We study the nature of the multicritical point in the three-dimensional O(3)+O(2) symmetric Landau-Ginzburg-Wilson theory, which describes the competition of two order parameters that are O(3) and O(2) symmetric, respectively. This study is relevant for the SO(5) theory of high-Tc superconductors, which predicts the existence of a multicritical point in the temperature-doping phase diagram, where the antiferromagnetic and superconducting transition lines meet. We investigate whether the O(3)+O(2) symmetry gets effectively enlarged to O(5) approaching the multicritical point. For this purpose, we study the stability of the O(5) fixed point. By means of a Monte Carlo simulation, we show that the O(5) fixed point is unstable with respect to the spin-4 quartic perturbation with the crossover exponent $\phi_{4,4}=0.180(15)$, in substantial agreement with recent field-theoretical results. This estimate is much larger than the one-loop $\epsilon$-expansion estimate $\phi_{4,4}=1/26$, which has often been used in the literature to discuss the multicritical behavior within the SO(5) theory. Therefore, no symmetry enlargement is generically expected at the multicritical transition. We also perform a five-loop field-theoretical analysis of the renormalization-group flow. It shows that bicritical systems are not in the attraction domain of the stable decoupled fixed point. Thus, in these systems--high-Tc cuprates should belong to this class--the multicritical point corresponds to a first-order transition.

The RP(2) gauge model is studied in 2D. We use Monte-Carlo renormalization techniques for blocking the mean spin-spin interaction, <A>, and the mean gauge field plaquette, <P>. The presence of the O(3) renormalized trajectory is verified and is consistent with the known three-loop beta-function. The first-order `vorticity' transition observed by Solomon et al. is confirmed, and the location of the terminating critical point is established. New scaling flows in (<A>,<P>) are observed associated with a large exponent kappa in the range 4~5. The scaling flows give rise to a strong cross-over effect between regions of high and low vorticity and are likely to induce an apparent signal for scaling in the cross-over region which we propose explains the scaling observed for RP(2), RP(3) and SO(4)-matrix models. The signal for this `pseudo' scaling will occur for the RP(2) spin model in the cross-over region which is the region in which computer simulations are done. We find that the RP(2) spin model is in the same universality class as the O(3) spin model but that it is likely to require a very large correlation length before the true scaling of this class sets in. We conjecture that the scaling flows are due either to the influence of a nearby new renormalized trajectory or to the ghost of the Kosterlitz-Thouless trajectory in the associated XY model.