Free-energy distribution functions for the randomly forced directed polymer

TL;DR

This paper investigates the distribution functions of free energies for a 1+1-dimensional random directed polymer, introducing approximations and techniques to analyze short-scale, finite-temperature behavior with finite-ranged disorder.

Contribution

It develops exact solutions for free-energy distributions using novel approximations and techniques, clarifying the effects of disorder correlators and boundary conditions in the directed polymer model.

Findings

Derived free-energy distribution functions for fixed and free boundary conditions.

Analyzed the effects of different disorder correlator approximations.

Identified issues with harmonic approximation and proposed solutions.

Abstract

We study the $1+1$-dimensional random directed polymer problem, i.e., an elastic string $\phi(x)$ subject to a Gaussian random potential $V(\phi,x)$ and confined within a plane. We mainly concentrate on the short-scale and finite-temperature behavior of this problem described by a short- but finite-ranged disorder correlator $U(\phi)$ and introduce two types of approximations amenable to exact solutions. Expanding the disorder potential $V(\phi,x) \approx V_0(x) + f(x) \phi(x)$ at short distances, we study the random force (or Larkin) problem with $V_0(x) = 0$ as well as the shifted random force problem including the random offset $V_0(x)$; as such, these models remain well defined at all scales. Alternatively, we analyze the harmonic approximation to the correlator $U(\phi)$ in a consistent manner. Using direct averaging as well as the replica technique, we derive the distribution functions ${\cal P}_{L,y}(F)$ and ${\cal P}_L(F)$ of free energies $F$ of a polymer of length $L$ for both fixed ($\phi(L) = y$) and free boundary conditions on the displacement field $\phi(x)$ and determine the mean displacement correlators on the distance $L$. The inconsistencies encountered in the analysis of the harmonic approximation to the correlator are traced back to its non-spectral correlator; we discuss how to implement this approximation in a proper way and present a general criterion for physically admissible disorder correlators $U(\phi)$.