From Processors to Reservoirs: The Stability Gate and the Homeostatic Double Flip in Galaxy Evolution

Galaxy evolution models often treat galaxies as passive processors of inflow, with metallicity and star formation set by one,way scaling relations and instantaneous state variables. Here a different picture is tested: mature galaxies may behave as regulated reservoirs that organise the balance between dynamical depth, enrichment structure, chemical memory and regeneration. A compact, dimensionless state vector is constructed from four observables: an energy or depth proxy H (velocity dispersion or sSFR in simulations), a stability proxy S (structural compactness or an enrichment,maturity coordinate), a chemical memory proxy M (metallicity at fixed mass), and a regeneration proxy R (specific star formation rate or its analogue). Galaxies are ordered by a stability coordinate and split into “infant” and “adult” regimes at the median S. In SDSS DR8 a sharp stability gate is detected, and two coupled inversions appear across it when controlling for stellar mass: (i) the partial correlation between depth and chemical memory changes sign between infants and adults, consistent with an “energy eraser” regime giving way to a retention regime, and (ii) the stability,regeneration relation changes from suppressive to supportive. Cross,catalogue comparisons show that the depth, memory inversion is recovered in EAGLE, while GAMA shows a weaker, same,sign trend and IllustrisTNG remains strongly negative in both regimes, suggesting sensitivity to tracer choice and feedback implementation rather than a trivial selection artefact. Spatially resolved MaNGA measurements provide a sanity check that systems classified as “adult” preferentially host stable inner regions with non,negligible Hα emission. Together these results favour a picture in which at least a subset of mature galaxies behave as regulated reservoirs rather than simple processors, and motivate effective models that encode history dependence in addition to instantaneous scaling relations.