Abstract
Recent research by Krzakala, Ricci-Tersenghi, and Zdeborova established that the random field Ising model (RFIM) with non-negative interactions and arbitrary external magnetic fields lacks a static spin-glass phase on any lattice. This paper generalizes that proof to the Ginzburg–Landau model—a soft scalar spin version of the Ising model—incorporating both random magnetic fields and random temperature parameters. By demonstrating that the spin glass susceptibility ($\chi_{SG}$) is mathematically bounded by the ferromagnetic susceptibility ($\chi_{F}$), we prove that a spin glass transition cannot occur independently of a ferromagnetic transition.
Introduction
In the study of statistical physics, disordered systems—materials where impurities or random fluctuations influence particle interactions—present significant theoretical challenges. The Random Field Ising Model (RFIM) has long served as the primary framework for understanding these systems. A persistent question in the field has been whether such systems can enter a "spin glass" phase: a state where spins are frozen into a fixed but disordered pattern, distinct from both the ordered alignment of ferromagnetism and the total randomness of paramagnetism.
While rigorous mathematical proofs have previously ruled out a spin glass phase in the standard Ising model (utilizing discrete $\pm1$ spins) under ferromagnetic conditions, some theoretical physicists using perturbative field theory suggested that more complex models might behave differently. Specifically, it was hypothesized that "soft spins" (continuous values) or random local temperature parameters might provide enough degrees of freedom to allow for a spin glass phase. This work rigorously addresses these hypotheses.
The Ginzburg-Landau Model
The research focuses on the scalar Ginzburg-Landau Hamiltonian, which describes soft spins $\phi_i$ as real numbers. The model includes terms for:
- Random Magnetic Fields: Localized external forces acting on individual spins.
- Random Temperature Parameters: Localized fluctuations in the coupling strength.
- Ferromagnetic Interactions: All interaction coefficients $J_{ij}$ are assumed to be non-negative.
The goal is to determine if the randomness in the field and temperature can induce a glassy state despite the underlying ferromagnetic nature of the interactions.
Methodology: Correlation Inequalities
The authors employ a purely mathematical approach, eschewing simulations for a rigorous proof based on correlation inequalities. The proof relies on two primary pillars:
1. Susceptibility Comparison
The study defines and compares two critical responses of the system:
- Ferromagnetic Susceptibility ($\chi_{F}$): A measure of the system's tendency to align spins globally.
- Spin Glass Susceptibility ($\chi_{SG}$): A measure of the system's tendency to freeze into a disordered, stable configuration.
2. Inductive Proof and Inequalities
By generalizing the 2nd Griffiths inequality and utilizing FKG (Fortuin-Kasteleyn-Ginibre) inequalities, the authors prove by induction on the system size $N$ that the correlation between any two spins in this model is always non-negative. This non-negativity is the linchpin of the proof, as it demonstrates that the system lacks the "frustration" necessary to create a spin glass.
Key Insights
The Ferromagnetic Ceiling
The central result of the paper is the derivation of the inequality: $$\chi_{SG}(h, N) \le \chi_{F}(h, N)$$ In statistical mechanics, a phase transition is marked by the divergence of susceptibility (where the value approaches infinity). Because $\chi_{SG}$ is mathematically trapped below $\chi_{F}$, a spin glass transition is impossible unless the system is already undergoing a ferromagnetic transition. Therefore, a standalone spin glass phase cannot exist.
The Role of Frustration
The proof highlights that without "explicit frustration"—situations where the geometry or signs of the interactions prevent spins from satisfying all bonds—a glassy state cannot gain a foothold. By maintaining non-negative $J_{ij}$, the model remains fundamentally driven toward alignment, even if the random fields and temperatures create local disorder.
Soft Spins and Field Theory
This result is significant because it contradicts several high-profile studies that used perturbative field theory to predict a spin glass phase in soft-spin models. The rigorous proof suggests that those earlier predictions were likely artifacts of mathematical approximations, such as the premature truncation of a series expansion.
Conclusion and Limitations
This work provides a definitive "no" to the existence of a spin glass phase in ferromagnetic scalar Ginzburg-Landau models. However, the authors note specific boundaries to this proof:
- Scalar vs. Vector Spins: The proof applies to scalar (one-dimensional) spins. In vector spin models (2D or 3D arrows), negative correlations can emerge even with positive couplings, potentially allowing for different phase behavior.
- Interaction Complexity: The proof assumes pairwise interactions. Systems with 3-spin interactions or higher-order terms may still exhibit spin glass phases on certain lattices.
Ultimately, this research reinforces the power of correlation inequalities in providing definitive answers to long-standing debates in disordered systems, correcting the course of previous field-theoretic assumptions.
Summary updated