⚛️ quantum physics

Sharp error tolerance of the monogamy-of-entanglement game

This paper improves the error threshold for the BB84 monogamy-of-entanglement game from approximately 3.7% to a sharp threshold of roughly 14.64%, thereby strengthening security reductions and noise tolerances for various quantum cryptographic protocols.

Original author: Llorenç Escolà-Farràs, Florian Speelman

Published 2026-11-09

📖 5 min read🧠 Deep dive

CC BY 4.0

Original author: Llorenç Escolà-Farràs, Florian Speelman

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

In the strange and counterintuitive world of quantum physics, there is a rule that limits how much two people can know about a shared secret at the same time. This rule, known as the "monogamy of entanglement," suggests that if two particles are deeply linked, they cannot be equally linked to a third. Imagine two people, Alice and Bob, who share a special connection. If a third person, Charlie, tries to learn what Alice and Bob are doing, the very act of Charlie trying to know something forces the connection between Alice and Bob to weaken. This principle is not just a theoretical curiosity; it is the bedrock of modern quantum cryptography. It allows scientists to build communication systems where any attempt to spy on the message inevitably leaves a detectable trace, guaranteeing that the information remains secure.

For years, researchers have used a specific game based on this principle to test the limits of quantum security. In this game, a referee sends a series of quantum signals to two separated players. The players must guess the outcome of a measurement without talking to each other. If they try to share information beforehand, the laws of physics limit how often they can both be right. For a long time, scientists believed they knew exactly how much noise or error the players could tolerate before the game became impossible to win. However, a new study by Llorenç Escolà-Farràs and Florian Speelman has rewritten this understanding, revealing that the system is far more robust than previously thought.

The researchers focused on a scenario where the players are allowed to make a certain number of mistakes. In the real world, quantum systems are never perfect; they are subject to noise and interference. To make quantum cryptography useful, we need to know exactly how much error a system can handle before it becomes insecure. The old calculations suggested that if the players made more than about 3.7 percent errors, they would be caught. The new paper demonstrates that this limit was far too conservative. By using a more sophisticated mathematical approach, the authors proved that the players can actually tolerate errors up to approximately 14.64 percent. This is a significant jump, nearly quadrupling the amount of noise a secure system can withstand.

To understand how they reached this conclusion, one must look at how the players try to win. In the simplest strategy, the players prepare their quantum states in a specific way that gives them the best possible chance of guessing correctly, even if they are wrong sometimes. This strategy, known as the Breidbart strategy, naturally produces an error rate of about 14.64 percent. The old mathematical tools used to analyze the game were too blunt to distinguish between a strategy attempting to gain an advantage and this natural limit. They relied on a counting method that assumed the players' errors were distributed in a very specific, restrictive way. The new study replaces this old method with a sharper tool that looks at the probability of the players succeeding together across many rounds of the game.

The authors show that as long as the players' error rate stays below that 14.64 percent threshold, the chance of them winning the game by gaining an advantage drops so quickly that it becomes effectively zero. If they try to push the error rate higher, they can win almost every time, but that is only because the rules of the game have changed to allow for that much noise. The critical finding is that 14.64 percent is the precise tipping point. Below it, security is guaranteed; above it, the game becomes trivial. This result is not just a theoretical improvement; it changes the practical design of quantum security protocols.

The implications of this discovery extend to several real-world applications. In quantum position verification, a technique used to prove that a person is physically located at a specific place, the new limit means the system can function correctly even in much noisier environments than previously believed. Similarly, for quantum key distribution, which allows two parties to share a secret code, the new threshold means that the system can tolerate more interference from the environment without needing to discard the data. The study also addresses a more complex version of the game where the two players do not need to give identical answers, but only need to be correct within their own individual error limits. Here, the researchers found that the sum of their error tolerances cannot exceed the same 14.64 percent threshold, further tightening the security guarantees.

What makes this work particularly powerful is that it does not just suggest a new limit; it proves that the limit is sharp. The authors demonstrate that the strategy which naturally produces the 14.64 percent error rate is the best possible way to play the game. This means the new bound is not just a safe estimate, but the exact mathematical boundary of what is possible. By refining the tools used to analyze these quantum games, the researchers have strengthened the foundation of quantum cryptography. They have shown that the universe is more forgiving of noise than we thought, allowing for more practical and resilient quantum communication systems. This work ensures that the security of future quantum networks is built on a more accurate and robust understanding of the fundamental laws of physics.

Technical Summary: Sharp Error Tolerance of the Monogamy-of-Entanglement Game

Problem Statement
The paper addresses the error tolerance limits of the BB84 monogamy-of-entanglement (MoE) game under parallel repetition. In this game, two separated players (Bob and Charlie) share a quantum state with a referee (Alice). Alice measures $n$ qubits in randomly chosen computational or Hadamard bases and asks the players to guess the outcomes without communicating. While the optimal winning probability for error-free parallel repetition is known to be $\lambda^n$ (where $\lambda = \frac{1}{2} + \frac{1}{2\sqrt{2}}$ ), the behavior of the game when players are allowed a bounded number of errors ( $\gamma n$ ) has been less precise.

Previous security analyses, specifically the counting argument in [TFKW13], established an exponential decay in winning probability for error rates $\gamma$ below approximately 3.7%. However, a specific strategy known as the product Breidbart strategy allows players to achieve an error rate of $\gamma^* = 1 - \lambda \approx 14.64\%$ while maintaining a winning probability that does not decay exponentially. This created a significant gap between the certified security threshold (3.7%) and the strategy-based threshold (14.64%). The central problem is to determine the sharp error threshold for parallel repetition and to characterize the asymptotic decay rate of the winning probability below this threshold.

Methodology
The authors employ a combination of strong parallel repetition bounds and a generalized concentration inequality to derive tighter bounds on the winning probability.

Generalized Chernoff Bound: The authors utilize a bound by Panconesi and Srinivasan [PS97], as used by Impagliazzo and Kabanets [IK10]. Unlike standard Chernoff bounds that require independent random variables, this generalized version requires only an upper bound on the probability of simultaneous success for any fixed subset of coordinates.
Strong Parallel Repetition: The paper leverages the known strong parallel repetition result for the error-free BB84 MoE game [TFKW13], which states that the probability of winning on any subset of $s$ coordinates is bounded by $\lambda^s$ .
Synthesis: By defining indicator variables for correct answers and applying the generalized Chernoff bound, the authors replace the independence assumption with the subset success bound derived from strong parallel repetition. This allows them to bound the probability that the number of errors exceeds a certain threshold.

Key Contributions and Results

Sharp Error Threshold: The paper establishes that the sharp error threshold for the BB84 MoE game with identical answer strings is $\gamma^* = 1 - \lambda = \frac{1}{2} - \frac{1}{2\sqrt{2}} \approx 14.64\%$ .
Tight Asymptotic Bound: For any error rate $\gamma < \gamma^*$ , the winning probability for $n$ parallel repetitions is bounded by:
$\Pr[\text{win}] \leq 2^{-n D(\gamma \parallel \gamma^*)}$
where $D$ is the binary relative entropy (Kullback–Leibler divergence) with base-two logarithms. This bound is shown to be tight, as the product Breidbart strategy achieves this asymptotic exponent.
Unequal Error Tolerances: The authors extend the analysis to scenarios where players have different error tolerances ( $\gamma_B$ and $\gamma_C$ ) and are not required to provide identical answer strings. They prove that exponential decay occurs whenever $\gamma_B + \gamma_C < \gamma^*$ . Specifically, if one player must guess exactly ( $\gamma_C = 0$ ), the other player can tolerate any error rate $\gamma_B < \gamma^*$ . However, the paper notes that unlike the identical-answer case, this result does not establish the sharp threshold or optimal exponent for the unequal-error game.
Generalization: The argument is extended to a broader class of monogamy-of-entanglement games where the optimal winning probability for $s$ parallel repetitions is bounded by $p^s$ . The resulting error threshold is $1-p$ .

Significance and Applications
The paper claims that these results strengthen existing security reductions for cryptographic protocols based on the BB84 MoE game. By replacing the loose counting bound ( $2h_2(\gamma)\lambda$ ) with the sharp relative entropy bound, the certified noise tolerances for several protocols are significantly improved:

Position Verification: The certified error tolerance for protocols in the random-oracle model and parallel position verification increases from ~3.7% to any fixed $\gamma < 14.64\%$ .
Spacetime-Constrained Oblivious Transfer (SCOT): The certified tolerance increases from ~1.5% to any fixed $\gamma < 7.32\%$ (where $\gamma^*/2$ applies to the sum of tolerances).
Relativistic Bit Commitment: The security bounds are tightened using the new decay rate.
One-Sided Device-Independent Key Distribution: The guessing estimate for an adversary is sharpened, improving the secrecy bound, though the paper notes that 14.64% is not a key-rate threshold itself, as other factors (parameter estimation, error correction) still apply.

Conclusion
The authors conclude that the product Breidbart strategy is optimal for the BB84 MoE game under parallel repetition with errors. The derived bounds provide the exact asymptotic exponent and the sharp threshold, resolving the gap between previous security proofs and achievable strategies. The paper notes that the main proof idea was assisted by an AI model, but the development into a full note was motivated by the wide applicability of MoE games in quantum cryptography.

Technical Summary: Sharp Error Tolerance of the Monogamy-of-Entanglement Game

More like this