Estimating sparse precision matrices

Abstract

We apply a method recently introduced to the statistical literature to directly estimate the precision matrix from an ensemble of samples drawn from a corresponding Gaussian distribution. Motivated by the observation that cosmological precision matrices are often approximately sparse, the method allows one to exploit this sparsity of the precision matrix to more quickly converge to an asymptotic 1/sqrt(Nsim) rate while simultaneously providing an error model for all of the terms. Such an estimate can be used as the starting point for further regularization efforts which can improve upon the 1/sqrt(Nsim) limit above, and incorporating such additional steps is straightforward within this framework. We demonstrate the technique with toy models and with an example motivated by large-scale structure two-point analysis, showing significant improvements in the rate of convergence.For the large-scale structure example we find errors on the precision matrix which are factors of 5 smaller than for the sample precision matrix for thousands of simulations or, alternatively, convergence to the same error level with more than an order of magnitude fewer simulations.

11 pages, 14 figures, submitted to MNRAS

References in corpus (4)

• Why your model parameter confidences might be too optimistic -- unbiased estimation of the inverse covariance matrix
• Parameter inference with estimated covariance matrices
• Analytic model for the matter power spectrum, its covariance matrix, and baryonic effects
• Improving the precision matrix for precision cosmology

Cited by in corpus (22)

• Weak lensing for precision cosmology
• Cosmological constraints from BOSS with analytic covariance matrices
• 2D-FFTLog: Efficient computation of real space covariance matrices for galaxy clustering and weak lensing
• Characterizing Signal Loss in the 21 cm Reionization Power Spectrum: A Revised Study of PAPER-64
• Fewer Mocks and Less Noise: Reducing the Dimensionality of Cosmological Observables with Subspace Projections
• Perturbation theory approach to predict the covariance matrices of the galaxy power spectrum and bispectrum in redshift space
• Massive data compression for parameter-dependent covariance matrices
• Cosmic Shear Covariance Matrix in $w$CDM: Cosmology Matters
• Large Covariance Matrices: Smooth Models from the 2-Point Correlation Function
• Optimal 1D Ly$\alpha$ Forest Power Spectrum Estimation -- II. KODIAQ, SQUAD & XQ-100
• Non-linear shrinkage estimation of large-scale structure covariance
• Precision matrix expansion - efficient use of numerical simulations in estimating errors on cosmological parameters
• Galaxy 2-Point Covariance Matrix Estimation for Next Generation Surveys
• Large Covariance Matrices: Accurate Models Without Mocks
• Robustness of the covariance matrix for galaxy clustering measurements
• Cosmological Model Parameter Dependence of the Matter Power Spectrum Covariance from the DEUS-PUR $Cosmo$ Simulations
• Fast Low-Rank Matrix Estimation without the Condition Number
• Euclid: Fast two-point correlation function covariance through linear construction
• Fast Algorithms for Learning Latent Variables in Graphical Models
• Minimax Estimation of Large Precision Matrices with Bandable Cholesky Factor
• Minimax Estimation of Bandable Precision Matrices
• Residual Smoothing: Using Mocks to Correct Model Covariance Matrices