[Paper Review] Replica Field Theory of the Dynamical Transition in Glassy Systems

Silvio Franz, Giorgio Parisi+2 more|arXiv (Cornell University)|May 26, 2011

Material Dynamics and Properties3 references3 citations

TL;DR

This paper develops a replica field theory for the dynamical transition in mean-field glassy systems, showing that the critical behavior at the Mode-Coupling temperature $T_d$ is governed by a replica-symmetric action with $n=1$ replica, leading to perturbative dimensional reduction. The theory accurately describes numerical simulations and identifies universal scaling laws for susceptibility divergences near $T_d$, with critical exponents matching those of the static theory and finite-size scaling collapse observed in simulations.

ABSTRACT

The critical behaviour of the dynamical transition of glassy system is controlled by a Replica Symmetric action with n=1 replicas. The most divergent diagrams in the loop expansion correspond at all orders to the solutions of a stochastic equation leading to perturbative dimensional reduction. The theory describe accurately numerical simulations of mean-field models.

Motivation & Objective

• To understand the critical behavior at the dynamical transition temperature $T_d$ in mean-field spin-glass models.
• To identify the relevant field theory governing the divergence of correlation lengths at $T_d$, despite activated effects in finite dimensions.
• To establish a connection between the static replica-symmetric field theory and the dynamic response near $T_d$, particularly through susceptibility scaling.
• To validate the perturbative and scaling predictions against numerical simulations of finite-size systems.
• To clarify the role of disconnected and connected correlations in characterizing state multiplicity and internal fluctuations at the transition.

Proposed method

• Formulate a replica-symmetric field theory with $n=1$ replica, focusing on the $m \times m$ block solution with $m=1$ at $T_d$, to describe the onset of exponentially many equilibrium states.
• Identify the relevant field-theory action (1) involving cubic interactions and mass terms, with $\phi_{ab}$ representing fluctuations around the mean-field overlap $q_d^{MF}$, and derive connected and disconnected correlation functions.
• Define three connected correlation functions $G_{SG}$, $G_{th}$, and $G_{q}$ from combinations of $G_1$, $G_2$, and $G_3$ to separate internal state fluctuations from inter-state fluctuations.
• Use finite-time scaling and reparameterization of correlation functions in terms of $C_{av}(t)$ to match perturbative expansions with static field theory predictions.
• Apply scaling ansatz to rescaled susceptibilities $\chi_{het}(C_{av})$ and $\chi_{th}(C_{av})$, leading to scaling laws (12) and (13) with $x = N^{-1/4}(C_{av} - C_p)$, and test data collapse numerically.
• Compare perturbative predictions with numerical data in the $\beta$ and $\alpha$ regimes, showing agreement in the scaling region and evidence that non-perturbative effects appear only on longer time scales.

Experimental results

Research questions

• RQ1 What is the correct field-theory description of the dynamical transition at $T_d$ in mean-field glassy systems, particularly regarding replica symmetry and critical exponents?
• RQ2 How do disconnected and connected correlation functions relate to inter-state and intra-state fluctuations in the critical region near $T_d$?
• RQ3 Can the perturbative expansion of the replica field theory be matched to numerical simulations of finite-size systems, and what scaling laws emerge?
• RQ4 What is the relationship between the dynamic susceptibility and the static field theory action at $T_d$, and how do they scale with system size?
• RQ5 Are non-perturbative effects observable on the same time scale as the critical dynamics, or do they dominate only in the late $\alpha$ regime?

Key findings

• The critical behavior at $T_d$ is governed by a replica-symmetric field theory with $n=1$ replica, confirming that the most divergent diagrams correspond to perturbative dimensional reduction.
• The analytical solution of the model yields $T_d = 1.3420(5)$ and $C_p = 0.750(5)$, which are confirmed by numerical data and used as reference points for scaling analysis.
• Rescaled susceptibilities $N^{-1/2}\chi_{het}(C_{av})$ and $N^{-1/4}\chi_{th}(C_{av})$ collapse onto universal scaling functions $f_{het}(x)$ and $f_{th}(x)$ when plotted against $x = N^{-1/4}(C_{av} - C_p)$, confirming the scaling ansatz.
• The scaling functions diverge as $x \to -\infty$ and decay as $x^{-2}$ and $x^{-1}$ for $f_{het}(x)$ and $f_{th}(x)$, respectively, matching the perturbative predictions (10) and (11).
• Numerical data show that the $\alpha$ regime, where $C_{av} < C_p$, is well described by $\chi_4(C_{av}) \approx N C_{av}(C_p - C_{av})$, suggesting a sharp jump in correlation from plateau to uncorrelated values.
• The time scale of the critical region diverges as $N^{1/4a}$, while the $\alpha$ regime likely scales as $N^{1/4a + 1/4b}$, indicating that non-perturbative effects appear only on longer time scales than the critical dynamics.