Research

Quasar clustering measurements are a powerful probe of the physical processes governing the growth of supermassive black holes and their coevolution with host galaxies. Wide-field surveys reveal a dramatic increase in the clustering strength of quasars at high redshift, with measurements at \(z\approx4\) that have been challenging to reproduce theoretically — even assuming quasars are hosted by the most massive dark matter halos in the early Universe. In the first project of my Ph.D., we revisited this decade-long problem using new, large-volume N-body cosmological simulations. We developed a model that reproduces the observed quasar clustering and demographics through a novel method for computing the halo mass function and halo cross-correlation functions across multiple simulations. Using a conditional luminosity function framework, the model captures the stochastic relationship between quasar luminosity and halo mass and predicts key observables — the quasar auto-correlation and luminosity functions, host mass function, and duty cycle. I then extended this framework to the highest redshifts probed by JWST. Using NIRCam/WFSS observations from surveys such as EIGER and ASPIRE — which detected numerous [OIII]-emitting galaxies around bright high-z quasars and measured the quasar–galaxy cross-correlation function for the first time at \(z>6\) — we proposed a framework to jointly infer the properties of quasars and [OIII]-emitting galaxies from their clustering and luminosity functions. Applied to EIGER data, it constrains the luminosity–halo mass relation, the host halo mass function, and the duty cycle/occupation fraction of quasars and galaxies at \(z\approx6\).

JWST's NIRCam Wide Field Slitless Spectroscopy (WFSS) mode has, for the first time, enabled spectroscopic clustering measurements of quasars and galaxies deep into the Epoch of Reionization. Surveys such as EIGER and ASPIRE target bright high-z quasar fields and detect the [OIII]-emitting galaxies around them, revealing a wide range of large-scale environments — from sparse fields to spectacular overdensities. I have contributed to interpreting the results of several of these measurement campaigns, which deliver the quasar–galaxy cross-correlation function, the galaxy auto-correlation function, and constraints on host halo masses and duty cycles at \(z\approx6\). Interpreting these data with population models (see above) is key to understanding where early quasars live and how they grow. Robust inference also requires careful treatment of systematics such as cosmic variance and the contribution of satellites in the small volumes probed by JWST.

Cosmological hydrodynamical simulations follow the non-linear evolution of cosmic structure by modeling the many processes important to the physics of galaxies, stars, and black holes, and have become capable of reproducing a wide range of observed galaxy properties. Yet, while much effort has gone into the connection between SMBHs and their host galaxies through gas fueling and AGN feedback, only a few studies have focused on the global properties of SMBHs and their evolution over cosmic time. SMBHs are visible at cosmological distances when they shine as bright quasars — rare objects whose statistical properties can only be captured theoretically with large simulated volumes. In my research, I use state-of-the-art, large-volume cosmological hydrodynamical simulations to reproduce observational constraints on the SMBH population and quasar activity across cosmic time. In particular, in a project I supervised, we used the FLAMINGO simulations to study the quasar luminosity function and the quasar auto-correlation function. We study how well these simulations match observational constraints across redshift, and investigate which black holes contribute to the bright quasar population. This lets us test the ability of large-volume hydrodynamical simulations to capture the evolution of massive black holes and quasar activity across cosmic time.

The dramatic leap in sensitivity brought by JWST has led to unexpected discoveries, such as an abundant population of broad-line high-z AGN appearing as “little red dots” (LRDs) in JWST imaging. When corrected for obscuration, these LRDs have surprisingly large bolometric luminosities, comparable to UV-selected quasars studied for decades. This is remarkable: UV-luminous quasars are selected from wide-field \(1400\,\mathrm{deg}^2\) surveys, whereas JWST AGN are found in fields of no more than \(\sim300\text{--}600\, \mathrm{arcmin}^2\) — implying these AGN are far more abundant than comparably luminous unobscured quasars, and that our picture of SMBH growth in the early Universe needs revision. By comparing JWST AGN/LRDs to UV-selected quasars, we found that LRDs outnumber quasars by a large, rapidly redshift-evolving factor. This suggests the LRD population cannot be accommodated in the same halos as unobscured quasars, and that LRDs may represent a different evolutionary phase of early SMBHs — a hypothesis to be tested by constraining LRD clustering, for which we developed a successful mock analysis based on a quasar population model.

Investigating the environments of galaxies during the Epoch of Reionization is a pressing goal of modern astrophysics. Telescopes such as ALMA and NOEMA have opened a new window on the primordial Universe, revealing obscured star formation and ISM line emission at rest-frame FIR wavelengths up to \(z\sim7\). One of ALMA's most compelling findings is that a significant fraction of \(z>4\) galaxies is surrounded by extended (10–15 kpc) [C II]-emitting haloes not predicted even by advanced zoom-in simulations. As part of my Master's thesis, I worked with Andrea Ferrara, Andrea Pallottini, and the cosmology group at SNS on a plausible formation mechanism, focusing on the hypothesis that these halos are remnants of past or ongoing outflow activity. Using a semi-analytical model for an outflow undergoing catastrophic cooling in the inner halo, we computed the abundance of singly ionized carbon and the resulting [C II] emission, and compared with data from the ALMA ALPINE program. We concluded that outflows are a promising explanation for the observed [C II] halos, pointing to star-formation-driven feedback already in place well into the Epoch of Reionization.

Constraining the strength of gas turbulence in protoplanetary discs is crucial for understanding gas accretion and planet formation. A promising probe is the vertical scale height of the dust, which is coupled to the gas — now measurable thanks to the very high-resolution observations of ALMA, which reveal dark gaps and emission rings in disc images. As shown by Pinte et al. 2016, these features can be exploited to recover the 3-D morphology of discs: due to projection effects, a gap is partly filled by emission from neighbouring regions — more so along the minor axis than the major axis — so the gap contrast between the two axes traces the disc's vertical structure. With Giovanni Rosotti and Benoît Tabone, we applied this method to high-resolution DSHARP observations, building a radiative transfer model to reproduce the observed gap contrast for different dust scale heights. The best-fitting scale heights are generally small, implying low levels of gas turbulence — an important step toward understanding the physics that governs planet formation.

With the first direct detection of gravitational waves in 2015, the era of gravitational-wave astronomy began, already delivering a wealth of information about compact binary mergers of neutron stars and black holes. The next generation of detectors — Cosmic Explorer and the Einstein Telescope — is expected to observe hundreds of thousands of binary coalescences per year. With these leaps in sensitivity come new challenges: signals will be so frequent that they begin to overlap in the time domain, where standard data-analysis pipelines are not guaranteed to work and may yield biased parameter estimates. With Bangalore Sathyaprakash, Surabhi Sachdev, and Anuradha Gupta, we quantified the biases that arise when current inference pipelines are used in the presence of multiple overlapping signals. We showed that, by setting a prior on the coalescence time, the properties of overlapping signals can be recovered correctly with existing infrastructure — provided the coalescence times are more than \(\sim1\text{--}2\) seconds apart. Signals within \(\sim0.5\) seconds of each other, however, suffer significant biases, motivating new strategies and algorithms.