Intrinsic alignments of galaxies in multiple projections

Abstract

Intrinsic alignments of galaxies are measured and modelled to gain cosmological information, to further understand the interactions between galaxies and to mitigate their effects on gravitational weak lensing studies. Hydrodynamical simulations are often used to constrain priors or calibrate models. Therefore, obtaining the maximum amount of information possible from these simulations is imperative. In this work, we have combined the information of shapes projected over two or three axes ($x,y,z$), for intrinsic alignment signals ($w_{g+},\ \tildeξ_{g+,2}$), showing a consistent gain in signal-to-noise ratio (SNR) for all cases studied using TNG300-1. The gain in SNR is found to be higher for the addition of the second projection than for the third, and higher for shapes calculated using the reduced inertia tensor rather than the simple one. The two shape samples studied, $n_\star>300$ and $\mathrm{log}(M_\star \ h/\mathrm{M_\odot})>10.5$, where the latter has a much higher signal amplitude, show similar gains in SNR when more projections are added. We also model the correlation functions with the non-linear alignment model. The SNR gains from the measurements are higher but consistent with the constraints on the non-linear alignment amplitude $A_{\rm IA}$ and galaxy bias $b_{\rm g}$. Using multiple projection axes increases SNR overall, enabling more efficient use of numerically expensive hydrodynamical simulations.

Figures (13)

• Figure 1: Correlation functions, $r_\mathrm{p} w_\mathrm{g+}$ (top) and $r^2 \tilde{\xi}_\mathrm{g+,2}$ (bottom), in TNG300 for two shape samples: $n_\star>300$ (continuous lines) and $\mathrm{log}(M_\star \ h/\mathrm{M_\odot})>10.5$ (dashed lines). The shapes are projected over the $x$ (light blue), $y$ (medium blue), and $z$ (dark blue) axes and measured using the simple inertia tensor.
• Figure 2: Normalised combined covariance matrices for $w_\mathrm{g+}$ (top panel) and $\tilde{\xi}_\mathrm{g+,2}$ (bottom panel) for the $n_\star>300$ shape sample. Darker colours indicate higher values. Each block of the combined covariance matrix is labelled according to the definitions in Sect. \ref{['sect:cov']}.
• Figure 3: S/Ns for the $w_\mathrm{g+}$ (top) and $\tilde{\xi}_\mathrm{g+,2}$ (bottom) of combining one vs two (unfilled) and two vs three (filled) projections. The colours of the markers correspond to the different shape samples: $n_\star>300$ is red and $\mathrm{log}(M_\star h/\mathrm{M_\odot})>10.5$ is blue. Shapes measured using the simple, reduced inertia tensor are depicted by circles, crosses, respectively. The black line denotes the boundary marking where both S/N values are equal.
• Figure 4: Correlation functions, $r_\mathrm{p} w_\mathrm{g+}$ (top) and $r^2 \tilde{\xi}_\mathrm{g+,2}$ (bottom) in TNG300 for two shape samples: $n_\star>300$ (downward triangle) and $\mathrm{log}(M_\star \ h/\mathrm{M_\odot})>10.5$ (upward triangle). The shapes are projected over the $x$ (light blue), $y$ (medium blue) and $z$ (dark blue) axes and measured using the simple inertia tensor, whereas the joint NLA fit is shown as a solid (dashed) line for the cut in $n_\star$ ($M_\star$). The second panel in both plots shows the fit residual for the $y$-projection. Grey regions indicate excluded regions for the fit. Note: the data points are slightly horizontally displaced for easier readability.
• Figure 5: Same as Fig. \ref{['Fig:SNR']}, but for the S/N defined in Eq. (\ref{['eq:SNR fitted scaling']}) obtained from the non-linear alignment model fit.