首页

Full-shape clustering measurements from DESI DR1 leverage information from the linear to quasi-nonlinear scales to deliver percent-level constraints on cosmological parameters. However, FS analyses relying on broad priors frequently suffer from projection effects, where marginalization over poorly constrained nuisance parameters can bias the inferred posteriors. To address this, we introduce physically motivated Halo Occupation Distribution (HOD) priors—calibrated from high-fidelity mocks—directly into the Bayesian treatment of nuisance parameters. By anchoring marginalization to realistic galaxy–halo connections, this HOD-informed prior sharpens posterior constraints on σ₈ while mitigating projection biases that would otherwise skew cosmological inference. In parallel, we implement a fully frequentist pipeline based on profile-likelihood scans, which by design eschews priors on nuisance parameters and guarantees correct coverage. This prior-free approach offers an independent, likelihood-based perspective on the same data and remains inherently immune to high-dimensional marginalization effects. Applied to the identical DESI DR1 data set, these complementary strategies yield consistent cosmological results, demonstrating that informed Bayesian priors and prior-agnostic frequentist intervals can jointly underpin a robust full-shape cosmology program.

BIO

Hanyu Zhang is a WCA postdoctoral fellow at University of Waterloo. He completed his Ph.D in physics at Kansas State University under the supervision of Dr. Lado Samushia. Hanyu is actively involved in the DESI collaboration, contributing to various key projects and working groups.