1: \begin{abstract}

2: %  In this paper we propose a general framework for estimating sparse eigenvectors in the standard eigenvalue problem for positive semidefinite matrices based on the orthogonal iteration, and establish numerical convergence results. %analysis for such framework. 

3: %  We propose two variants of the Truncated Orthogonal Iteration to compute multiple leading eigenvectors with sparsity constraints. We present numerical experiments for the proposed algorithms, and [achieves better efficiency, performs better in some test problems compared with others] show they are competitive (Is this all that you show?  That doesn't sound very exciting.) compared with existing sparse eigenvector algorithms. We illustrate the efficiency and accuracy of these algorithms for the sparse Principal Components Analysis (PCA) application and present experiments on simulated datasets as well as real world datasets.

4: A wide range of problems in computational science and engineering 

5: require estimation of sparse eigenvectors for high dimensional systems. 

6: %To estimate sparse eigenvectors in the standard eigenvalue problem for a perturbed positive semidefinite matrix, 

7: Here, we propose two variants of the Truncated Orthogonal Iteration to compute multiple leading eigenvectors with sparsity constraints simultaneously. We establish numerical convergence results for the proposed algorithms using a perturbation framework, and extend our analysis to other existing alternatives for sparse eigenvector estimation. We then apply our algorithms to solve the sparse principle component analysis problem for a wide range of test datasets, from simple simulations to real-world datasets including MNIST, sea surface temperature and 20 newsgroups. In all these cases, we show that the new methods get state of the art results quickly and with minimal parameter tuning. 

8: 

9: \end{abstract}

10: