Bootstrapping transport in the Drude-Kadanoff-Martin model
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 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. 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.Where the paper says it
Where the paper says it
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 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. 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.Where the paper says it
Where the paper says it
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 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 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.Where the paper says it
Where the paper says it
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 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. 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].Where the paper says it
Where the paper says it
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 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. 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.Where the paper says it
Where the paper says it
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 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. 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). 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. 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.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
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 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. 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.Where the paper says it
Where the paper says it
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 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. 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).Where the paper says it
Where the paper says it
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 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. 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].Where the paper says it
Where the paper says it
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 (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.Where the paper says it
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 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]. 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].Where the paper says it
Where the paper says it
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 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. Therefore, the cutoff on the DKM model must occur below the microscopic energy
scale, Λ ≲ ε. This fact is illustrated in Fig.Where the paper says it
Where the paper says it
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 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. 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. 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).Where the paper says it
Where the paper says it
Where the paper says it
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 1
arXiv:2509. 18255v2 [hep-th] 14 May 2026Where the paper says it
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
- On the Planckian bound for heat diffusion in insulators
- Viscosity in Strongly Interacting Quantum Field Theories from Black Hole Physics
- Theory of universal incoherent metallic transport
- Planckian Dissipation in Metals
- Relativistic viscous hydrodynamics, conformal invariance, and holography
- An upper bound on transport
- Locality Bound for Dissipative Quantum Transport
- Rigorous bounds on transport from causality
- The Hydrohedron: Bootstrapping Relativistic Hydrodynamics
- The stickiness of sound: An absolute lower limit on viscosity and the breakdown of second order relativistic hydrodynamics
- Fluctuation bounds on charge and heat diffusion
- Thermalization, Viscosity and the Averaged Null Energy Condition
- A Bound on Thermalization from Diffusive Fluctuations
- Bounds on transport from univalence and pole-skipping
- Causality, Analyticity and an IR Obstruction to UV Completion
- The Chiral Lagrangian parameters, $\overline{\ell}_1$, $\overline{\ell}_2$, are determined by the $\rho$--resonance
- Consistency of the Chiral Pion-Pion Scattering Amplitudes with Axiomatic Constraints
- On Renormalization Group Flows in Four Dimensions
- The $a$-theorem and the Asymptotics of 4D Quantum Field Theory
- Snowmass White Paper: UV Constraints on IR Physics
Related papers
- The facts on quantum self-dual Einstein gravity and self-dual Yang-Mills theory
- Quantum Langlands correspondence for Painlev'e II
- Fortuitous BMN Operators from Multiple Cores
- Building far-from-equilibrium effective field theories using shift symmetries
- Inflation, Cosmic Perturbations and Non-Gaussianities
- Wavefunctions of AdS 3 Universes and T-deformed Torus Partition Functions