{"id":"arxiv/2608.27209","url":"https://arxiv.org/abs/2608.27209","tags":["math.ST"],"score":0,"title":"Connecting Riemannian Geometry and Statistical Inference for Correlation Matrices","domain":"math.ST","source":"arxiv","authors":["Argyn Kuketayev"],"reviews":0,"subject":"math.ST","summary":"The quotient-affine metric gives an intrinsic Riemannian geometry to full-rank correlation matrices, but its geodesic distance has no closed form and we are not aware of an analytic asymptotic null distribution for it. We connect this geometry, introduced in 2019, with Jennrich's 1970 asymptotic test for equality of correlation matrices. The quadratic form underlying Jennrich's statistic is exactly one half of the quotient-affine metric tensor. The identity arises because eliminating marginal standard deviations from Gaussian Fisher information performs the same projection as quotienting out…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-08-27","reactions":{}}