Impact of relativistic effects on cosmological parameter estimation

Future surveys will access large volumes of space and hence very long wavelength fluctuations of the matter density and gravitational field. It has been argued that the set of secondary effects that affect the galaxy distribution, relativistic in nature, will bring new, complementary cosmological constraints. We study this claim in detail by focusing on a subset of wide-area future surveys: Stage-4 cosmic microwave background experiments and photometric redshift surveys. In particular, we look at the magnification lensing contribution to galaxy clustering and general-relativistic corrections to all observables. We quantify the amount of information encoded in these effects in terms of the tightening of the final cosmological constraints as well as the potential bias in inferred parameters associated with neglecting them. We do so for a wide range of cosmological parameters, covering neutrino masses, standard dark-energy parametrizations and scalar-tensor gravity theories. Our results show that, while the effect of lensing magnification to number counts does not contain a significant amount of information when galaxy clustering is combined with cosmic shear measurements, this contribution does play a significant role in biasing estimates on a host of parameter families if unaccounted for. Since the amplitude of the magnification term is controlled by the slope of the source number counts with apparent magnitude, $s(z)$, we also estimate the accuracy to which this quantity must be known to avoid systematic parameter biases, finding that future surveys will need to determine $s(z)$ to the $\sim 5\%–10\%$ level. On the contrary, large-scale general-relativistic corrections are irrelevant both in terms of information content and parameter bias for most cosmological parameters but significant for the level of primordial non-Gaussianity.

Figure 1

Forecast $1\sigma$ contours for $\mathrm{\Sigma }{m}_{\nu }$, $w_0$, and $w_a$ from LSST clustering only (orange ellipses) and $\text{LSST clustering}+\text{LSST shear}+\mathrm{S}4\text{(cyan ellipses)}$ in the fiducial case without lensing magnification or GR effects. The thin solid and dashed ellipses correspond to the $1\sigma$ contours after including the lensing contribution to the clustering power spectrum in the same two cases, respectively. The black circle and square show the bias associated with ignoring the presence of lensing magnification (again, in the same two cases). In all cases, the impact of GR effects is negligible, and therefore we have not included the corresponding ellipses in this figure.

Figure 2

Same as Fig. 1 for the Horndeski parameters $c_B$, $c_M$, and $c_T$.

Figure 3

Forecast distribution for $f_{\mathrm{NL}}$ for LSST red galaxies (red); blue galaxies (blue); the combination of both in a multitracer sense (black); and the combination of LSST galaxy clustering, LSST cosmic shear, and CMB S4 (orange). The bias on $f_{\mathrm{NL}}$ associated with the GR effects, corresponding to $f_{\mathrm{NL}}^{\mathrm{GR}}\simeq -0.7$, is shown as a vertical dashed line.

Figure 4

Relative degradation in the final constraints associated with removing all scales larger than a factor $\lambda$ times the comoving horizon at the source redshift (the associated angular scales at $z=1$ are shown in the upper twin $x$ axis). The results are shown for simple extensions to $\mathrm{\Lambda }\mathrm{CDM}$ (upper panel), scalar-tensor theories (middle panel), and primordial non-Gaussianity (lower panel). Except in the case of $f_{\mathrm{NL}}$, the information content of the largest scales is heavily suppressed due to cosmic variance.