Bathas G., Neuberger H.
A possible barrier at $z\approx 1$ for local algorithms.
(22K, TeX)
ABSTRACT. It is shown that a certain class of generalizations of over-relaxation
algorithms is incapable to further reduce the dynamical exponent z
below its standard over-relaxed value of z=1 approximately. Approximately means
that the mean field value is one, while the true value can be somewhat
different in certain dimensions. The generalizations are obtained by viewing
over-relaxation as a slightly deformed deterministic algorithm and should,
therefore, hold for Hybrid Monte Carlo algorithms as well.