Bootstrapping transport in the Drude-Kadanoff-Martin model

arXiv:2509.18255 · hep-th, cond-mat.str-el · Submitted 2025-09-22 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: High Energy Physics: Theory. The day in high-energy theory: gravity, strings, fields and the questions left open.

Kai: I'm Kai.

Mira: And I'm Mira, here with Lev.

Kai: We're looking at the paper: "Bootstrapping transport in the Drude-Kadanoff-Martin model".

Mira: Here's the gist of the paper. The Drude-Kadanoff-Martin model provides constraints on charge transport parameters by applying bootstrap logic to upper bounds on the retarded Green’s function for the charge density This approach establishes a version of the Mott-Ioffe-Regel bound by showing that systems with short mean free paths cannot have conventional Drude peaks Pruning redundant reasoning tokens The paper aims to obtain sharp constraints on transport parameters, specifically the current relaxation timescale τ, charge diffusivity D, and charge compressibility χ, by using upper bounds on the retarded Green’s function for the charge density The model is introduced as a simple low energy and long wavelength description of charge transport parameterized by these quantities The primary tools employed are upper bounds on the retarded Green’s function for the charge density in such a setting How the training signal is built The bootstrap logic is adapted by considering the retarded Green’s function for the charge density operator, GR(ω, k), which plays a role analogous to partial waves in scattering amplitudes The upper bounds on Im GR(ω, k) are obtained by requiring the microscopic ‘high energy’ completion of the theory to be a lattice model with local and bounded interactions Two such bounds are derived: 1.

Kai: The first bound is similar in flavor to the f-sum rule pertaining in lattice models, trading stronger frequency resolution for an inequality instead of an equality.

Kai: The second bound requires the microscopic Hamiltonian to be local and that local terms be bounded by a characteristic energy scale ε, leading to a stronger exponential falloff at large ω.

Paper summary: Kai: So, to summarize this paper "Bootstrapping transport in the Drude-Kadanoff-Martin model," they are using upper bounds on the retarded Green's function for charge density to get sharp constraints on parameters like tau, D, and chi. The core thesis is that you can use this bootstrap logic to connect a simple low energy and long wavelength description of charge transport—the Drude-Kadanoff-Martin model—to the actual microscopic energy and length scales of any lattice model with local and bounded interactions.

Mira: They claim that the Drude-Kadanoff-Martin model cannot be valid at microscopic energy scales because it clashes with the exponential suppression of spectral weight at high frequencies in a lattice model. That’s one big point.

Lev: And they show that under the assumption that low energy dynamics is captured by this model, they get a lower bound on the collective mean free path l, defined as root tau D. This bound relates to how fast quantum many-body thermalization happens and connects transport bounds to Planckian timescales.

Kai: Essentially, the paper shows that these microscopic constraints force the Drude-Kadanoff-Martin model to obey a version of the Mott-Ioffe-Regel bound, which is significant because it links it directly to issues surrounding Planckian dissipation and resistivity saturation.

Mira: The reason this matters is that it gives us a rigorous way to test these transport bounds using general physical principles. It’s not just some rough estimate; it’s built on the structure of how the Green's function behaves.

Lev: And they have two specific bounds derived from requiring the microscopic completion to be a lattice model with local and bounded interactions, one similar to an f-sum rule and another one that requires local terms to be bounded by a characteristic energy scale epsilon.

Kai: So it’s about using those bounds to constrain the parameters of this simple transport model, even though the model itself has limitations when you look at high frequencies.

Mira: Exactly. They are essentially using the structure of how Im GR(omega, k) is constrained by microscopic physics to refine what we can say about how charge moves in these effective theories.

Lev: And they show that extending the model with analyticity properties actually leads to a lower bound on diffusivity, which is l equals root D tau. This re-establishes the mean free path bound in a way that's sharply posed within the model.

Conclusion: Kai: So looking at "Bootstrapping transport in the Drude-Kadanoff-Martin model," the main point is they use upper bounds on the retarded Green's function for charge density to constrain transport parameters like tau, D, and chi by relating them back to microscopic lattice models.

Mira: They found that this forces a version of the Mott-Ioffe-Regel bound onto their simple transport description, and it does so by showing that systems with short mean free paths cannot have conventional Drude peaks in that model.

Lev: It's important because it connects these transport bounds to Planckian dissipation and resistivity saturation issues, which is a real physical connection.

Kai: The authors are essentially providing a rigorous method for establishing constraints on transport by using upper bounds on the retarded Green's function for the charge density, which enables that bootstrap program connecting low energy effective field theories to microscopic lattice models.

Mira: And they conclude by showing that 'good metals' like Fermi liquids and metals with saturating mean free paths at high temperature are compatible with conventional Drude peaks, based on the bounds established in this paper.

Lev: So it’s a result about establishing a lower bound on the mean free path, which is l equals root D tau. It’s a concrete limit derived from the model parameters that you can use for testing systems.

Subham Dutta Chowdhury, Sean A. Hartnoll, Aditya Hebbar, Ruby Khondaker

Abdus Salam International Centre for Theoretical Physics Trieste, University of Cambridge, Rudolf Peierls Centre for Theoretical Physics University of Oxford

hep-th, cond-mat.str-el

Submitted: 2025-09-22

Updated: 2026-11-03

Comments: 21 pages. 3 figures. v2: material re-arranged for clarity, no changes to content. v3: minor improvements to presentation

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 88/100

The gist: The Drude-Kadanoff-Martin model provides sharp constraints on transport parameters by applying bootstrap logic inspired by scattering amplitudes to upper bounds on the retarded Green’s function for…

Key concepts

Drude-Kadanoff-Martin (DKM) model
This is a simple, low energy and long wavelength description used to parameterize charge transport. It describes diffusive charge dynamics by including a single current relaxation timescale, which is then related to optical conductivity via the standard Drude form.
Retarded Green’s function for charge density (GR(ω, k))
This function is treated like partial waves in scattering amplitudes. Upper bounds on its imaginary part are derived by requiring that the microscopic theory, when completed at high energy, must be a lattice model with local and bounded interactions.
Mott-Ioffe-Regel (MIR) bound
This bound connects microscopic lengthscales to transport properties. The paper shows that the constraints derived from the DKM model imply a version of this bound, which states that systems with mean free paths shorter than a certain scale cannot have well-defined Drude peaks.

Terminology

Summary

The Drude-Kadanoff-Martin model provides sharp constraints on transport parameters by applying bootstrap logic inspired by scattering amplitudes to upper bounds on the retarded Green’s function for charge density. p. 0

Where the paper says it

We obtain sharp constraints on these parameters in terms of the microscopic energy and length scales of any underlying lattice model with local and bounded interactions. Our primary tools are upper bounds on the retarded Green’s function for the charge density in such a setting.

This approach allows researchers to identify core physical mechanisms and place fundamental bounds on transport coefficients in terms of microscopic energy and length scales. p. 0
Where the paper says it

These have a similar role to the uncertainty relation bounds in quantum mechanics. That is, they identify core physical mechanisms at work and give intuition about how complicated systems will behave.

Establishing Fundamental Constraints

The paper uses the bootstrap program for scattering amplitudes, adapting its logic to transport by considering the retarded Green’s function for the charge density operator. This object plays a role analogous to partial waves in S-matrix theory. p. 1

Where the paper says it

We will adapt the bootstrap logic to transport by considering the retarded Green’s function for the charge density operator, GR(ω, k). This object will play the role of the partial waves in the previous paragraph.

The upper bounds on the imaginary part of this Green’s function are obtained by requiring the microscopic ‘high energy’ completion to be a lattice model with local and bounded interactions. p. 2
Where the paper says it

Within the integration domain the Green’s function is determined by a low energy effective theory of charge transport. The upper bounds are obtained by requiring the microscopic ‘high energy’ completion of the theory to be a lattice model with local and bounded interactions.

Characterizing the Effective Model

The authors define the Drude-Kadanoff-Martin (DKM) model as an effective low energy and long wavelength description of charge dynamics. This minimal model consists of diffusive charge dynamics supplemented by a single current relaxation timescale. p. 2

Where the paper says it

In §3 we pick an effective low energy and long wavelength description of charge transport, which we call the Drude-Kadanoff-Martin (DKM) model. The model consists of diffusive charge dynamics supplemented by a single current relaxation timescale [59].

The optical conductivity derived from this model is given by the standard Drude form. p. 4
Where the paper says it

The optical conductivity is, using charge conservation ∇ · j + ˙n = 0, σ(Ω) = Im GR jj (Ω, 0) Ω = lim k→0 Ω Im GR(Ω, ⃗k) k 2 = χD 1 + τ 2Ω2 . (5) This is the standard Drude form and for this reason we will refer to (4) as the Drude-KadanoffMartin (DKM) model.

Deriving Bounds on Spectral Weight

Two upper bounds on the spectral weight Im GR(ω, k) are derived using physical inputs similar to the standard f-sum rule and operator growth bounds. These bounds constrain low energy transport parameters in terms of microscopic length and energy scales. p. 2

Where the paper says it

The model consists of diffusive charge dynamics supplemented by a single current relaxation timescale [59]. Using this model to compute the integrals, our bounds constrain low energy transport parameters in terms of microscopic length and energy scales.

The second bound is stronger at high frequencies, offering a more precise exponential falloff at large ω. p. 3
Where the paper says it

For ω ≤ ε the bound (3) is weaker than (2) but, as anticipated above, at large ω it is exponentially stronger. The large ω bound in (3) has a stronger exponential falloff, by a factor of e, than the one given in [57].

Bounding the Collective Mean Free Path

The main result is a bound on the collective mean free path l defined as √Dτ in terms of the microscopic lengthscale a. The resulting inequality (9) provides a precise and general constraint on l/a in terms of temperature, relaxation time, and other parameters. p. 7

Where the paper says it

The first of these is the collective mean free path in terms of the lattice spacing, ℓ/a. The second ratio is the current relaxation time divided by an inverse energy scale.

This bound is shown to imply a version of the Mott-Ioffe-Regel bound for systems with short mean free paths. p. 0
Where the paper says it

Secondly, under the assumption that the low energy dynamics is captured by the model, we obtain a lower bound on the collective mean free path ℓ ≡ √ τD. This bound is shown to imply a version of the Mott-Ioffe-Regel bound: systems with ℓ much shorter than the lattice length scale cannot have conventional Drude peaks.

Physical Regimes and Conclusions

The analysis examines five different regimes to illustrate the physical meaning of the derived inequality (11). The results show that systems with l shorter than the lattice spacing are not consistent with conventional Drude peaks. p. 9

Where the paper says it

In summary, we have shown that systems with ℓ shorter than the lattice spacing are not consistent with the DKM effective low energy description of transport. In particular, they cannot have conventional Drude peaks.

Conversely, ‘good metals’ including Fermi liquids and metals with saturating mean free paths at high temperature are compatible with conventional Drude peaks. p. 9
Where the paper says it

This pertains to both low temperature disordered systems with very short mean free paths, which are in practice insulators, and to high temperature Planckian bad metals. On the other hand, ‘good metals’ including Fermi liquids and metals with saturating mean free paths at high temperature are compatible with conventional Drude peaks (at least according to the bounds we have established here).

The version of the MIR bound established is therefore sharply posed within that model. p. 9
Where the paper says it

We should emphasise that ℓ here is a collective mean free path that is well-defined within the effective low energy DKM model (4). The version of the MIR bound that we have established is therefore sharply posed within that model.

p. 9
Where the paper says it

We should emphasise that ℓ here is a collective mean free path that is well-defined within the effective low energy DKM model (4). The version of the MIR bound that we have established is therefore sharply posed within that model.

Discussion and Future Directions

The work suggests that extending the approach beyond the DKM model to more general charge response forms requires additional input, such as analyticity in frequency or causality constraints. A dual S-matrix bootstrap approach is proposed as a way to obtain both upper and lower bounds by incorporating causality constraints. p. 1

Where the paper says it

Firstly, unitarity of the S-matrix implies that the imaginary part of the partial waves Tl(s) have both a lower and upper bound, 0 ≤ Im Tl(s) ≤ 2, for Mandelstam s variable in the physical region. Secondly, causality implies that Tl(s) is analytic in s over some domain.

Furthermore, incorporating fluctuation corrections could lead to nonlinear constraints on l that yield both upper and lower bounds. p. 1
Where the paper says it

However, the low energy amplitudes are part of a full S-matrix that is subject to several key constraints. Firstly, unitarity of the S-matrix implies that the imaginary part of the partial waves Tl(s) have both a lower and upper bound, 0 ≤ Im Tl(s) ≤ 2, for Mandelstam s variable in the physical region.

Methods for Obtaining Bounds

The derivation involves integrating the spectral weight over a region of (ω, k) space using two different methods. The first bound uses similar physical input to the standard f-sum rule, while the second bound requires that the microscopic Hamiltonian be local and bounded by a characteristic energy scale ε. p. 3

Where the paper says it

The second bound is stronger than (2) at high frequencies, but weaker at low frequencies. It requires that the microscopic Hamiltonian be local and that the local terms in the Hamiltonian be bounded by a characteristic energy scale ε, defined precisely in §7.

The computation of DKM spectral weight is simplified by restricting the momentum integral to k ≤ 1/l. p. 3
Where the paper says it

While there are two factors of the density in the Green’s function (1), one of the inverse factors of ν is removed by the continuum integral over space, ´ d dx = ν P α . In the methods section 7 we derive two upper bounds on the spectral weight Im GR(ω,⃗k).

Causality and Analyticity in Momentum Space

The exponential lightcone enforced by the Lieb-Robinson bound leads to a domain of analyticity in momentum space for complex ω and k. The Fourier transform of the Green’s function must then be analytic in momentum space for complex ω and ⃗k such that Im(ω) > v Im(⃗k) and Im(⃗k) < κ. p. 1

Where the paper says it

This object will play the role of the partial waves in the previous paragraph. The retarded Green’s function has positive imaginary part at real frequencies, 0 ≤ Im GR(ω, k), and is analytic in the upper-half complex ω plane, see e.

This causality constraint is key to obtaining a lower bound on diffusivity, re-instating l = √Dτ. p. 1
Where the paper says it

Such causality arises either as a consequence of Lorentz invariance or from the Lieb-Robinson velocity of local lattice Hamiltonians [20,21]. The simplest transport process, diffusion, is acausal unless the diffusivity obeys an upper bound [22–26].

Continuum UV Completions

The paper notes that both bounds (2) and (3) used the discreteness of the lattice completion in an essential way. A continuum version of bound (3) may be possible, but will need to make sense of the composite operator n(x) squared. p. 8

Where the paper says it

(super-Planckian T τ ≪ 1) (12) This inequality gives a version of the Mott-Ioffe-Regel bound on the mean free path [4, 5]. Conversely, we may conclude that a system with a short low-temperature mean free path, violating (12), cannot have a well-defined Drude peak in the sense we have described.

Relativistic hydrodynamics typically requires a Drude-like relaxation time τ in order to render the theory causal.

Solving Strongly Correlated Transport

A more ambitious goal is to ‘solve’ a specific model by strongly bounding observables on both sides. Constraints from fluctuations, which are inversely proportional to classical transport coefficients, offer another route to additional bounds. p. 1

Where the paper says it

If the fluctuations become too large the system is unable to self-consistently establish local thermal equilibrium. This leads to bounds on transport coefficients [27–30].

Packaging microscopic information recursively into Lanczos coefficients can systematically strengthen the bounds in §2. p. 1
Where the paper says it

This leads to bounds on transport coefficients [27–30]. Other approaches include using analyticity properties of hydrodynamic correlation functions [31] and uncertainty relations applied to microscopic observables in a thermal state [32].

Computation of the DKM Spectral Weight

The DKM spectral weight integral is simplified by using an approximation for Fd(z, λ) when z ≥ 1. This leads to the final bound (9) when maximizing the bound (2) over all frequencies ω. p. 9

Where the paper says it

High temperature bad metals are typically Planckian [13, 15, 16], illustrated in the middle plot of Fig. Using the high temperature behaviour χ ∼ k d F /T together with Planckian dissipation τ ∼ 1/T in the bound (9) leads to a ≲ ℓ.

This maximization shows that any tension with the bound must therefore come at intermediate frequencies, as illustrated in Fig. p. 5
Where the paper says it

Therefore, the cutoff on the DKM model must occur below the microscopic energy scale, Λ ≲ ε. This fact is illustrated in Fig.

2.

Summary of Findings

Systems with l shorter than the lattice spacing are not consistent with the DKM effective low energy description of transport. In particular, they cannot have conventional Drude peaks. p. 9

Where the paper says it

In summary, we have shown that systems with ℓ shorter than the lattice spacing are not consistent with the DKM effective low energy description of transport. In particular, they cannot have conventional Drude peaks.

This pertains to both low temperature disordered systems with very short mean free paths, which are in practice insulators, and to high temperature Planckian bad metals. p. 9
Where the paper says it

In particular, they cannot have conventional Drude peaks. This pertains to both low temperature disordered systems with very short mean free paths, which are in practice insulators, and to high temperature Planckian bad metals.

On the other hand, ‘good metals’ including Fermi liquids and metals with saturating mean free paths at high temperature are compatible with conventional Drude peaks. p. 9
Where the paper says it

This pertains to both low temperature disordered systems with very short mean free paths, which are in practice insulators, and to high temperature Planckian bad metals. On the other hand, ‘good metals’ including Fermi liquids and metals with saturating mean free paths at high temperature are compatible with conventional Drude peaks (at least according to the bounds we have established here).

The version of the MIR bound established is therefore sharply posed within that model.

Acknowledgements

This work has been partially supported by STFC consolidated grant ST/T000694/1. SAH is partially supported by Simons Investigator award. SDC & AH are supported by “Exotic High Energy Phenomenology” (X-HEP), a project funded by the European Union – Grant Agreement n. p. 0

Where the paper says it

1 arXiv:2509. 18255v2 [hep-th] 14 May 2026

101039756. Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or the ERC Executive Agency (ERCEA).

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

1

Start with constraints derived from upper bounds on spectral weight, such as equation (3) of Section 2, to determine if a given effective low energy description like the Drude-Kadanoff-Martin model can be a microscopic model. This allows AI systems to immediately reject inconsistent microscopic descriptions based on the required exponential falloff at high frequencies.

2

Implement lower bounds on the collective mean free path, such as equation (9), to establish a version of the Mott-Ioffe-Regel bound for transport parameters. This capability enables AI systems to identify bad metals or systems with short mean free paths that cannot support conventional Drude peaks in their effective low energy description.

3

Develop a mechanism to systematically improve bounds by incorporating information from microscopic theory, such as Lanczos coefficients, into the bootstrap program. This allows the AI system to recursively strengthen constraints on transport parameters at low frequencies, leading to more precise bounds.

Abstract

The Drude-Kadanoff-Martin model is a simple low energy and long wavelength description of charge transport, parameterised by the current relaxation timescale τ, charge diffusivity D and charge compressibility χ. We obtain sharp constraints on these parameters in terms of the microscopic energy and length scales of any underlying lattice model with local and bounded interactions. Our primary tools are upper bounds on the retarded Green's function for the charge density in such a setting. We first note that the Drude-Kadanoff-Martin model cannot pertain at microscopic energy scales because it is inconsistent with the exponential suppression of spectral weight at the highest frequencies in a lattice model. Secondly, under the assumption that the low energy dynamics is captured by the model, we obtain a lower bound on the collective mean free path sqrt τD. This bound is shown to imply a version of the Mott-Ioffe-Regel bound: systems with much shorter than the lattice length scale cannot have conventional Drude peaks, in the sense of being captured by the DKM model.

Sources

Related papers