1: \begin{abstract}

2: This paper considers sparse spiked covariance matrix models in the high-dimensional setting and studies the minimax estimation of the covariance matrix and the principal subspace as well as the minimax rank detection. 

3: %Each of these three problems is important in its own right. 

4: The optimal rate of convergence for estimating the spiked covariance matrix under the spectral norm is established, which requires significantly different techniques from those for estimating other structured covariance matrices such as bandable or sparse covariance matrices. 

5: %This  rate is quite different from those for estimating other structured covariance matrices such as bandable or sparse covariance matrices. 

6: We also establish the minimax rate under the spectral norm for estimating the principal subspace, the primary object of interest in principal component analysis. In addition,  the optimal rate for the rank detection boundary is obtained. 

7: This result also resolves a gap between the upper and lower bounds in a recent paper by \citet{BR12} where the special case of rank one is considered.  

8: 

9: \end{abstract}