Cosmology and neutrino mass with the Minimum Spanning Tree.

The information content of the minimum spanning tree (MST), used to capture higher-order statistics and other information from the cosmic web, is compared to that of the power spectrum for a $\nu\Lambda$CDM model. The measurements are made in redshift space using haloes from the Quijote simulation of mass $\geq 3.2\times 10^{13}\,h^{-1}{\rm M}_{\odot}$ in a box of length $L_{\rm box}=1\,h^{-1}{\rm Gpc}$. The power spectrum multipoles (monopole and quadrupole) are computed for Fourier modes in the range $0.006 < k < 0.5\, h{\rm Mpc}^{-1}$. For comparison the MST is measured with a minimum length scale of $l_{\min}\simeq13\,h^{-1}{\rm Mpc}$. Combining the MST and power spectrum allows for many of the individual degeneracies to be broken; on its own the MST provides tighter constraints on the sum of neutrino masses $M_{\nu}$, Hubble constant $h$, spectral tilt $n_{\rm s}$, and baryon energy density $\Omega_{\rm b}$ but the power spectrum alone provides tighter constraints on $\Omega_{\rm m}$ and $\sigma_{8}$. The power spectrum on its own gives a standard deviation of $0.25\,{\rm eV}$ on $M_{\nu}$ while the combination of power spectrum and MST gives $0.11\,{\rm eV}$. There is similar improvement of a factor of two for $h$, $n_{\rm s}$, and $\Omega_{\rm b}$. These improvements appear to be driven by the MST's sensitivity to small scale clustering, where the effect of neutrino free-streaming becomes relevant. The MST is shown to be a powerful tool for cosmology and neutrino mass studies, and therefore could play a pivotal role in ongoing and future galaxy redshift surveys (such as DES, DESI, Euclid, and Rubin-LSST).