MIT

Using a weak-disorder scheme and real-space renormalization-group techniques, we obtain analytical results for the critical behavior of various q-state Potts models with correlated disordered exchange interactions along d1 of d spatial dimensions on hierarchical (Migdal-Kadanoff) lattices. Our results indicate qualitative differences between the cases d-d1=1 (for which we find nonphysical random fixed points, suggesting the existence of nonperturbative fixed distributions) and d-d1>1 (for which we do find acceptable perturbartive random fixed points), in agreement with previous numerical calculations by Andelman and Aharony. We also rederive a criterion for relevance of correlated disorder, which generalizes the usual Harris criterion.

A relatively new class of materials has been found in which the basic assumption of Landau Fermi-liquid theory?that at low energies the electrons in a metal should behave essentially as a collection of weakly interacting particles?is violated. These non-Fermi-liquid systems exhibit unusual temperature dependences in their low-temperature properties, including several examples in which the specific heat divided by temperature shows a singular log T temperature dependence over more than two orders of magnitude, from the lowest measured temperatures in the milliKelvin regime to temperatures over 10 K. These anomalous properties, with their often pure power-law or logarithmic temperature dependences over broad temperature ranges and inherent low characteristic energies, have attracted active theoretical interest from the first experimental report in 1991. This article first describes the various theoretical approaches to trying to understand the source of strong temperature- and frequency-dependent electron-electron interactions in non-Fermi-liquid systems. It then discusses the current experimental body of knowledge, including a compilation of data on non-Fermi-liquid behavior in over 50 systems. The disparate data reveal some interesting correlations and trends and serve to point up a number of areas where further theoretical and experimental work is needed. ©2001 The American Physical Society

In the most popular approach to the numerical study of the Anderson metal-insulator transition the transfer matrix (TM) method is combined with finite-size scaling ideas. This approach requires large computer resources to overcome the statistical fluctuations and to accumulate data for a sufficient range of different values of disorder or energy. In this paper we present an alternative approach in which the basic TM is extended to calculate the derivative with respect to disorder. By so doing we are able to concentrate on a single value of energy or disorder and, potentially, to calculate the critical behaviour much more efficiently and independently of the assumed range of the critical regime. We present some initial results which illustrate both the advantages and the drawbacks of the method.