(See Suzuki and Takano, Phys. Lett. Vol.A, vol.69A, no.6, p.426, 1979). It is shown that recently proposed recursions for the spin- 1 / 2 Heisenberg model lead to spurious behaviour when the anisotropy is XY-like, and that the results for the isotropic case depend very strongly on the length scale change factor.
The spin-(1/2 anisotropic Heisenberg model is studied by a modified Migdal-Kadanoff method, called the cluster-decimation approximation. The free energy for a small cluster of spins with an appropriate boundary condition is assumed to be preserved in the bond-moving step and the approximate decimation of Suzuki and Takano is adopted in the site-decimation step. Critical temperatures obtained by the present method are in general much better than those determined by preserving the high-temperature series expansion of the free energy. In particular the XY fixed point and the isotropic Heisenberg fixed point for the square lattice are not found in the present calculation.
The critical behaviour of the kinetic Ising model is analysed in the framework of the Migdal dynamic renormalisation group, for arbitrary dimensions. The dynamic exponent z, correlation length exponent nu and critical fixed point K* are computed for d=1 to 6, in the Suzuki approximation. By analytic continuation, the case d=1+ epsilon is considered within this approximation. It is shown that already to leading order in epsilon , the predictions for K*, nu and z given by dynamics differ from those given by statics. Accordingly, the two-dimensional case is revisited by taking into account higher-order correlation functions. This modification of the renormalisation-group transformation improves significantly the preceding results.