Consistent clustering and lensing of SDSS-III BOSS galaxies with an extended abundance matching formalism

TL;DR

This paper introduces an extended abundance matching model called SHAMe that successfully resolves the lensing-is-low problem in galaxy clustering and lensing data from SDSS-III BOSS, emphasizing the importance of realistic galaxy-halo connection models.

Contribution

The paper presents SHAMe, an extended abundance matching formalism that accurately reproduces galaxy clustering and lensing, solving the longstanding lensing-is-low problem in large-scale structure analyses.

Findings

SHAMe matches clustering and lensing in mock and real data.

The model's success depends on realistic galaxy-halo connection assumptions.

Differences in assembly bias levels relate to cosmological parameters.

Abstract

Several analyses have shown that LCDM-based models cannot jointly describe the clustering (GC) and galaxy-galaxy lensing (GGL) of galaxies in the SDSS-III BOSS survey, which is commonly known as the 'lensing-is-low problem'. In this work, we show that an extension of Subhalo Abundance Matching, dubbed SHAMe, successfully solves this problem. First, we show that this model accurately reproduces the GC and GGL of a mock galaxy sample in the TNG300 hydrodynamic simulation with analogous properties to BOSS galaxies. Then, we switch our attention to observed BOSS galaxies at z=0.31-0.43, and we attempt to reproduce their GC and GGL by evaluating SHAMe on two different simulations: one adopting best-fitting cosmological parameters from Planck and the other from weak gravitational lensing surveys (Low S8), where the amplitude of matter fluctuations is lower for the latter. We find excellent agreement between SHAMe predictions and observations for both cosmologies, indicating that the lensing-is-low problem originates from approximations in previous theoretical descriptions of the data. The main difference between SHAMe results in these cosmologies is the level of galaxy assembly bias, which is approximately 20 and 10% for Planck and Low S8, respectively. These results highlight the dangers of employing oversimplified models to analyse current large-scale structure datasets, and the need for realistic yet flexible descriptions of the galaxy-halo connection.