Quantum Dynamical Entropy and non-Markovianity: a collisional model perspective

This paper demonstrates that the Alicki-Lindblad-Fannes (ALF) dynamical entropy, derived from multi-time correlation functions in a collisional model with a classical spin chain environment, serves as a quantitative measure linking the statistical properties of the environment to the activation and super-activation of non-Markovian memory effects in open quantum systems.

The Gist

The Big Picture: A Quantum Detective Story

Imagine you are trying to figure out why a cup of coffee is cooling down.

• The Old Way (Markovian): You just look at the coffee. You see it getting colder and assume it's just losing heat to the room. You think, "Okay, the past doesn't matter; it's just cooling down right now." This is called Markovian behavior—no memory.
• The New Way (Non-Markovian): But what if the room has a weird, bouncy floor? The coffee cools, but then the floor bounces the heat back into the cup, making it warm up for a second before cooling again. The coffee "remembers" the heat it lost. This is Non-Markovian behavior—there is memory.

The Problem: In the quantum world, we usually only look at the "coffee" (the open system). We can't see the "floor" (the environment). Because we can't see the environment, we often miss the fact that information is flowing back and forth. We think the system is just forgetting things, but it might actually be remembering them.

The Solution in this Paper: The authors, Giovanni Nichele and Fabio Benatti, developed a new "detective tool" called ALF Entropy (named after scientists Alicki, Lindblad, and Fannes). This tool doesn't just look at the coffee; it looks at the pattern of measurements we take over time to see if the system is truly forgetting or if it's holding onto secrets from the environment.

---

The Analogy: The "Blindfolded Juggler"

To understand their method, imagine a juggler (the Quantum System) trying to juggle balls in a room full of mirrors (the Environment).

1. The Setup (Collisional Model):
   The juggler doesn't just juggle alone. Every second, a new mirror slides in, the juggler bumps into it, and then the mirror slides away. This is the Collisional Model. The juggler interacts with a fresh piece of the environment every tick of the clock.

2. The Blindfold (Reduced Dynamics):
   Usually, we only watch the juggler's hands. We can't see the mirrors. If the juggler drops a ball, we assume it's gone forever. This is the standard way physicists study open systems.

3. The New Tool (ALF Entropy):
   The authors say, "Wait! Let's look at the sequence of drops and catches."

   • High Entropy (Forgetting): If the juggler is just dropping balls randomly and never catching them, the pattern is chaotic. We learn a lot of new information every second because the future is unpredictable. This is like a system with no memory.
   • Low Entropy (Remembering): If the juggler starts catching balls in a perfect, rhythmic pattern, the future becomes predictable. We stop learning new information because we can guess what happens next. This is low entropy.

The Twist: In the quantum world, if the entropy drops to near zero, it doesn't mean the juggler is perfect. It might mean the mirrors are bouncing the balls back to the juggler in a very specific way. The system is "remembering" the environment so well that it stops generating new randomness.

The "Super-Activation" Surprise

The paper discovers something truly weird, which they call Super-Activation.

Imagine you have two identical jugglers.

• Juggler A is doing great; he never drops a ball.
• Juggler B is also doing great; he never drops a ball.

If you watch them separately, you see no "memory effects" or weird bouncing. They look perfectly normal.

But, if you put them side-by-side and let them interact with the same weird room of mirrors, something magical happens. Suddenly, they start dropping balls and catching them in a chaotic, bouncing rhythm that neither of them could do alone.

The paper shows that memory effects can be "super-activated." A system that looks like it has no memory (no information flowing back) can suddenly show massive memory effects when you look at it in a specific way (like looking at two copies together). The "ALF Entropy" tool is the only one sensitive enough to spot this hidden connection.

Why Does This Matter?

1. Better Quantum Computers: Quantum computers are very fragile. They lose information to their environment (noise). If we can understand how information flows back (memory), we might be able to use that "echo" to fix errors or protect our data.
2. New Definition of "Memory": The paper argues that the old way of defining memory (just checking if the system gets "less predictable") isn't enough. We need to look at the information rate (the entropy). If the information rate drops, it means the environment is talking back to the system.
3. The "GNS" Construction: The authors use a fancy mathematical trick (GNS) to imagine the environment as a "shadow twin" of the system. By studying this twin, they can see the memory effects that are invisible when looking at the system alone.

Summary in One Sentence

This paper introduces a new way to measure how much a quantum system is "remembering" its environment by analyzing the patterns of information over time, revealing that even systems that look like they are forgetting everything might actually be deeply connected to their surroundings in ways we couldn't see before.

Technical Summary

1. Problem Statement

The paper addresses the challenge of characterizing non-Markovianity in open quantum systems. Traditional approaches often rely on the reduced dynamics of the system alone (e.g., checking for Complete Positive (CP) divisibility or information back-flow via trace distance revivals). However, the authors argue that these reduced descriptions lose crucial statistical information contained in multi-time correlations.

The core problem is to find a measure that:

1. Captures the full statistical properties of the joint system-environment dynamics.
2. Distinguishes between Markovian and non-Markovian behaviors more robustly than reduced dynamics alone.
3. Quantifies the "information back-flow" from the environment to the system in a way that accounts for the measurement process inherent in quantum mechanics.

2. Methodology

The authors employ an algebraic approach to quantum dynamical systems, utilizing Collisional Models and Quantum Dynamical Entropy.

• Collisional Models: The open system ($S$) interacts sequentially with individual sites of an infinite environment ($E$), modeled as a classical spin chain. The global dynamics is a unitary automorphism $\Theta$ composed of a local interaction $\Phi$ and a shift operator $\sigma_E$.
• Quantum Regression (QR): Instead of looking at the reduced map $\Lambda_t$, the authors focus on multi-time correlation functions of system observables. They define a system as QR-Markovian if these correlations can be expressed entirely in terms of a sequence of Completely Positive and Unity-preserving (CPU) propagators acting only on the system.
• Alicki-Lindblad-Fannes (ALF) Entropy: The authors adapt the ALF entropy (originally for reversible systems) to open, dissipative systems.
  • They construct a coarse-grained density matrix $\rho_S[X^{(n)}]$ based on repeated measurements (POVMs) on the system at discrete time steps.
  • The Open-System ALF Entropy ($h_S(\Theta)$) is defined as the asymptotic rate of the von Neumann entropy of this coarse-grained matrix.
• GNS Construction: To provide an operational interpretation, the authors map the algebraic setup to the Gelfand-Naimark-Segal (GNS) Hilbert space. This allows them to view the entropy production as the entropy exchanged with a measurement apparatus during repeated measurements intertwined with unitary evolution.

3. Key Contributions

1. Definition of Open-System ALF Entropy: The paper extends the ALF entropy framework to dissipative open systems, defining it specifically over the subalgebra of the open system ($A_S \otimes 1_E$), rather than the full system-environment algebra.
2. QR-Markovianity as a Benchmark: The authors prove that if a system satisfies the Quantum Regression formula (QR), the open-system ALF entropy is strictly positive (unless the dynamics is unitary/isometric). Conversely, a decrease in this entropy rate below the QR-Markovian bound serves as a witness for information back-flow.
3. Upper Bound Theorem: They derive a general upper bound for the ALF entropy of a system coupled to a quantum spin chain:
   $$h_S(\Theta) \leq S_{\omega_E} + \log(D)$$
   where $S_{\omega_E}$ is the mean entropy rate of the environment. This implies that stronger correlations in the environment (lower $S_{\omega_E}$) lead to lower entropy production in the system.
4. Super-Activation of Memory Effects: The paper demonstrates that memory effects (non-Markovianity) can be "super-activated." A system might show no back-flow in its reduced dynamics (P-divisible) but exhibit strong back-flow when viewed through the GNS dilation (a two-qubit system), which the ALF entropy captures.

4. Key Results

The authors explicitly compute the ALF entropy for a qubit coupled to a classical Markovian spin chain with specific correlation parameters ($\Delta$).

• Entropy vs. Correlations: The calculated dynamical entropy $h_S(\Theta)$ is a monotonically decreasing function of the environmental correlations ($\Delta$).
• Vanishing Entropy in Highly Non-Markovian Regimes: In the limit of maximal environmental correlations ($\Delta = p = 1/2$), the system's dynamical entropy drops to zero.
  • Paradox: In this regime, the reduced dynamics is highly non-Markovian (showing trace-norm revivals/back-flow), yet the entropy production is zero, mimicking a reversible closed system.
  • Interpretation: Zero entropy production implies that repeated measurements on the system yield no new information asymptotically. The system's state is so predictable (due to strong memory) that the "ignorance" does not increase, effectively signaling that information has flowed back from the environment.
• Divisibility vs. QR: The study reveals a hierarchy:
  • CP-divisibility $\implies$ P-divisibility $\implies$ No back-flow (in reduced dynamics).
  • However, the system can be P-divisible (no back-flow in reduced trace distance) while violating QR.
  • In the GNS representation (two-qubit dilation), the dynamics can lose P-divisibility even when the single-qubit reduced dynamics remains P-divisible, demonstrating super-activation of memory effects.

5. Significance

• Beyond Reduced Dynamics: The work establishes that reduced dynamics (CPTP maps) are insufficient to fully characterize non-Markovianity. Multi-time statistics and the ALF entropy provide a more complete picture of the system-environment interplay.
• Operational Meaning of Entropy: By linking ALF entropy to the GNS construction and measurement apparatus, the paper provides a physical interpretation: Dynamical entropy measures the rate of information extractable from the system. A decrease in this rate indicates that the environment is "feeding" information back to the system, making the system's future more predictable than a Markovian process would suggest.
• Super-Activation: The paper provides a concrete example where memory effects are invisible to standard reduced dynamics but become apparent when considering the system's purification (GNS dilation). This has implications for quantum information processing, where memory effects might be hidden in standard metrics but crucial for error correction or state discrimination.
• Theoretical Framework: It bridges the gap between algebraic quantum statistical mechanics (Kolmogorov-Sinai entropy) and modern open quantum system theory (collisional models, process tensors), offering a rigorous tool to analyze non-Markovianity in complex environments.

In summary, Nichele and Benatti demonstrate that Quantum Dynamical Entropy serves as a powerful, operationally meaningful witness for non-Markovianity. They show that high environmental correlations can suppress entropy production to zero, signaling a regime where information back-flow is so dominant that the system behaves as if it were reversible, despite being an open, dissipative system.