Posterior-Level Reconstruction of the Remnant-Mass Fraction in BNS and NSBH Mergers: A GWTC-4.0/O4a Test of the ECI Scaling

We reconstruct posterior distributions of the dimensionless remnant-mass parameter δ for compact mergers involving neutron stars. The parameter is defined as δ = Mₒᵤₜ/Mᵦ for BNS systems and as δ = Mₒᵤₜ/Mᵦᵉᶠᶠ for NSBH systems, where Mₒᵤₜ is the baryonic mass remaining outside the final black hole or compact remnant. Public LVK/IGWN posterior samples are used for GW170817, GW190425, GW200105, GW200115, GW230518_125908, and GW230529_181500. For GW170817, Mₒᵤₜ is introduced through the same model-informed remnant-matter range as in Ref. [1]. For GW190425, no full Mₒᵤₜ posterior is available, and the event is treated as a censored upper-limit case. For NSBH events, the posterior-derived field «baryonic_torus_mass_source» is used as a model-derived proxy for Mₒᵤₜ when present. The reconstructed distributions separate the sample into finite-δ, weak-disruption, zero-dominated, and censored regimes. GW170817 and GW230529_181500 occupy the upper finite-δ branch. GW200105 gives a lower full-posterior median with a nonzero positive branch. GW200115 and GW230518_125908 are zero-dominated. GW190425 enters only through δᵁᴸ. A supplementary rescaling test Mₒᵤₜ → κMₒᵤₜ with κ = 0.5, 1.0, and 2.0 shows that the relative ordering of GW230529_181500 and GW170817 depends on the global NSBH remnant-mass normalization. The separation between the finite-δ, weak-disruption, and zero-dominated regimes is not erased by this rescaling. The result is therefore a posterior-level, prescription-conditioned event-level test of the δ hierarchy, not a population-level or prescription-marginalized inference.