Towards a full w CDM map-based analysis for weak lensing surveys

With the rapid advance of wide-field surveys it is increasingly important to perform combined cosmological probe analyses. We present a new pipeline for simulation-based multi-probe analyses, which combines tomographic large-scale structure (LSS) probes (weak lensing and galaxy clustering) with cosmic microwave background (CMB) primary and lensing data. These are combined at the C ℓ-level, yielding 12 distinct auto- and cross-correlations. The pipeline is based on UFalconv2, a framework to generate fast, self-consistent map-level realizations of cosmological probes from input lightcones, which is applied to the CosmoGridV1 N-body simulation suite. It includes a non-Gaussian simulation-based covariance for the LSS tracers, several data compression schemes, and a neural network emulator for accelerated theoretical predictions. We validate the pipeline by comparing the simulations to these predictions, and our derived constraints to earlier analyses. We apply our framework to a simulated 12×2 pt tomographic analysis of KiDS, BOSS, and Planck, and forecast constraints for a ΛCDM model with a variable neutrino mass. We find that, while the neutrino mass constraints are driven by the CMB data, the addition of LSS data helps to break degeneracies and improves the constraint by up to 35%. For a fiducial Mν = 0.15 eV, a full combination of the above CMB+LSS data would enable a 3σ constraint on the neutrino mass. We explore data compression schemes and find that MOPED outperforms PCA and is made robust using the derivatives afforded by our automatically differentiable emulator. We also study the impact of an internal lensing tension in the CMB data, parametrized by AL , on the neutrino mass constraint, finding that the addition of LSS to CMB data including all cross-correlations is able to mitigate the impact of this systematic. UFalconv2 and a MOPED compressed Planck CMB primary + CMB lensing likelihood are made publicly available.[UFalconv2: https://cosmology.ethz.ch/research/software-lab/UFalcon.html, compressed Planck CMB primary + CMB lensing likelihood: https://github.com/alexreevesy/planck_compressed.]

Section: Data Compressionmentioning

confidence: 99%

“…We used functionalities provided by numpy [146], scipy [147], and matplotlib [148] for this work. Job arrays were submitted using esub-epipe [122,149]. Corner plots were created with trianglechain [8,83].…”

Section: Acknowledgmentsmentioning

confidence: 99%

12 × 2 pt combined probes: pipeline, neutrino mass, and data compression

Reeves,

Nicola,

Refregier

et al. 2024

“…We thank the High Performance Computing group at ETH Zurich for support with the Euler cluster, which was heavily used throughout this work. We submitted jobarrays to the cluster using esub-epipe [75,76]. We used functionalities from several Python packages: numpy [77], scipy [78], and matplotlib [79].…”

Section: Jcap05(2024)049mentioning

confidence: 99%

Simulation-based inference of deep fields: galaxy population model and redshift distributions

Moser,

Kacprzak,

Fischbacher

et al. 2024

Accurate redshift calibration is required to obtain unbiased cosmological information from large-scale galaxy surveys. In a forward modelling approach, the redshift distribution n(z) of a galaxy sample is measured using a parametric galaxy population model constrained by observations. We use a model that captures the redshift evolution of the galaxy luminosity functions, colours, and morphology, for red and blue samples. We constrain this model via simulation-based inference, using factorized Approximate Bayesian Computation (ABC) at the image level. We apply this framework to HSC deep field images, complemented with photometric redshifts from COSMOS2020. The simulated telescope images include realistic observational and instrumental effects. By applying the same processing and selection to real data and simulations, we obtain a sample of n(z) distributions from the ABC posterior. The photometric properties of the simulated galaxies are in good agreement with those from the real data, including magnitude, colour and redshift joint distributions. We compare the posterior n(z) from our simulations to the COSMOS2020 redshift distributions obtained via template fitting photometric data spanning the wavelength range from UV to IR. We mitigate sample variance in COSMOS by applying a reweighting technique. We thus obtain a good agreement between the simulated and observed redshift distributions, with a difference in the mean at the 1σ level up to a magnitude of 24 in the i band. We discuss how our forward model can be applied to current and future surveys and be further extended. The ABC posterior and further material will be made publicly available at https://cosmology.ethz.ch/research/software-lab/ufig.html.

“…Various features have been proposed to extract the information beyond 2-pt, including the bispectrum [20][21][22][23] and trispectrum [24], higher order moments of mass maps [25][26][27][28], Minkowski functionals [29][30][31][32], weak lensing voids [33,34], and wavelet decomposition coefficients [35]. The feature that has been most extensively studied is the shear peaks counts [19,[36][37][38][39][40] The peak counts were used to make measurements from surveys [41][42][43][44][45] and recently by Zürcher et al [46, hereafter [46]]. Several non-Gaussian analyses combine features from different probes: the density split statistics [47][48][49], and Minkowski functionals [50].…”

Section: Inference With Non-gaussian Featuresmentioning

confidence: 99%

CosmoGridV1: a simulated 𝗐CDM theory prediction for map-level cosmological inference

Kacprzak

Fluri

Schneider

et al. 2023

We present CosmoGridV1: a large set of lightcone simulations for map-level cosmological inference with probes of large scale structure. It is designed for cosmological parameter measurement based on Stage-III photometric surveys with non-Gaussian statistics and machine learning. CosmoGridV1 spans the wCDM model by varying Ωm, σ 8, w 0, H 0, n s, Ω b , and assumes three degenerate neutrinos with fixed ∑ mν = 0.06 eV. This space is covered by 2500 grid points on a Sobol sequence. At each grid point, we run 7 simulations with PkdGrav3 and store 69 particle maps at nside = 2048 up to z = 3.5, as well as halo catalog snapshots. The fiducial cosmology has 200 independent simulations, along with their stencil derivatives. An important part of CosmoGridV1 is the benchmark set of 28 simulations, which include larger boxes, higher particle counts, and higher redshift resolution of shells. They allow for testing if new types of analyses are sensitive to choices made in CosmoGridV1. We add baryon feedback effects on the map level, using shell-based baryon correction model. The shells are used to create maps of weak gravitational lensing, intrinsic alignment, and galaxy clustering, using the UFalcon code. The main part of CosmoGridV1 are the raw particle count shells that can be used to create full-sky maps for a given n(z). We also release projected maps for a Stage-III forecast, as well as maps used previously in KiDS-1000 deep learning constraints with CosmoGridV1. The data is available at http://www.cosmogrid.ai/.