Let $A \in \mathbb{Z}^{n \times n}$ be a positive semidefinite sparse matrix. I am looking for asymptotically fast algorithms to compute for computing the nullspace basis of a $A$ (or just random elements in the nullspace). I wonder whether there are methods that can exploit the fact that $A$ is positive semidefinite sparse matrixto achieve better perfomance.