Projecting Unequal Time Fields and Correlators of Large Scale Structure

TL;DR

This paper introduces a new formalism for analyzing large scale structure data that accounts for unequal time effects at the field level, leading to more accurate cross-bin and single bin correlator corrections.

Contribution

The paper develops the unequal time field level projection method, providing first order correction terms for single and multi-tracer power spectra, improving upon previous equal time approximations.

Findings

New first order correction terms for power spectra and correlators.

Corrections can significantly affect cross-bin correlator estimates.

Method can be extended to include full redshift bin integration.

Abstract

Many large scale structure surveys sort their observations into redshift bins and treat every tracer as being located at the mean redshift of its bin, a treatment which we refer to as the equal time approximation. Recently, a new method was developed which allows for the estimation and correction of errors introduced by this approximation, which we refer to as the unequal time correlator-level projection. For single tracer power spectra, corrections arise at second order and above in a series expansion, with first order terms surviving only in multi-tracer analyses. In this paper we develop a new method which we refer to as the unequal time field level projection. This formalism projects the fields individually onto the celestial sphere, displaced from individual reference times, before defining their correlators. This method introduces new, first order correction terms even in the case of single tracer power spectra. Specifically, new first order terms are introduced which apply to both cross-bin and single bin correlators. All of these new corrections originate with derivatives over combinations of a delta function, a cross-bin phase term, and the power spectrum itself and stem from the introduction of two unequal time Fourier transforms into the analysis. We analyse these corrections in the context of a linearly biased power spectrum divided between two redshift bins and find that they can lead to non-trivial corrections, particularly to cross-bin correlators. We also show that these terms can be replicated by appropriately extending the correlator-level analysis to include a second Fourier transform which allows for a full redshift bin integration.