{"id":"arxiv/2610.09105","url":"https://arxiv.org/abs/2610.09105","tags":["math.ST"],"score":0,"title":"Local asymptotic normality for spectral-tail likelihoods under fixed-domain sampling","domain":"math.ST","source":"arxiv","authors":["Alexandre B. Simas","Jonas Wallin"],"reviews":0,"subject":"math.ST","summary":"We establish local asymptotic normality (LAN) for joint inference on scale and roughness in stationary Gaussian random fields observed on increasingly dense regular grids in a fixed domain. Scale and roughness are represented by the coefficient and decay exponent of the leading spectral tail. Under spectral regularity conditions, we identify the joint normalization governing the LAN expansion and prove consistency and asymptotic normality of the maximum likelihood estimators. For \\(N\\) observations, the estimation errors are of order \\(N^{-1/2}\\) for roughness and \\((\\log N)/\\sqrt N\\) for…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-10-06","reactions":{}}