22
9

Optimal Inference with a Multidimensional Multiscale Statistic

Abstract

We observe a stochastic process YY on [0,1]d[0,1]^d (d1d\geq 1) satisfying dY(t)=n1/2f(t)dtdY(t)=n^{1/2}f(t)dt + dW(t)dW(t), t[0,1]dt \in [0,1]^d, where n1n \geq 1 is a given scale parameter (`sample size'), WW is the standard Brownian sheet on [0,1]d[0,1]^d and fL1([0,1]d)f \in L_1([0,1]^d) is the unknown function of interest. We propose a multivariate multiscale statistic in this setting and prove its almost sure finiteness; this extends the work of D\"umbgen and Spokoiny (2001) who proposed the analogous statistic for d=1d=1. We use the proposed multiscale statistic to construct optimal tests for testing f=0f=0 versus (i) appropriate H\"{o}lder classes of functions, and (ii) alternatives of the form f=μnIBnf=\mu_n \mathbb{I}_{B_n}, where BnB_n is an axis-aligned hyperrectangle in [0,1]d[0,1]^d and μnR\mu_n \in \mathbb{R}; μn\mu_n and BnB_n unknown. In the process we generalize Theorem 6.1 of D\"umbgen and Spokoiny (2001) about stochastic processes with sub-Gaussian increments on a pseudometric space, which is of independent interest.

View on arXiv
Comments on this paper