Complexity of stoquastic frustration-free Hamiltonians

We study several problems related to properties of non-negative matrices that arise at the boundary between quantum and classical probabilistic computation. Our results are twofold. First, we identify a large class of quantum Hamiltonians describing systems of qubits for which the adiabatic evolution can be efficiently simulated on a classical probabilistic computer. These are stoquastic local Hamiltonians with a "frustration free" ground-state. A Hamiltonian belongs to this class iff it can be represented as $H=\sum_a H_a$ where (1) every term $H_a$ acts non-trivially on a constant number of qubits, (2) every term $H_a$ has real non-positive off-diagonal matrix elements in the standard basis, and (3) the ground-state of $H$ is a ground-state of every term $H_a$. Secondly, we generalize the Cook-Levin theorem proving NP-completeness of the satisfiability problem to the complexity class MA -- a probabilistic analogue of NP. Specifically, we construct a quantum version of the k-SAT problem which we call "stoquastic k-SAT" such that stoquastic k-SAT is contained in MA for any constant $k$, and any promise problem in MA is Karp-reducible to stoquastic 6-SAT. This result provides the first non-trivial example of a MA-complete promise problem.