Cosmological parameter estimation via iterative emulation of likelihoods

Abstract

The interpretation of cosmological observables requires the use of increasingly sophisticated theoretical models. Since these models are becoming computationally very expensive and display non-trivial uncertainties, the use of standard Bayesian algorithms for cosmological inferences, such as MCMC, might become inadequate. Here, we propose a new approach to parameter estimation based on an iterative Gaussian emulation of the target likelihood function. This requires a minimal number of likelihood evaluations and naturally accommodates for stochasticity in theoretical models. We apply the algorithm to estimate 9 parameters from the monopole and quadrupole of a mock power spectrum in redshift space. We obtain accurate posterior distribution functions with approximately 100 times fewer likelihood evaluations than an affine invariant MCMC, roughly independently from the dimensionality of the problem. We anticipate that our parameter estimation algorithm will accelerate the adoption of more accurate theoretical models in data analysis, enabling more comprehensive exploitation of cosmological observables.

arXiv comment12 pages, 5 figures

References in corpus (48)

• Observational Evidence from Supernovae for an Accelerating Universe and a Cosmological Constant
• emcee: The MCMC Hammer
• Planck 2015 results. XIII. Cosmological parameters
• The clustering of galaxies in the completed SDSS-III Baryon Oscillation Spectroscopic Survey: cosmological analysis of the DR12 galaxy sample
• The Cosmic Linear Anisotropy Solving System (CLASS) II: Approximation schemes
• Bayes in the sky: Bayesian inference and model selection in cosmology
• Dark Energy Survey Year 1 Results: Cosmological Constraints from Galaxy Clustering and Weak Lensing
• Power Spectrum Analysis of Three-Dimensional Redshift Surveys
• The Coyote Universe II: Cosmological Models and Precision Emulation of the Nonlinear Matter Power Spectrum
• FastPM: a new scheme for fast simulations of dark matter and halos
• Euclid preparation: II. The EuclidEmulator -- A tool to compute the cosmology dependence of the nonlinear matter power spectrum
• RegPT: Direct and fast calculation of regularized cosmological power spectrum at two-loop order
• II: Bayesian Methods for Cosmological Parameter Estimation from Cosmic Microwave Background Measurements
• Cosmology Constraints from the Weak Lensing Peak Counts and the Power Spectrum in CFHTLenS
• The clustering of galaxies in the SDSS-III Baryon Oscillation Spectroscopic Survey: single-probe measurements from CMASS anisotropic galaxy clustering
• Neutrino masses and cosmological parameters from a Euclid-like survey: Markov Chain Monte Carlo forecasts including theoretical errors
• Gaussian covariance matrices for anisotropic galaxy clustering measurements
• One simulation to fit them all - changing the background parameters of a cosmological N-body simulation
• The Aemulus Project III: Emulation of the Galaxy Correlation Function
• The Aemulus Project II: Emulating the Halo Mass Function
• The Initial Conditions of the Universe from Constrained Simulations
• Emulating the CFHTLenS Weak Lensing data: Cosmological Constraints from moments and Minkowski functionals
• Dynamical dark energy versus $\Lambda =$const. in light of observations
• Cosmic Emulation: Fast Predictions for the Galaxy Power Spectrum
• Modelling the large scale structure of the Universe as a function of cosmology and baryonic physics
• Approximate Bayesian Computation for Forward Modeling in Cosmology
• Cosmology in the era of Euclid and the Square Kilometre Array
• Estimation of cosmological parameters using adaptive importance sampling
• Bayesian Optimization for Likelihood-Free Inference of Simulator-Based Statistical Models
• cosmoabc: Likelihood-free inference via Population Monte Carlo Approximate Bayesian Computation
• Bayesian optimisation for likelihood-free cosmological inference
• Bayesian emulator optimisation for cosmology: application to the Lyman-alpha forest
• Galaxy Formation: Bayesian History Matching for the Observable Universe
• Model independent inference of the expansion history and implications for the growth of structure
• Evaluating machine learning techniques for predicting power spectra from reionization simulations
• On the road to percent accuracy II: calibration of the non-linear matter power spectrum for arbitrary cosmologies
• How to add massive neutrinos to your $\Lambda$CDM simulation -- extending cosmology rescaling algorithms
• MultiModeCode: An efficient numerical solver for multifield inflation
• BAMBI: blind accelerated multimodal Bayesian inference
• Euclid-era cosmology for everyone: Neural net assisted MCMC sampling for the joint 3x2 likelihood
• KiDS-450: Enhancing cosmic shear with clipping transformations
• Reconstructing Probability Distributions with Gaussian Processes
• Higher Order Hamiltonian Monte Carlo Sampling for Cosmological Large-Scale Structure Analysis
• A New Approach to the Optimal Target Selection Problem
• Learn-As-You-Go Acceleration of Cosmological Parameter Estimates
• Model dispersion with PRISM; an alternative to MCMC for rapid analysis of models
• On Sparse variational methods and the Kullback-Leibler divergence between stochastic processes
• Gaussbock: Fast parallel-iterative cosmological parameter estimation with Bayesian nonparametrics

Cited by in corpus (28)

• The BACCO Simulation Project: Exploiting the full power of large-scale structure for cosmology
• Large-scale dark matter simulations
• Simultaneous modelling of matter power spectrum and bispectrum in the presence of baryons
• CONNECT: A neural network based framework for emulating cosmological observables and cosmological parameter inference
• 3\%-accurate predictions for the clustering of dark matter, haloes and subhaloes, over a wide range of cosmologies and scales
• The Bacco Simulation Project: Bacco Hybrid Lagrangian Bias Expansion Model in Redshift Space
• Accelerating Large-Scale-Structure data analyses by emulating Boltzmann solvers and Lagrangian Perturbation Theory
• Machine Learning for Observational Cosmology
• Multi-Fidelity Emulation for the Matter Power Spectrum using Gaussian Processes
• Cosmological parameter estimation with Genetic Algorithms
• Fast and robust Bayesian Inference using Gaussian Processes with GPry
• Modelling galaxy clustering in redshift space with a Lagrangian bias formalism and $N$-body simulations
• The Aemulus Project VI: Emulation of beyond-standard galaxy clustering statistics to improve cosmological constraints
• LINNA: Likelihood Inference Neural Network Accelerator
• Combined full shape analysis of BOSS galaxies and eBOSS quasars using an iterative emulator
• Full forward model of galaxy clustering statistics with AbacusSummit lightcones
• Accelerating cosmological inference with Gaussian processes and neural networks -- an application to LSST Y1 weak lensing and galaxy clustering
• Cosmological Parameter Estimation and Inference using Deep Summaries
• Constraining CMB temperature evolution with Sunyaev-Zel'dovich galaxy clusters from the Atacama Cosmology Telescope
• Halo Occupation Distribution of Emission Line Galaxies: fitting method with Gaussian Processes
• The BAHAMAS project: Evaluating the accuracy of the halo model in predicting the non-linear matter power spectrum
• Fast and effortless computation of profile likelihoods using CONNECT
• Galaxy Clustering in the Mira-Titan Universe I: Emulators for the redshift space galaxy correlation function and galaxy-galaxy lensing
• Deep Learning and genetic algorithms for cosmological Bayesian inference speed-up
• MF-Box: Multi-fidelity and multi-scale emulation for the matter power spectrum
• The BACCO Simulation Project: A baryonification emulator with Neural Networks
• Boosting MonteCarlo sampling with a non-Gaussian fit
• Using Statistical Emulation and Knowledge of Grain-Surface Diffusion for Bayesian Inference of Reaction Rate Parameters: An Application to a Glycine Network