Null space for sparse matrix

Computes a null space basis, similar to null, but exploiting sparsity.

The function implements the algorithm proposed in

M. Khorramizadeh and N. Mahdavi-Amiri, "An efficient algorithm for sparse null space basis problem using ABS methods," Numerical Algorithms, vol. 62, no. 3, pp. 469–485, Jun. 2012.

It efficiently computes a null space basis for sparse matrices, exploiting and preserving sparsity. Contrary to Matlab's builtin null function, the basis will in general not be orthogonal.