Large deviations of spectral determinants of matrix-valued random Schrödinger operators and Dyson Brownian motion in cubic potentials

Basic Information

• Paper ID: 2511.00954
• Title: Large deviations of spectral determinants of matrix-valued random Schrödinger operators and Dyson Brownian motion in cubic potentials
• Authors: Yan Fyodorov (King's College London), Pierre Le Doussal (ENS Paris), Alexander Ossipov (King's College London)
• Classification: math-ph, cond-mat.dis-nn, cond-mat.stat-mech, math.MP, math.PR
• Submission Date: November 2, 2025
• Paper Link: https://arxiv.org/abs/2511.00954

Abstract

This paper investigates the moments of spectral determinants $\overline{|\det(H-E)|^q}$ of matrix-valued random Schrödinger operators and their large deviation properties, where $H$ contains the Laplacian and random potential. In one dimension, these are $N \times N$ matrix-valued random Schrödinger operators, where $\log|\det(H-E)|$ equals the sum of $N$ Lyapunov exponents. By mapping to random matrix Riccati equations, the authors establish connections between spectral properties of these operators and the $N$-particle total flux of Dyson Brownian motion (DBM) in cubic potentials. The barrier crossing probabilities of DBM for large but finite $N$ are computed, yielding estimates for exponential tails of the average density of states below the spectral edge of matrix Schrödinger operators. The barrier behaves as $\sim N(-E)^{3/2}$ at large negative energies and vanishes as $\sim N(E^*-E)^{5/4}$ near the edge.

Research Background and Motivation

Research Questions

This paper investigates the statistical properties of spectral determinants of matrix-valued random Schrödinger operators, particularly focusing on:

1. Moments of spectral determinants $|\det(H-E)|^q$ and their growth rates $\Sigma_q$
2. Large deviation properties of the logarithmic spectral determinant $\log|\det(H-E)|$
3. Connections to Lyapunov exponents and Dyson Brownian motion in the one-dimensional case

Significance

1. Physical Applications: These operators naturally arise in multiple physical contexts:
   • Anderson localization: models of multi-channel quasi-one-dimensional systems (wires or strips)
   • Hessian matrices of disordered elastic systems: describing glassy properties
   • DMPK method: mathematical foundation of disordered conductor theory
2. Mathematical Significance:
   • Extends research on spectral determinant moments in random matrix theory
   • Generalization from scalar case ($N=1$) involving generalized Lyapunov exponents
   • Connection to Kac-Rice formula for counting stationary points of energy functionals
3. Complexity Theory: For $q=1$, the quantity $\Sigma_1$ is interpreted as the annealed complexity of stationary points, describing the abundance of mechanical equilibrium states in disordered systems

Limitations of Existing Methods

1. The scalar case ($N=1$) has been studied, but research on matrix-valued cases ($N>1$) is limited
2. Previous work primarily focused on the $q=1$ case
3. Lack of systematic research on general $q$ values and large deviation properties
4. Deep connections with Dyson Brownian motion have not been fully explored

Research Motivation

The authors aim to:

1. Develop two complementary methods for computing $\Sigma_q$: matrix Riccati method and saddle-point method
2. Establish precise correspondence between matrix Schrödinger operators and DBM in cubic potentials
3. Understand the relationship between Lifshitz tails at the spectral edge and DBM barrier crossing
4. Provide independent complexity calculations for disordered elastic systems

Core Contributions

1. Dual Methodology: Develops two independent and complementary methods for computing moment growth rates $\Sigma_q$:
   • Matrix Riccati equation method ($d=1$, arbitrary $N$)
   • Saddle-point method (arbitrary $d$, large $N$ limit)
2. DBM Connection: Establishes profound connections between matrix Schrödinger operators and Dyson Brownian motion in cubic potentials:
   • Proves spectral edge $E^*$ corresponds to DBM phase transition point
   • Establishes correspondence between density of states and DBM particle flux
   • Provides independent verification of self-averaging properties
3. Large Deviation Theory: Obtains rate function $\Phi(e)$ through Legendre transform, where $e = \frac{1}{LdN}\log|\det(H-E)|$: $\Phi(e) = \frac{1}{2}\left(f^{-1}(\delta e + f(\mu/J)) - \mu/J\right)^2$
4. Barrier Crossing Analysis: Computes barrier crossing probabilities of DBM for finite $N$:
   • Near edge: $U \simeq \frac{4}{5}\sqrt{2 \times 3^{1/4}}\beta^{1/6}(E^*-E)^{5/4}$
   • Large negative energies: $U \simeq \frac{4}{3}|E|^{3/2}$
   • Provides Lifshitz tail estimates: $\rho_K(\alpha) \sim \exp(-NU/\tilde{J}^2)$
5. Phase Diagram Characterization: Identifies two phases:
   • Simple phase ($\mu > \mu_b$): zero complexity region (for $q=1$)
   • Complex phase ($\mu < \mu_b$): non-zero complexity, corresponding to spectral interior
   • Phase boundary: $\mu_b = \mu_c - (q-1)J^2\int_k \frac{1}{\mu_c - t\Delta(k)}$

Detailed Methods

Task Definition

Input:

• Matrix-valued random operator $H = K + X + \mu I$, where:
  • $K$: structured operator containing Laplacian and GOE random matrix
  • $X$: diagonal Gaussian noise
  • Parameters: $N$ (matrix dimension), $L$ (system size), $J$ (disorder strength), $c$ (noise correlation parameter)

Output:

1. Moment growth rate: $\Sigma_q = \lim_{L\to\infty} \frac{1}{NL^d}\log\overline{Y_q}$, where $Y_q = |\det(H-E)|^q$
2. Large deviation rate function: $\Phi(e)$ satisfying $P(e) \sim e^{-NL^d\Phi(e)}$
3. Sum of Lyapunov exponents ($d=1$): $Ne = \sum_{j=1}^N \gamma_j$

Method One: Matrix Riccati Method ($d=1$)

Core Idea

Represents the functional determinant of the one-dimensional continuous model as a trace integral of solutions to matrix Riccati equations:

$\det(H-E) = y(L), \quad |y(L)| = e^{\int_0^L d\tau \text{Tr}Z(\tau)}$

where $Z(\tau)$ satisfies the random matrix Riccati equation: $\partial_\tau Z = -E - Z^2 + \tilde{W}(\tau), \quad Z(0) = +\infty$

Eigenvalue Evolution

Introducing eigenvalues $\lambda_i(\tau)$ of $Z$, they satisfy random evolution equations: $d\lambda_i(\tau) = -(E + \lambda_i(\tau)^2)d\tau + \frac{\tilde{J}^2 d\tau}{N}\sum_{j\neq i}\frac{1}{\lambda_i(\tau)-\lambda_j(\tau)} + \sqrt{\frac{2\tilde{J}^2}{N}}dB_i(\tau) + \tilde{J}\tilde{\xi}(\tau)d\tau$

This is precisely the Dyson Brownian motion equation in cubic potential, where:

• Cubic potential: $V(\lambda) = E\lambda + \lambda^3/3$
• Logarithmic repulsive interaction between particles
• Brownian noise and common noise $\tilde{\xi}(\tau)$

Large $N$ Limit: Allez-Dumaz Theory

Define Stieltjes transform of empirical density: $G(z,\tau) = \frac{1}{N}\sum_i \frac{1}{\lambda_i(\tau)-z}$

In the large $N$ limit, $G$ satisfies evolution equation (decomposing $\tilde{\xi}(\tau) = \bar{\xi} + \eta(\tau)/\sqrt{N}$): $\partial_\tau G(z,\tau) = \partial_z[z + (E+z^2)G(z,\tau)] + \frac{1}{2}\tilde{J}^2\partial_z G(z,\tau)^2$

The stationary solution satisfies algebraic equation: $z + (E+z^2)G(z) + \frac{1}{2}\tilde{J}^2 G(z)^2 = \mathcal{J}$

where the integral constant $\mathcal{J}(E)$ has imaginary part giving particle flux: $j = \frac{1}{\pi}\text{Im}\mathcal{J}(E)$

Phase Transition Structure

• Confined phase ($E < E^* = -\frac{3}{4}(2\tilde{J}^2)^{2/3}$): $\mathcal{J}$ is real, zero flux, density has finite support
• Flowing phase ($E > E^*$): $\mathcal{J}$ has imaginary part, non-zero flux, density support is entire real axis

Key identification: $E^* = -\mu_c$ is precisely the spectral edge of the matrix Schrödinger operator

Connection to Density of States

Stationary average trace: $\langle \text{Tr}Z \rangle_{st} = -\text{Re}\mathcal{J}(E-\tilde{J}\bar{\xi})$

Using Gelfand-Yaglom relation: $\lim_{N,L\to\infty} \frac{1}{NL}\log|y(L)| = -\text{Re}\mathcal{J}(E-\tilde{J}\bar{\xi})$

Proves: $\frac{d}{dE}\text{Re}\mathcal{J}(E-\tilde{J}\bar{\xi}) = -\text{PV}\int d\alpha \frac{\rho_K(\alpha)}{\alpha - E + \tilde{J}\bar{\xi}}$

where $\rho_K(\alpha)$ is the average spectral density of operator $K$.

Method Two: Saddle-Point Method (Arbitrary $d$, Large $N$)

Self-Averaging Property

Key observation: In the large $N$ limit, the logarithmic determinant exhibits self-averaging: $\langle e^{q\log|\det[K+X+(\mu-E)I]|}\rangle_{\text{GOE's}} \approx e^{q\langle\text{Tr}\log|K+X+(\mu-E)I|\rangle_{\text{GOE's}}}$

This simplifies the problem to: $Y_q|_{N\gg 1} \sim \prod_x \int_{\mathbb{R}} \frac{d\xi(x)}{\sqrt{2\pi/N}} e^{-NS[\xi]}$

where the action is: $S[\xi] = \sum_x \frac{1}{2c}\xi(x)^2 - \frac{q}{N}\langle\text{Tr}(\log|K+X+(\mu-E)I| - \log|-\Delta+\mu|)\rangle_{\text{GOE's}}$

Saddle-Point Equations

Minimization of action yields saddle-point condition: $\xi^*_q = qf'(\xi^*_q + \mu/J), \quad f(\xi) := \int d\alpha \ln|\alpha+J\xi|\rho_K(\alpha)$

Key property: The saddle point $\xi^*_q$ is independent of position $x$ (proven by convexity argument)

Pastur-Type Self-Consistent Equation

Average spectral density $\rho_K(\alpha)$ is determined through resolvent $ir_\lambda$: $\rho_K(\lambda) = \frac{1}{\pi}\text{Im}(ir_\lambda)|_{\text{Im}\lambda=0^-}$

where $ir_\lambda$ satisfies: $ir_\lambda = \int_k \frac{1}{\lambda + t\Delta(k) - ir_\lambda J^2}$

Phase Structure

Introducing real and imaginary parts $ir_\lambda = x_\lambda + iy_\lambda$, at saddle point $\lambda = -J\xi^*_q - \mu$:

Simple phase ($y=0$): $\xi^*_q = Jq\int_k \frac{1}{\mu_q - t\Delta(k)}, \quad \mu_q = \mu + (1-\frac{1}{q})J\xi^*_q$

For $q=1$: $\Sigma_1 = 0$ (zero complexity)

Complex phase ($y>0$): $1 = J^2\int_k \frac{1}{(\mu_q - t\Delta(k))^2 + J^4 y^2}$$\xi^*_q = Jq\int_k \frac{\mu_q - t\Delta(k)}{(\mu_q - t\Delta(k))^2 + J^4 y^2}$

Phase boundary determined by Larkin mass: $1 = J^2\int_k \frac{1}{(\mu_c - t\Delta(k))^2}$

Growth Rates and Rate Functions

Growth rate: $\Sigma_q = -\frac{1}{2}(\xi^*_q)^2 + qf(\xi^*_q + \mu/J) - q\int_k \log(\mu - t\Delta(k))$

Large deviation rate function (via Legendre transform): $\Phi(e) = \max_q[qe - \Sigma_q] = \frac{1}{2}(\xi^*_q)^2$

where the relationship between $e$ and $q$ is: $e = \partial_q\Sigma_q = f(\xi^*_q + \mu/J) - \int_k \log(\mu - t\Delta(k))$

Technical Innovations

1. Method Equivalence: Proves that two methods (with different limit orderings) yield identical results:
   • Riccati/DBM method: $\lim_{N\to\infty}\lim_{L\to\infty}$
   • Saddle-point method: $\lim_{L\to\infty}\lim_{N\to\infty}$
   • This is non-trivial for matrix-valued operators, as local statistics are sensitive to limit ordering
2. New Barrier Crossing Mechanism: Identifies new universality class near critical point
   • Standard Lifshitz tail: $\exp(-c|E|^{3/2})$ ($N=1$)
   • New exponent: $\exp(-N(E^*-E)^{5/4}/\tilde{J}^2)$ (large $N$, near edge)
   • Arises from unconventional density vanishing at critical point ($\propto (E^*-E)^{3/2}$ rather than typical $\sqrt{E^*-E}$)
3. Zero-Mode Dominance: Proves that in large $N$ limit, fluctuations of $\log\det(H-E)$ are primarily controlled by zero mode $\bar{\xi}$ (exists only when $c>0$)

Experimental Setup

This is a purely theoretical work with no numerical experiments. All results are obtained through analytical calculations and asymptotic analysis.

Model Parameters

• Dimension: $d=0, 1$ and arbitrary $d$
• Matrix size: $N$ (focus on $N\to\infty$ limit)
• System size: $L$ (or $M=L^d$)
• Disorder strength: $J$ (or $\tilde{J}$)
• Correlation parameter: $c\geq 0$ ($c=0$ or $c=1$)

Analytical Verification

Authors verify result consistency through:

1. Comparison of results from two methods at $d=1$
2. Comparison with known scalar case ($N=1$)
3. Comparison with Allez-Dumaz 1 results on DBM
4. Checking physical meaning at special points (e.g., $q=0, 1$)

Analytical Results

Main Analytical Results

$d=0$ Case (Single Particle)

Corresponds to single GOE matrix plus diagonal noise. Spectral density is semicircle: $\rho_K(\alpha) = \frac{1}{2\pi J^2}\sqrt{4J^2-\alpha^2}$

Simple phase ($\mu > (2-q)J$): $\xi^*_q = \frac{2Jq}{\mu + \sqrt{\mu^2 + 4J^2(q-1)}}$$\Sigma_q = -\frac{q(\mu^2 - \mu\sqrt{\mu^2+4J^2(q-1)})}{4J^2(q-1)} + q\log\left(1+\sqrt{1+\frac{4J^2(q-1)}{\mu^2}}\right) - \frac{q}{2}(1+2\log 2)$

Complex phase ($0 \leq \mu \leq (2-q)J$): $\mu_q = \frac{\mu}{2-q}, \quad \xi^*_q = \frac{q}{2-q}\frac{\mu}{J}, \quad y^2 = \frac{1}{J^2}\left(1 - \frac{1}{(2-q)^2}\frac{\mu^2}{J^2}\right)$$\Sigma_q = \frac{q}{2(2-q)}\frac{\mu^2}{J^2} + q\log\frac{J}{\mu} - \frac{q}{2}$

Rate function: For $\mu < 2J$, $\Phi(e) = \frac{\mu^2}{J^2}\varphi\left(\frac{J^2}{\mu^2}\delta e\right), \quad \varphi(x) = 1 + 2x - \sqrt{1+4x}$

$d=1$ Case (Continuous Model)

Larkin mass: $\mu_c = (J/2)^{4/3}$

Phase boundary: $\mu_b = (3-2q)(J/2)^{4/3}$

Key observation: At $q\to 0$, $\mu_b(q=0) = 3(J/2)^{4/3} = -E^*$, exactly the DBM phase transition point!

Simple phase ($\mu > \mu_b$): $\mu_q$ satisfies cubic equation $\mu_q - \frac{J^2}{2}(q-1)\frac{1}{\mu_q^{1/2}} = \mu$

Growth rate: $\Sigma_q = q\left(\mu_q^{1/2} - \mu^{1/2} - \frac{J^2(q-1)}{8\mu_q}\right)$

At phase boundary: $\Sigma_q|_{\mu=\mu_b} = q\left(1 - \sqrt{3-2q} - \frac{q-1}{2}\right)(J/2)^{2/3}$

Complex phase ($\mu < \mu_b$): More complex equations involving integrals $I_1(\hat{y}), I_2(\hat{y})$: $I_1(\hat{y}) = \int_0^\infty \frac{dz}{2\pi\sqrt{z}}\frac{1}{(1+z)^2+\hat{y}^2}, \quad I_2(\hat{y}) = \int_0^\infty \frac{dz}{2\pi\sqrt{z}}\frac{1+z}{(1+z)^2+\hat{y}^2}$

Using identities: $I_2(\hat{y}) \mp i\hat{y}I_1(\hat{y}) = \frac{1}{2\sqrt{1\pm i\hat{y}}}$

Barrier Crossing Results

Effective barrier (confined phase, $a > a^*$): $U = \int_{\gamma_-}^{z_a} d\lambda (\lambda - z_a)\sqrt{(\gamma_- - \lambda)(\gamma_+ - \lambda)}$

where $z_a$ is double root, $\gamma_\pm = -z_a \pm \sqrt{2(a-z_a^2)}$ are other roots.

Explicit calculation gives: $U = \frac{2}{3}a\sqrt{6z_a^2 - 2a} - \beta\sinh^{-1}\left(\sqrt{\frac{4(-z_a)^{3/2}}{\sqrt{\beta}}} - \frac{\sqrt{2}}{2^{3/4}}\right)$

Near-critical behavior ($a \to a^*_+$): $U \simeq \frac{4}{5}\sqrt{2\times 3^{1/4}}\beta^{1/6}(a-a^*)^{5/4}$

Corresponding to energy: $E \to E^*_-$$U \simeq \frac{4}{5}\sqrt{2\times 3^{1/4}}\beta^{1/6}(E^*-E)^{5/4}$

Large negative energies ($a = -E \to +\infty$): $U \simeq \frac{2}{3}a\sqrt{4a} = \frac{4}{3}|E|^{3/2}$

This recovers the Lifshitz tail exponent $3/2$ from the $N=1$ scalar case.

Density of states estimate: $\rho_K(\alpha) \sim \tau_{\text{crossing}}^{-1} \sim \exp\left(-\frac{NU}{\tilde{J}^2}\right)$

where crossing time $\tau_{\text{crossing}} \sim \exp(NU/\tilde{J}^2)$ (effective temperature $T = \tilde{J}^2/N$).

Key Findings

1. New Universality Class: The exponent $5/4$ is characteristic of the large $N$ near-critical region, arising from:
   • Unconventional vanishing of droplet density at critical point: $\rho(\lambda) \propto (\lambda-\gamma_-)^{3/2}\sqrt{\gamma_+-\lambda}$
   • Different from standard semicircle edge with $1/2$ exponent
2. Spectral Edge Correspondence: Rigorously proves $E^* = -3(J/2)^{4/3} = -\mu_c = \alpha^*$ DBM phase transition point = Schrödinger operator spectral edge
3. Flux-Density Connection: Establishes precise correspondence $\rho_K(\alpha) = \frac{1}{\pi}\frac{d}{dE}\text{Im}\mathcal{J}(E)\bigg|_{E=\alpha}$ Generalizes oscillation theorem from $N=1$ to large $N$
4. Self-Averaging Verification: Consistency of two methods independently verifies self-averaging property (for $q=1$ proven by 9,10)
5. Complexity: For $q=1$, $\Sigma_1$ gives total complexity of stationary points in disordered elastic systems, providing independent derivation

Related Work

Spectral Determinants of Random Matrices

• Classical random matrices: Hermitian/unitary matrix characteristic polynomial moments extensively studied 25-35
  • Connections to Riemann zeta function
  • Fisher-Hartwig singularities, Toeplitz determinants
  • Gaussian free field, freezing transition, Gaussian multiplicative chaos
• Non-Hermitian case: Recent extensions 46-51
• Sparse/banded structures: Research in 52-55

Contribution: Spectral determinants of matrix-valued Schrödinger operators previously understudied; this work systematically develops the theory

Scalar Schrödinger Operators ($N=1$)

• Generalized Lyapunov exponents $\Lambda(q)$: 56-58
• Elastic string ($d=1$): Fyodorov et al. 11 for arbitrary $q>-1$
• This work generalizes to matrix-valued case ($N>1$)

Disordered Elastic Systems

• Hessian matrices: 7,8,9,10 studied operators as Hessians of energy functionals
• Complexity for $q=1$:
  • 8: $N\to\infty$, arbitrary $d$
  • 10: Rigorous proof
  • This work: Independent derivation, extension to general $q$

Dyson Brownian Motion

• Standard DBM: Dyson 63, no external potential
• DBM in cubic potential: Allez-Dumaz 1 discovered phase transition
  • This work establishes precise correspondence with matrix Schrödinger operators
  • Computes finite $N$ barrier crossing

DMPK Method

• Multi-channel disordered conductors: Dorokhov 2, Mello-Pereyra-Kumar 3
• Mathematical aspects: 5,6
• Instanton calculations: 13
• This work provides new perspective

Conclusions and Discussion

Main Conclusions

1. Dual Methodology Success:
   • Matrix Riccati/DBM method ($d=1$, arbitrary $N$)
   • Saddle-point method (arbitrary $d$, large $N$)
   • Both yield consistent results at $d=1$, $N\to\infty$
2. Profound Physical-Mathematical Connections:
   • Matrix Schrödinger operator spectral properties ↔ DBM in cubic potential
   • Spectral edge ↔ DBM phase transition point
   • Density of states ↔ particle flux
   • Lifshitz tail ↔ barrier crossing
3. New Universality: The $5/4$ exponent in near-critical region is unique to large $N$ systems
4. Complete Large Deviation Theory:
   • Growth rates $\Sigma_q$ for arbitrary $q$
   • Rate function $\Phi(e)$ via Legendre transform
   • Phase diagram (simple vs. complex phase)
5. Physical Applications:
   • Complexity of disordered elastic systems ($q=1$)
   • Density of states tails in quasi-one-dimensional conductors
   • Matrix generalization of Anderson localization

Limitations

1. Dimension Restrictions:
   • Matrix Riccati method only for $d=1$
   • Higher dimensions only via saddle-point method (large $N$ limit)
2. Parameter Range:
   • Main results for $q > -1$
   • $q \leq -1$ requires additional regularization (not addressed)
3. Finite $N$ Effects:
   • Most explicit results in $N\to\infty$ limit
   • Barrier crossing gives finite $N$ estimates, but based on single-particle approximation
4. Numerical Verification:
   • Pure analytical work, lacks numerical simulation verification
   • Particularly for barrier crossing formulas and $5/4$ exponent
5. Complex Phase Details:
   • Complete analytical solution for $d=1$ complex phase is intricate
   • Paper only provides equation forms

Future Directions

1. Finite $N$ Corrections:
   • Systematic $1/N$ expansion
   • Beyond single-particle barrier crossing approximation
   • Collective excitation contributions
2. Numerical Verification:
   • Direct numerical simulation of matrix Schrödinger operators
   • DBM simulation to verify barrier crossing formulas
   • Test universality of $5/4$ exponent
3. Generalizations:
   • Other symmetry classes (GUE, GSE)
   • Non-Hermitian extensions
   • More general potentials (beyond cubic)
4. Applications:
   • Quantitative predictions for specific physical systems
   • Comparison with experimental data (e.g., mesoscopic conductors)
   • Complexity landscape of glassy systems
5. Mathematical Rigor:
   • Rigorous proof of self-averaging for general $q$
   • Rigorous saddle-point uniqueness arguments
   • Rigorous large deviation principle establishment
6. Boundary Effects:
   • Finite-size corrections
   • Boundary condition effects
   • Open system generalizations

Deep Evaluation

Strengths

1. Methodological Innovation:
   • Dual approach: Two independent methods mutually verify results, enhancing credibility
   • Interdisciplinary bridge: Connects seemingly unrelated fields (random operators, DBM, disordered systems)
   • Technical depth: Matrix Riccati method application demonstrates sophisticated techniques
2. Physical Insights:
   • Phase transition correspondence: Identifying precise correspondence between DBM phase transition and spectral edge is important discovery
   • New universality class: The $5/4$ exponent reveals new physics in large $N$ systems
   • Unified framework: Incorporates multiple physical problems (Anderson localization, elastic systems, conductors) into unified theory
3. Mathematical Rigor:
   • Detailed derivations with clear logic
   • Consistency checks with known results ($N=1$, Allez-Dumaz)
   • Multiple independent arguments (e.g., three ways to identify spectral edge)
4. Completeness:
   • Covers arbitrary $d$ (saddle-point method) and $d=1$ (Riccati method)
   • Systematic development from $d=0$ (single particle) to general dimension
   • Comprehensive analysis including phase diagrams, large deviations, finite $N$ effects
5. Presentation Quality:
   • Clear structure with sufficient introduction
   • Technical details in appendices, main text focuses on core ideas
   • Good balance between physical intuition and mathematical rigor

Weaknesses

1. Missing Numerical Verification:
   • All results purely analytical, no numerical simulation support
   • Particularly the new $5/4$ exponent prediction needs numerical validation
   • Single-particle barrier crossing approximation validity unverified
2. Incomplete Finite $N$ Theory:
   • Main results in $N\to\infty$ limit
   • Barrier crossing gives finite $N$ estimates but based on heuristic arguments
   • Lacks systematic $1/N$ expansion
3. High Technical Barrier:
   • Requires deep background in random matrix theory, stochastic processes, large deviations
   • Some steps (e.g., self-averaging) rely on proofs in other literature
   • Complete analytical solution for $d=1$ complex phase not fully provided
4. Insufficient Application Discussion:
   • Multiple physical applications mentioned but lack concrete quantitative analysis
   • Unclear connection to experimentally measurable quantities
   • Missing estimates for realistic material parameters
5. Generalization Limitations:
   • Primarily GOE symmetry
   • Only $c=0$ and $c=1$ cases, intermediate values not discussed
   • Non-Gaussian noise generalization unclear

Impact

1. Theoretical Contribution:
   • Pioneering: First systematic study of large deviations of spectral determinants for matrix-valued Schrödinger operators
   • Methodology: Matrix Riccati method may inspire research on other problems
   • Conceptual Innovation: DBM-random operator connection opens new research directions
2. Cross-Disciplinary Impact:
   • Random Matrix Theory: Extends spectral determinant research to new operator classes
   • Disordered Systems: Provides new tools for complexity calculations
   • Mathematical Physics: Connects different mathematical structures (Riccati equations, Pastur equations, DBM)
3. Practical Value:
   • Provides theoretical predictions for mesoscopic physics
   • Complexity theory for glassy systems
   • May inspire numerical algorithms (via DBM connection)
4. Reproducibility:
   • High: Analytical derivations detailed, verifiable step-by-step
   • Main equations and results explicitly given
   • Numerical implementation requires specialized knowledge

Applicable Scenarios

1. Direct Applications:
   • Density of states calculation for multi-channel disordered conductors
   • Equilibrium statistics of disordered elastic systems (strings, membranes)
   • Matrix generalization of Anderson localization
2. Method Borrowing:
   • Research on other matrix-valued random operators
   • Large deviation analysis of random systems
   • Problems involving Riccati equations
3. Conceptual Generalization:
   • Applying DBM ideas to other physical problems
   • Exploring saddle-point-zero-mode structure in other systems
   • Investigating phase transition correspondences
4. Theory Development:
   • Generalization to other symmetry classes or dimensions
   • Finite $N$ theory development
   • Connection with replica symmetry breaking theory

Key References

1 R. Allez and L. Dumaz. Random matrices in non-confining potentials. J. Stat. Phys. 160(3), 681–714 (2015) - DBM cubic potential phase transition

7 Y.V. Fyodorov, P. Le Doussal. Manifolds pinned by a high-dimensional random landscape: Hessian at the global energy minimum. J. Stat. Phys 179(1), 176–215 (2020) - Large N limit density of states

8 Y.V. Fyodorov and P. Le Doussal. Manifolds in a high-dimensional random landscape: Complexity of stationary points and depinning. Phys. Rev. E 101(2), 020101 (2020) - Complexity calculation

10 G. Ben Arous, P. Bourgade, and B. McKenna. Landscape complexity beyond invariance and the elastic manifold. Commun. Pure Applied Math. 77(2) 1302–1352 (2024) - Rigorous q=1 proof

11 Y.V. Fyodorov et al. Exponential number of equilibria and depinning threshold for a directed polymer in a random potential. Annals of Physics 397, 1–64 (2018) - N=1 case

62 A. Ossipov. Gelfand-Yaglom formula for functional determinants in higher dimensions. J Phys A 51(49), 495201 (2018) - Matrix Riccati method

Overall Assessment: This is a high-quality theoretical physics/mathematical physics paper making important contributions to spectral theory of matrix-valued random operators. By establishing profound connections with Dyson Brownian motion, the authors not only solve specific technical problems but reveal intrinsic relationships between different mathematical structures. The newly discovered $5/4$ exponent and phase transition correspondence have significant physical implications. Main limitations are lack of numerical verification and incomplete finite $N$ theory. This work is likely to become an important reference in the field and inspire subsequent research.