In the past few years I worked on different aspects of cosmology and structure formation, all linked by my interest in computational astrophysics and in how numerical simulations and theoretical predictions can be compared to observational results. In particular, I see gravitational lensing as a natural observational counterpart of my computational work, since it is a powerful tool to trace the mass distribution of galaxies, offering us a way to “see” the dark matter distribution. I run and analyse numerical simulations with different dark matter models - including CDM, WDM and SIDM - and state-of-the-art baryonic physics. My work includes a detailed parametrisation of halo properties in all these different scenarios and the comparison with gravitational lensing observations, through realistic mock images.

    Numerical simulations play a central role in testing dark matter models: cosmological and galaxy formation simulations run with different physical models allow us to make prediction on the abundance and spatial distribution of dark matter haloes and the galaxies they host, which can then be compared with observational data. For this purpose, we need both large scale and high-resolution re-simulations: in the formers, the galaxy formation model can be tested against observables such as the galaxy stellar mass function and we can calculate the number density of dark matter haloes that could act as a lens or as a line-of-sight perturber; since the possible resolution of these simulations is limited by the computational resources, higher resolution re-simulations of individual haloes can tell us more on the satellite distribution and can be used to generate realistic mock images.

    In particular, a very promising way of testing dark matter models is substructure lensing, in which low mass dark perturber are detected as surface brightness perturbations in gravitational arcs. A fundamental prediction of the cold dark matter paradigm is the existence of a large number of small mass dark (sub)haloes, while their abundance is lower in alternative models, such as warm or self- interacting dark matter. Confirming or not the existence of such objects is one of the most important empirical tests for dark matter. While current lensing data are available in limited number and allow to detect objects of mass ~ 10^9 M☉, future observational campaign will lead to an increase in the sample size (DES, LSST, Euclid) and resolution (~ 10^7 M☉ with JWST and hopefully ~ 10^6 M☉ with ALMA or global VLBI).

    I am part of the SHARP collaboration.

    Below you can find a summary of my most recent works

Updated predictions for CDM vs SIDM in elliptical galaxies

We study the effect of self-interacting dark matter (SIDM) and baryons on the shapes of early-type galaxies (ETGs) and their dark matter haloes, comparing them to the predictions of the standard cold dark matter (CDM) scenario. We use a sample of five zoom-in simulations of haloes hosting ETGs ($M_{\text vir}\sim 10^{13}M_{\odot}$ and $M_{*}\sim10^{11}M_{\odot}$), simulated in CDM and a SIDM model with constant cross-section of $\sigma_T/m_\chi = 1\ \mathrm{cm}^2 \mathrm{g}^{-1}$, with and without the inclusion of baryonic physics. We measure the three-dimensional and projected shapes of the dark matter haloes and their baryonic content by means of the inertia tensor and compare our measurements to the results of gravitational lensing and X-ray observations. We find that the inclusion of baryons greatly reduces the differences between CDM and a SIDM and thus the ability to draw constraints on the basis of shapes. We find that lensing measurements clearly reject the predictions from CDM dark-matter-only simulations, whereas they show a different degree of preference for the CDM and SIDM hydro scenarios, and cannot discard the SIDM dark-matter-only case. The shapes of the X-ray emitting gas are also comparable to observational results in both hydro runs, with CDM predicting higher elongations only in the very center. Contrary to previous claims at the scale of elliptical galaxies, we conclude that both CDM and our SIDM model with $\sigma_T/m_\chi = 1\ \mathrm{cm}^2 \mathrm{g}^{-1}$ are able to explain observed distributions of halo shapes, once baryons are included in the simulations.

Detailed sensitivity maps to detect low-mass haloes

This paper aims to quantify how the lowest halo mass that can be detected with galaxy-galaxy strong gravitational lensing depends on the quality of the observations and the characteristics of the observed lens systems. Using simulated data, we measure the lowest detectable NFW mass at each location of the lens plane, in the form of detailed sensitivity maps. In summary, we find that: (i) the lowest detectable mass Mlow decreases linearly as the signal-to-noise ratio (SNR) increases and the sensitive area is larger when we decrease the noise; (ii) a moderate increase in angular resolution (0.07" vs 0.09") and pixel scale (0.01" vs 0.04") improves the sensitivity by on average 0.25 dex in halo mass, with more significant improvement around the most sensitive regions; (iii) the sensitivity to low-mass objects is largest for bright and complex lensed galaxies located inside the caustic curves and lensed into larger Einstein rings (i.e rE ≥ 1.0″). We find that for the sensitive mock images considered in this work, the minimum mass that we can detect at the redshift of the lens lies between 1.5 × 108 and 3 × 109M⊙. We derive analytic relations between Mlow, the SNR and resolution and discuss the impact of the lensing configuration and source structure. Our results start to fill the gap between approximate predictions and real data and demonstrate the challenging nature of calculating precise forecasts for gravitational imaging. In light of our findings, we discuss possible strategies for designing strong lensing surveys and the prospects for HST, Keck, ALMA, Euclid and other future observations.

Modelling the line-of-sight contribution in substructure lensing

We investigate how Einstein rings and magnified arcs are affected by small-mass dark-matter haloes placed along the line of sight to gravitational lens systems. By comparing the gravitational signature of line-of-sight haloes with that of substructures within the lensing galaxy, we derive a mass-redshift relation that allows us to rescale the detection threshold (i.e. lowest detectable mass) for substructures to a detection threshold for line-of-sight haloes at any redshift. We then quantify the line-of-sight contribution to the total number density of low-mass objects that can be detected through strong gravitational lensing. Finally, we assess the degeneracy between substructures and line-of-sight haloes of different mass and redshift to provide a statistical interpretation of current and future detections, with the aim of distinguishing between cold dark matter and warm dark matter. We find that line-of-sight haloes statistically dominate with respect to substructures, by an amount that strongly depends on the source and lens redshifts, and on the chosen dark-matter model. Substructures represent about 30 percent of the total number of perturbers for low lens and source redshifts (as for the SLACS lenses), but less than 10 per cent for high-redshift systems. We also find that for data with high enough signal-to-noise ratio and angular resolution, the non-linear effects arising from a double-lens-plane configuration are such that one is able to observationally recover the line-of-sight halo redshift with an absolute error precision of 0.15 at the 68 per cent confidence level.

The interplay of self-interacting dark matter and baryons

We use high-resolution hydrodynamical simulation to test the difference of halo properties in cold dark matter (CDM) and a self-interacting dark matter (SIDM) scenario with a constant cross-section of σT/mχ = 1 cm2 g-1. We find that the interplay between dark matter self-interaction and baryonic physics induces a complex evolution of the halo properties, which depends on the halo mass and morphological type, as well as on the halo mass accretion history. While high-mass haloes, selected as analogues of early-type galaxies, show cored profiles in the SIDM run, systems of intermediate mass and with a significant disc component can develop a profile that is similar or cuspier than in CDM. The final properties of SIDM haloes - measured at z = 0.2 - correlate with the halo concentration and formation time, suggesting that the differences between different systems are due to the fact that we are observing the impact of self-interaction. We also search for signatures of SIDM in the lensing signal of the main haloes and find hints of potential differences in the distribution of Einstein radii, which suggests that future wide-field survey might be able to distinguish between CDM and SIDM models on this basis. Finally, we find that the subhalo abundances are not altered in the adopted SIDM model with respect to CDM.

The impact of baryonic physics on subhaloes

We investigate the impact of baryonic physics on the subhalo population by analysing the results of two recent hydrodynamical simulations (EAGLE and Illustris), which have very similar configuration, but a different model of baryonic physics. We concentrate on haloes with a mass between 10^12.5 and 10^14 M☉/h and redshift between 0.2 and 0.5, comparing with observational results and subhalo detections in early-type galaxy lenses. We compare the number and the spatial distribution of subhaloes in the fully hydro runs and in their dark-matter-only (DMO) counterparts, focusing on the differences between the two simulations. We find that the presence of baryons reduces the number of subhaloes, especially at the low-mass end (≤1010 M☉ h-1), by different amounts depending on the model. The variations in the subhalo mass function are strongly dependent on those in the halo mass function, which is shifted by the effect of stellar and AGN feedback. Finally, we search for analogues of the observed lenses (Sloan Lens ACS) in the simulations, selecting them in velocity dispersion and dynamical properties. We use the selected galaxies to quantify detection expectations based on the subhalo populations in the different simulations, calculating the detection probability and the predicted values for the projected dark matter fraction in subhaloes fDM and the slope of the mass function α. We compare these values with those derived from subhalo detections in observations and conclude that the DMO and hydro EAGLE runs are both compatible with observational results, while results from the hydro Illustris run do not lie within the errors.

Substructure lensing in sterile neutrino cosmologies

We use high-resolution hydrodynamical simulations run with the EAGLE model of galaxy formation to study the differences between the properties of - and subsequently the lensing signal from - subhaloes of massive elliptical galaxies at redshift 0.2, in Cold and Sterile Neutrino (SN) Dark matter models. We focus on the two 7 keV SN models that bracket the range of matter power spectra compatible with resonantly-produced SN as the source of the observed 3.5 keV line. We derive an accurate parametrisation for the subhalo mass function in these two SN models relative to CDM, as well as the subhalo spatial distribution, density profile, and projected number density and the dark matter fraction in subhaloes. We create mock lensing maps from the simulated haloes to study the differences in the lensing signal in the framework of subhalo detection. We find that subhalo convergence is well described by a log-normal distribution and that signal of subhaloes in the power spectrum is lower in SN models with respect to CDM, at a level of 10 to 80 per cent, depending on the scale. However, the scatter between different projections is large and might make the use of power-spectrum studies on the typical scales of current lensing images very difficult. Moreover, in the framework of individual detections through gravitational imaging a sample of ~30 lenses with an average sensitivity of M_sub=5 x 10^7 M☉ would be required to discriminate between CDM and the considered sterile neutrino models.

The universality of the virial halo mass function

The abundance of galaxy clusters can constrain both the geometry and structure growth in our Universe. However, this probe could be significantly complicated by recent claims of nonuniversality in the halo mass function – non-trivial dependences with respect to the cosmological model and redshift. In this work we analyse the dependance of the mass function on the way haloes are identified and establish if this can cause departures from universality. In order to explore this, we use a set of different dark-matter-only cosmological simulations (Le SBARBINE simulations) adopting the cosmological parameters from the latest analysis of the Planck satellite’s data. This suite of simulations is complemented by a lower resolution set carried out with different cosmological parameters. We identify dark matter haloes using a Spherical Overdensity algorithm with varying overdensity thresholds (virial, 2000rho_c, 1000rho_c, 500rho_c, 200rho_c and 200rho_b) at several redshifts. We notice that previous claims of nonuniversality largely orginate from a pseudo-evolution of the mass function. When expressed in terms of the rescaled variable ⌫, the mass function for virial haloes is universal at the 5-10 % level, over the whole range of redshifts and cosmologies explored here. For every other overdensity considered, the departures of universailty are significantly larger, reaching 60% in some cases. We provide fitting functions for the halo mass function parameters as a function of density which predict, with a few percent accuracy, the halo mass function for a wide range of halo definitions, redshifts and cosmological models. We then present how the departures from universality associated with other halo definitions can be derived by combining the universality of the virial definition with the expected shape of the density profile of dark matter halos.