PRIN 2022: PROGETTI DI RICERCA DI RILEVANTE INTERESSE NAZIONALE – Prot. 2022NY2ZRS 001 - CUP: J53D23001650006
Research Unit 1: M. Moresco (PI, Unibo), R. Metcalf (Unibo), M. Guidi (Unibo)
Research Unit 2: B. Granett (INAF-OAB), I. Risso (INAF-OAB)
Observational cosmology has entered a new era driven by a new generation of spectroscopic galaxy redshift surveys aimed at understanding the accelerated expansion of the Universe. The large-scale structure (LSS) traced by galaxies encodes the imprint of primordial density fluctuations through their clustering, providing key information on the matter content of the Universe and its dark components.
Over the past decade, galaxy clustering has become one of the most powerful cosmological probes. Ongoing surveys, such as the Dark Energy Spectroscopic Instrument (DESI) and the ESA Euclid mission launched in 2023, have delivered unprecedented three-dimensional maps of the Universe. Despite the dramatic improvement in measurement precision, the nature of the cosmological constant Λ and dark energy has remained elusive, highlighting the need to extract the maximum amount of cosmological information from clustering measurements.
In this project, we developed a comprehensive framework for the full-shape analysis of two-point and three-point correlation functions (2PCF and 3PCF). We implemented an end-to-end clustering analysis pipeline that enables, for the first time, the joint exploitation of 2PCF and 3PCF to directly constrain cosmological parameters from spectroscopic galaxy redshift surveys, while rigorously accounting for observational systematic effects.
The framework includes realistic forward modeling of survey-specific systematics that bias clustering measurements, together with the development and application of mitigation strategies. We implemented fast and accurate emulators to significantly accelerate the evaluation of 3PCF models and performed cosmological parameter inference through a full Bayesian likelihood analysis. By jointly exploiting lower- and higher-order clustering statistics, our analysis accessed information on the expansion history of the Universe and the growth of cosmic structures that is not fully captured by standard two-point analyses.
We tested and validated the framework on state-of-the-art simulated datasets and applied it to existing spectroscopic survey data for which a combined 2PCF–3PCF cosmological analysis had not previously been fully explored. By unifying lower- and higher-order statistics within a single end-to-end pipeline, this work established a robust and scalable methodology for extracting maximal cosmological information from large-scale structure data and enabling their full scientific exploitation.
A schematic view of our project is presented in this figure.
WP1 and WP2 are organised to allow us to fully exploit survey data with detailed knowledge of the galaxy samples. In particular, we assessed survey systematics, developed optimal summary statistics, and built emulators to efficiently make cosmological inferences with two-point and higher-order statistics.
The main results of the project can be summarized as follows:
Forward modelling of observational systematics. A forward-modelling code was developed to imprint survey geometry and observational systematic effects into mock galaxy catalogues, providing fast and robust forecasts for spectroscopic surveys. The framework was applied to the Euclid spectroscopic survey and contributed to the preparation of its clustering analyses.
Spectroscopic redshift pipeline and performance characterization. The project contributed to the development of the Euclid spectroscopic reduction pipeline and to the characterization of redshift measurement accuracy, precision, and failure modes. Validation tests and pixel-level simulations were performed to quantify the impact of redshift uncertainties on cosmological analyses.
Impact and mitigation of redshift interlopers. The impact of catastrophic redshift errors and interlopers on two-point and three-point correlation function measurements was assessed for Euclid-like surveys. Mitigation strategies were developed and validated, showing that most of the cosmological information can be recovered.
Modelling and emulation of higher-order clustering. Algorithms for measuring the galaxy three-point correlation function were designed, implemented, and validated within the Euclid Science Ground Segment. Both exact and accelerated estimators were developed to enable higher-order clustering measurements on survey-scale datasets. Theoretical modelling of the anisotropic halo three-point correlation function was also advanced, and new emulators were developed to allow fast cosmological inference from three-point statistics. Methods for denoising clustering covariance matrices were also introduced.
Joint two-point and three-point cosmological analyses. Full-shape joint analyses of the two-point and three-point correlation functions were performed on Euclid simulations and BOSS data, delivering the first cosmological constraints based on combined 2PCF+3PCF measurements.
Higher-order statistics for beyond-standard cosmology. The imprint of massive neutrinos on the three-point correlation function was investigated for the first time. The results show that configuration-dependent 3PCF measurements can break the degeneracy between neutrino mass and σ₈, enabling competitive constraints on neutrino masses.
Complementary cosmological probes beyond the standard approaches. Independent and synergistic approaches were developed beyond galaxy clustering, including cosmic chronometers based on age-dating of stellar populations and clusters, and gravitational-wave standard sirens. These methods provide independent measurements of the expansion history of the Universe and strengthen the connection between large-scale structure analyses and future multi-messenger cosmology.
The research of Work Package 1 (WP1), carried out mostly by RU2, dealt with the characterisation of galaxy redshift survey data, including mitigation of systematic effects and survey forward modelling with mock galaxy catalogues, applying the techniques to the most advanced surveys, in particular Euclid.
Complex survey selection effects, such as those present in Euclid's slitless spectroscopy, are challenging to model. In Euclid Collaboration: Granett, de la Torre, Moresco et al. (in prep), we develop simulators to apply the systematic effects to mock galaxy catalogues, such that the impact can be propagated to the cosmological inference.
In Euclid Collaboration: Monaco, Elkhashab, Granett et al. (2025) we study the impact of angular effects on the cosmological constraints.
In the figure, we show a simulated survey area for the Euclid mission (obtained by the code developed in Euclid Collaboration: Granett, de la Torre, Moresco et al. in prep). The color indicates the density of target galaxies, which is modulated by instrumental noise, and astrophysical foregrounds including zodiacal emission.
Catastrophic errors in spectroscopic redshift measurements can introduce interlopers that severely impact the analysis and bias cosmological parameter inference. It is crucial to accurately quantify and control these systematics to ensure robust cosmological constraints from current and future spectroscopic surveys. In Euclid Collaboration: Le Brun, Bethermin, Moresco et al. (2025), we presented the code developed for the Euclid mission to perform spectroscopic measurements, which has been tested to deliver highly accurate redshift measurements when applying an optimized quality selection, with negligible bias and percent-level precision. We demonstrated that this selection can allow us to achieve a high redshift success rate in the key cosmological redshift range, enabling the construction of large and robust samples for cosmological analyses.
In the figure from Le Brun, Bethermin, Moresco et al. (2025) is shown the comparison between the redshifts measured on the same sources from Euclid data (y axis) and from a ground-based survey (DESI, x axis). Along the diagonal are found the correctly measured redshifts, while other diagonal lines represent redshift interlopers, characterized in this work.
We developed and validated a modeling approach that explicitly incorporates the contribution of redshift interlopers into the galaxy clustering likelihood. In Euclid Collaboration: Risso et al. (2025), using realistic Euclid simulations, we quantified how the effects of different interloper populations propagate through the two-point correlation function into cosmological parameter estimation, with particular focus on redshift-space distortions and baryon acoustic oscillations. A central outcome of this work is the demonstration that the impact of redshift interlopers can be successfully mitigated, for the survey area corresponding to the first Euclid Data Release (DR1), using a simple model that accounts only for the overall purity of the spectroscopic sample. However, we showed that the same modeling assumptions may not be sufficient for the full survey at the end of the mission, when the observed volume will be six times larger than that of DR1.
In Principi, Veropalumbo, Moresco et al. (2026, in prep), we extended this analysis to the 3PCF, assessing for the first time the impact of interlopers in the measurement of cosmological parameters with the 3PCF. We derived and tested new estimators to quantify the cross-correlation terms, and implemented mitigation strategies that allow us to retrieve unbiased parameters also in the presence of interlopers within an error smaller than 0.3 sigma.
In the figure from Risso et al. (2025), the cosmological parameter derived from an unbiased redshift sample is shown in grey, and the impact of redshift interlopers, when not correctly modeled, is shown in red. The green data show how it is possible to properly recover the correct parameters when the mitigation strategy is adopted.
The research of Work Package 2 (WP2), carried out mostly by RU1, focused on the cosmological analysis, developing the estimators needed to measure the main clustering statistics (including the combination of 2PCF and 3PCF), applying them to the real and simulated data provided by WP1, and extracting cosmological information, assessing the impact of systematics on the results, and developing mitigation strategies.
To fully exploit the statistical power of next-generation galaxy surveys, it is fundamental to develop fast and robust algorithms capable of measuring higher-order correlation functions on extremely large datasets. In Euclid Collaboration: Veropalumbo, Moresco et al. (2026), we presented the code developed within the Euclid Science Ground Segment to measure the galaxy 3PCF, implementing and testing different estimators that achieve the required accuracy while remaining computationally feasible for the full survey. Extensive validation tests demonstrate the robustness and precision of the implementation, while performance analyses confirm the scalability of the code to Euclid data volume.
In Euclid Collaboration: Guidi, Veropalumbo, Moresco et al. (2025), we have further developed a novel emulator that, for the first time, enables a full-shape cosmological analysis by compressing the information from the combined two-point and three-point correlation functions (2PCF + 3PCF) directly into targeted cosmological parameters. This approach exploits the complementary information encoded in higher-order statistics, allowing a significant gain in constraining power beyond standard two-point analyses. In the reference paper, we validated the method using Flagship simulations that realistically reproduce Euclid galaxy clustering in real space, demonstrating that the inclusion of the 3PCF leads to a substantial tightening of cosmological constraints. The methodology has subsequently been extended to redshift space with a dedicated modelling code (Melcorr), and applied to observational data to extract cosmological constraints from the joint 2PCF + 3PCF analysis of BOSS DR12 galaxies (Guidi, Moresco et al., in prep)
This figure (from Euclid Collaboration: Guidi, Veropalumbo, Moresco et al. 2025) shows marginalized 1D and 2D posterior distributions for various cosmological parameters. The grey contours correspond to a standard 2PCF analysis, while coloured contours show results from the joint 2PCF+3PCF analysis for different minimum scale cuts in the two- and three-point functions. In all cases, the inclusion of the 3PCF produces a marked tightening of the posterior distributions and a partial breaking of parameter degeneracies. Moreover, reducing the minimum scales included in the 3PCF further enhances the constraining power, highlighting the additional cosmological information carried by higher-order statistics beyond the two-point function alone.
In parallel, we developed Mod3l, a novel code to model the galaxy three-point correlation function in both its isotropic and anisotropic components (Farina, Veropalumbo, Branchini, Guidi, 2025). We first investigated the impact of anisotropic information through a template-fitting analysis, modelling redshift-space distortions as deviations around a fixed theoretical template. This study showed that the anisotropic contribution of the 3PCF provides a significant and complementary source of information when combined with the isotropic component alone. Building on these results, we subsequently developed a dedicated emulator trained on Mod3l modelling predictions (Nagainis, Guidi, Moresco et al., in prep) enabling for the first time a full-shape cosmological analysis that consistently includes both the isotropic and anisotropic 3PCF contributions. This framework allows a robust and efficient extraction of cosmological information from higher-order anisotropic clustering statistics.
In the figure (from Nagainis, Guidi, Moresco et al, in prep): Fisher forecast on the cosmological parameters As, Omega_m, and h obtained from a full-shape analysis of the matter three-point correlation function. The figure compares constraints derived from the isotropic 3PCF alone with those obtained from the combined isotropic + anisotropic 3PCF. The inclusion of anisotropic information leads to a tightening of the confidence regions, highlighting the additional cosmological sensitivity carried by redshift-space distortions in the three-point statistics.
After delivering the first cosmological measurements from a joint 2PCF+3PCF analysis on both Euclid simulated data and BOSS observations, in Labate, Guidi, Moresco et al. (2025) we further explored the potential of three-point statistics to constrain cosmological parameters beyond the standard model. In particular, we investigated how the 3PCF can be used to probe the imprint of massive neutrinos on large-scale structure. Using N-body simulations with varying neutrino masses, we demonstrated that neutrino free-streaming produces distinctive configuration-dependent signatures in the 3PCF, especially for elongated and squeezed triangle shapes, and that these signatures differ markedly from those induced by changes in σ8. This enables the 3PCF to be a powerful tool to break the Mν–σ8 degeneracy and opens the way to competitive neutrino-mass constraints from ongoing spectroscopic surveys such as Euclid and DESI.
In the figure (from Labate, Guidi, Moresco et al. 2025) are shown, as an illustrative example, 3PCF in different configurations (the various plots) presenting measurements obtained from simulations with different neutrino masses, compared to a fiducial ΛCDM (different colors). The lower plots show the detectabilities of different neutrino masses in the 3PCF at different scales, where the yellow shaded area shows the position of the BAO peak.
Maximizing the return of upcoming galaxy surveys requires going beyond standard cosmological approaches (e.g., see Moresco et al. 2022) and enriching the statistical description of large-scale structure. To this end, within the scope of this project, we developed and tested alternative methodologies that extend the information content of galaxy clustering beyond conventional two-point analyses. A strong synergy between these novel methodological approaches and LSS analyses has recently been investigated in several works in the literature (e.g., see Pierra et al. 2025), suggesting how their joint exploitation could be of fundamental importance for precision cosmology.
Example of the spherical harmonic multipoles of the velocity–density–density bispectrum in configuration space. The multipoles are shown as combinations of theoretical templates derived from perturbation theory. This representation illustrates how the orientation-dependent information of the velocity vector contributes additional constraints beyond standard two-point velocity–density analyses, highlighting the potential of this new observable to break degeneracies and improve cosmological parameter inference. From Guidi, Farina, Nusser, Branchini (in prep).
Alternative indicators
We have developed a novel modelling framework for a new observable: the peculiar velocity–density–density cross-correlation function. Both in Fourier space, and in configuration space the signal is decomposed into spherical harmonic multipoles as a function of the separation and the orientation of the velocity vector. The theoretical prediction is obtained using perturbation theory and the modelling code Modv3l, providing a fully consistent model for the three-point velocity–density correlations. This approach opens a new observational window, allowing us to break degeneracies that are present in standard two-point velocity–density analyses, and offers a complementary probe of cosmological parameters.
The figure, adapted from Moresco (2024), illustrates the expansion rate of the Universe as a function of redshift derived using the cosmic chronometers method. White points show current measurements from real data, while blue and yellow points indicate forecasts for upcoming spectroscopic surveys. The shaded region represents the reconstructed H(z) expected to be achievable with future observations.
Ages as cosmological probes
A new and powerful frontier in observational cosmology is emerging through the use of time itself as a cosmological probe, in the form of cosmic clocks and cosmic chronometers (Moresco 2024, Moresco 2026). By measuring the ages of the oldest stellar populations, globular clusters, and galaxy clusters across cosmic time, these approaches provide a direct and independent determination of the expansion history of the Universe. Recent advances, from age-dating of lensed globular clusters with JWST (Tomasetti, Moresco et al. 2025a) to Galactic archaeology with Gaia (Tomasetti, Chiappini, Nepal, Moresco et al. 2025) and the development of cluster-based cosmic chronometers (Tomasetti, Moresco et al. 2025b), demonstrate that precise cosmic-age measurements are now achievable out to significant redshifts. Coupled with large-scale structure analyses, these time-based probes offer a complementary and highly synergistic route to tighten cosmological constraints and test the consistency of the standard cosmological model. Altogether, this opens a genuinely new observational window on cosmic expansion, enabling cross-checks of standard probes and offering a promising avenue to address current tensions in cosmology.
The figure (from Tagliazucchi, Moresco et al. 2026) shows the measurement of the Hubble constant obtained on the latest binary black holes (BBH) dataset from the LIGO-Virgo-KAGRA collaboration, comparing the constraint obtained with a standard modelling of the BBH mass function (in dashed grey) and the improvements that can be obtained with the developed data-driven semiparametric parametrization.
Gravitational waves and standard sirens
Gravitational waves from compact binary mergers provide an independent and self-calibrating probe of the cosmic expansion through the standard siren method. Fully exploiting this opportunity requires substantial methodological advances to cope with the computational demands of Bayesian inference, the incompleteness of galaxy catalogues used for statistical host identification, and astrophysical uncertainties in source population modelling. Recent developments, including GPU-accelerated inference pipelines (Tagliazucchi, Moresco et al. 2025), Bayesian frameworks accounting for galaxy-catalog incompleteness (Borghi, Moresco et al. 2025), and data-driven semiparametric modelling of gravitational-wave source populations (Tagliazucchi, Moresco et al. 2026), have significantly improved the robustness and precision of standard siren cosmology. These advances are particularly important in view of forthcoming deep spectroscopic surveys, as they enable a natural and powerful synergy between large-scale structure analyses and future gravitational-wave observations, ultimately leading to joint constraints on the expansion history of the Universe.