Limitations on quantum key repeaters for all key correlated states

1. Problem Statement

The paper addresses a fundamental open problem in quantum information theory: determining the maximum rate at which a secure cryptographic key can be generated between two distant nodes (Alice and Bob) in a Quantum Key Repeater network.

• Context: Quantum repeaters are essential for the future Quantum Internet to overcome signal attenuation in optical fibers without violating the no-cloning theorem. Unlike classical repeaters that amplify signals, quantum repeaters use entanglement swapping and distillation.
• The Specific Challenge: Previous bounds on the key repeater rate (specifically for "key correlated states") relied on the assumption that the "key-attacked state" (the state resulting from measuring the key part in the computational basis) is separable (i.e., contains no entanglement and is totally insecure).
• The Bottleneck: Checking whether a general quantum state is separable is an NP-hard problem. This makes previous bounds computationally intractable for arbitrary mixed states. The authors aim to relax this assumption to provide a bound that holds for a broader class of states without requiring the separability check.

2. Methodology

The authors employ a combination of information-theoretic bounds, random matrix theory, and resource theories.

• Relaxation of Separability Assumption: Instead of requiring the key-attacked state $\hat{\rho}$ to be separable, the authors utilize the sandwiched Rényi relative entropy of entanglement ($\tilde{E}_\alpha$). This allows them to measure the "distance" of a state from the set of separable states without explicitly solving the separability problem.
• Strong Converse Bounds: The derivation relies on the strong converse bound for private key distillation. This principle states that if one attempts to distill a key rate exceeding a certain threshold, the fidelity of the output state with a maximally entangled state (singlet) drops exponentially to zero.
• Technical Lemmas:
  • Lemma 1: Establishes an upper bound on the fidelity of a state (after a repeater protocol) with a singlet, penalized by the sandwiched Rényi entropy.
  • Lemma 2: Relates the relative entropy distance from a "cone" of states (mixtures of the target state and separable states) to the sandwiched Rényi entropy.
• Random Matrix Theory: For the analysis of random private bits, the authors utilize the Hilbert-Schmidt distribution and properties of Ginibre matrices (random matrices with i.i.d. Gaussian entries). They analyze the asymptotic behavior of eigenvalues and entropy as the dimension of the shielding system ($d_s$) approaches infinity.
• Resource Theory of Private Randomness: The paper applies results from the distillation of private randomness (independent states) to bound the key content of generic local independent bits.

3. Key Contributions

A. A Generalized Bound on Key Repeater Rates

The primary contribution is a new upper bound on the one-way key repeater rate $R_{C_1C_2 \to A:B}$ for arbitrary key correlated states.

• The Bound:
  $$R \leq \inf_{\alpha \in (1, \infty)} \left[ \frac{\alpha}{\alpha - 1} E_D^{\to}(\rho) + \min\{\tilde{E}_\alpha(\hat{\rho}), \tilde{E}_\alpha(\hat{\rho}')\} \right]$$
  Where:
  • $E_D^{\to}$ is the one-way distillable entanglement.
  • $\hat{\rho}$ is the key-attacked state.
  • $\tilde{E}_\alpha$ is the sandwiched Rényi relative entropy of entanglement.
• Significance: This bound does not require $\hat{\rho}$ to be separable. It generalizes previous results (Christandl & Ferrara) by replacing the strict separability condition with a computable entanglement measure.
• Limit Case: As $\alpha \to \infty$, the bound recovers the previous result where the second term vanishes if $\hat{\rho}$ is separable (since $\tilde{E}_\infty$ is zero for separable states).

B. Bound for Random Private Bits

The authors analyze "random private bits" (states containing ideal keys). They prove that for a randomly chosen private bit $\gamma_{rand}$:
$$R \leq 2 E_D^{\to}(\gamma_{rand}) + 1$$
This implies that the "extra" key generated by the repeater beyond twice the distillable entanglement is bounded by a constant (1 bit), regardless of the dimension of the shielding system.

C. Bound on Distillable Key of Random States

Using techniques from random matrix theory, they derive a tighter bound on the distillable key ($K_D$) of a random private bit:
$$K_D(\gamma_{rand}) \leq 1 + \frac{1}{4 \ln 2} \approx 1.360674$$
This result is independent of the shield dimension and suggests that generic random private bits do not possess arbitrarily large distillable key content.

D. Private Randomness in Independent States

The paper extends the analysis to local independent states (states containing ideal private randomness). They show that the achievable rate of localizable private randomness $R_A$ for a generic independent bit approaches:
$$R_A \leq 1 + \frac{1}{2 \ln 2}$$
This holds across four different scenarios involving free/no communication and free/no noise.

4. Key Results

1. Generalization: The new bound (Eq. 6) is valid for a strictly larger class of states than previous bounds because it avoids the NP-hard separability problem.
2. Constant Penalty: For random private bits, the repeater rate is limited to $2 E_D^{\to} + 1$. The "shield" (the part of the state protecting the key) does not allow for an unbounded increase in the repeater rate as its dimension grows; the excess is capped at 1 bit.
3. Mutual Information Limit: The mutual information of a random private state is bounded by $2 + \frac{1}{2 \ln 2}$. Since the distillable key is bounded by half the mutual information (via squashed entanglement), this leads to the $1.36$ bound on $K_D$.
4. Achievability: The derived bounds on private randomness are shown to be achievable for asymptotically large shielding dimensions in all considered communication/noise scenarios.

5. Significance

• Theoretical Advancement: The work bridges the gap between the theoretical limits of quantum repeaters and the computational complexity of verifying state properties. By using Rényi entropies, it provides a practical tool for bounding key rates in complex, noisy quantum networks.
• Quantum Internet Security: As the Quantum Internet relies on repeaters to distribute keys over long distances, understanding the maximum possible key rate is crucial for protocol design. This paper establishes that even with complex shielding and randomization, the key rate cannot exceed specific limits determined by distillable entanglement and a small constant.
• Randomness Generation: The results on private randomness provide insights into the fundamental limits of generating secure random numbers in distributed quantum systems, showing that generic states have a finite "randomness capacity" relative to their dimension.
• Future Directions: The authors suggest that their approach, based on strong-converse bounds, can be extended to two-way repeater protocols and other strong-converse bounds (e.g., for privacy amplification), potentially leading to even tighter limits.

In summary, this paper provides a robust, computationally feasible framework for bounding the performance of quantum key repeaters, demonstrating that the "boost" provided by quantum repeaters is strictly limited and does not scale infinitely with the complexity of the quantum states involved.