🔢 mathematics

Memory dimension detection in open quantum dynamics via pseudo-control

This paper introduces a pseudo-control protocol that experimentally characterizes the memory dimension of open quantum dynamics by encoding environmental information into an interference matrix, enabling the certification of memory lower bounds without requiring direct access to the environment or full process-tensor tomography.

Original authors: Chang Liu, Alexander Yosifov, Ximing Wang, Zhenhuan Liu, Jinzhao Sun

Published 2026-10-01

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Chang Liu, Alexander Yosifov, Ximing Wang, Zhenhuan Liu, Jinzhao Sun

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the quantum world, nothing exists in total isolation. Every tiny particle, from an electron to a photon, is constantly bumping into its surroundings, exchanging whispers of information with the vast, chaotic environment around it. When a system is open to these interactions, its behavior is shaped not just by its current state, but by the history of those exchanges. This history is known as memory. Just as a person might remember a conversation from an hour ago and let it influence their next reply, a quantum system retains traces of its past interactions with the environment, which then steer its future evolution. Scientists call this phenomenon non-Markovian dynamics. The crucial question is not just whether this memory exists, but how much of it there is. How large is the "storage space" in the environment required to hold these traces? This size, known as the memory dimension, determines how complex the environment is and how difficult it will be to predict or control the system's behavior. Understanding this dimension is vital for building reliable quantum computers and communication networks, yet measuring it has historically been a nearly impossible task because the environment is usually too vast and inaccessible to observe directly.

A team of researchers has now developed a new method to measure this hidden memory without ever needing to look inside the environment itself. Instead of trying to map the entire environment, which would require reconstructing a massive, multi-step process, they devised a clever trick using a concept called pseudo-control. Imagine a system that interacts with its environment twice, with a pause in between. During the first interaction, the system leaves a mark on the environment. During the pause, the environment evolves on its own, potentially spreading or storing that mark. In the second interaction, the system interacts again. The researchers created a protocol where the system is placed in a superposition, effectively experiencing both the first and second interactions simultaneously in a coherent way. By carefully comparing the outcomes of these two parallel histories, they can detect the subtle interference patterns that arise only if the environment retained information from the first event to influence the second.

The core of their discovery lies in analyzing a specific mathematical structure called an interference matrix, which is built from the measurement results of the system and a reference partner. The researchers proved that the complexity of this matrix, specifically its rank, acts as a certificate for the memory dimension. If the matrix is complex enough, it mathematically guarantees that the environment must have a memory of at least a certain size. This provides a lower bound, meaning the actual memory could be larger, but it cannot be smaller than what the measurement reveals. Crucially, this method does not require full knowledge of the environment or the ability to control every single particle within it. It only requires measuring the system's output after these two carefully timed interactions.

To test this idea, the team ran detailed simulations using models of many-body environments, which are complex systems made of many interacting parts, like a chain of atoms. They varied the duration of the interactions, the length of the pause between them, and the strength of the connection between the system and the environment. They found that the detectable memory changes dynamically. When the interaction is very short, little information is exchanged, and the memory appears small. As the interaction time increases, more information is encoded, and the measured memory dimension grows. However, if the interaction lasts too long in a dissipative environment—one where energy is lost to the surroundings—the memory can actually start to fade as the environment scrambles the information. The researchers also discovered that the geometry of the connection matters. When the system touches the environment at only one point, strong coupling can sometimes create a bottleneck, making it harder for information to spread through the rest of the environment, which paradoxically lowers the detectable memory dimension. In contrast, when the system is connected to the environment at multiple points simultaneously, this bottleneck disappears, and the memory dimension remains high even under strong coupling.

The study confirms that the ability to detect memory is not just a fixed property of the environment but depends on how the system interacts with it and how long it waits between interactions. By adjusting these parameters, the researchers could reveal different layers of the environment's structure. For instance, they showed that environments with chaotic, non-repeating internal dynamics retain memory differently than those with simple, predictable patterns. The method successfully distinguished between environments that merely store information and those that actively scramble it. The results suggest that this pseudo-control protocol offers a powerful new tool for characterizing open quantum dynamics. It allows scientists to probe the structure of system-environment interactions through their memory signatures, providing a way to certify the resources needed for simulation and control without the impossible burden of mapping the entire universe of environmental degrees of freedom. This approach opens a path to understanding how complex quantum systems evolve in the real world, where memory and noise are inextricably linked.

Technical Summary: Memory Dimension Detection in Open Quantum Dynamics via Pseudo-Control

Problem Statement
Characterizing open quantum dynamics requires identifying the environmental degrees of freedom that influence a system's evolution. The "memory dimension" ( $d_m$ ) is defined as the Hilbert-space dimension of the smallest effective memory required to reproduce the influence of the environment on the system. While this concept connects physical environmental structure to the resources needed for simulation and control, inferring $d_m$ in real-world scenarios is challenging. Environmental degrees of freedom are typically inaccessible, necessitating indirect inference from system-environment interactions. Existing methods often rely on full process-tensor tomography to reconstruct the complete multi-step input-output map, which is experimentally demanding and scales poorly. The central question addressed is whether accessible output measurements can certify a lower bound on the memory dimension without reconstructing the full multi-step map or accessing the environment directly.

Methodology: The Pseudo-Control Protocol
The authors introduce a "pseudo-control" protocol designed to encode information about the memory dimension into an experimentally accessible interference matrix ( $M_{RS}$ ). The protocol involves a system ( $S$ ) interacting with an environment ( $E$ ) in two distinct interactions ( $C_1$ and $C_2$ ), separated by a waiting interval ( $T$ ) during which the system is decoupled while the environment evolves.

Circuit Architecture: A control qubit ( $c$ ) is prepared in the state $|+\rangle$ . A reference system ( $R$ ) is prepared in a pure state $|\Psi\rangle_{RS}$ entangled with $S$ . Controlled-SWAP gates route the system $S$ through interaction $C_1$ if the control is $|0\rangle_c$ and through $C_2$ if the control is $|1\rangle_c$ . The bypassed interaction in each branch receives a fixed input state.
Interference Matrix Construction: After the interactions and tracing out all discarded outputs, the state retained on the control and the system-reference pair ( $cRS$ ) is measured. The off-diagonal block of this state, $M_{RS} := \omega_{RS}^{01}$ , constitutes the interference matrix.
Operational Reconstruction: $M_{RS}$ is not a direct observable but is reconstructed via measurements of the control qubit in the $X$ and $Y$ bases, followed by conditional tomography on the $RS$ subsystem. Specifically, unnormalized conditional operators $\Omega_\alpha$ are obtained for control outcomes $\alpha \in \{+, -, +i, -i\}$ , and $M_{RS}$ is recovered via the linear combination:
$M_{RS} = \frac{1}{2} [\Omega_+ - \Omega_- - i(\Omega_{+i} - \Omega_{-i})]$
Theoretical Bound: Using tensor-network methods and quantum comb formalism, the authors prove that if the rank of the interference matrix is $r = \text{rank}(M_{RS})$ , the memory dimension of any compatible realization of the process must satisfy:
$d_m \ge \lceil \sqrt{r} \rceil$
This bound is derived by analyzing the contraction patterns of the tensor network, showing that the rank of $M_{RS}$ is limited by the number of independent paths (bonds) crossing the cut between the two interactions, which are determined by the memory dimension $d_m$ .

Key Results
The protocol was applied to numerical simulations of various many-body environments to study how interaction time ( $\tau$ ), interval time ( $T$ ), coupling strength ( $\lambda$ ), and coupling geometry affect the detectable memory.

Interaction and Interval Dependence: The resolved rank $r_\epsilon$ (counting singular values above a threshold $\epsilon$ ) initially increases with interaction time $\tau$ as information is encoded. In dissipative models, $r_\epsilon$ subsequently decreases as $\tau$ increases further due to depolarization. Similarly, increasing the interval time $T$ generally leads to memory decay (lower $r_\epsilon$ ) in dissipative models, though coherent dynamics can temporarily redistribute information, causing local increases in rank.
Coupling Strength and Geometry:
- Boundary Contact: When the system couples only to the boundary of the environment (e.g., the last system qubit to the first environment qubit), the rank increases with coupling strength $\lambda$ initially but decreases at strong coupling. This suggests a "bottleneck" where strong local coupling prevents the rest of the environment from being accessed within the fixed interaction time.
- Ladder Contact: When every system site is coupled to the corresponding environment site (ladder geometry), the rank does not decrease at strong coupling. This confirms that the downturn in the boundary case is due to access limitations rather than intrinsic environmental dynamics.
Dissipation vs. Coherence: Comparing models with and without local depolarization revealed that dissipation suppresses the resolved rank, while coherent dynamics (even in non-integrable systems) allows for broader information spreading and higher detectable ranks.
Reference System Role: Using a maximally entangled reference system (maximizing Schmidt rank) raises the theoretical ceiling for the observable rank from $d_S$ to $d_S^2$ , enhancing the protocol's ability to detect higher memory dimensions.

Significance and Claims
The paper claims to provide a new operational technique for characterizing open quantum dynamics. Its primary significance lies in:

Efficiency: It certifies a lower bound on memory dimension without requiring direct access to the environment or full process-tensor tomography.
Probing Structure: The protocol reveals how system-environment interaction structures (geometry and strength) and internal environmental dynamics shape detectable memory signatures.
Operational Insight: It demonstrates that memory dimension is not a static property but depends on the specific parameters of the interaction (time, coupling) and the measurement strategy (reference system).

The authors conclude that this approach offers a focused method to probe the structure of system-environment interactions through their memory signatures, bridging the gap between physical environmental structure and the resources required for simulation and control. They note that future work could explore simpler circuit implementations to reduce experimental overhead and further clarify how information propagation in many-body environments dictates the required memory dimension.

More like this