Estimation and inference for precision matrices of nonstationary time series (Q2215745)

The paper under review deals with the problem of estimation and inference for precision matrices of nonstationary time series. A class of nonstationary (locally stationary) and nonlinear time series is introduced and the theoretical properties of its covariance and precision matrices are studied. The authors consistently estimate the precision matrices and provide convergent rates for these estimators. Next, they propose two adaptive tests using some statistics from their estimation procedure, where the convergence rates and powers of the tests are adaptive to the strength of the temporal dependence and the smoothness of the underlying data generating mechanism. Monte Carlo simulations, technical proofs, and auxiliary lemmas are provided in the Supplementary Material.