Predictable Information Across Galactic and Cosmological Scales: An Inquiry

Abstract

A previous rotation-curve study found that galaxy discrepancies contain predictive information that transfers robustly across galaxies. We ask whether comparable organization appears in the large-scale peculiar-velocity field. Using 36,506 Cosmicflows-4 galaxies within 270 Mpc, we withhold each of 64 three-dimensional spatial regions and predict its galaxies from the remaining regions. First- and second-order spatial derivatives together reduce held-out root-mean-square prediction error by only about 2% beyond bulk motion. No comparable predictive strength emerges under the models tested. This contrast redirects attention to the galaxy-scale phenomenon: an explanation must account not only for the rotation-curve discrepancy itself, but also for why its structure remains so predictable across galaxies.