Random fields, topology, and the Imry-Ma argument

We consider $n$-component fixed-length order parameter interacting with a weak random field in $d=1,2,3$ dimensions. Relaxation from the initially ordered state and spin-spin correlation functions have been studied on lattices containing hundreds of millions sites. At $n-1 d$, when topological objects are absent, the final, lowest-energy, state is independent of the initial condition. It is characterized by the exponential decay of correlations that agrees quantitatively with the theory based upon the Imry-Ma argument. In the borderline case of $n-1=d$, when topological structures are non-singular, the system possesses a weak metastability with the Imry-Ma state likely to be the global energy minimum.