINDER algorithm to mine these data systematically for stellar streams, the gossamer filaments of stars that trace the orbits of disrupted clusters and dwarf galaxies. Using this tool, together with extensive spectroscopic follow-up from major observatories around the world, we have discovered and characterised 87 stellar streams in the Milky Way's halo, the largest homogeneous catalog of such structures ever assembled. Among them is C-19, the most metal-poor stellar system ever found, a relic of the primordial Universe formed from gas with almost no previous chemical enrichment, offering a direct glimpse of star formation at the very dawn of cosmic time.

We have used this atlas of streams as a set of dynamical tracers to construct a detailed, self-consistent model of the Milky Way's gravitational potential, and thus its dark matter distribution, to a level of precision not previously achieved. These streams are effectively seismometers, sensitive to perturbations from the Large Magellanic Cloud, from dark matter subhalos, and from the overall mass distribution of the Galaxy. The picture that emerges is of a Galaxy whose halo is a mosaic of debris from roughly half a dozen major mergers, and whose dark matter halo is slightly flattened and broadly consistent with theoretical predictions, but with enough dynamical and chemical nuance to keep the field busy for years.

Since Kepler inferred his laws of planetary motion by staring at tables of numbers, could an artificial intelligence do something analogous, but at the scale of modern astrophysical surveys?

Teaching machines to discover physical laws
Alongside this observational and dynamical work, I have been pursuing what I regard as perhaps the most ambitious thread of my career: teaching machines to discover physical laws autonomously from data. Since Kepler inferred his laws of planetary motion by staring at tables of numbers, could an artificial intelligence do something analogous, but at the scale of modern astrophysical surveys? To explore this, Wassim Tenachi, Foivos Diakogiannis and I developed PhySO (Physical Symbolic Optimization), an engine that builds mathematical equations symbol-by-symbol, similar to how language models build sentences word-by-word, but constrained throughout by the physical rules any real law must obey. We have recently taken this further with NestyNet, a high-precision neural network framework capable of approximating complex functions with extraordinary accuracy and computing their derivatives analytically. Our approach yields transparent, human-readable equations, the kind that can be inspected, challenged and built upon. Whether it will one day suggest a refinement to the law of gravity, or uncover an unexpected principle unifying galaxy scaling relations, remains to be seen. But I am convinced that as the data deluge from the next generation of instruments arrives, machines designed to follow physical constraints will be helping us to find the deeper discoveries.

Driven by a sense of wonder
Looking back, what has driven me most consistently is a sense of wonder at the complexity and beauty of the structures that emerge from simple physical laws, and an impatience to understand them. I have been fortunate to work at a time when the data available to astronomers have undergone a genuine revolution in quality and quantity, and I hope that the methods and results my team has contributed have helped to make the most of that opportunity. I am also deeply grateful to the many students, postdoctoral fellows and collaborators who have shared this work with me; their energy and ideas have been indispensable.

I have been fortunate to work at a time when the data available to astronomers have undergone a genuine revolution in quality and quantity.

The sense of scientific wonder I have alluded to was instilled by my mother Georgina, and I inherited from Jorge, my father, an ambition (or is it a curse?) of wanting to try to figure everything out by oneself and not be intimidated by seemingly impossible tasks. My profoundest thanks to my wife Hélène for her love, support and fascinating conversations over the years despite the demands of her own career. And I am eternally grateful for the love of my children Neil, Caela and Oliver, and my siblings Ramiro, Catherine and Alonso.