Isotropy of the z ≈ 0.65 expansion anomaly: a test of the Vaidya-FRW framework

In a companion paper [1], we derived H = Ṁ/M from Vaidya-FRW junction conditions, detecting a localized growth-rate enhancement at zc ≈ 0.65 at 2.9σ in combined supernova, BAO, and CMB data. Because the interior scale factor couples to the total horizon mass M(v), a scalar with no angular dependence, the framework requires this signal to be spatially isotropic. We test this prediction by splitting Pantheon+ supernovae into hemispheres aligned with three CMB anomaly axes and fitting independently. Unlike previous hemisphere analyses, which identify preferred directions from the data and compare to CMB anomalies post hoc, the splitting axes here are pre-specified, the test targets a localized redshift feature rather than global parameters, and Bayesian model comparison is performed independently within each hemisphere. All inter-hemisphere tensions are below 0.35σ. The signal is detected at 2.6–2.8σ in every hemisphere. Nested sampling cross-checks confirm these results and yield Bayesian evidence favoring the Gaussian clump over w₀wa in all six hemispheres (geometric mean Bayes factor 11.4:1). The Gaussian clump parameters are 2–16× more stable across sky cuts than w₀wa, whose best-fit values fragment by Δwa = 0.16–0.36 between hemispheres. Both results, isotropy of the signal and fragmentation of smooth parameterizations, were anticipated in [1].