⚛️ quantum physics

Fully quantum fidelity-based entropic continuity bounds

This paper introduces dimension-independent continuity bounds for quantum conditional entropies based on fidelity, offering improved performance over traditional trace distance methods for applications like quantum key distribution, while also extending these results to infinite-dimensional systems and various Rényi entropy variants.

Original authors: Matthew Simon Tan, Ernest Y. -Z. Tan

Published 2026-10-20

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Matthew Simon Tan, Ernest Y. -Z. Tan

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 quiet, high-stakes world of quantum cryptography, the safety of a secret message depends on a fundamental question: how much uncertainty exists in the data? Imagine two people trying to share a secret key over a public channel. To ensure no eavesdropper has listened in, they must measure the "randomness" of their shared quantum states. If the randomness is high, the key is secure; if it is low, the secret might be compromised. For decades, scientists have used mathematical tools called entropy to measure this uncertainty. However, real-world devices are never perfect. They produce states that are slightly flawed or "noisy" compared to the ideal theoretical models. The critical challenge has always been determining how much the calculated security changes when the actual state is just a tiny bit different from the perfect one. If a small physical error leads to a massive, unpredictable drop in the estimated security, the entire system becomes unreliable.

For years, researchers relied on a specific way of measuring how different two quantum states are, known as trace distance. While useful, this method often acted like a magnifying glass that distorted reality, suggesting that tiny, harmless imperfections caused huge, catastrophic drops in security estimates. This led to overly pessimistic conclusions, forcing engineers to design systems that were far more conservative than necessary. A more natural way to measure the closeness of quantum states exists, called fidelity, which better reflects how these states actually behave in many practical scenarios. Yet, the mathematical rules that connect this better measurement to security estimates were missing or incomplete, particularly for the complex, infinite-dimensional systems often found in real-world quantum networks.

In a recent study, researchers Matthew Simon Tan and Ernest Y.-Z. Tan have filled this gap by developing a new set of mathematical limits that connect quantum security to this more natural measure of closeness. They proved that if two quantum states are close according to fidelity, their security estimates will also be close, and they did so without the results exploding into nonsense when the systems become very large or even infinite in size. Their work provides a much tighter, more accurate bound for the uncertainty in these systems. This means that for applications like quantum key distribution, scientists can now calculate security rates that are far more realistic and less pessimistic than before. Instead of assuming a small error destroys the key, their new formulas show that the impact is manageable and predictable, allowing for more efficient and robust quantum communication systems.

The researchers focused on a specific family of entropy measures that are crucial for analyzing the security of quantum protocols over a finite number of rounds. They tackled the problem by constructing a smooth path between two different quantum states and calculating how the entropy changes as one moves along this path. By using a sophisticated mathematical technique involving the geometry of quantum states, they were able to bound the rate of change. A key breakthrough was their ability to handle situations where the system being measured has an infinite number of possible states, a common scenario in real-world optics and communication channels. Previous methods often failed or required assumptions that the system was small and finite, but this new approach works regardless of the system's size, provided the states are close enough in the fidelity sense.

The study also addressed a specific limitation found in earlier work: many existing formulas only worked well when one part of the system was classical, meaning it behaved like a standard computer bit. The new results apply even when both parts of the system are fully quantum, which is the general case for most advanced cryptographic protocols. Furthermore, the authors showed that when the system does happen to be classical in one part, the bounds can be made even sharper, offering even greater precision for specific types of hardware. They verified that their formulas hold true even when the "conditioning" system—the part of the system that helps decode the message—has a countably infinite number of dimensions, a scenario that was previously difficult to analyze rigorously.

Beyond the core entropy bounds, the paper also explored variations of these formulas that are directly useful for the numerical calculations used by engineers designing quantum networks. In these practical applications, the quantity of interest is often not the entropy itself, but a specific exponential function of it. The researchers provided tailored versions of their bounds for these functions, ensuring that the improved accuracy carries through to the final security calculations. By offering these specific variants, they have given the community the tools needed to implement these tighter bounds in real-world optimization software, potentially leading to higher key rates and more secure connections.

The work stands as a rigorous mathematical proof rather than a simulation or a suggestion. The authors have demonstrated that these new bounds are mathematically sound and hold under the strictest conditions, including infinite-dimensional systems. They explicitly showed that the older methods based on trace distance were unnecessarily pessimistic in many cases, and their new fidelity-based approach corrects this without sacrificing mathematical rigor. While they noted that for very small system sizes, the improvement over a specific older method might be slight, the new formulas are consistently tighter for larger systems and, crucially, they apply to the general quantum case where previous methods failed. This provides a solid foundation for the next generation of quantum security proofs, ensuring that the theoretical limits of secrecy match the practical realities of building quantum networks.

Technical Summary: Fully Quantum Fidelity-Based Entropic Continuity Bounds

Problem Statement
In quantum information theory, particularly in the security proofs of Quantum Key Distribution (QKD), it is often necessary to bound the difference in entropy between two quantum states that are "close" to one another. This is formalized through entropic continuity bounds. Existing bounds typically rely on the trace distance as the metric of closeness. However, trace-distance-based bounds suffer from a significant limitation: they grow rapidly with the distance, leading to overly pessimistic estimates of entropy changes. Furthermore, in many QKD scenarios (e.g., device imperfections or dimension-reduction techniques), the natural measure of closeness is fidelity, not trace distance.

While fidelity-based bounds exist, they are currently limited by one or more of the following constraints:

They apply only to states with classical conditioning registers (limiting their utility in fully quantum QKD protocols).
They depend on the dimension of the conditioning system (rendering them unusable when the dimension is unknown or infinite).
They apply only to specific structured states.

This work addresses the need for continuity bounds that are based on fidelity, apply to fully quantum states (where the conditioning system $B$ can be quantum), and are independent of the dimension of the conditioning system.

Methodology
The authors develop a framework to derive continuity bounds for the sandwiched Rényi conditional entropy ( $\tilde{H}^\uparrow_\alpha(A|B)$ ) and the von Neumann conditional entropy ( $H(A|B)$ ). The core methodology involves:

Path Integration: Utilizing the fundamental theorem of calculus, the difference in entropy between two states $\rho_{AB}$ and $\sigma_{AB}$ is bounded by integrating the derivative of the entropy along a smooth path connecting the two states.
Uhlmann Purifications: To minimize the length of the integration path, the authors employ the "reduced great-circle path" between Uhlmann purifications of the states. The length of this path corresponds to the angular distance $\Lambda(\rho, \sigma) = \arccos F(\rho, \sigma)$ , a metric derived from fidelity.
Danskin's Theorem: To handle the optimization over the conditioning system $B$ inherent in the definition of conditional entropy, the authors apply Danskin's theorem. This allows the derivative of the entropy to be expressed in terms of the derivative of the divergence at the optimizer.
Variance Bounds: The derivative is bounded by an "entropy variance" term. The authors derive dimension-independent upper bounds for this variance (specifically the $\alpha$ $α$ -variance) by relating it to exponentiated Petz-Rényi divergences.
- For general states, the bound depends on the dimension of the system $A$ ( $d_A$ ).
- For classical-quantum states (where $A$ is classical), the bound is sharpened, depending on $\sqrt{d_A}$ .
Limiting Arguments: The results are extended to cover:
- The limit $\alpha \to 1$ (recovering von Neumann entropy).
- Countably infinite-dimensional conditioning systems ( $B$ ) via approximation by finite-dimensional sequences.
- The "endpoints" $\alpha = 1/2$ and $\alpha = \infty$ .

Key Contributions and Results

Dimension-Independent Fidelity Bounds: The paper presents a continuity bound for $\tilde{H}^\uparrow_\alpha(A|B)$ that depends on the angular distance $\Lambda$ and the dimension of the unconditioned system $A$ only. Crucially, the bound is independent of the dimension of the conditioning system $B$ .
- Main Result (Theorem 1): For $\alpha \in (1/2, 1) \cup (1, \infty)$ and $q = (\alpha-1)/\alpha$ :
  $|\tilde{H}^\uparrow_\alpha(A|B)_\rho - \tilde{H}^\uparrow_\alpha(A|B)_\sigma| \leq 2 \Lambda(\rho_{AB}, \sigma_{AB}) \frac{\sinh(|q| \text{arccosh}(d_A^\chi))}{|q|}$
  where $\chi = 1/2$ if states are classical on $A$ , and $\chi = 1$ otherwise.
- In the limit $\alpha \to 1$ , this yields a bound for the von Neumann entropy involving $\ln(d_A^\chi + \sqrt{d_A^{2\chi}-1})$ .
Sharpening for Classical-Quantum States: The authors demonstrate that if the state is classical on register $A$ , the bound can be strictly improved by replacing $d_A$ with $\sqrt{d_A}$ in the argument of the hyperbolic functions.
Infinite-Dimensional Validity: The authors verify that these bounds hold even when the conditioning system $B$ has countably infinite dimension (separable Hilbert space), a critical feature for QKD applications involving continuous variables or unbounded dimensions.
Variants for QKD Applications: Recognizing that QKD security proofs often utilize exponentiated forms of Rényi entropies (e.g., $R_\alpha(A|B) = e^{-q \tilde{H}^\uparrow_\alpha(A|B)}$ ), the authors provide specific continuity bounds for these quantities (Corollaries 3 and 4).
- Corollary 3 provides a concave, non-decreasing upper bound on $R_\alpha(A|B)_\rho$ in terms of $R_\alpha(A|B)_\sigma$ , facilitating numerical optimization in security proofs.
- Corollary 4 provides a direct additive bound on the difference $|R_\alpha(A|B)_\rho - R_\alpha(A|B)_\sigma|$ .

Significance and Claims
The authors claim that their results offer improved performance for QKD security proofs compared to previous trace-distance-based bounds. By utilizing fidelity (the natural metric in many QKD contexts) and removing the dependency on the conditioning system's dimension, the bounds avoid the "unnecessarily pessimistic" entropy changes associated with trace distance.

The paper explicitly notes that while the bounds for the von Neumann entropy ( $\alpha=1$ ) are slightly weaker than a specific existing bound [23] for very small dimensions ( $d_A=2,3$ ) in the classical-classical case, they provide tighter results for $d_A \geq 4$ and, more importantly, extend to the fully quantum case where the conditioning system $B$ is quantum. The work aims to provide a rigorous, versatile toolset for analyzing finite-key QKD scenarios where states are perturbed or approximated.

Technical Summary: Fully Quantum Fidelity-Based Entropic Continuity Bounds

More like this