1: \begin{abstract}

2:   We introduce a new sparse estimator of the covariance

3:   matrix for high-dimensional models 

4:   in which the variables have a known ordering.  Our

5:   estimator, which is the solution to a convex optimization problem,

6:   is equivalently expressed as an estimator which tapers the sample covariance matrix by a

7:   Toeplitz, sparsely-banded, data-adaptive matrix.  As a result of this adaptivity, the convex banding 

8:   estimator enjoys theoretical optimality properties not attained by

9:   previous banding or tapered estimators.  In particular, our  convex banding estimator

10:   is minimax rate adaptive  in Frobenius and operator norms, up to log factors, over  commonly-studied classes of

11:   covariance matrices, and  over more general classes.  Furthermore, it correctly recovers

12:   the bandwidth when the true covariance is exactly banded.   Our convex formulation

13:   admits a simple and efficient algorithm. Empirical studies

14:   demonstrate its practical effectiveness and illustrate that  

15:    our exactly-banded estimator works well even when the true

16:   covariance matrix is only close to a banded matrix, confirming  our theoretical results. 

17:   Our method compares favorably with all existing methods, in terms of accuracy and speed. 

18:   We  illustrate the practical merits of the convex banding estimator by showing that 

19:   it  can be used to improve the performance of

20:   discriminant analysis for classifying sound recordings.\\

21: \noindent{\bf Keywords:} {\em covariance; banded;

22:   structured sparsity;

23: positive definite; hierarchical group lasso; high-dimensional; convex;

24: adaptive.}

25: \end{abstract}

26: