Is there any suggestion about how could one construct a model that uses semidefinite programming that minimizes sum of k smallest eigenvalues of Laplacian matrix?
I found two papers that have done something for the just bound not for optimization.

minimizing sum of $k$-largest is a convex problem; the sum of $k$-smallest eigenvalues is concave, and therefore, not really that conducive to minimization...
– SuvritFeb 4 '13 at 22:35

Thanks a lot for the answer. i tried another way using the maximization of k-largest eigenvalues.
– RoyehFeb 23 '13 at 0:03