Extracting the Alcock-Paczynski signal from voids: A novel approach via reconstruction

Abstract

The void galaxy cross correlation function is a powerful tool to extract cosmological information. Through the void galaxy cross correlation function, cosmic voids, the underdense regions in the galaxy distribution, are used for refined deductions of the Universe's content by correcting apparent geometric distortions. This study proposes a novel procedure for optimally extracting the Alcock Paczynski, AP, signal from cosmic voids through a cosmological reconstruction technique. Employing cosmological reconstruction, specifically using the Zel'dovich approximation, we estimate the true positions of galaxies from their redshift space locations, reducing distortions introduced by peculiar velocities. Unlike previous analyses, we identify voids and measure the void galaxy cross correlation function directly in reconstructed space. This approach enables us, for the first time, to include in our analysis small nonlinear voids, typically discarded in previous studies, thus enhancing the statistical power of void studies and significantly improving their cosmological constraining power. Reconstruction is particularly effective even at small scales for voids, due to their clean and dynamically simple environment. This ability to recover information encoded on small scales significantly enhances the precision of the analysis, leading to a 23 % improvement in the constraints on the AP parameters compared to previous methods where the analysis is performed in redshift space and, consequently, to a better estimate of the derived cosmological parameters. Our analysis also includes a comprehensive set of consistency checks, demonstrating its robustness. We expect this methodology to yield a substantial gain in constraining power when applied to data from modern large scale structure surveys.

Figures (14)

• Figure 1: Top Panel: number density of voids per unit radius interval, i.e., the void size function, of the VIDE voids found in real (green stars), redshift (orange triangles), and reconstructed (blue dots) space as a function of their effective radius $R$, averaged over 100 mocks; shaded bands show the $1\sigma$ standard deviation across mocks. The dashed line marks the mean tracer separation of the halo simulations. Bottom Panel: residuals divided by $1\sigma$ standard deviation relative to real space, redshift - real (orange) and reconstructed - real (blue), showing reconstructed space agrees more closely with real space. The grey band indicates the $[-1,1]\ \sigma$ interval.
• Figure 2: Monopole (left), quadrupole (center) and hexadecapole (right) of the average VGCF measured for the 100 mocks, in redshift space (orange triangles), reconstructed space (blue dots) and real space (green stars) together with the associated best-fit models (blue dashed lines for reconstructed space, orange dash dotted lines for redshift space, and green solid line for real space) estimated via the likelihood analysis presented in Section \ref{['sec: likelihood']}. Error bars represent the standard deviation among the 100 mocks. In the left panel, the monopole represents the density profile of halos inside the void region, and it is quite similar in all the three cases (real, redshift and reconstructed space). In the middle panel, the quadrupole is visibly influenced by RSD in redshift space (orange), while in reconstructed space (blue) and real space (green), it is distortion-free and consistent with zero. The hexadecapole is 0 in both redshift and reconstructed space. Here $r$ is a dimensionless quantity representing the physical separation between the center of the void and the halo, normalized by the void effective radius $R$.
• Figure 3: Posterior probability distribution of the model parameters that enter in Eq. \ref{['eq: xi model RSD MQ']}, obtained via MCMC from the data vector in redshift space shown in orange in Fig. \ref{['fig: multipoles bestfit rsd recon']}. Dark and light-shaded areas represent $68\%$ and $95\%$ confidence regions, and dashed lines indicate fiducial values of the RSD and AP parameters $\beta$ and $\varepsilon$. The top of each column states the mean and standard deviation of the 1-dimensional marginal distributions.
• Figure 4: Posterior probability distribution of the model parameters that enter in Eq. \ref{['eq: xi model RSD MQ']}, obtained via MCMC from the data in reconstructed space, shown in blue curve in Fig. \ref{['fig: multipoles bestfit rsd recon']}. Dark and light-shaded areas represent $68\%$ and $95\%$ confidence regions, and dashed lines indicate fiducial values of the RSD and AP parameters. The top of each column states the mean and standard deviation of the 1-dimensional marginal distributions.
• Figure 5: Top Panel: comparison of the values of the AP parameter $\varepsilon$ with its error $\sigma_\varepsilon$ (error bars), obtained with the fitting procedure described in Section \ref{['sec: likelihood']}, for the analysis with tracers and voids in redshift-space (orange dots) and the analysis with tracers and voids in reconstructed space (blue triangles). Each dot represents the $\varepsilon$ value as a function of the minimum radius $R_\mathrm{min}$ for the voids in that specific subsample, expressed in mean tracer separation (mps) units, used for computing VGCF. Bottom Panel: improvement factor $I$, i.e., the ratio between $\sigma_\varepsilon$ estimated in redshift space and reconstructed space.