Periodic Reporting for period 4 - GREATDIGINTHESKY (Accelerating Galactic Archaeology)
Periodo di rendicontazione: 2024-04-01 al 2024-09-30
The “GreatDigInTheSky” project aimed to use the relic structures in the nearby universe to measure the acceleration field around our Galaxy to try to understand the nature of gravity and the properties of dark matter. One of the great puzzles in physics today is that our direct observations seem to be “missing” most of the universe’s mass. The leading theory is that this material is composed of some, as yet unidentified, elementary particle, that interacts only very weakly (or not at all) with normal matter. While this so-called “cold dark matter” theory is in good agreement with observations on cosmological scales, some important tensions appear on the physical scale of galaxies. Many competing theories have been developed to address these tensions, positing different physical properties for the dark matter particle. Yet others propose that dark matter does not exist, and that it is our theories of gravity that are at fault.
The recent revolution in the quality of astronomical data, especially for the Milky Way, now permits a careful reassessment of dark matter and gravity theories. The data of particular importance are the astrometric observations from the European Space Agency's Gaia mission. Our team has developed methods to use these Gaia data to identify a large number of star streams in the Milky Way. These structures appear as long bands of stars on the sky that share share similar distances and velocity. These star streams are relics from ancient dwarf galaxies and star clusters that were slowly wrecked over cosmic time due to the tidal forces of the Milky Way. Their importance is that the stars that they are composed of have very similar orbits, and as such they can be used to put stringent constraints on the acceleration field (i.e. on the force field) of our Galaxy.
Since theories of gravity and dark matter are effectively recipes for the force field, we worked to build methods that would allow us to distinguish which model works best, and which ones can be ruled out. Needless to say, finding the solution to this very fundamental problem of physics will help improve our understanding of the universe, of mass, and of the force of gravity.
Our project was successful in finding a large sample of 87 stellar stream structures and in developing a state-of-the-art toolset to analyse them. We were able to put stringent constraints on the distribution of mass in the Milky Way. We showed that wide binary stars pose a strong challenge to Modified Newtonian Dynamics. However, during the course of the project we realized that that the dynamics of stellar streams could be more complicated than we initially envisaged due to their possible partial dissolution within the dark sub-halo within which they were formed, before they were incorporated into the Milky Way. This complexity, while making the modeling even more challenging, actually provides new opportunities to probe the hierarchical assembly of our Galaxy. We are currently developing data-driven approaches to account for these effects in our dynamical models, with final analyses forthcoming that will exploit this added layer of information about galaxy formation.
The recent revolution in the quality of astronomical data, especially for the Milky Way, now permits a careful reassessment of dark matter and gravity theories. The data of particular importance are the astrometric observations from the European Space Agency's Gaia mission. Our team has developed methods to use these Gaia data to identify a large number of star streams in the Milky Way. These structures appear as long bands of stars on the sky that share share similar distances and velocity. These star streams are relics from ancient dwarf galaxies and star clusters that were slowly wrecked over cosmic time due to the tidal forces of the Milky Way. Their importance is that the stars that they are composed of have very similar orbits, and as such they can be used to put stringent constraints on the acceleration field (i.e. on the force field) of our Galaxy.
Since theories of gravity and dark matter are effectively recipes for the force field, we worked to build methods that would allow us to distinguish which model works best, and which ones can be ruled out. Needless to say, finding the solution to this very fundamental problem of physics will help improve our understanding of the universe, of mass, and of the force of gravity.
Our project was successful in finding a large sample of 87 stellar stream structures and in developing a state-of-the-art toolset to analyse them. We were able to put stringent constraints on the distribution of mass in the Milky Way. We showed that wide binary stars pose a strong challenge to Modified Newtonian Dynamics. However, during the course of the project we realized that that the dynamics of stellar streams could be more complicated than we initially envisaged due to their possible partial dissolution within the dark sub-halo within which they were formed, before they were incorporated into the Milky Way. This complexity, while making the modeling even more challenging, actually provides new opportunities to probe the hierarchical assembly of our Galaxy. We are currently developing data-driven approaches to account for these effects in our dynamical models, with final analyses forthcoming that will exploit this added layer of information about galaxy formation.
A major component of our work was the development of algorithms to detect stellar streams, which is extremely challenging due to their low contrast over the normal stellar populations in our Galaxy. We detected 87 stellar streams in the Gaia/ESA space mission sky survey, the majority of which were discovered by our team. The resulting kinematic and chemical information was analysed to understand the orbits and origin of the streams, both those from low-mass globular clusters and from much larger satellite galaxies.
These star streams tell us directly about the structures that fell into the Milky Way, thereby contributing to its buildup and formation. Indeed, using an objective measure of correlation between infalling satellites and star streams, our team was able to identify six families of accreted structures, testifying that there were at least six different main accretion events that occurred in the Milky Way's distant past. Another highlight of our work was the discovery of the most metal-poor structure yet known (i.e. the closest in chemical makeup to primordial matter), which as it so happens is now a stellar stream.
We devoted a large effort to building the computational machinery to model the Milky Way and its stellar streams in such a way as to remain as agnostic as possible to the underlying physics of dark matter and gravitation. To this end, we developed novel methods to map the acceleration field of the Galaxy by modeling a large sample of stellar streams in a conjoint manner. A major methodological advance was our development of new ways to analyze non-equilibrium dynamics in the Galaxy. Rather than seeing perturbations and time-varying structures as a nuisance, we showed how they can be used as powerful probes of the underlying gravitational potential.
The team also made significant progress in numerical modeling techniques. We developed and released new simulation codes capable of modeling galaxy formation in modified gravity theories, allowing for detailed comparisons between standard cold dark matter and alternative gravity models. Our analysis codes employ neural networks and machine learning to let the data reveal preferred relationships with minimal theoretical assumptions, and we created new tools for analyzing both simulations and real Milky Way data, including methods to compute action-angle coordinates in general gravitational potentials.
The project has been highly productive and influential, resulting in 106 refereed publications that have garnered 2673 citations to date.
These star streams tell us directly about the structures that fell into the Milky Way, thereby contributing to its buildup and formation. Indeed, using an objective measure of correlation between infalling satellites and star streams, our team was able to identify six families of accreted structures, testifying that there were at least six different main accretion events that occurred in the Milky Way's distant past. Another highlight of our work was the discovery of the most metal-poor structure yet known (i.e. the closest in chemical makeup to primordial matter), which as it so happens is now a stellar stream.
We devoted a large effort to building the computational machinery to model the Milky Way and its stellar streams in such a way as to remain as agnostic as possible to the underlying physics of dark matter and gravitation. To this end, we developed novel methods to map the acceleration field of the Galaxy by modeling a large sample of stellar streams in a conjoint manner. A major methodological advance was our development of new ways to analyze non-equilibrium dynamics in the Galaxy. Rather than seeing perturbations and time-varying structures as a nuisance, we showed how they can be used as powerful probes of the underlying gravitational potential.
The team also made significant progress in numerical modeling techniques. We developed and released new simulation codes capable of modeling galaxy formation in modified gravity theories, allowing for detailed comparisons between standard cold dark matter and alternative gravity models. Our analysis codes employ neural networks and machine learning to let the data reveal preferred relationships with minimal theoretical assumptions, and we created new tools for analyzing both simulations and real Milky Way data, including methods to compute action-angle coordinates in general gravitational potentials.
The project has been highly productive and influential, resulting in 106 refereed publications that have garnered 2673 citations to date.
Our project has advanced several research frontiers significantly beyond the state of the art. The development of the STREAMFINDER algorithm led to the detection of 87 stellar streams in the Milky Way, with the majority being new discoveries by our team. This represents the most comprehensive catalog of such structures to date and provides the community an unprecedented probe of our Galaxy's gravitational field.
A major breakthrough was the discovery of the C-19 stream, which is the most metal-poor structure ever found. This fossil from the early universe provides unique constraints on galaxy formation in the primitive Milky Way. We also made fundamental advances in understanding how the lowest-mass dark matter subhalos evolve under tidal forces, showing that their cuspy density profiles can protect dwarf galaxies from complete disruption, potentially leading to a population of extremely faint "microgalaxies" that have yet to be detected.
In terms of methodology, we developed novel machine learning approaches that advanced the field beyond the state of the art. Our "ACTIONFINDER" tool provides an innovative way to derive orbital parameters without prior knowledge of the gravitational potential. We also created "PhySO", a symbolic regression framework that enforces physical unit consistency and can treat multiple datasets simultaneously. PhySO represents a significant step forward for interpretable machine learning in physics.
An unanticipated breakthrough came from our investigation of primordial non-Gaussianities, which appears to offer a solution to the long-standing "S8 tension" in cosmology, suggesting that the existence of this tension is evidence for novel physics in the early universe.
A major breakthrough was the discovery of the C-19 stream, which is the most metal-poor structure ever found. This fossil from the early universe provides unique constraints on galaxy formation in the primitive Milky Way. We also made fundamental advances in understanding how the lowest-mass dark matter subhalos evolve under tidal forces, showing that their cuspy density profiles can protect dwarf galaxies from complete disruption, potentially leading to a population of extremely faint "microgalaxies" that have yet to be detected.
In terms of methodology, we developed novel machine learning approaches that advanced the field beyond the state of the art. Our "ACTIONFINDER" tool provides an innovative way to derive orbital parameters without prior knowledge of the gravitational potential. We also created "PhySO", a symbolic regression framework that enforces physical unit consistency and can treat multiple datasets simultaneously. PhySO represents a significant step forward for interpretable machine learning in physics.
An unanticipated breakthrough came from our investigation of primordial non-Gaussianities, which appears to offer a solution to the long-standing "S8 tension" in cosmology, suggesting that the existence of this tension is evidence for novel physics in the early universe.