Fluids in random media and dimensional augmentation

John Cardy
2023-09-13
Abstract:We propose a solution to the puzzle of dimensional reduction in the random field Ising model, inverting the question and asking: to what random problem in $D=d+2$ dimensions does a pure system in $d$ dimensions correspond? We consider two models: a continuum binary fluid, and a lattice gas which maps exactly onto an Ising model. In both cases we show that the mean density and other observables are equal to those of a similar model in $D$ dimensions, but with interactions and correlated disorder in the extra two dimensions of range $\propto l$, in the limit as $l\to\infty$. There is no conflict with rigorous results that the finite range model with locally correlated disorder orders in $D=3$. Our arguments avoid the use of replicas and perturbative field theory, instead being based on convergent cluster expansions, which, for the lattice gas, may be extended all the way to the critical point by virtue of the Lee-Yang theorem. Although the results may be viewed as a consequence of Parisi-Sourlas supersymmetry, they follow more directly from Kirchhoff's matrix-tree theorem.
Statistical Mechanics,Disordered Systems and Neural Networks
What problem does this paper attempt to address?