Testing the mass–SFR structure of galaxies around a weak-lensing-selected dark matter footprint in COSMOS-Web

ABSTRACT

We test whether a weak-lensing-selected dark matter (DM) region in COSMOS-Web is associated with measurable changes in the stellar mass–star formation rate (SFR) distribution of galaxies. Using the HSC-SSP PDR3 Mizuki catalogue, we construct a mass-complete Subaru sample and compare the DM region with adjacent UP and DOWN controls matched in photometric redshift, stellar mass, and local projected density. We propagate photo-z uncertainties through Monte Carlo realizations, validate the curved ridge classification with mock catalogues and label shuffle tests, and use image-derived weak-lensing proxies only as a geometric/tomographic consistency check. An independent COSMOS2020 sample is analysed as a catalogue-level consistency check within the available sky footprint of the DM region, rather than as a geometrical replication of the Subaru UP–DM–DOWN experiment. The baryonic density field shows its strongest spatial agreement with the JWST E-mode proxy near $z\simeq 0.55$–0.60, suggesting that the DM region is not an arbitrary aperture. This evidence is supportive rather than definitive because it is based on image-derived proxies, not calibrated numerical convergence maps. In the controlled Subaru analysis, the matched DM, UP, and DOWN samples have nearly identical low-SFR tail fractions, $f_{\rm tail}\simeq 0.062$, with no statistically meaningful DM region-specific excess. COSMOS2020 gives the same qualitative result. Observed colour diagnostics show an ordered ridge–sub-ridge–tail sequence, with the tail population reddest, but this red tail is not unique to the DM region. We conclude that the DM region is tomographically interesting, but the data do not support a robust DM region-specific restructuring of the mass–SFR plane once redshift, stellar mass, local density, photo-z uncertainty, and classification effects are controlled.