NPAC Technical Report SCCS-527
On Generalized Cluster Algorithms for Frustrated Spin Models
Paul Coddington, Leping Han
Submitted October 01 1993
Abstract
Standard Monte Carlo cluster algorithms have proven to be very effective for
many different spin models, however they fail for frustrated spin systems.
Recently a generalized cluster algorithm was introduced that works
extremely well for the fully frustrated Ising model on a square lattice,
by placing bonds between sites based on information from plaquettes rather
than links of the lattice. Here we study some properties of this algorithm
and some variants of it. We introduce a practical methodology for constructing
a generalized cluster algorithm for a given spin model, and investigate apply
this method to some other frustrated Ising models.
We find that such algorithms work well for simple fully frustrated
Ising models in two dimensions, but appear to work poorly or not at all
for more complex models such as spin glasses.