Excitation Flow, Positivity, and Fisher Information for Open Subsystems of an $N$-Qubit Network

This paper derives closed-form propagators for open subsystems of an $N$-qubit network with a single excitation, demonstrating that a single transition amplitude governs excitation flow, positivity, entanglement, and Fisher information, while revealing that positivity coincides with complete positivity and is determined solely by the direction of excitation flow toward a fixed point.

The Gist

Imagine a large, closed room filled with $N$ light switches (qubits). In this room, there is exactly one light bulb that is turned on, while all the others are off. The switches are all connected to each other in a complex web, allowing the "on" energy to jump from one switch to another.

The paper by Tommy Chin and Sarah Shandera studies what happens when an observer can only look at a small group of these switches (a subsystem) while the rest of the room remains hidden. They ask: Can we predict how the light moves in the small group just by watching it? And how much can we learn about the whole room just by looking at this small piece?

Here is a breakdown of their findings using simple analogies:

1. The "Flow" of Light Determines the Rules

The researchers found that the entire behavior of any small group of switches is controlled by a single number: the direction the light is flowing.

• Outward Flow (Good News): When the light spreads out from the "on" switch to the others, the physics behaves nicely. The math used to describe the system is "positive" and "completely positive." In everyday terms, this means the rules of probability hold up perfectly; you can't get negative probabilities, and the system behaves predictably.
• Backward Flow (Bad News): Eventually, the light bounces back toward the original switch. When this happens, the math breaks down. The system becomes "non-positive." It's like trying to describe a movie played in reverse where the rules of cause and effect seem to glitch.

The Big Surprise: Usually, in quantum physics, there is a difference between "positive" (rules work for the group) and "completely positive" (rules work even if the group is entangled with something else). The authors found that in this specific network, these two concepts are identical. If the light is flowing outward, everything is fine. If it's flowing backward, everything breaks. It doesn't matter how big your group of switches is; the rule is the same.

2. The "Fixed Point" and the Rubber Band

The authors describe the system as having a "fixed point"—a resting state the system wants to return to.

• Think of the system state as a rubber band attached to a fixed point.
• When the light flows outward, the rubber band contracts. The system is pulled closer to its resting state. This is the "safe" zone where the math works.
• When the light flows backward, the rubber band stretches. The system is pushed away from the resting state. This is the "danger" zone where the math becomes weird (non-positive).

The "Ghost" Zone:
The researchers discovered a strange phenomenon with single switches. There is a specific range of states (a "band" of possibilities) that could mathematically exist without breaking the rules of probability. However, the actual physical light in the room never visits this zone. It's like a hallway that exists on the map but is physically blocked off; the light can never walk down it, even though the doors are theoretically open.

3. Entanglement vs. The Rules

You might think that if the switches are highly "entangled" (deeply connected in a spooky quantum way), the math would break.

• The Finding: The authors found no direct link between how "entangled" the switches are and whether the math breaks.
• The math breaks solely based on which way the light is moving (flowing out or flowing back). You could have high entanglement and perfect math, or low entanglement and broken math. The "flow" is the only thing that matters.

4. Learning About the Whole Room (Fisher Information)

Finally, the paper asks: If I only watch a small group of switches, how well can I guess the rules of the whole room (like how fast the light jumps or how many switches there are)?

They measure this using "Fisher Information," which is like a "sensitivity meter."

• The State Contribution: This is what you learn just by looking at the current position of the light. This information is limited and bounces up and down.
• The Process Contribution: This is what you learn by watching the light move over time. This information grows steadily the longer you watch.

The Connection to the Rules:
The sensitivity meter hits its lowest point exactly when the light is flowing backward (when the math is broken/non-positive). It hits its highest point when the light is flowing outward (when the math is perfect).

• Analogy: Imagine trying to guess the speed of a car by watching it drive. You learn the most when the car is driving smoothly forward (positive math). You learn the least when the car is skidding or reversing (non-positive math), even though the car is still moving.

Summary

The paper shows that for this specific type of quantum network:

1. One rule controls everything: The direction of energy flow determines if the math works or breaks.
2. No middle ground: The system is either "safe" (contracting) or "unsafe" (expanding); there is no gray area.
3. Hidden truths: There are mathematical possibilities that the physical system never actually explores.
4. Learning limits: We learn the most about the global system when the local physics is "well-behaved" (positive), and the least when it is "broken" (non-positive).

This work provides a new way for observers to understand complex quantum systems without needing to control them or know their starting point, relying instead on the "ensemble" of all possible observations.

Technical Summary

Technical Summary: Excitation Flow, Positivity, and Fisher Information for Open Subsystems of an N-Qubit Network

Problem Statement
Observers of autonomous quantum systems lack control over initial states, coupling strengths, or system-environment boundaries, unlike experimentalists who engineer these conditions. Standard open-system dynamics frameworks often assume factorized initial states or completely positive trace-preserving (CPTP) maps, which fail when subsystems are correlated with their environment or when observation windows do not align with the system's natural temporal scales. Consequently, the resulting dynamics are generically non-Markovian and may not be positive or completely positive (CP). This paper addresses the challenge of inferring global properties and characterizing dynamics from partial observations of correlated subsystems within a closed, finite-dimensional quantum network, without privileged initial times or reference states.

Methodology
The authors analyze a closed network of $N$ qubits evolving under a global unitary generated by a homogeneous all-to-all XXZ Hamiltonian that conserves a single excitation (charge $q=1$). The network evolves from a generating product state $|\psi_{\text{gen}}\rangle = |10\dots0\rangle$, where a single excitation is localized on one qubit.

1. Analytical Derivation: The authors derive closed-form propagators $\Phi_{S_i}(t_1, t_2)$ for any $K$-qubit subsystem $S_i$ over an arbitrary time interval $[t_1, t_2]$. They distinguish two dynamical classes: Class 1 (subsystems containing the initially excited qubit) and Class 0 (subsystems excluding it).
2. Operator-Sum Representation: The propagators are expressed in a unified operator-sum form involving block-diagonal and off-block-diagonal operators. The off-block-diagonal terms mediate excitation flow between charge sectors ($q=0$ and $q=1$).
3. Positivity Analysis: The authors test the positivity and complete positivity of these propagators using the Choi matrix and the trace distance criterion relative to fixed points.
4. Information Theoretic Analysis: They compute the quantum Fisher information (QFI) for global parameters (coupling strength $J$ and total qubit number $N$) and decompose it into classical (eigenvalue) and quantum (eigenvector) contributions, as well as state and process contributions over an observation window.

Key Contributions and Results

• Unified Control via Excitation Flow: A single transition amplitude, $u_d(t)$ (or its non-unitary counterpart $\phi_\tau$ in the propagator), simultaneously governs excitation flow between subsystems, the entanglement entropy of every subsystem, the positivity of every propagator, and the QFI for global parameters.
• Coincidence of Positivity and Complete Positivity: For this model, positivity and complete positivity coincide. A propagator is positive and CP if and only if the excitation flow parameter $\phi_\tau(t_1, t_2) \geq 0$. This condition is equivalent to the propagator contracting the subsystem state toward its fixed point (reducing the trace distance to the fixed point).
  • $\phi_\tau > 0$: Excitation disperses away from the excited qubit; the map is CP.
  • $\phi_\tau < 0$: Excitation backflows toward the excited qubit; the map is non-positive and non-CP.
  • This criterion holds independently of subsystem size $K$, coherence, or entanglement structure.
• Ensemble Constraints: The ensemble of propagators collectively constrains global properties inaccessible to any single subsystem. For single-qubit subsystems, the authors identify a "band of states" that lies within the positivity domain of every possible propagator in the ensemble but is never visited by the physical dynamics of the system. This contrasts with CP-divisible dynamics, where any propagator preserves positivity on the full state space.
• Fisher Information Decomposition:
  • State vs. Process: The QFI for global parameters decomposes into a bounded "state" contribution (carried by the state at the start of the window $t_1$) and a "process" contribution (generated by the propagator over $[t_1, t_2]$). The process contribution grows secularly as $t^2$, dominating at long times.
  • Positivity and QFI: The total Fisher information is minimal when future propagators are non-positive/non-CP (during excitation backflow) and near its maximum when they are positive/CP (during excitation dispersal).
  • Non-Additivity: Due to correlations between subsystems and the rest of the network, the Fisher information is non-additive; the information from a partition cannot be determined solely from individual subsystem contributions.
• Entanglement vs. Positivity: The authors demonstrate that the positivity of a propagator is determined solely by the direction of change of the trace distance to the fixed point, not by the magnitude of entanglement entropy or its rate of change. Entanglement entropy depends non-monotonically on the excitation probability, whereas positivity depends strictly on the flow direction.

Significance and Claims
The paper posits that for an observer of an autonomous system, the natural framework is an ensemble of open-system dynamics rather than a single system-environment split. The results provide a benchmark for this framework, showing that:

1. Non-Markovian dynamics naturally include non-positive and non-CP propagators, which are determined by the direction of excitation flow rather than initial correlations alone.
2. The ensemble of propagators imposes consistency constraints that allow an observer to infer global properties (such as network size $N$ and coupling $J$) even when individual subsystems provide incomplete information.
3. The relationship between non-Markovianity (signaled by non-CP maps) and information retrieval (Fisher information) is complex; while non-CP dynamics occur, they do not systematically correlate with the extraction of information in the manner suggested by some prior literature on non-Markovianity.

The work suggests that a general observer framework must abandon privileged initial times and fixed boundaries, instead relying on the consistency of ensembles of propagators to define the space of compatible global quantum systems.