Abstract
For phase transitions in disordered systems, an exact theorem provides a bound on the finite size correlation length exponent: . It is believed that the intrinsic satisfies the same bound. We argue that the standard averaging introduces a noise and a new diverging length scale. For self-averaging breaks down, disconnecting from , and the bound applies only for the latter. We illustrate these ideas on two exact examples, with . We propose a new method of disorder averaging, which is able to capture the intrinsic exponents.
- Received 18 April 1997
DOI:https://doi.org/10.1103/PhysRevLett.79.5130
©1997 American Physical Society