Generalized cluster algorithms for frustrated spin models
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 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.
- Publication:
-
Physical Review B
- Pub Date:
- August 1994
- DOI:
- 10.1103/PhysRevB.50.3058
- arXiv:
- arXiv:cond-mat/9402030
- Bibcode:
- 1994PhRvB..50.3058C
- Keywords:
-
- 05.50.+q;
- 75.40.Mg;
- Lattice theory and statistics;
- Numerical simulation studies;
- Condensed Matter
- E-Print:
- 34 pages in RevTeX. No figures included. A compressed postscript file for the paper with figures can be obtained via anonymous ftp to minerva.npac.syr.edu in users/paulc/papers/SCCS-527.ps.Z. Syracuse University NPAC technical report SCCS-527