Lagrangian displacement field estimators in cosmology

The late-time nonlinear Lagrangian displacement field is highly correlated with the initial field, so reconstructing it could enable us to extract primordial cosmological information. Our previous work [A. Ota et al., Phys. Rev. D 104, 123508 (2021)PRVDAQ2470-001010.1103/PhysRevD.104.123508] carefully studied the displacement field reconstructed from the late-time density field using the iterative method proposed by Schmittfull et al. [Phys. Rev. D 96, 023505 (2017)PRVDAQ2470-001010.1103/PhysRevD.96.023505] and found that it does not fully converge to the true, underlying displacement field (e.g., ∼8% offset at k∼0.2 h Mpc-1 at z=0.6). We also constructed the Lagrangian perturbation theory model for the reconstructed field, but the model could not explain the discrepancy between the true and the reconstructed fields in the previous work. The main sources of the discrepancy were speculated to be a numerical artifact in the displacement estimator due to the discreteness of the sample. In this paper, we develop two new estimators of the displacement fields to reduce such a numerical discreteness effect, the normalized momentum estimator and the rescaled resumed estimator. We show that the discrepancy Ota et al. [Phys. Rev. D 104, 123508 (2021)PRVDAQ2470-001010.1103/PhysRevD.104.123508] reported is not due to the numerical artifacts. We conclude that the method from Schmittfull et al. [Phys. Rev. D 96, 023505 (2017)PRVDAQ2470-001010.1103/PhysRevD.96.023505] cannot fully reconstruct the shape of the nonlinear displacement field at the redshift we studied, while it is still an efficient baryon acoustic oscillation reconstruction method. In parallel, by properly accounting for the UV-sensitive term in a reconstruction procedure with an effective field theory approach, we improve the theoretical model for the reconstructed displacement field, by almost 5 times, from ∼15% to the level of a few percent at k∼0.2 h Mpc-1 at the redshift z=0.6.