🔬 mesoscale physics

Dynamical Keldysh model for colored noise and finite relaxation rate

This paper extends the Keldysh model of quantum dots under Gaussian noise to include finite correlation times, demonstrating that finite memory fills the pseudogap and narrows the spectral line in multi-component noise systems.

Original author: F. Z. Bendjeddou, D. V. Efremov, M. N. Kiselev

Published 2026-11-06

📖 4 min read☕ Coffee break read

CC BY 4.0

Original author: F. Z. Bendjeddou, D. V. Efremov, M. N. Kiselev

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the author. For technical accuracy, refer to the original paper. Read full disclaimer

In the microscopic world of quantum dots, tiny islands of matter where electrons are trapped, the environment is rarely still. These electrons are constantly buffeted by electric noise from the gates that control them and the surrounding materials. This noise is not a single, uniform hum; it fluctuates in time, sometimes changing rapidly and other times lingering for long periods. Physicists have long understood two extreme ways this noise behaves. In one scenario, the noise changes so quickly that it acts like white noise, simply causing pure dephasing where each level acquires a Lorentzian broadening. In the opposite scenario, the noise is so slow that it appears frozen, splitting the electron's energy level into distinct states separated by a gap, rather than locking it into a single unchanging state. Between these two extremes lies a complex middle ground where the noise has a memory, changing at a rate that is neither instant nor eternal. Understanding how electrons respond to this specific type of "colored" noise, which has a finite memory, is crucial for predicting the behavior of future nanodevices.

A team of researchers has now solved a long-standing puzzle regarding how this noise affects systems with multiple interacting components. In simple quantum dots, the noise might just shift the energy level up or down. However, in more complex arrangements, such as double dots or clusters, the noise can act in different directions simultaneously, coupling through mathematical structures that do not behave like ordinary numbers. When the noise is frozen, this multi-directional action creates a distinct gap in the density of states, a region where the electron is unlikely to exist, surrounded by two peaks of high probability. This gap is a known feature of the "Keldysh model," a theoretical framework used to describe such systems. The question that remained unanswered was what happens when the noise is not frozen but has a finite speed of change. Does the gap simply fill in smoothly, or does something more dramatic occur?

The researchers tackled this by developing an exact mathematical solution for a model where the noise components are independent, random fluctuations that decay over a specific time scale. They treated the problem by mapping the complex, time-dependent interactions of the electron onto a set of damped oscillators, a technique that allowed them to track the evolution of the system with perfect precision. Their work reveals that the presence of any finite memory in the noise, no matter how small, is enough to begin filling the gap. As the noise becomes faster, the two separate peaks that flank the gap first move outward and then merge into a single, central peak. This transition is not a sudden phase change but a smooth evolution of the system's shape. The researchers found that the point at which the two peaks fully merge into one depends on the strength of the noise and the number of components involved, occurring at a specific correlation rate where the noise fluctuations become comparable to the energy splitting caused by the noise itself.

For very slow noise, the gap does not fill in linearly. Instead, the researchers discovered that the density of states at the center of the gap grows according to a specific fractional power of the noise speed. This indicates the emergence of a new energy scale that is much larger than the simple correlation time of the noise. In this regime, the electron sees the noise field diffusing around it, and the filling of the gap is governed by how far the field can wander during the time the electron is near the center. This behavior is distinct from the fast-noise limit, where the system behaves like a simple, motionally narrowed line, similar to how rapid shaking can blur a spinning object into a single image. The study confirms that the complex combinatorial factors previously proposed to describe these systems are exact, validating a specific mathematical approach that had been used as an approximation.

The findings provide a complete picture of how the spectral line shape of a quantum system evolves from a frozen state to a fast, white-noise state. The researchers show that by measuring the shape of the electron's energy distribution—whether it shows two distinct peaks, a filled gap, or a single central peak—one can directly infer the correlation time of the noise affecting the device. This offers a practical way to characterize the noise environment in weakly coupled quantum dots without needing to measure the noise source directly. The work settles the theoretical status of these models, proving that the complex interactions of multi-component noise can be reduced to a single, exact description that captures the full range of physical behaviors from the slowest to the fastest fluctuations.

Technical Summary: Dynamical Keldysh Model for Colored Noise and Finite Relaxation Rate

Problem Statement
The paper addresses the spectral properties of quantum dots and gated nanostructures subjected to electric noise with finite correlation times. While the fast-noise (Markovian) limit results in pure dephasing with Lorentzian broadening, and the slow-noise (frozen field) limit maps onto the exact Keldysh model, the intermediate regime of colored noise with finite memory remains analytically challenging for multi-component systems. Specifically, for $d \ge 2$ components where noise couples via non-commuting (anticommuting) matrices, the time-ordering of vertices prevents standard Abelian solutions. In the frozen limit ( $\gamma \to 0$ ), such systems exhibit a pseudogap in the density of states (DOS) proportional to $|\varepsilon - \varepsilon_0|^{d-1}$ . The central problem is to determine how this pseudogap evolves as the noise correlation time $1/\gamma$ becomes finite, bridging the gap between the static Keldysh solution and the dynamic white-noise limit.

Methodology
The authors present an exact solution for the $d$ -component dynamical Keldysh model driven by Gaussian noise with an Ornstein–Uhlenbeck correlation function, $D(t) = W^2 e^{-\gamma|t|}$ .

Auxiliary Boson Representation: To handle the non-commuting noise vertices $M_j$ (where $\{M_j, M_k\} = 2\delta_{jk}$ ) and the resulting time-ordering issues, the authors map the averaged time-evolution operator $\langle U(t) \rangle$ onto a system of $d$ auxiliary bosonic modes. The noise vertices are represented by operators $B = \sum_j M_j a_j$ , where $a_j$ are bosonic annihilation operators.
Effective Hamiltonian: The problem is transformed into finding the vacuum resolvent of a non-Hermitian Hamiltonian $H = W(B + B^\dagger) - i\gamma N$ , where $N$ is the number operator. The term $-i\gamma N$ accounts for the finite correlation time (damping) of the auxiliary modes.
Continued Fraction Solution: By exploiting the commutation relations of the bosonic operators and the anticommutation of the matrices, the authors demonstrate that the Green function reduces to a single exact continued fraction. The coefficients of this fraction are determined by the combinatorial factors associated with the dimension $d$ (commensurate, incommensurate, or Heisenberg cases).

Key Results
The study yields the complete evolution of the Density of States (DOS) $\rho_d(x, \alpha)$ as a function of the dimensionless correlation rate $\alpha = \gamma/W$ :

Filling of the Pseudogap: For any finite $\gamma > 0$ , the pseudogap present in the frozen limit is filled. The DOS at the band center grows from zero, and the two side peaks (characteristic of the static limit) eventually merge into a single central peak.
Critical Correlation Rate: The transition from a double-peak structure to a single central peak occurs at a critical rate $\gamma_c$ . For $d=2$ , $\alpha_c^{(2)} \approx 1.25$ , and for $d=3$ , $\alpha_c^{(3)} \approx 1.90$ . Near this critical point, the peak position and relative depth vanish smoothly, indicating a crossover rather than a phase transition.
Fast Noise Limit ( $\alpha \gg 1$ ): In the regime of rapid fluctuations, the DOS converges to a Lorentzian profile with a half-width of $dW^2/\gamma$ , consistent with motional narrowing. However, the spectral tails decay as $x^{-4}$ , causing higher-order moments to diverge, unlike the frozen-field case.
Slow Noise Limit ( $\alpha \to 0$ ): The filling of the pseudogap is singular. The central DOS scales as $\rho_d(0, \alpha) \propto \alpha^{(d-1)/3}$ . This fractional exponent reveals a new energy scale $E^* = (\gamma W^2)^{1/3}$ , which is much larger than the correlation rate $\gamma$ . The pseudogap is effectively smeared on this scale $E^*$ , not on $\gamma$ .

Significance and Claims
The paper claims to provide the first exact solution for the dynamical Keldysh model with colored noise for arbitrary $d$ and finite correlation time.

Validation of Sadovskii's Models: The derivation proves that the continued fractions proposed by Sadovskii for pseudogap states are exact in the time domain, not just in the static limit. The coefficients of the derived fraction reproduce the specific combinatorial factors for $d=1, 2, 3$ .
Unified Description: The work unifies the description of noise-induced spectral broadening, connecting the static Keldysh model, the intermediate colored noise regime, and the fast-noise limit within a single analytical framework.
Experimental Relevance: The authors note that the line shape of the DOS, which can be probed via tunneling spectroscopy in weakly coupled dots, serves as a direct measure of the noise correlation time. The transition from a split peak structure to a single narrowed peak provides a signature for the crossover from slow to fast noise dynamics.

The paper concludes that finite memory fundamentally alters the spectral properties of multi-component noise systems, closing the pseudogap and introducing a distinct scaling behavior in the slow-noise regime that differs significantly from simple exponential damping.

Technical Summary: Dynamical Keldysh Model for Colored Noise and Finite Relaxation Rate

More like this