Boundary-driven magnetization transport in the spin-$1/2$ XXZ chain: Role of the system-bath coupling strength and timescales

This study reveals a fundamental discrepancy between closed-system linear-response and open-system boundary-driven approaches to magnetization transport in the spin-1/2 XXZ chain, demonstrating that while their diffusion coefficients agree at intermediate times, they diverge in the thermodynamic limit due to an unavoidable conflict between the order of taking long-time and large-system limits.

The Gist

The Big Picture: Two Ways to Watch Traffic

Imagine you are trying to understand how fast cars (representing magnetism or spin) move through a long, crowded highway (representing a quantum material).

Physicists have two main ways to study this traffic:

1. The "Closed Highway" Method (The Idealist): You imagine the highway is a perfect, isolated loop with no exits or entrances. You watch how a tiny bump in traffic naturally spreads out over time. This is based on Linear Response Theory (or the Kubo formula). It's like watching a ripple in a still pond to understand how water moves.
2. The "Open Highway" Method (The Realist): You imagine the highway has on-ramps and off-ramps at both ends. You force cars in at one end and take them out at the other, creating a constant flow of traffic. This is the Boundary-Driven method using "Lindblad baths." It's like setting up a toll booth system to force traffic to move so you can measure the speed.

The Big Question: Do these two methods give you the same answer about how fast the cars are really moving?

The Discovery: A Surprising Mismatch

The authors of this paper decided to test this on a specific type of quantum highway (the Spin-1/2 XXZ chain). They ran massive computer simulations to compare the results.

What they found was shocking:
The two methods did not agree.

• The "Closed" method gave a steady, reliable speed for the traffic.
• The "Open" method gave a speed that changed wildly depending on how hard they pushed the cars in at the on-ramp (the coupling strength).

It's as if you measured the speed of a river by looking at a calm lake (Method 1) and got 5 mph, but then you measured it by turning on a firehose at the source (Method 2) and got 10 mph, 20 mph, or 5 mph depending on how hard you turned the nozzle.

This is a problem because the "speed of traffic" (the diffusion constant) is a fundamental property of the road itself. It shouldn't change just because you changed how you measured it.

The Detective Work: Why is this happening?

The authors dug deeper to find the culprit. They realized the issue wasn't that the "Open" method was broken, but that they were looking at the wrong time.

Think of it like this:

• The Short-Term (The "Sweet Spot"): When you first turn on the traffic flow, there is a brief window where the cars are moving smoothly. During this time, the "Open" method agrees perfectly with the "Closed" method. It's like the first few seconds after a traffic light turns green; everyone moves at the natural speed of the road.
• The Long-Term (The "Mess"): If you wait too long, the cars start piling up against the off-ramp, creating a traffic jam that depends entirely on how fast you are forcing cars in. The "Open" method eventually settles into a "Steady State" (NESS), but this state is contaminated by the size of the highway.

The "Order of Limits" Problem

The paper explains this using a concept called the "Order of Limits."

Imagine you are trying to measure the speed of a river, but your measuring tape is only 10 feet long.

• Method A (The Open System): You wait until the water is flowing perfectly steady (Long Time), then you try to imagine the river is infinitely long (Large System).
  • Result: Because you waited so long, the water has already piled up against the end of your 10-foot tape. Your measurement is wrong because the tape is too short.
• Method B (The Closed System): You imagine the river is infinitely long first, and then you look at the flow.
  • Result: You get the true, natural speed of the river.

The authors found that in the "Open" method, the "traffic jam" (finite-size effects) takes a very long time to form (scaling with the square of the system size, $L^2$). However, the "natural flow" (the plateau where the methods agree) only lasts for a shorter time (scaling with the size, $L$).

Because computers can only simulate highways of a certain length, the "Open" method usually forces us to wait until the traffic jam forms. We are stuck measuring the jam, not the natural flow.

The Takeaway

1. The "Open" method is tricky: If you use boundary-driven methods to measure how fast things move in quantum materials, you might get the wrong answer if you wait too long. The result will depend on how hard you push the system, which shouldn't happen for a real material property.
2. There is a "Golden Window": If you look at the data early enough (before the traffic jam forms), the "Open" method actually works perfectly and matches the "Closed" method.
3. The Lesson: When studying quantum transport, you have to be careful about when you take your measurement. If you wait for the system to settle into a steady state, you might be measuring the artifacts of your experiment rather than the physics of the material.

In short: The "Open" method is a powerful tool, but it's like trying to measure the speed of a car by watching it drive into a wall. If you watch it hit the wall, you get a messy result. If you watch it drive freely for a short while before it hits the wall, you get the true speed. The authors showed us exactly how long that "free drive" window lasts.

Technical Summary

1. Problem Statement

The study addresses a fundamental discrepancy in the theoretical understanding of transport in quantum many-body systems. Two primary frameworks are used to calculate transport coefficients (specifically the DC diffusion constant, $D_{dc}$):

1. Closed Systems: Based on Linear Response Theory (LRT) and the Kubo formula, analyzing equilibrium correlation functions.
2. Open Systems: Based on boundary-driven dynamics governed by the Gorini–Kossakowski–Sudarshan–Lindblad (GKSL) master equation, where the system is coupled to external baths to induce a Non-Equilibrium Steady State (NESS).

While recent studies suggested a correspondence between the dynamical evolution of density profiles in these two frameworks, it remained unclear whether they yield equivalent transport coefficients. Specifically, the paper investigates whether the $D_{dc}$ extracted from boundary-driven open systems converges to the $D_{dc}$ obtained from closed-system Kubo calculations in the thermodynamic limit, and how these values depend on the system-bath coupling strength ($\gamma$).

2. Methodology

Model System

The authors investigate the spin-1/2 XXZ chain with open boundary conditions:
$$H = J \sum_{r=1}^{L-1} [S^x_r S^x_{r+1} + S^y_r S^y_{r+1} + \Delta S^z_r S^z_{r+1}]$$
They study two regimes:

• Integrable: $\Delta > 1$ (specifically $\Delta = 1.5$ and $3.0$), where transport is diffusive.
• Non-integrable: A perturbed Hamiltonian $H'$ including next-nearest-neighbor interactions ($\Delta' > 0$) to break integrability.
• Conditions: Infinite-temperature limit ($\beta \to 0$) and weak driving ($\mu \ll 1$).

Open System Setup

The system is coupled to Lindblad baths at the boundaries (sites 1 and $L$) via jump operators $L_j$ with rates $\alpha_j$ dependent on coupling strength $\gamma$ and driving $\mu$. The dynamics are governed by:
$$\dot{\rho}(t) = -i[H, \rho(t)] + \sum_j \alpha_j \left( L_j \rho L_j^\dagger - \frac{1}{2} \{L_j^\dagger L_j, \rho\} \right)$$

Numerical Techniques

To solve the GKSL master equation for finite systems ($L=10$ to $60$), the authors employ two state-of-the-art methods:

1. Stochastic Unraveling (SU): Monte Carlo wavefunction technique averaging over $N_{traj} = 50,000$ trajectories.
2. Matrix Product States (MPS): Time-Evolving Block Decimation (TEBD) applied to the vectorized density matrix with bond dimension $\chi = 150$.

Comparison Metrics

• Closed System: $D_{dc}$ calculated via the Kubo formula using the recursion method (Lanczos coefficients) in the thermodynamic limit.
• Open System:
  • Steady State: $D_{dc}^{NESS} = -j_{ness} / \nabla S^z_{ness}$ (derived from the linear slope of the magnetization profile in the NESS).
  • Time-Dependent: $D_r(t) = -\langle j_r(t) \rangle / \langle \nabla S^z_r(t) \rangle$ (local diffusion constant at site $r$).

3. Key Results

A. Mismatch in DC Diffusion Constants

• Strong Coupling Dependence: The $D_{dc}$ extracted from the NESS of the open system shows a strong dependence on the system-bath coupling strength $\gamma$. This is unphysical, as a bulk material property should be independent of boundary coupling details.
• Disagreement with Kubo: The NESS-based $D_{dc}$ values significantly mismatch the closed-system Kubo results. Neither the magnitude nor the $\gamma$-dependence vanishes as the system size $L$ increases (up to $L=60$).

B. Time-Dependent Analysis and Finite-Size Effects

By analyzing the full time dependence of the local diffusion coefficient $D_r(t)$, the authors identify a specific dynamical regime:

• The Plateau: For short-to-intermediate times ($t < t_{fs}$), $D_r(t)$ reaches a plateau.
  • This plateau value is nearly independent of $\gamma$.
  • It shows excellent agreement with the closed-system Kubo result.
• Timescales:
  • The duration of this plateau scales linearly with system size: $t_{fs} \propto L$.
  • The time required to reach the NESS scales quadratically: $t_{NESS} \propto L^2$.
• The "Wrong" Order of Limits: The discrepancy arises because standard boundary-driven studies effectively take the limit $L \to \infty$ after $t \to \infty$ (reaching NESS). However, the correct physical limit for bulk transport is $t \to \infty$ before $L \to \infty$.
  • In the NESS limit ($t \to \infty$), finite-size effects (reflected in the boundary coupling $\gamma$) dominate the result.
  • In the pre-NESS plateau ($t \ll t_{NESS}$), the system behaves as if it were in the thermodynamic limit, recovering the correct Kubo value.

C. Integrable vs. Non-Integrable

These findings hold true for both the integrable ($\Delta' = 0$) and non-integrable ($\Delta' = 0.5$) cases. The qualitative behavior of the plateau and the subsequent deviation in the NESS is consistent across both regimes.

4. Significance and Contributions

1. Resolution of the Equivalence Question: The paper clarifies that while closed and open systems describe the same physics, extracting steady-state transport coefficients from open systems is not reliable for determining bulk diffusion constants in the thermodynamic limit.
2. Diagnostic Tool: The strong dependence of $D_{dc}$ on $\gamma$ in the NESS is identified as a diagnostic signature of finite-size effects, rather than a physical property of the material.
3. Methodological Insight: The study demonstrates that time-dependent diffusion coefficients $D(t)$ extracted from open systems (specifically within the pre-NESS plateau) provide a robust and accurate method to determine transport coefficients that match the Kubo formalism.
4. Limit of Validity: It highlights a critical limitation of the boundary-driven approach: it fails to capture the correct DC limit if the simulation runs long enough to reach the NESS, due to the non-commutativity of the limits $\lim_{L \to \infty}$ and $\lim_{t \to \infty}$.

Conclusion

The authors conclude that the boundary-driven open-system approach is highly valuable for studying transient dynamics and scaling of steady-state currents, but direct extraction of DC diffusion constants from the NESS is flawed due to unavoidable finite-size contamination. To obtain accurate bulk transport coefficients, one must analyze the time-dependent diffusion coefficient within the finite-size scaling window (the plateau) before the system settles into the boundary-dominated NESS.