Cosmological parameters from weak lensing power spectrum and bispectrum tomography: including the non-Gaussian errors

TL;DR

This study demonstrates that incorporating the weak lensing bispectrum with the power spectrum significantly enhances cosmological parameter constraints, especially for dark energy, by capturing non-Gaussian information and cross-covariances.

Contribution

We quantitatively assess the added value of the bispectrum in weak lensing tomography, including non-Gaussian errors and cross-covariances, showing a 60% improvement in dark energy constraints.

Findings

Adding bispectrum improves dark energy figure-of-merit by 60%.

Including non-Gaussian errors reduces the expected information gain.

Bispectrum provides complementary information to the power spectrum.

Abstract

We re-examine a genuine power of weak lensing bispectrum tomography for constraining cosmological parameters, when combined with the power spectrum tomography, based on the Fisher information matrix formalism. To account for the full information at two- and three-point levels, we include all the power spectrum and bispectrum information built from all-available combinations of tomographic redshift bins, multipole bins and different triangle configurations over a range of angular scales (up to lmax=2000 as our fiducial choice). For the parameter forecast, we use the halo model approach in Kayo, Takada & Jain (2013) to model the non-Gaussian error covariances as well as the cross-covariance between the power spectrum and the bispectrum, including the halo sample variance or the nonlinear version of beat-coupling. We find that adding the bispectrum information leads to about 60% improvement in the dark energy figure-of-merit compared to the lensing power spectrum tomography alone, for three redshift-bin tomography and a Subaru-type survey probing galaxies at typical redshift of zs~1. The improvement is equivalent to a 1.6 larger survey area. Thus our results show that the bispectrum or more generally any three-point correlation based statistics carries complementary information on cosmological parameters to the power spectrum. However, the improvement is modest compared to the previous claim derived using the Gaussian error assumption, and therefore our results imply less additional information in even higher-order moments such as the four-point correlation function.