1: \begin{abstract} We study the minimization of  a quadratic over  Stiefel manifolds (the set of all orthogonal $r$-frames in $\IR^n$), which   has applications in high-dimensional

2: semi-supervised classification tasks. To reduce the computational complexity, sequential subspace methods(SSM) are employed to 

3: convert the high-dimensional minimization problems to low-dimensional ones.

4: In this paper, we are interested in attaining an optimal solution of good quality, i.e.,  

5: a ``qualified" critical point.  Qualified critical points are those critical points, at which the associated multiplier matrix meets some upper bound condition. 

6: These critical points enjoy the global optimality in special  quadratic problems. 

7: For a general quadratic,

8: %To attain an optimal solution of good quality, 

9:  SSM computes a sequence of  ``qualified critical points"   in its low-dimensional ``surrogate regularized models".  %One approximate qualified solution can be obtained by solving a sequence of local dimensional surrogate regularized models.

10: The convergence  to a qualified critical point is ensured, whenever

11: each SSM subspace is constructed by the following  vectors:

12: (i) a set of orthogonal unit vectors associated with the current iterate,  (ii) a set of vectors corresponding to the gradient of the objective, and (iii) a set of eigenvectors associated with the smallest $r$ eigenvalues of the system matrix. In addition, when Newton direction vectors are included in subspaces,  the convergence of SSM can be accelerated significantly. 

13: 

14: \end{abstract}

15: