Boosting galaxy clustering analyses with nonperturbative modeling of redshift-space distortions

Redshift-space distortions (RSD), caused by the peculiar velocities of galaxies, are a key modeling challenge in galaxy clustering analyses, limiting the scales from which cosmological information can be reliably extracted. Unlike dynamical or galaxy bias effects, RSD imprint features are sensitive to nonlinearities across all scales. This paper explores an alternative to the state-of-the-art analytical approach—the effective field theory (EFT)—by partially preserving this nonperturbative nature of RSD. We compare its effectiveness against the EFT in analyzing power spectrum and bispectrum multipoles from synthetic samples of luminous red galaxies at redshifts 0.3 and 0.57, using the projected sensitivity of a Stage-IV galaxy survey. Our results demonstrate that this distinct treatment of RSD improves the robustness of model predictions for both statistics, extending the validity range of the EFT from approximately 0.2h Mpc⁻¹ to 0.35h Mpc⁻¹ for the one-loop power spectrum and from 0.1h Mpc⁻¹ to 0.14h Mpc⁻¹ for the tree-level bispectrum. This leads to a significant enhancement in the precision of cosmological parameter constraints, with uncertainties on the Hubble rate, matter density, and scalar amplitude of fluctuations reduced by 20%–40% for the power spectrum multipoles alone compared to the EFT, and by 25%–50% for joint analyses with the bispectrum. The RSD treatment proposed here may thus play a crucial role in maximizing the scientific return of current and future galaxy surveys. To support this advancement, all models for the power spectrum and bispectrum used in this work are made available through an extended version of the Python package COMET.