It's All About the Environment: Local $f_{\rm NL}$ from a Dark Matter Conditioned Multitracer Analysis

Abstract

Primordial non-Gaussianity (PNG) offers a powerful test of inflationary physics beyond the simplest single-field scenarios. In large-scale structure, biased tracers respond to local-type PNG with a distinctive scale-dependent signature that is strongest on large scales, but single-tracer measurements are fundamentally limited by cosmic variance. Multitracer methods can mitigate this limitation; however, many existing proposals require splitting tracers using a wide range of halo masses which are often not available, or using secondary halo properties that are not directly observable. Here we develop and validate an alternative strategy based on splitting a tracer sample by its large-scale dark matter environment. We derive a theoretical framework for the $f_{\rm NL}$ response of environmentally selected tracers and test it in $N$-body simulations, confirming consistency between separate-universe predictions and direct power-spectrum inference. To enable observational implementation, we show that a simple linear-theory reconstruction of the dark matter environment from halos yields unbiased $f_{\rm NL}$ constraints in $\mathtt{Quijote}$ simulations. Forecasts for the DESI LRG sample indicate that environmental multitracing can improve constraints on local PNG by a factor of 2-3 relative to conventional single-tracer analyses for high density tracers, comparable to gains achieved when halo formation time is assumed to be perfectly known. We develop an analytic model explaining why the gains are smaller for a halo environment split and why separate-universe predictions fail in this case. Unlike approaches that require modeling assembly bias or inferring secondary halo properties, our method relies only on linear theory. These results establish environmental multitracer techniques as a promising route toward $\sigma(f_{\rm NL})\sim 1$ with upcoming galaxy surveys.