The correlation between CMB lensing measurements from the CMB 2- and 4-point functions is computed for a possible CMB-Stage 4 experiment, including CMB polarization. Julien Peloton released a light version of the code used for the paper in his lenscov package. It computes full CMB and CMB lensing power spectrum covariances and other quantities useful for CMB lensing.

FFT-PT: Reducing the 2-loop large-scale structure power spectrum to low-dimensional, radial integrals

Modeling the probability that two galaxies are separated by a certain distance is a computationally involved task because of nonlinearities involved in the formation of large-scale structures. We present a method to evaluate this in a simpler way by reducing otherwise 5-dimensional integrals to a series of low-dimensional integrals.

Improving photometric redshifts with Lyα tomography

Based on the fact that galaxies form predominantly in overdensities, we show how photometric redshifts of galaxies can be improved by using flux information along the line of sight inferred from Lyα tomography.

A bias to CMB lensing measurements from the bispectrum of large-scale structure

The deflection of CMB photons by gravitational lensing depends on the statistical properties of the structures encountered along the line of sight. We investigate how non-Gaussianity of these structures caused by gravitational clustering affects measurements of the lensing power spectrum.

Fast Large Scale Structure Perturbation Theory using 1D FFTs

Perturbative model predictions for large-scale structure observables often involve convolution integrals in Fourier space. We show how these can be evaluated in a fast and elegant way using one-dimensional FFTs.

The matter bispectrum of large-scale structure: three-dimensional comparison between theoretical models and numerical simulations

A comprehensive comparison of various models for the dark matter bispectrum against simulations.

Eulerian BAO Reconstructions and N-Point Statistics

Eulerian BAO reconstructions are introduced as a new class of reconstruction algorithms. Rather than displacing objects, they transform the Eulerian mass density. The reconstructed power spectrum is shown to be a combination of specific 3- and 4-point statistics. Five new BAO reconstruction algorithms are developed and tested against simulations, finding that two can compete with the standard algorithm. Various Eulerian and Lagrangian reconstructions are shown to be perturbatively equivalent.

View more

Delensing the CMB with the Cosmic Infrared Background

Since the Cosmic Infrared Background (CIB) is very correlated with the large-scale structure responsible for gravitationally lensing the CMB, it can be used to delens the CMB. The paper finds that even if the correlation coefficient is not known accurately, delensing with the CIB will be useful for ground- based CMB experiments in the near future.

Near Optimal and Simple Bispectrum Estimators for Large-Scale Structure

Rather than measuring the bispectrum for thousands of triangles and comparing the measurement of each triangle with the theory prediction, the theory prediction is used as the starting point to derive optimally compressed bispectrum statistics that retains full sensitivity to cosmological parameters. They are given by cross-correlating the mass density with three auxiliary fields that are quadratic in the density. This is simple, fast and nearly optimal, and allows for simple covariance estimation. The method is extensively tested on N-body simulations for dark matter and halos, finding that the shift-term cross-spectrum is most sensitive to linear halo bias, and shot noise requires non-Poissonian corrections.

View more

The One-Loop Matter Bispectrum in the Effective Field Theory of Large Scale Structures

The Effective Field Theory of Large-Scale Structure (EFTofLSS) is used to model the matter bispectrum. The predictions are compared against bispectrum measurements in N-body simulations, which serves as a robust check of the EFTofLSS framework. The resulting bispectrum model performs better than any previous perturbative model.

On the Joint Analysis of CMB Temperature and Lensing-Reconstruction Power Spectra

Gravitational lensing of the CMB can be reconstructed from the observed CMB fluctuations. This paper shows how to combine the reconstruction power spectrum with the CMB power spectrum by working out their covariance in theory and simulations. In addition, an accurate approximation for the likelihood of the reconstruction power spectrum is established. These results were used in the official Planck analyses in 2013 and 2015.

View more

Fast Estimation of LSS Bispectra with the Modal Method

The paper presents a new modal bispectrum estimator that projects the data bispectrum on an orthonormalized basis of separable polynomial bispectrum templates. This algorithm is orders of magnitude faster than any previous approach in the literature. It is applied to a large suite of N-body simulations with Gaussian and non-Gaussian initial conditions of various types. Simulation results are modeled with phenomenological fitting formulas and a time-shift model.

Universal Non-Gaussian Initial Conditions for N-body Simulations

Based on an expansion of primordial bispectra in separable polynomial modes, an algorithm is developed and implemented to setup N-body simulations for general non-Gaussian initial conditions in a fast way. Until today, this is the most general and fastest way to setup non-Gaussian N-body simulations for arbitrary primordial bispectra from general inflation models.

Covariate Shift and Local Learning by Distribution Matching

This undergraduate work presents a way to deal with situations in machine learning where training and test data have different statistical properties. The proposed solution transforms the data to match statistical properties in a well-defined way. It is validated on various real-world data sets.