The directed-loop algorithm (DLA) is one of the most robust algorithms for quantum Monte-Carlo simulation, and enjoys very broad applicability. Updates of world-line configurations in DLA are done by a worm, which consists of a pair of discontinuity points moving stochastically on world-lines and altering the state on the line just behind itself. The direction of motion of a discontinuity point is altered only by scattering at vertices that are placed between world-lines or on a single world-line with density determined by the Hamiltonian. However, when one applies the method to a system such as the Bose-Hubbard model with t<<U (t is the hopping amplitude and U is the on-site energy), the efficiency of the method is low because of high density of vertices due to large U. We improve DLA by omitting the vertices that express the effect of U in this paper. The effect of U is reflected in other procedures. We demonstrate the efficiency of the new method by applying it to the interacting dilute Bose gas system in a discrete space that has the aforesaid difficulty.
|
|
|
