EMU and Euclid: Detection of a radio-optical galaxy clustering cross-correlation signal between the Evolutionary Map of the Universe and Euclid

TL;DR

This work reports the first detection of the harmonic-space cross-spectrum between EMU radio-continuum sources and Euclid galaxies, achieving high significance and strong agreement with ΛCDM-based predictions. By using two independent EMU source finders and Euclid Q1 samples, the authors demonstrate robustness to catalog construction and establish a multi-tracer framework to probe the EMU redshift distribution through clustering redshifts. They test several n_EMU(z) and b_EMU(z) models, constrain an overall amplitude A_b ≈ 1, and perform tomographic analysis with three Euclid redshift bins to reconstruct p_EMU(z) under different parameterisations, finding model-dependent results due to degeneracies with bias. The results validate the methodology for combining EMU and Euclid data and outline a clear path toward more powerful, self-consistent multi-tracer cosmology with larger sky overlap and inclusion of EMU auto-spectra and CMB-lensing cross-correlations.

Abstract

Synergies between large-scale radio-continuum and optical/near-infrared galaxy surveys are a powerful tool for cosmology. Cross-correlating these surveys can constrain the redshift distribution of radio sources, mitigate systematic effects, and place constraints on cosmological models. We perform the first measurement of the clustering cross-spectrum between radio-continuum sources in the Evolutionary Map of the Universe (EMU) survey and galaxies from the ESA Euclid satellite mission's Q1 release. Our goal is to detect and characterise the cross-correlation signal, test its robustness against systematic effects, and compare our measurements with theoretical predictions. We use data from the Australian SKA Pathfinder's EMU Main Survey, which overlaps with the Euclid Deep Field South. We generate two radio-source catalogues using different source finders to create galaxy maps. We measure the harmonic-space cross-correlation signal using a pseudo-spectrum estimator. The measured signal is compared to theoretical predictions based on a ΛCDM cosmology, using several models for the EMU source redshift distribution and bias. We report detection above 8σ of the cross-correlation signal consistent across all tested models and data sets. The measured cross-spectra from the two radio catalogues are in excellent agreement, demonstrating that the cross-correlation is robust against the choice of source-finding algorithm. The measured signal also agrees with theoretical models developed from previous cross-correlation studies and simulations. This pathfinder study establishes a statistically significant cross-correlation between EMU and Euclid. The robustness of the signal is a crucial validation of the methodology, paving the way for future large-scale analyses leveraging the full power of this synergy to constrain cosmological parameters and our understanding of galaxy evolution.

Figures (11)

• Figure 1: Weighted galaxy overdensity maps for (top panels: left for pgal0.90 sample and right from pstar0.05 sample) and EMU (bottom panels: left for the Selavy source finder and right for the PyBDSF source finder) obtained with \ref{['eq:delta_g_v2']}.
• Figure 2: Top panel: Galaxy redshift distributions used in this analysis. For (red curves), we show the 1bin and 3bin configurations with solid and dotted lines, respectively. For EMU (blue curves) we plot two different models from Saraf:EMUxDES and 2008MNRAS.388.1335W. Note that the unbinned redshift distributions are normalised to their peak, with the binned ones being normalised to the peak of the corresponding unbinned distribution, to enhance readability. Bottom panel: Fiducial biases adopted, for galaxies from Flagship-2 (red curve) and for EMU sources considering both the constant bias model (blue, dashed line) and constant-amplitude one (blue, solid curves).
• Figure 3: Measured harmonic-space cross-spectra between and EMU. The black markers represent the measured cross-spectra for the two EMU baseline maps (left panelPyBDSF and right panelSelavy) combined with the two maps. Coloured lines show the theory predictions for the cross spectrum using the fiducial redshift distributions and bias models discussed in \ref{['ssec:theory_met.theory']}. Note that the grey shadowed areas represent the multipoles excluded from our analysis. In each panel, full and empty markers are for pgal0.90 and pstar0.05 respectively.
• Figure 4: Best-fit overall amplitude, $A^{\ast}_b$, for the one-parameter fit, as in \ref{['eq:Ab', 'eq:sigma_Ab']}. The figure demonstrates the high degree of internal consistency among all modelling choices (colour and line-style code as in \ref{['fig:cls_fidu']}) and data subsets (upper/lower panel for PyBDSF/Selavy; filled/empty markers for pgal0.90/pstar0.05). The main conclusion is that the best-fit $A^{\ast}_b$ is highly robust against the choice of EMU catalogue and $\ell_{\max}$ scale cut (marker shapes), allowing us to interpret $A^{\ast}_b$ as a measure of the overall rescaling required for the FS2 simulation's bias prescription.
• Figure 5: $A_b$-$C^\times_\mathrm{shot}$ posterior contours corresponding to the results in \ref{['tab:ab_crossSn']}, i.e., before applying a prior on $C^\times_\mathrm{shot}$. We only present results for PyBDSF here for clarity. The parameter correlations obtained with Selavy are consistent with the contours presented here. The grey shadowed area represents the region of parameter space excluded by the prior on $C_{\rm shot}^{ \times}$ that is, for unphysical negative values of the cross-shot noise.