PRIN 2022 PNRR grant funded by the Italian Ministry of University and Research (MUR)

The Cosmic Mystery of the Metals

Any chemical element (e.g. carbon, oxygen, nitrogen, silicon, calcium, iron, ...) heavier than Hydrogen and Helium is called a Metal in astronomy. Metals are the building blocks of planetary systems including our world (the planet Earth) and ourselves (living organisms like the human being), thus vital ingredients for the origin of life in the Universe. We know that stars are the great furnaces of the cosmos, which generate the metals inside galaxies. But how did these metals escape their home galaxies and spread throughout the vast emptiness of space? This question points to one of the biggest unsolved puzzles in astrophysics.

To investigate the answer, first we have to learn what is the IGM. Most of the Universe's normal matter (the baryons) do not lie in shining stars and galaxies. Instead, the majority of baryons exist as very-large-scale filaments: the cosmic network of diffuse gas flowing between galaxies. This matter in between the galaxies is called the Intergalactic Medium (IGM); a subset of which is the region in the immediate vicinity (2 or 3 times the virial radius) of each galaxy called the Circumgalactic Medium (CGM). The IGM and the CGM are the hiding places for the majority of cosmic baryons. To completely understand how galaxies live and grow, we must explore this hidden large-scale reservoir of baryons and their metal content.

Energetic processes inside galaxies like exploding stars, supernovae and supermassive black holes drive powerful galactic outflows, or gas outflowing at large velocities. These forceful winds is one of the mechanisms to expel metals out from galaxies into the CGM and the IGM. However, the physics of these feedback processes are still poorly understood.

Our PROJECT : Probing the Cosmic Metal Content

Shedding light on some of the above unknowns is the mission of our project: "Next-generation Computing and Data Technologies to probe the Cosmic Metal content". Over the next 2 years, we will combine the power of massively-parallel supercomputers and artificial intelligence to investigate the unknown distribution of metals around galaxies in the early Universe, looking back in cosmic time 10 to 12 billion years ago.

Our STRATEGY : A Two-Pronged Attack

To shed light on the cosmic mystery of metals, our project integrates two parallel objectives:

🎯 GOAL 1: Building Digital Universes Inside Supercomputers

We cannot travel back in time, so we build our own digital Universes inside supercomputers. We execute Cosmological Hydrodynamical Simulations: huge numerical simulations that begin with the faint resonances of the Big Bang (the primordial density fluctuations at the epoch of the CMBR); thereby use the laws of physics to calculate how dark matter, baryons and energy evolve over 13.8 billion-years age of the Universe to create the large-scale structures we observe around Earth today.

One of the goals of our project is to perform and exploit a new generation of cosmological hydrodynamical simulations. We will:

• Develop new codes that run on state-of-the-art High-Performance Computing (HPC) platforms.

• Include new physics in our numerical models: the evolution of cosmic dust, and a more realistic model for the feedback from supermassive black holes.

• Generate large statistical samples of simulated data that predicts how galaxies and the IGM gas influence each other.

🎯 GOAL 2: Taming the Data Deluge with AI

Modern Astronomy is a Big Data science. Current telescopes like the ESO VLT and Keck, together with the future giants like the 39-meter Extremely Large Telescope (ELT), are producing an avalanche of information; the quantity of which is only going to escalate in the upcoming years. This data volume and amount is far too large to be analyzed manually.

Our novel tactics for seeing the invisible IGM is Absorption Spectroscopy, using beacons of light from extremely-distant bright astronomical objects called Quasars. As the quasar light travels to us being emitted from large distances, it passes through the IGM/CGM gas, and the metals in the IGM/CGM gas absorb specific colors of light, leaving a fingerprint or absorption line in the spectrum. By detecting and measuring these fingerprints, we can infer the chemical composition of the IGM/CGM gas. The challenge for the astronomical distances is that a single quasar spectrum can contain hundreds of these absorption features.

The other goal of our project is to develop a new, modern toolkit for the fast and reliable analysis of spectroscopic big data. We will:

• Use advanced Machine Learning (ML) techniques to process spectral data.

• Build a new toolkit that can automatically identify and classify the absorption fingerprints of different metal ions.

• Develop a novel GPU-accelerated fitting code to rapidly extract physical measurements (like density and temperature) from the classified features.

• Enhance the existing AstroCook package, implementing sophisticated new techniques for precision spectroscopy.

The PAYOFF: From Code to Cosmology

The final step of our project is to bring the two goals together. We will apply our new ML toolkit (Goal 2) to analyze both our new simulations (Goal 1) and vast amounts of real observational data from modern telescopes and large surveys.

This will allow us to perform a direct apples-to-apples comparison between astronomical observations of the real Universe and numerical simulation predicted digital Universes. It will permit, for the first time, to rigorously test our theories of galactic feedback and answer our driving questions: How, exactly, do galaxies pollute their environments - the CGM and the IGM? How far away from their home galaxies do the metals travel? And what role do supermassive black holes play in this cosmic cycle of metals?

IMPACT Beyond Astronomy

Our project will contribute to pushing the frontiers of our knowledge about how metals distribute from the galaxies to the CGM and IGM. At the same time, the technology we develop during the project has applications right here up close on Earth. The ML-powered toolkit (our Goal 2) for analyzing absorption spectra will be robust and flexible. This same technique can be applied in many Earth-based industries, including:

• Medical and pharmaceutical applications.

• Agricultural analysis (e.g., testing soil composition).

• Forensic applications.

The new code libraries and AI tools forged to explore the far-off Universe will be ready to be deployed for current on-Earth industry-based applications. This will be a step forward toward creating a data-driven economy right here on Earth.