Lossless, Scalable Implicit Likelihood Inference for Cosmological Fields
We present a comparison of simulation-based inference to full, field-based analytical inference in cosmological data analysis. We demonstrate speed and optimality of summaries computed using Information Maximizing Neural Networks. A single iteration of IMNN training enhances robustness to the choice of fiducial model.
CARPool Covariance: Fast, unbiased covariance estimation for large-scale structure observables
We develop a matrix generalization of Convergence Acceleration by Regression and Pooling (CARPool) to combine a small number of expensive simulations with fast surrogates and obtain low-noise estimates of the covariance matrix that are unbiased by construction.
The GIGANTES dataset: precision cosmology from voids in the machine learning era
We present GIGANTES, the most extensive and realistic void catalog suite ever released -- containing over 1 billion cosmic voids covering a volume larger than the observable Universe, more than 20 TB of data,...
Solving high-dimensional parameter inference: marginal posterior densities and Moment Networks
We propose direct estimation of lower-dimensional marginal distributions, bypassing high-dimensional density estimation or high-dimensional MCMC.. By evaluating the two-dimensional marginal posteriors we can unveil the full-dimensional parameter covariance structure. In addition, we construct a simple hierarchy of fast neural regression models, called Moment Networks, that compute increasing moments of any desired lower-dimensional marginal posterior density.
CARPool: fast, accurate computation of large-scale structure statistics by pairing costly and cheap cosmological simulations
We propose a general method that exploits the correlation between simulations and surrogates to compute fast, reduced-variance statistics of large-scale structure observables without model error at the cost of only a few simulations. We call this approach Convergence Acceleration by Regression and Pooling (CARPool).
The Quijote Simulations
The QUIJOTE simulations are a set of 44,100 full N-body simulations spanning more than 7000 cosmological models in the {Ωm,Ωb,h,ns,σ8,Mν,w} hyperplane.
Cosmology Intertwined IV: The Age of the Universe and its Curvature
A precise measurement of the curvature of the Universe is of primeval importance for cosmology since it could not only confirm the paradigm of primordial inflation but also help in discriminating between different early Universe scenarios.
Cosmology Intertwined III: fσ8 and S8
The standard Λ Cold Dark Matter cosmological model provides a wonderful fit to current cosmological data, but a few tensions and anomalies became statistically significant with the latest data analyses.
Cosmology Intertwined II: The Hubble Constant Tension
The current cosmological probes have provided a fantastic confirmation of the standard Λ Cold Dark Matter cosmological model, that has been constrained with unprecedented accuracy. However, with the increase of the experimental sensitivity a few statistically significant tensions between different independent cosmological datasets emerged.
Cosmology Intertwined I: Perspectives for the Next Decade
The standard Λ Cold Dark Matter cosmological model provides an amazing description of a wide range of astrophysical and astronomical data. However, there are a few big open questions, that make the standard model look like a first-order approximation to a more realistic scenario that still needs to be fully understood.
Precision cosmology with voids in the final BOSS data
We report novel cosmological constraints obtained from cosmic voids in the final BOSS DR12 dataset. They arise from the joint analysis of geometric and dynamic distortions of average void shapes in redshift space.
Cosmology at Home
Cosmology@Home lets you volunteer your spare computer time (like when your screen saver is on) to help better understand our universe by searching for the range of theoretical models that agree with cutting-edge cosmological and particle physics data.