The four-dimensional Anderson model: a case study for critical SPDEs

Abstract: We study the weakly coupled elliptic Anderson model with spatial white noise on the four-dimensional torus, which provides a basic example of a critical SPDE requiring renormalization at arbitrarily high orders. With coupling $λ|\log\varepsilon|<sup>{-\frac12}$ where $λ&gt;0$ is sufficiently small, we prove that the Green's function of the corresponding random Schrödinger operator, suitably centered and rescaled, converges to a centered Gaussian random field with explicit covariance. The main difficulty is that, for such critical models, one must expand up to order $|\log\varepsilon|$, while the perturbative expansion contains factorially many pairings and a growing number of renormalization terms. To overcome this, we construct a truncated renormalized parametrix and prove sharp high-order bounds for its remainder. A central ingredient is a multiscale analysis based on a new version of Hepp trees, combined with new estimates for summations over permutations. These estimates reveal a precise balance between logarithmic losses from scale summation and factorial gains from the structure of primitive pairings. The methods developed here are intended as a first step toward a general theory for critical SPDEs with weak couplings.

• The paper establishes that the normalized deviation of the Green's function converges to a Gaussian random field with an explicitly computed covariance.
• It employs a logarithmic truncation of the chaos expansion to rigorously control factorial combinatorial divergences.
• The study introduces a renormalized parametrix and generalized Hepp trees, setting a new benchmark for the analysis of critical SPDEs.

The Four-Dimensional Anderson Model as a Paradigm for Critical SPDEs

Introduction and Model Formulation

This paper develops a comprehensive analysis of the elliptic Anderson model on the four-dimensional torus $[-\pi, \pi]^4$, focusing on the weakly coupled regime where the random potential is spatial white noise. The coupling constant $\lambda/\sqrt{|\log \varepsilon|}$ is chosen to ensure criticality, in contrast to subcritical regimes where conventional solution techniques for SPDEs are well-developed. The Anderson Hamiltonian is thus given by

$$L_\varepsilon = 1 - \Delta - \lambda_\varepsilon \cdot \xi_\varepsilon + c_\varepsilon\ ,\quad \lambda_\varepsilon = \lambda / \sqrt{|\log \varepsilon|}\ ,$$

where $\xi_\varepsilon$ denotes mollified white noise and $c_\varepsilon$ aggregates renormalization constants determined by the stochastic expansion. The main observable is the Green's function $G_\varepsilon(x, y) = L_\varepsilon^{-1}(x, y)$.

A principal result establishes that, after appropriate centering and normalization, the field $$\lambda_\varepsilon^{-1} \left[ G_\varepsilon(x, y) - G(x, y) \right]$$ converges in law (as $\varepsilon \to 0$) to a Gaussian random field with covariance independent of the noise mollifier, given explicitly by

$$\operatorname{Cov}\big((x_1, y_1), (x_2, y_2)\big) = \frac{2 \pi^2}{2 \pi^2 - \lambda^2} \int_{\mathbb{T}^4} G(x_1 - z) G(y_1 - z) G(x_2 - z) G(y_2 - z)\,dz.$$

This construction provides, for the first time, a rigorous treatment of a critical SPDE in which an infinite sequence of renormalization terms is required and controlled.

Context: Subcritical vs. Critical SPDEs

Much previous work on SPDEs has addressed subcritical equations, where solution theories (regularity structures, renormalization group, paracontrolled calculus, etc.) produce well-posedness via finite expansions and renormalization up to finite order. In contrast, critical equations, such as the four-dimensional Anderson model, $\Phi^4_4$, four-dimensional Yang-Mills, or the 2D KPZ equation, are scaling invariant, requiring infinitely many renormalization corrections, and high-order terms are not regularized by the perturbation series.

This work highlights the essential differences between the four-dimensional Anderson model and celebrated parabolic critical models such as the 2D stochastic heat equation (SHE). For 2D SHE, the "one-sided" nature of the heat kernel eliminates the need for renormalization, and factorial combinatorial divergences do not emerge; thus, full infinite-order expansions are tractable term-by-term. The Anderson model, by contrast, presents $\lambda/\sqrt{|\log \varepsilon|}$0 divergences in the expansion of high-order chaos, making direct infinite series summation infeasible.

Technical Contributions

Parametrix Construction and Truncation at Logarithmic Depth

The authors circumvent the divergent chaos expansion by constructing a renormalized parametrix—i.e., a well-chosen finite expansion up to $\lambda/\sqrt{|\log \varepsilon|}$1 terms—such that the remainder can be sharply estimated. High-order resolvent expansions are evaluated, and the renormalization constants are systematically defined via the combinatorics of pairings and subtraction of divergent subinterval contributions.

The central technical insights are:

• Finite (Logarithmic) Truncation: Expanding chaos series only up to order $\lambda/\sqrt{|\log \varepsilon|}$2 matches the scale at which factorial combinatorial losses can be controlled by compensative logarithmic gains. This principle appears to be universal for critical SPDEs.
• Hepp Trees and Multiscale Analysis: The work utilizes a generalization of Hepp trees to encode the metric structure of spatial configurations at various scales. This gives rise to a precise classification of the relevant diagrams and their associated scale hierarchies.
• Sharp Combinatorial Estimates: Summations over pairings and permutations of expansion indices reveal a fine balance: for each class of pairings, the logarithmic losses due to scale sums can be exactly compensated against the factorial gains from non-primitive (i.e., block-structured or bubble-inserted) pairings.

Diagrammatics and Renormalization

The algebraic and combinatorial machinery developed for the main term analysis is also extended to the analysis of renormalization constants. The renormalization counterterms in the four-dimensional Anderson model are organized via nested intervals and pairings, akin yet more intricate than in BPHZ-type subtractions. The sharp control of high-order renormalization constants emerges as a corollary of the main high-moment estimates.

Numerical and Theoretical Results

• The main result provides explicit scaling limits for the Green function fluctuations: central Gaussian with explicit covariance.
• The required renormalization constant $\lambda/\sqrt{|\log \varepsilon|}$3 is an infinite sum (truncated at depth $\lambda/\sqrt{|\log \varepsilon|}$4), with each term controlled as in the main expansion.
• For small enough $\lambda/\sqrt{|\log \varepsilon|}$5, the method provides control; however, generalizing up to the conjectured Gaussian-non-Gaussian threshold $\lambda/\sqrt{|\log \varepsilon|}$6 appears a substantial open problem.

Broader Implications

The methodology developed in this work extends, in principle, to a broad class of linear and nonlinear critical SPDEs (e.g., $\lambda/\sqrt{|\log \varepsilon|}$7, critical Yang–Mills, etc.) where combinatorial factorial divergences and infinite-order renormalizations are generic. In particular, the duality between combinatorial explosion and probabilistic cancellation, as mediated by the geometry encoded in (generalized) Hepp trees, is expected to be a central technical tool for future research in stochastic quantization and high-dimensional disordered systems.

The techniques may also inform recent developments in kinetic theory and wave turbulence, despite the analytic structure (PDE vs. random walks) being different; in both cases, expansions up to logarithmic order and sharp enumeration of diagrammatic contributions play decisive roles.

Directions for Future Work

• Extension to nonlinear models: The principles and estimates here could, with significant further effort, be generalized to nonlinear critical SPDEs, confronting additional diagrammatic complexities.
• Sharp threshold phenomenon: Establishing the precise phase transition from Gaussian to non-Gaussian fluctuations in this setting (as done for KPZ/SHE) is a compelling, but technically more demanding, open problem.
• Deeper connections to kinetic scaling limits: Comparative analysis with criticality in dispersive and kinetic theories could lead to the unification of techniques and insights between SPDEs and classical kinetic models.
• Rigorous universality: The explicit covariance structure obtained hints at universality phenomena that may be proven more generally for classes of critical models.

Conclusion

This paper provides a rigorous and quantitative framework for the analysis of critical SPDEs, exemplified by the four-dimensional Anderson model. It demonstrates that infinite renormalization and factorial combinatorics can be tamed via parametrix truncation, precision combinatorial analysis, and multiresolution (Hepp tree) methods. The results open the door for systematic study and eventual solution theory for a wide range of critical SPDEs, moving beyond the classical subcritical paradigm. The analytic-combinatorial machinery developed here constitutes a foundational step in this program.

Reference: "The four-dimensional Anderson model: a case study for critical SPDEs" (2607.10105)