Weak lensing, Big Magnification: Recovering hidden lensing information from galaxy positions

Abstract

Galaxy clustering and weak gravitational lensing are powerful cosmological probes, traditionally combined through galaxy clustering and galaxy-galaxy lensing in a $2\times2$pt analysis. An alternative is the foreground-background clustering cross-correlation induced by lensing magnification, but measurements have achieved lower signal-to-noise than shear-based lensing. We show that, for deep photometric samples, cancellation between galaxies with positive and negative magnification responses can suppress this signal. Using DES Year 6 photometry and Balrog synthetic-source injections, we explore high-redshift ($z\simeq0.7$-$2.5$) samples split by the sign and amplitude of their magnification response. This separation reduces cancellation and can increase both signal-to-noise and cosmological constraining power. We call the combination of foreground clustering and foreground-background cross-correlations with response-selected background samples BigMag. Forecasts with DES-like simulated data vectors show that, for an optimistic sample configuration and sufficiently precise response calibration, BigMag can achieve constraining power comparable to traditional $2\times2$pt using galaxy positions alone. The same samples also enhance inverse galaxy-galaxy lensing (IGGL), a position-shear correlation dominated by magnification when the position sample lies behind the shape sample. Combining BigMag and IGGL further self-calibrates the magnification responses and yields marginalized $S_8$ constraints comparable to traditional $3\times2$pt analyses, and tighter for the more precise calibration scenarios considered here. The framework also enables near-zero-response samples for traditional $3\times2$pt and geometric lensing-ratio analyses. These results motivate further optimization and validation of galaxy selections to realize magnification as a complementary cosmological probe. [Abrg.]