Publications

Comparison of satellite galaxy distributions in the MillenniumTNG hydrodynamic simulation versus its dark-matter-only counterpart. Quantifies how baryonic effects shift the spatial and kinematic properties of satellites, with implications for cosmological modelling of galaxy clustering.

Develops new probabilistic estimators for measuring how dark-matter halo shapes align with the surrounding large-scale tidal field. These estimators reveal universal relationships between shape-bias parameters and halo properties, offering a cheap and precise way to constrain the physical scenarios driving the emergence of galaxy shape correlations.

Introduces probabilistic estimators for galaxy bias parameters derived from Lagrangian environment distributions, and proposes the “cumulant bias” formulation as an alternative description of how the galaxy density field responds to large-scale perturbations. The estimators are shown to be efficient tools for investigating bias parameters in hydrodynamical simulations.

Presents HYMALAIA, a hybrid Lagrangian model for predicting the intrinsic alignments of galaxies. The model combines a perturbative expansion of Lagrangian shapes with non-linear displacement fields from N-body simulations, accurately capturing the shape–tidal field correlations that must be modelled in weak-lensing surveys to avoid biased cosmological inference.

Presents an emulator based on a second-order Lagrangian bias expansion displaced to Eulerian space using N-body simulations. The model predicts galaxy power spectra in redshift space with sub-percent accuracy and is applied to constrain cosmological parameters from galaxy clustering data.

Applies fixing-and-pairing variance-reduction techniques to 21-cm signal simulations of the Epoch of Reionization. The method achieves 5–10% precision while halving the required simulation volume and reducing computational cost by a factor of four, enabling more efficient exploration of reionization parameter space.

Studies the relationships among Lagrangian bias expansion parameters across ~8000 halo and galaxy samples with different clustering properties, using the BACCO simulation suite. The resulting priors on bias parameters can be used in Bayesian cosmological analyses to limit parameter degeneracies and improve constraints.

Cosmological simulations play an increasingly important role in analysing the observed large-scale structure of the Universe. Recently, they have been particularly important in building hybrid models that combine a perturbative bias expansion with displacement fields extracted from N-body simulations to describe the clustering of biased tracers. Here, we show that simulations that employ a technique referred to as “Fixing-and-pairing” (F&P) can dramatically improve the statistical precision of such hybrid models. Specifically, by numerical and analytical means, we show that F&P simulations provide unbiased estimates for all statistics employed by hybrid models while reducing, by up to two orders of magnitude, their uncertainty on large scales. This roughly implies that an EUCLID-like survey could be analysed using simulations of 2 Gpc a side — a 20% of the survey volume. Our work establishes the robustness of F&P for current hybrid theoretical models for galaxy clustering, an important step towards achieving an optimal exploitation of large-scale structure measurements.