8bf03717f3b4c0cf.tex
1: \begin{abstract}
2: Effective theories for random critical points are usually
3: non-unitary, and thus may contain relevant operators with negative 
4: scaling dimensions. To study the consequences of the existence of
5: negative dimensional operators, we consider the random-bond $XY$ model.
6: It has been argued that the $XY$ model on a square lattice, when
7: weakly perturbed by random phases, 
8: has a quasi-long-range ordered phase (the random spin wave phase)
9: at sufficiently low temperatures.
10: We show that infinitely many relevant perturbations to the
11: proposed critical action for the random spin wave phase
12: were omitted in all previous treatments.
13: The physical origin of these perturbations is intimately related to
14: the existence of broadly distributed correlation functions.
15: We find that those relevant perturbations do enter the
16: Renormalization Group equations, and affect critical behavior.
17: This raises the possibility that the random $XY$ model has no
18: quasi-long-range ordered phase and no Kosterlitz-Thouless (KT) 
19: phase transition.
20: \end{abstract}
21: