⚛️ quantum physics

Correlated Catalysis Removes the Strong Converse for Entanglement Distillation

This paper demonstrates that correlated catalysis eliminates the strong converse for entanglement distillation in all distillable bipartite states, enabling non-vanishing fidelity at any rate exceeding the distillable entanglement limit without increasing the maximum reliable rate.

Original author: Piotr Masajada, Aby Philip, Alexander Streltsov

Published 2026-11-03

📖 4 min read🧠 Deep dive

CC BY 4.0

Original author: Piotr Masajada, Aby Philip, Alexander Streltsov

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 author. For technical accuracy, refer to the original paper. Read full disclaimer

Quantum entanglement is a peculiar connection between particles that defies our everyday experience of how the world works. When two particles are entangled, the state of one instantly influences the state of the other, no matter how far apart they are. This phenomenon is not just a theoretical curiosity; it is the fuel that powers future technologies like ultra-secure communication and quantum computers. However, creating perfect entanglement is difficult. In the real world, noise and interference corrupt these delicate connections, leaving scientists with imperfect, "noisy" pairs. To make them useful, researchers must perform a process called entanglement distillation. This is a method of taking many noisy pairs and, using only local actions and classical communication, squeezing out a smaller number of nearly perfect ones. For decades, a fundamental rule seemed to govern this process: if you tried to extract entanglement faster than a specific maximum speed, the quality of the output would collapse completely, dropping to zero. This was known as the "strong converse," a hard wall that suggested there was no middle ground between success and total failure.

A team of researchers at the Institute of Fundamental Technological Research in Poland has now shown that this hard wall is not as solid as once believed. By introducing a clever trick involving a "catalyst," they demonstrated that it is possible to distill entanglement at rates far above the traditional limit without the quality vanishing. In their work, the scientists used an auxiliary quantum system, which they call a catalyst, to assist the distillation process. Think of this catalyst like a specialized tool that helps a worker assemble a complex machine; the tool is used during the assembly, but when the job is done, the tool is returned in exactly the same condition it started in, ready to be used again. In the quantum version, this tool can be an entangled state that helps transform the noisy input into a better output, yet it emerges from the process unchanged and uncorrelated with the final product.

The researchers found that when this catalyst is allowed to share a tiny, controlled amount of information with the output system, the rules of the game change. Previously, it was thought that if you tried to distill entanglement at a rate higher than the maximum reliable speed, the fidelity—the measure of how close the output is to a perfect state—would drop to zero as the number of particles increased. The new study proves that this is not the case. Instead, the fidelity settles at a steady, non-zero value. Specifically, if you try to distill at a rate that is twice the maximum reliable speed, the quality of the output will settle at half the maximum possible perfection. If you try at ten times the speed, the quality settles at one-tenth. The output never disappears; it simply scales down in a predictable way. This means that the strong converse, the idea that exceeding the limit leads to total failure, has been removed for every type of entangled state that can be distilled.

This discovery does not mean that scientists can now create perfect entanglement at any speed they like. The maximum rate at which they can produce perfect, error-free entanglement remains the same. What has changed is the behavior when they push beyond that limit. Before this work, pushing past the limit was a dead end. Now, it is a path that leads to a lower-quality, but still usable, result. The researchers also calculated exactly how large the catalyst needs to be to maintain this effect. They found that to keep the quality from dropping to zero, the catalyst must grow in size as the number of input particles increases. The size of this catalyst depends on how closely the researchers want to control the tiny correlations between the catalyst and the output. If they want these correlations to be extremely small, the catalyst must be larger, but it does not need to be infinitely large.

The implications of this finding are significant for the future of quantum information. It suggests that the resources required to manipulate quantum states are more flexible than previously thought. The team constructed specific protocols to prove that these results are not just theoretical possibilities but can be achieved with concrete steps. They showed that even when the catalyst is returned exactly as it was, and the connection between the catalyst and the output fades away as the process scales up, the benefit remains. This work overturns a long-held assumption about the limits of quantum communication and resource management. It reveals a landscape where the transition from success to failure is not a cliff edge, but a gradual slope, offering new ways to think about how we might harness the power of the quantum world.

Technical Summary: Correlated Catalysis Removes the Strong Converse for Entanglement Distillation

Problem Statement
Entanglement distillation is a fundamental protocol in quantum information theory that converts noisy shared quantum states into maximally entangled pairs using Local Operations and Classical Communication (LOCC). The performance of this process is characterized by the distillable entanglement, $E_D(\rho)$ , which defines the maximum asymptotic rate at which maximally entangled pairs can be extracted with vanishing error.

A critical question in the asymptotic regime is the behavior of the protocol when the target distillation rate $R$ exceeds $E_D(\rho)$ . For several classes of states (e.g., bipartite pure states and maximally correlated states), the "strong converse" property holds: if $R > E_D(\rho)$ , the fidelity of the output state with the target maximally entangled state vanishes asymptotically as the number of input copies $n \to \infty$ . This implies that allowing a fixed non-zero error does not increase the achievable distillation rate. However, it remains an open question whether the strong converse holds for arbitrary bipartite states under unassisted LOCC.

This paper investigates whether the strong converse property persists when entanglement distillation is assisted by a correlated catalyst. In this framework, an auxiliary entangled state (the catalyst) is provided to assist the transformation. Unlike standard catalysis where the catalyst must be returned uncorrelated with the output, correlated catalysis allows the catalyst to be returned in its exact marginal state while potentially becoming correlated with the output system.

Methodology
The authors analyze the performance of distillation protocols under the following constraints:

Input: $n$ copies of a bipartite source state $\rho$ .
Catalyst: A bipartite catalyst state $\omega_C$ with finite dimensions.
Output: An approximation of $m = \lfloor Rn \rfloor$ Bell pairs.
Catalytic Condition: The catalyst marginal must be returned exactly ( $\Omega_C = \omega_C$ ), and the mutual information between the output system $S$ and the catalyst $C$ must satisfy $I(S:C) \le \delta_n$ , where $\delta_n \to 0$ as $n \to \infty$ .

The study employs two primary approaches:

Upper Bounds: The authors derive upper bounds on the achievable fidelity using the properties of distillable entanglement and the max-relative entropy of entanglement ( $E_{\max}$ ) of the catalyst.
Constructive Protocols: They design explicit correlated catalytic LOCC protocols. These protocols utilize a two-stage process:
1. Amplification: A correlated catalytic transformation converts $n$ copies of $\rho$ into a mixture containing a larger number of source copies ( $e_m > n$ ) and a separable state. This stage utilizes a "batch catalyst" construction involving permutation-invariant states and a classical clock to manage correlations.
2. Distillation: Standard LOCC distillation is applied to the amplified output to generate the target Bell pairs.

The authors also establish lower bounds on the required catalyst resources (specifically the max-relative entropy of entanglement and local dimensions) necessary to sustain non-vanishing fidelity above the reliable rate.

Key Contributions and Results

Removal of the Strong Converse:
The central result (Theorem 2) demonstrates that correlated catalysis removes the strong converse for every distillable bipartite state. For any target rate $R > 0$ , the optimal asymptotic fidelity $F$ is given by:
$\lim_{n \to \infty} F = \min\left(1, \frac{E_D(\rho)}{R}\right)$
Consequently, for any rate $R > E_D(\rho)$ , the fidelity does not vanish but approaches $E_D(\rho)/R$ . This implies that the strong-converse threshold $\tilde{D}_{cc}(\rho)$ is infinite, even when the output-catalyst correlations vanish asymptotically ( $\delta_n \to 0$ ).
Catalyst Resource Bounds:
The paper characterizes the resources required to achieve these fidelities.
- Lower Bounds: If non-catalytic distillation obeys an exponential strong converse (which holds for pure and maximally correlated states), maintaining a fixed fidelity $f_0 > 0$ at a rate $R > E_D(\rho)$ requires the catalyst's max-relative entropy of entanglement to grow at least linearly with $n$ . If the mutual information constraint $I(S:C) \le \delta_n$ is imposed, the required entanglement scales as $\Omega(n \log_2(1/\delta_n))$ .
- Dimension Scaling: This translates to a lower bound on the local catalyst dimensions $D_n$ : $\log_2 D_n = \Omega(n \log_2(1/\delta_n))$ .
- Achievability: Theorem 6 constructs a protocol that achieves a fidelity $f_0$ with catalyst dimensions scaling as $O(n \log_2(1/\delta_n))$ . This matches the lower bound, establishing that the optimal local catalyst memory cost is $\Theta(n \log_2(1/\delta_n))$ qubits per side for fixed fidelity requirements strictly below the optimal asymptotic fidelity.
Distillable Rate vs. Strong Converse Threshold:
The authors clarify that while correlated catalysis allows non-vanishing fidelity at rates exceeding $E_D(\rho)$ , it does not increase the reliable distillation rate. The reliable rate (where fidelity approaches 1) remains $E_D(\rho)$ . The advantage is solely the ability to maintain a positive, non-vanishing fidelity at any finite rate, effectively removing the sharp transition to zero fidelity characteristic of the strong converse.

Significance and Claims
The paper claims to resolve the open question of whether the strong converse holds for entanglement distillation assisted by correlated catalysts. The authors assert that for every distillable bipartite state, the strong converse does not hold in the presence of correlated catalysis, even under strict conditions where the catalyst is returned exactly and correlations vanish asymptotically.

The work highlights a fundamental distinction between the reliable distillation rate and the strong-converse threshold in the catalytic regime. While the reliable rate is invariant ( $E_D(\rho)$ ), the strong-converse threshold becomes infinite. This contrasts with other catalytic enhancements (such as in coherence manipulation under time-translation covariant operations) where the reliable rate itself can be increased.

The authors note that their construction for optimal asymptotic fidelity (Theorem 2) requires a large catalyst memory, while their second construction (Theorem 6) optimizes memory cost for fixed fidelity. They explicitly state that the joint dependence of the optimal memory cost on the fidelity gap and the allowed correlations remains an open question, particularly as the fidelity approaches the theoretical limit $E_D(\rho)/R$ .

The results are derived for finite-dimensional catalysts and rely on the existence of an exponential strong converse for the underlying non-catalytic distillation of the source state (a condition satisfied by pure and maximally correlated states). The paper concludes that correlated catalysis fundamentally alters the asymptotic landscape of entanglement distillation by eliminating the strong converse barrier, allowing for reliable extraction of entanglement at any rate, provided one accepts a fidelity proportional to the ratio of distillable entanglement to the target rate.

More like this