Constraining primordial non-Gaussianity with cosmological weak lensing: shear and flexion

Abstract

We examine the cosmological constraining power of future large-scale weak lensing surveys on the model of \emph{Euclid}, with particular reference to primordial non-Gaussianity. Our analysis considers several different estimators of the projected matter power spectrum, based on both shear and flexion, for which we review the covariances and Fisher matrices. The bounds provided by cosmic shear alone for the local bispectrum shape, marginalized over $\sigma_8$, are at the level of $\Delta f_\mathrm{NL} \sim 100$. We consider three additional bispectrum shapes, for which the cosmic shear constraints range from $\Delta f_\mathrm{NL}\sim 340$ (equilateral shape) up to $\Delta f_\mathrm{NL}\sim 500$ (orthogonal shape). The competitiveness of cosmic flexion constraints against cosmic shear ones depends on the galaxy intrinsic flexion noise, that is still virtually unconstrained. Adopting the very high value that has been occasionally used in the literature results in the flexion contribution being basically negligible with respect to the shear one, and for realistic configurations the former does not improve significantly the constraining power of the latter. Since the flexion noise decreases with decreasing scale, by extending the analysis up to $\ell_\mathrm{max} = 20,000$ cosmic flexion, while being still subdominant, improves the shear constraints by $\sim 10%$ when added. However on such small scales the highly non-linear clustering of matter and the impact of baryonic physics make any error estimation uncertain. By considering lower, and possibly more realistic, values of the flexion intrinsic shape noise results in flexion constraining power being a factor of $\sim 2$ better than that of shear, and the bounds on $\sigma_8$ and $f_\mathrm{NL}$ being improved by a factor of $\sim 3$ upon their combination. (abridged)