Quantum randomness under unambiguous attacks
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Quantum Radio. Generated commentary on the latest quantum physics and condensed matter papers.
Kai: We're looking at the paper: "Quantum randomness under unambiguous attacks".
Mira: Here's the gist of the paper. Intrinsic randomness is generated when a quantum state is measured in any basis in which it is not diagonal <ref:2601.16343#pg2>.
Kai: The paper introduces and solves problems concerning the unambiguous randomness of a quantum state and measurement, as well as randomness with a fixed rate of inconclusive outcomes <ref:2601.16343#pg3>.
Mira: These results are significant because they quantify the knowledge gained by an eavesdropper about a secret key without causing disturbance in specific quantum key distribution scenarios <ref:2601.16343#pg6>.
Kai: Defining Unambiguous Randomness and its Bounds The maximal unambiguous randomness of any state is found to be proportional to the smallest eigenvalue of the state <ref:2601.16343#pg5>.
Title and authors: Kai: So we're looking at this paper called "Quantum randomness under unambiguous attacks" by Fionnuala Curran and her team. It sounds like it's diving into how much randomness you can actually extract from a quantum state when someone is trying to guess your measurement outcomes.
Mira: Yeah, the title itself really sets the stage, focusing on this specific idea of unambiguous randomness versus just regular guessing probabilities. It’s about quantifying what an eavesdropper knows without making any errors in her guesses.
Lev: From my side, I'm interested in how they set up these optimization problems for unambiguous and fixed rate of inconclusive outcomes randomness because that has direct implications for the overhead on real hardware when we try to implement these concepts.
Kai: Exactly. It moves beyond just asking the basic question about intrinsic randomness and starts looking at scenarios where the eavesdropper can either be perfect or she gets to choose when to stop guessing, which is a more realistic setup for some protocols than assuming perfect accuracy.
Mira: And the paper does this by introducing two main scenarios: one with no errors but an inconclusive outcome allowed, and another with a fixed rate of inconclusive outcomes, or FRIO randomness.
Lev: That’s where I see the engineering challenge—modeling that fixed rate of inconclusive outcomes requires you to know exactly how much noise is present in the system to predict if that outcome will be conclusive or not.
Kai: Right, so it’s not just about how good the eavesdropper is at guessing, but also about what kind of measurement setup we have and what kind of noise we are dealing with.
Mira: It really frames the problem as a trade-off: you can get more certainty by allowing less certainty in your measurements, or vice versa.
Lev: And that trade-off is where the real work lies for error correction; if you can't predict the inconclusive rate accurately, your security estimates become unreliable.
The paper's summary: Kai: The paper itself tackles solving two big problems: first, finding the maximal unambiguous randomness for any quantum state optimized over all possible measurements, and second, solving both problems for dimension two systems and isotropically noisy states measured in an unbiased basis of any dimension.
Mira: That’s a pretty broad scope. They find that the maximal unambiguous randomness is actually directly proportional to the smallest eigenvalue of the state, which is a very specific piece of information about the state itself.
Lev: If that proportionality holds, it suggests that for certain states, you don't need to check every possible measurement basis; you just need to look at what the state is doing in its own natural direction.
Kai: It’s interesting because they also tackle the fixed rate of inconclusive outcomes randomness, or FRIO randomness, which is when you fix a parameter Q for how often the eavesdropper can be inconclusive.
Mira: And they give an expression for that FRIO randomness involving Q and other state parameters, which is useful for practical scenarios where you can't guarantee perfect guessing.
Lev: But I wonder about the practical application here; if we’re building a real quantum key distribution system, how much does this mathematical expression actually simplify the actual hardware requirements compared to just using standard bounds?
Kai: Well, Section three point two applies their results for dimension two systems and noisy states measured in an unbiased basis to describe an eavesdropper's attack on a noisy QRNG used for selecting random bases in a prepare-and-measure quantum key distribution protocol <ref:2601.16343#pg1,measured in an unbiased basis>.
Mira: That’s the core application—quantifying the knowledge gained by an eavesdropper about the secret key when you're using a noisy device to generate random bases.
Lev: So it means we can use this math to bound how much information an attacker can get without triggering an alarm in a system that already has some imperfections.
The paper's improvements: Kai: One key improvement they suggest is the way they handle noisy states, where they define a noisy state as one affected by an amount epsilon of isotropic noise.
Mira: And their result for the maximal unambiguous randomness of this noisy state is quite clean: it’s given by epsilon. That's a very direct relationship between the noise level and your private randomness.
Lev: That result, P star UD rho equals epsilon, seems quite strong if it holds across different dimensions, but I need to know what constraints they put on epsilon for that to be true in practice on real hardware.
Kai: They also point out something important about shared noise—when the noise is shared between the state and the measurement, an eavesdropper can achieve a perfect guessing probability if that noise level epsilon is greater than some critical value called epsilon crit.
Mira: That’s a bit of a warning for us; it suggests that if your devices share correlated noise, you might actually generate zero private randomness above a certain point, which is definitely something to watch out for in our testing.
Lev: So what they are saying is that correct modeling of noise isn't just about the state alone; it’s about how the state and measurement interact with each other.
Kai: They also show that in an unbiased state and measurement pair, for any parameter T, dimension d, and total noise delta, the guessing probability of a correlated eavesdropper is strictly higher when both the state and the measurement are noisy than when only the state is noisy.
Mira: That’s a subtle but important distinction; it shows that joint correlations with both a noisy state and a noisy measurement always outperform an eavesdropper correlated only to just the noisy state given the same amount of global noise.
Conclusion: Kai: So to wrap up, this paper on "Quantum randomness under unambiguous attacks" provides bounds on how much knowledge an eavesdropper can gain about a secret key without causing any disturbance in a quantum key distribution protocol.
Mira: The main point is that they quantified this knowledge by using unambiguous randomness and FRIO randomness, giving us concrete expressions for these values based on the state's properties and the noise level.
Lev: And for me, I’m focused on how these results translate to running on real hardware; we need to know if those bounds are achievable when we have physical systems with actual imperfections.
Kai: Yeah, they show that beyond a critical error value that is no greater than a half, the joint-noise eavesdropper achieves a perfect guessing probability, which really limits the hope for private randomness in those specific scenarios.
Mira: So it’s less about finding absolute zero randomness and more about understanding exactly where the noise starts to destroy your security guarantees.
Lev: I just want to say that modeling noise correctly is crucial because if you overestimate the amount of private randomness generated, you might be surprised when the actual rate falls short.
Fionnuala Curran
ICFO-Institut de Cióries Fotóniques · The Barcelona Institute of Science and Technology
quant-ph
Submitted: 2026-01-22
Updated: 2026-11-09
Comments: 11.5 + 29 pages, 3 figures. Comments welcome! v3: discussion added about relevance to quantum cryptography. v4: new title, results and presentation
License: http://creativecommons.org/licenses/by/4.0/
Importance score: 76/100
The gist: Intrinsic randomness is generated when a quantum state is measured in any basis in which it is not diagonal <ref:2601.16343#pg2>.
Key concepts
- Unambiguous Randomness
- This measures the maximum private randomness extractable from a quantum state. It is defined by minimizing an eavesdropper's guessing probability, allowing them to output an inconclusive result. It quantifies knowledge gained without causing disturbance.
- FRIO Randomness
- This scenario models an eavesdropper who is allowed to output a fixed rate of inconclusive outcomes. The paper provides formulas showing how this randomness changes based on the state and measurement parameters, helping to set bounds for private randomness.
- Noisy States
- A noisy state is a pure quantum state affected by an amount $\epsilon$ of isotropic noise. The maximal unambiguous randomness of such a noisy state is directly proportional to this noise level $\epsilon$. This shows how environmental noise limits the security of quantum systems.
Terminology
Summary
Intrinsic randomness is generated when a quantum state is measured in any basis in which it is not diagonal <ref:2601. p. 0 Unambiguous randomness from a quantum state
Fionnuala Curran
ICFO-Institut de Ciències Fotòniques, The Barcelona Institute of Science and Technology, Av. Carl Friedrich
Gauss 3, 08860 Castelldefels (Barcelona), Spain
Intrinsic randomness is generated when a quantum state is measured in any
basis in which it is not diagonal. What if the eavesdropper is never wrong, but
can sometimes return an inconclusive outcome? Inspired by analogous concepts
in quantum state discrimination, we introduce the unambiguous randomness
of a quantum state and measurement, and, relaxing the assumption of perfect
accuracy, randomness with a fixed rate of inconclusive outcomes. We also solve both problems for any state and projective
measurement in dimension two, and for an isotropically noisy state measured
in an unbiased basis of any dimension. In the former case, when the setup is
used to choose random bases for a prepare-and-measure quantum key distribution protocol, the unambiguous randomness quantifies the knowledge gained
by an eavesdropper about the secret key, without causing any disturbance.Where the paper says it
Where the paper says it
Where the paper says it
Defining Unambiguous Randomness and its Bounds
The maximal unambiguous randomness of any state is found to be proportional to the smallest eigenvalue of the state <ref:2601.16343#pg5>. This concept is quantified by the optimal guessing probability of an eavesdropper who makes no errors, but who is allowed to output an additional, inconclusive outcome <ref:2601. p. 1 But what if she
wanted a higher, or even perfect, level of accuracy? Inspired by an analogous concept in quantum state discrimination [9–12], we introduce
the unambiguous randomness of a quantum state and measurement pair, as quantified
by the optimal guessing probability of an eavesdropper who makes no errors, but who
is allowed to output an additional, inconclusive outcome. Inspired by an analogous concept in quantum state discrimination [9–12], we introduce
the unambiguous randomness of a quantum state and measurement pair, as quantified
by the optimal guessing probability of an eavesdropper who makes no errors, but who
is allowed to output an additional, inconclusive outcome. The maximal unambiguous
randomness of the state is then found by minimizing this guessing probability over all
measurement bases. We solve the
maximal unambiguous randomness of any quantum state, optimised over all
projective measurements, and find that it’s proportional to the smallest eigenvalue of the state. We also solve both problems for any state and projective
measurement in dimension two, and for an isotropically noisy state measured
in an unbiased basis of any dimension. curran@icfo. eu
1
arXiv:2601. curran@icfo. eu
1
arXiv:2601.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Intermediate Scenarios and Qubit Analysis
The paper introduces a scenario where the eavesdropper maximizes her guessing probability subject to a fixed rate of inconclusive outcomes (FRIO) <ref:2601.16343#pg3>. The FRIO randomness of a quantum state and measurement pair, for a fixed parameter Q, is given by an expression involving Q and other state parameters <ref:2601. p. 1 Further, again adapting ideas from quantum state discrimination [21, 22], we present
an intermediate scenario wherein the eavesdropper maximizes her guessing probability subject to a fixed rate of inconclusive outcomes (FRIO), a term coined in [23]. The FRIO
randomness of a quantum state and measurement pair, for a fixed parameter Q, is then
given by the optimal guessing probability of an eavesdropper who returns an inconclusive
outcome with probability Q. 2 Qubits
In dimension two, we can represent the state ρ and the measurement elements Mj by the
Bloch vectors ⃗r and ±⃗m, where
ρ =
1
2
(1 + ⃗r · ⃗σ) , Mj =
1
2
1 + (−1)j+1 ⃗m · ⃗σ
for all j ∈ {1, 2} , (14)
where ⃗σ is a vector of Pauli matrices, ⃗σ = (σx, σy, σz). Since M is a rank-one measurement,
|⃗m|
2 = 1, and since ρ is mixed, |⃗r |
2 < 1. An eavesdropper’s optimal decomposition of a qubit state ⃗r, given a
measurement in the basis {±⃗m} and FRIO parameter Q. 2 FRIO randomness
The maximum probability Qmax of an inconclusive outcome given a state ⃗r and a measurement {±⃗m} is
Qmax(⃗r, {±⃗m}) = 1 − P UD(⃗r, {±⃗m}), (26)
so from Theorem 3, we find
Qmax(⃗r, {±⃗m}) =
p , m + p ≤ 1 ,
p
2+(1−m)
2
2(1−m)
, m + p > 1 .Where the paper says it
Where the paper says it
Where the paper says it
Application in Quantum Key Distribution
The unambiguous randomness quantifies the knowledge gained by an eavesdropper about the secret key without being detected when a noisy QRNG is used to generate random bases in a prepare-and-measure quantum key distribution (QKD) protocol <ref:2601.16343#pg10>. The analysis shows that this randomness characterizes the knowledge gained by an eavesdropper about the secret key without being detected <ref:2601. p. 1 We show that, when a noisy QRNG is used to generate random bases in a prepareand-measure quantum key distribution (QKD) protocol (e. , BB84 [20]), the unambiguous
randomness characterises the knowledge gained by an eavesdropper about the secret key
without being detected.Where the paper says it
Noisy States and Joint Noise Effects
In any finite dimension d, a noisy state is defined as a pure state affected by an amount ε of isotropic noise <ref:2601.16343#pg5>. The maximal unambiguous randomness of the noisy state is given by Corollary 4. p. 5 1 Unambiguous randomness
Corollary 2. The maximal unambiguous randomness of a qubit state ⃗r, with r = |⃗r |,
is
P
∗
UD(⃗r ) = 1 − r , (17)
and is achieved if and only if the measurement is unbiased to the diagonal basis of the
state. curran@icfo. eu
1
arXiv:2601. In fact, beyond a critical error value that is
no greater than a half, the joint-noise eavesdropper achieves a perfect guessing probability,
ending all hopes of private randomness. This case study warns that, even in a wholly
device-dependent setting, with perfect statistical randomness and stringent constraints on
the guessing accuracy, an eavesdropper might still gain some (or even total) information
2Where the paper says it
Where the paper says it
Where the paper says it
Comparison of Noise Scenarios
The paper shows that in the case of an unbiased state and measurement pair, for any parameter T, dimension d and total noise δ, the guessing probability of a correlated eavesdropper is strictly higher when the state and measurement are both noisy, rather than the state alone <ref:2601.16343#pg46>. This comparison demonstrates that joint correlations with both a noisy state and a noisy measurement always outperform an eavesdropper correlated only to a noisy state given the same amount of global noise <ref:2601. p. 1 The
eavesdropper, now correlated to both, can perform a joint decomposition of the state
and the measurement [26–28]. While experimental noise is typically ascribed to a single
device, we show that such a joint-noise eavesdropper always outperforms her single-noise
counterpart given the same amount of global noise.Where the paper says it
Conclusion on Security Implications
The work concludes by showing that the unambiguous randomness quantifies the knowledge gained by an eavesdropper about the secret key without causing any disturbance <ref:2601.16343#pg10>. The results provide bounds on the randomness generated in device-dependent scenarios, given an eavesdropper with arbitrarily high guessing accuracy <ref:2601. p. 1 In fact, beyond a critical error value that is
no greater than a half, the joint-noise eavesdropper achieves a perfect guessing probability,
ending all hopes of private randomness. This case study warns that, even in a wholly
device-dependent setting, with perfect statistical randomness and stringent constraints on
the guessing accuracy, an eavesdropper might still gain some (or even total) information
2 In fact, beyond a critical error value that is
no greater than a half, the joint-noise eavesdropper achieves a perfect guessing probability,
ending all hopes of private randomness. This case study warns that, even in a wholly
device-dependent setting, with perfect statistical randomness and stringent constraints on
the guessing accuracy, an eavesdropper might still gain some (or even total) information
2 In
this case, we may as well work with this measurement instead and take ρ to be full rank. 2Following [7], we consider only projective measurements, since we allow Alice no access to additional
quantum states or other sources of randomness.Where the paper says it
Where the paper says it
Where the paper says it
Pruning redundant reasoning tokens
The maximal unambiguous randomness of any state is given by the guessing probability P∗ UD(ρ) = dλmin(ρ) <ref:2601.16343#pg10>. This bound is saturated, for example, when the state umin⟩, the eigenvector of the state ρ corresponding to its minimum eigenvalue, satisfies ⟨uminmj ⟩2 = 1/d for all j <ref:2601. p. 4 We show in Appendix C that the bound (4) can be achieved by measuring in a basis {|mj ⟩}
for which |umin⟩, the eigenvector of the state ρ corresponding to its minimum eigenvalue,
satisfies
|⟨umin|mj ⟩|2 =
1
d
for all j ∈ {1, . (6)
This includes bases that are unbiased with respect to the eigenbasis {|uj ⟩} of ρ.Where the paper says it
How the training signal is built
The decomposition in (54) can be converted to a standard SDP by introducing the matrix L = X j pj mj⟩⟨mj (55). The dual problem is then minimized by minimizing G, w tr(Gρ) − wQ subject to G ≽ w1 <ref:2601. p. 3 1
The maximal unambiguous randomness of a state, given by P
∗
UD(ρ), is then found by
minimizing P UD(ρ, M) over all measurement bases. 2
When Eve has a fixed probability Q of giving an inconclusive outcome, her guessing
probability P Q(ρ, M) is given by the optimisation problem
P Q(ρ, M) = maximize
{pµ, ρµ}
X
d
j=1
pj ⟨mj |ρj |mj ⟩
subject to ρµ ≽ 0 for all µ ∈ {0, 1, .Where the paper says it
Where the gains come from
The optimal guessing probability when T < Tcrit is given by P T(⃗r, ±⃗m) = 1/2 (−Q + q(1 − Q) squared − (p − Q) 2) <ref:2601.16343#pg31>. This expression interpolates between the decomposition at the critical point Qcrit and the decomposition at Qmax <ref:2601. p. 7 (27)
(From hereon we assume the state and measurement are fixed, and drop the explicit dependence of Qmax on ⃗r and {±⃗m}. ) We also introduce a critical value for Q,
Qcrit =
1
2
1 − r
2
1 − p
.Where the paper says it
Summary of Results
The paper solves both problems for any state and projective measurement in dimension two, and for an isotropically noisy state measured in an unbiased basis of any dimension <ref:2601.16343#pg3>. The results show that beyond a critical error value that is no greater than a half, the joint-noise eavesdropper achieves a perfect guessing probability, ending all hopes of private randomness <ref:2601. p. 1 This holds true given any upper
bound imposed on the probability of error. In fact, beyond a critical error value that is
no greater than a half, the joint-noise eavesdropper achieves a perfect guessing probability,
ending all hopes of private randomness.Where the paper says it
E Noisy states
The maximal unambiguous randomnss of the noisy state ρε is given by P∗ UD(ρε) = ε <ref:2601.16343#pg9>. This follows immediately from Theorem 1, noting that the state ρε has d−1 degenerate ‘smallest’ eigenvalues with value ε/d <ref:2601.16343#pg9>.
E.2 Fixed rate of inconclusive outcomes
The FRIO randomness generated by measuring a qubit state ⃗r, of the form (15), in the basis ±⃗m is given by P Q(⃗r, ±⃗m) = 1/2 [1 − Q + q(1 − Q) squared − (p - Q) 2] <ref:2601.16343#pg31>. This expression is valid when T ≤ Tmax <ref:2601. p. 0 curran@icfo. eu
1
arXiv:2601.Where the paper says it
F Noisy states and measurements
The FRIO randomness, for a fixed rate Q of inconclusive outcomes, is given by P Q(ρε, M) = 1/d squared q d (1 − Q) − ε (d − 1) + (d − 1) √ε /2 <ref:2601.16343#pg38>. This result is derived from a lower bound using a specific decomposition <ref:2601. p. 3 The maximal unambiguous randomness of any state ρ is given by the guessing probability
P
∗
UD(ρ) = dλmin(ρ). (4)
We can prove the lower bound by considering the decomposition
p0ρ0 = ρ − λmin(ρ)1 , pjρj = λmin(ρ)|mj ⟩⟨mj | for all j ∈ {1, .Where the paper says it
The paper shows that in the case of an unbiased state and measurement pair, for any parameter T, dimension d and total noise δ, the guessing probability of a correlated eavesdropper is strictly higher when the state and measurement are both noisy, rather than the state alone. The results provide bounds on the randomness generated in device-dependent scenarios, given an eavesdropper with arbitrarily high guessing accuracy <ref:2601.16343#pg5>.
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Precise quantification of eavesdropper knowledge in quantum key distribution can be achieved by using unambiguous randomness to quantify
the knowledge gained by an eavesdropper about the secret key without causing any disturbance.
-
AI systems can be improved by modeling adversarial scenarios where an eavesdropper is allowed to return inconclusive outcomes, leading to a lower bound on her guessing probability, which helps determine when
all hopes of private randomness
are ended. -
For noisy quantum random number generators (QRNGs), AI can identify a
critical noise parameter beyond which the joint-noise eavesdropper achieves a perfect guessing probability,
allowing systems to detect when private randomness is lost due to correlated noise. -
In prepare-and-measure quantum key distribution protocols, AI can optimize the eavesdropper's attack by choosing an inconclusive outcome rate Q such that
her disturbance of the intercepted states coincides with the maximum error rate allowed by the communicating parties.
-
AI can distinguish between single and joint noise scenarios in device-dependent settings, proving that
an eavesdropper with joint correlations to both a noisy state and a noisy measurement is outperformed by an eavesdropper with joint correlations to both a noisy state and a noisy measurement.
Abstract
Intrinsic randomness is generated when a quantum state is measured in any basis in which it is not diagonal. In an adversarial scenario, we quantify this randomness by the probability that a correlated eavesdropper could correctly guess the measurement outcomes. What if the eavesdropper is never wrong, but can sometimes return an inconclusive outcome? Inspired by analogous concepts in quantum state discrimination, we consider the randomness generated by a measurement setup when an eavesdropper guesses unambiguously, and, relaxing the assumption of perfect accuracy, with a fixed rate of inconclusive outcomes. We solve the maximal randomness of any quantum state under an unambiguous or FRIO attack, optimized over all projective measurements. We also solve both problems for any state and projective measurement in dimension two. If a noisy setup is used to choose random bases for a prepare-and-measure quantum key distribution protocol, an unambiguous attack may allow an eavesdropper to gain knowledge about the secret key, without causing any disturbance. Moreover, while experimental noise is typically ascribed to a single device, we find, given a state measured in an unbiased basis of any dimension, that an eavesdropper correlated only to a noisy state is outperformed by an eavesdropper with joint correlations to both a noisy state and a noisy measurement. In fact, we identify a critical noise parameter beyond which the joint-noise eavesdropper achieves a perfect guessing probability, ending all hopes of private randomness.
Sources
- Quantum Random Number Generators
- A Comprehensive Review of Quantum Random Number Generators: Concepts, Classification and the Origin of Randomness
- Quantum randomness extraction for various levels of characterization of the devices
- The operational meaning of min- and max-entropy
- Security of Quantum Key Distribution
- Sampling of min-entropy relative to quantum knowledge
- Maximal intrinsic randomness of a quantum state
- How much secure randomness is in a quantum state?
- Unambiguous state discrimination in quantum cryptography with weak coherent states
- Sequential attacks against differential-phase-shift quantum key distribution with weak coherent states
- Zero-Error Attacks and Detection Statistics in the Coherent One-Way Protocol for Quantum Cryptography
- MHz-rate semi-device-independent quantum random number generators based on unambiguous state discrimination
- General Strategies for Discrimination of Quantum States
- Optimal discrimination of quantum states with a fixed rate of inconclusive outcomes
- Scalable generation and detection of on-demand W states in nanophotonic circuits
- True randomness from realistic quantum devices
- Intrinsic randomness under general quantum measurements
- Quantifying the intrinsic randomness of quantum measurements
- Unambiguous Discrimination Between Linearly-Independent Quantum States
- Optimal state discrimination with a fixed rate of inconclusive results: Analytical solutions and relation to state discrimination with a fixed error rate
Related papers
- Computational regimes in matrix-product-state-based quantum trajectory simulations
- Toward chemically accurate quantum simulations on quantum hardware: A quantum-HPC hybrid approach
- Transversal surface-code game with reconfigurable qubits
- Fulqrum: a generalized framework for quantum subspace diagonalization
- Flux-Activated Resonant Control of a Bosonic Quantum Memory
- Machine learning the effects of many quantum measurements