A forward modeling approach to analyzing galaxy clustering with SimBIG.

Area of Science:

• Cosmology
• Astrophysics
• Computational Science

Background:

• Galaxy clustering is a key probe of the Universe's expansion history and composition.
• Extracting cosmological information from small, nonlinear scales is challenging for traditional methods.

Purpose of the Study:

• To apply the SimBIG simulation-based inference framework to the BOSS CMASS galaxy sample.
• To derive cosmological constraints using galaxy clustering power spectrum data.
• To demonstrate the advantage of SimBIG in accessing information on nonlinear scales.

Main Methods:

• Utilized the SimBIG forward modeling framework with 20,000 simulated galaxy samples.
• Employed high-fidelity Quijote N-body simulations and incorporated detailed survey realism.
• Performed simulation-based inference by training normalizing flows to infer cosmological parameters.

Main Results:

• Derived significant constraints on Lambda-CDM cosmological parameters, specifically Omega_m and sigma_8.
• Achieved a 27% improvement in precision for the sigma_8 constraint compared to standard analyses.
• Demonstrated that SimBIG effectively exploits information on nonlinear scales beyond current analytic model limits.

Conclusions:

• SimBIG offers a powerful framework for extracting cosmological information from galaxy clustering, especially on nonlinear scales.
• The method provides statistically significant gains, comparable to much larger galaxy samples.
• Future analyses using SimBIG with other summary statistics are expected to yield further improvements in cosmological constraints.