Efficient Estimation of Linear Functionals of Principal Components

We study principal component analysis (PCA) for mean zero i.i.d. Gaussian observations $X_1,\dots, X_n$ in a separable Hilbert space $\mathbb{H}$ with unknown covariance operator $\Sigma.$ The complexity of the problem is characterized by its effective rank ${\bf r}(\Sigma):= \frac{{\rm tr}(\Sigma)}{\|\Sigma\|},$ where ${\rm tr}(\Sigma)$ denotes the trace of $\Sigma$ and $\|\Sigma\|$ denotes its operator norm. This framework includes, in particular, high-dimensional spiked covariance models as well as some models in functional PCA and kernel PCA in machine learning. We develop a method of bias reduction in the problem of estimation of linear functionals of eigenvectors of $\Sigma.$ Under the assumption that ${\bf r}(\Sigma)=o(n),$ we establish the asymptotic normality and asymptotic properties of the risk of the resulting estimators and prove matching minimax lower bounds, showing their semi-parametric optimality.