Cosmological structure formation with N-body simulations : the path to percent accuracy with scale-free models

Abstract:

Formation des structures cosmologiques avec des simulations à N-corps : tester la précision avec des modèles sans échelle Les simulations à N-corps sont actuellement la seule technique disponible pour résoudre le clustering à échelles non-linéaires. Avec les mesures à venir des satellites de phase IV (telles que Euclid, DESI ou LSST), qui apporteront une précision sans précédent, comprendre la résolution et les limites des simulations est devenue une nécessité urgente. Dans cette thèse, nous exploitons une nouvelle technique pour évaluer la résolution des simulations à N-corps dans le régime non-linéaire de formation de structures. Pour cela, nous utilisons un ensemble particulier de cosmologies avec une évolution d'Einstein de-Sitter (Ωm = 1) et un spectre de perturbations en loi de puissance (Pk ∝ kn), connues sous le nom de cosmologies sans échelle. Une propriété importante est leur évolution auto-similaire (c'est-à-dire, à des coordonnées correctement redimensionnées, toute statistique de regroupement est constante dans le temps). Il s'agit d'un outil excellent pour déterminer la résolution à laquelle nous pouvons mesurer les dites statistiques. Nous exploitons le fait que tous les écarts par rapport à un comportement auto-similaire doivent être dus à des échelles non physiques introduites par le système à N corps, et donc l'extrapolation à la limite du continuum n'est plus une bonne approximation pour la simulation. Comme échantillon de test pour les principaux résultats de cette thèse, nous exploitons une suite de grandes simulations à N-corps (jusqu'à N = 4096^3) réalisées avec Abacus. Nous exécutons une variété d'indices spectraux n, pour faciliter l'extrapolation de nos résultats aux cosmologies de type ΛCDM. Nous exécutons également des ensembles de simulations différant par un seul paramètre de discrétisation, afin d'étudier comment la résolution pourrait en dépendre. Nous commençons par présenter l'analyse des statistiques de champ de matière. Dans un premier temps, nous étudions l'échelle résolue minimale pour le spectre de puissance, et sa dépendance sur la distance interparticulaire de la configuration initiale. Nous continuons en examinant la résolution de vitesses par paires, et leur connexion à celle de la corrélation de densité à 2 points. De plus, comme sous-produit de ces études, nous avons également pu revisiter l'hypothèse de clustering stable, en estimant sa compatibilité avec les données. Enfin, nous fournissons une analyse des statistiques de halo pour différents détecteurs de halo populaires (FoF, Rockstar et CompaSO). Nous étudions la convergence de la fonction de masse du halo, le halo-halo vitesse relative par paires et leur fonction de corrélation à deux points, déterminant les limites de résolution en fonction de l'échelle et du nombre de particules par halo.

Disk dominated galaxies retain their shapes below $z = 1.0$

Authors:

Kai Hoffmann, Clotilde Laigle, Nora Elisa Chisari, Pau Tallada, Yohan Dubois, Julien Devriendt

Abstract:

The high abundance of disk galaxies without a large central bulge challenges predictions of current hydrodynamic simulations of galaxy formation. We aim to shed light on the formation of these objects by studying the redshift and mass dependence of their 3D shape distribution in the COSMOS galaxy survey. This distribution is inferred from the observed distribution of 2D shapes, using a reconstruction method which we test using hydrodynamic simulations. We find a moderate bias for the inferred average disk circularity and relative thickness with respect to the disk radius, but a large bias on the dispersion of these quantities. Applying the 3D shape reconstruction method on COSMOS data, we find no significant dependence of the inferred 3D shape distribution on redshift. The relative disk thickness shows a significant mass dependence which can be accounted for by the scaling of disk radius with galaxy mass. We conclude that the shapes of disk dominated galaxies are overall not subject to disruptive merging or feedback events below redshift $z=1.0$. This favours a scenario where these disks form early and subsequently undergo a tranquil evolution in isolation. In addition, our study shows that the observed 2D shapes of disk dominated galaxies can be well fitted using an ellipsoidal model for the galaxy 3D morphology combined with a Gaussian model for the 3D axes ratio distribution, confirming findings from similar work reported in the literature. Such an approach allows to build realistic mock catalogs with intrinsic galaxy shapes that will be essential for the study of intrinsic galaxy alignment as a contaminant of weak lensing surveys.

Early-type galaxy spin evolution in the Horizon-AGN simulation

The Astrophysical Journal University of Chicago Press

Authors:

H Choi, SK Yi, Y Dubois, T Kimm, JEG Devriendt, C Pichon

Abstract:

Using the Horizon-AGN simulation data, we study the relative role of mergers and environmental effects in shaping the spin of early-type galaxies (ETGs) after $z \simeq 1$. We follow the spin evolution of 10,037 color-selected ETGs more massive than 10$^{10} \rm \, M_{\odot}$ that are divided into four groups: cluster centrals (3%), cluster satellites (33%), group centrals (5%), and field ETGs (59%). We find a strong mass dependence of the slow rotator fraction, $f_{\rm SR}$, and the mean spin of massive ETGs. Although we do not find a clear environmental dependence of $f_{\rm SR}$, a weak trend is seen in the mean value of spin parameter driven by the satellite ETGs as they gradually lose their spin as their environment becomes denser. Galaxy mergers appear to be the main cause of total spin changes in 94% of central ETGs of halos with $M_{vir} > 10^{12.5}\rm M_{\odot}$, but only 22% of satellite and field ETGs. We find that non-merger induced tidal perturbations better correlate with the galaxy spin-down in satellite ETGs than mergers. Given that the majority of ETGs are not central in dense environments, we conclude that non-merger tidal perturbation effects played a key role in the spin evolution of ETGs observed in the local ($z < 1$) universe.

Enhanced constraints on large-scale structure from secondary CMB anisotropies

Abstract:

The large-scale structure of the Universe encodes invaluable information about the fundamental cosmological parameters, the physics of structure formation, and the thermodynamic history of the Universe. In this thesis, we explore how secondary anisotropies of the Cosmic Microwave Background (CMB), particularly the thermal Sunyaev-Zeldovich (tSZ) effect and CMB lensing, can improve constraints on the large-scale structure. We develop a framework to cross-correlate tSZ maps from the Planck satellite with the distribution of galaxies at low redshift using tomographic bins with data from the 2MASS Photometric Redshift catalogue and WISE x SuperCOSMOS. These cross-correlations enable precise measurements of the bias-weighted gas pressure, ⟨𝑏Pe⟩, and the hydrostatic mass bias parameter, 1−bH, as a function of redshift.

This thesis is primarily based on two complementary studies employing galaxy clustering (𝛿𝑔 × 𝛿𝑔), galaxy-tSZ cross-correlations (𝛿𝑔 × y), and galaxy-CMB lensing cross-correlations (𝛿𝑔 × K) to constrain cosmological and thermodynamic parameters across six redshift bins (𝑧 ∈ [0.1, 0.6]).

In the first study, we use a combination of 𝛿𝑔 × 𝛿𝑔 and 𝛿𝑔 × y to improve constraints on the thermal history of the Universe. We achieve ~6 \% precision on 1-bH across the six redshift bins, finding consistency with previous results and no evidence for significant redshift dependence. Our best-fit value of 1−𝑏H = 0.75 ± 0.03 aligns well with joint analyses of Planck cluster counts and CMB anisotropies calibrated with CMB lensing. Additionally, our constraints on ⟨𝑏𝑃𝑒⟩, accurate to ~10% per bin, represent the most precise measurements to date, providing a robust test of baryonic feedback and models of energy injection.

The second study incorporatess 𝛿𝑔 × K  to enhance our tomographic analysis of structure growth and gas thermodynamics. Using CMB lensing as an additional tracer of large-scale structure, we constrain the amplitude of fluctuations of the matter power spectrum, 𝜎8, to 6% across all redshift bins, the hydrostatic mass bias, 1 - bH to ~ 18 %, the bias-weighted average electron pressure, ⟨𝑏𝑃𝑒⟩, to ~12%, the thermal energy density, Ωth  to ~ 10%, as well as TAGN, a single parameter quantifying the intensive thermodynamic properties of haloes. We perform multiple robustness checks to verify the stability of our models, and we report broad agreement with previous results in the literature.

This work builds on the use of secondary CMB anisotropies as a probe of non-linear physics, and of the interplay between dark matter and baryonic matter in haloes. We incorporate novel methods for combining tSZ data with other probes, in an attempt to refine models of halo bias and constrain 𝜎8. The inclusion of 𝛿𝑔 × K helps to break degeneracies between key parameters of the theoretical framework used. Hence, we highlight the importance of secondary CMB anisotropies as a complementary tool for understanding the large-scale structure, offering new insights into the thermal evolution of the Universe, as well as the growth of structure.

Euclid preparation. TBD. The effect of linear redshift-space distortions in photometric galaxy clustering and its cross-correlation with cosmic shear

Submitted in A&A

Authors:

Euclid Collaboration: K.Tanidis, V.F.Cardone, M.Martinelli, I.Tutusaus, S.Camera et al.

Abstract:

Cosmological surveys planned for the current decade will provide us with unparalleled observations of the distribution of galaxies on cosmic scales, by means of which we can probe the underlying large-scale structure (LSS) of the Universe. This will allow us to test the concordance cosmological model and its extensions. However, precision pushes us to high levels of accuracy in the theoretical modelling of the LSS observables, in order not to introduce biases in the estimation of cosmological parameters. In particular, effects such as redshift-space distortions (RSD) can become relevant in the computation of harmonic-space power spectra even for the clustering of the photometrically selected galaxies, as it has been previously shown in literature studies. In this work, we investigate the contribution of linear RSD, as formulated in the Limber approximation by arXiv:1902.07226, in forecast cosmological analyses with the photometric galaxy sample of the Euclid survey, in order to assess their impact and quantify the bias on the measurement of cosmological parameters that neglecting such an effect would cause. We perform this task by producing mock power spectra for photometric galaxy clustering and weak lensing, as expected to be obtained from the Euclid survey. We then use a Markov chain Monte Carlo approach to obtain the posterior distributions of cosmological parameters from such simulated observations. We find that neglecting the linear RSD leads to significant biases both when using galaxy correlations alone and when these are combined with cosmic shear, in the so-called 3×2pt approach. Such biases can be as large as 5σ-equivalent when assuming an underlying ΛCDM cosmology. When extending the cosmological model to include the equation-of-state parameters of dark energy, we find that the extension parameters can be shifted by more than 1σ.