⚛️ quantum physics

Detection Efficiency Bounds in (Semi-)Device-Independent Scenarios

This review article comprehensively examines the critical role of detection efficiency in demonstrating non-classicality across various device-independent and semi-device-independent scenarios, analyzing how the detection loophole affects different causal structures such as Bell, instrumental, prepare-and-measure, and bilocality setups.

Original authors: Tailan S. Sarubi, Santiago Zamora, Moisés Alves, Vinícius F. Alves, Gandhi Viswanathan, Rafael Chaves

Published 2026-03-24

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Tailan S. Sarubi, Santiago Zamora, Moisés Alves, Vinícius F. Alves, Gandhi Viswanathan, Rafael Chaves

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

Imagine you are a detective trying to prove that two people, Alice and Bob, are communicating using a secret, magical code that defies the laws of physics. You suspect they are using "quantum entanglement," a spooky connection where their actions are perfectly linked, no matter how far apart they are.

However, there's a catch: your detective equipment is broken.

Your detectors (the cameras or sensors watching Alice and Bob) are lazy. Sometimes they miss a signal. They don't see the particle arrive. In the world of quantum physics, this is called the Detection Loophole.

If your detectors miss too many signals, a clever "classical" trickster could argue: "Hey, you only saw the signals that looked magical! You missed the boring ones that would have proven they were just using normal, old-fashioned tricks. You can't prove they are using magic because your camera is too slow!"

This paper is a massive guidebook for detectives. It asks: "How good does our camera have to be before we can finally say, 'Okay, the trickster is wrong, this is definitely quantum magic'?"

Here is the breakdown of their findings, explained through everyday analogies.

1. The Classic Case: The Bell Scenario (The Standard Test)

This is the original test. Alice and Bob are in separate rooms. They flip coins (measure particles) and compare results.

The Problem: If your cameras miss more than 1/3 of the flips, the trickster can hide the "boring" results and fake a magical connection.
The Solution: You need your cameras to be at least 67% efficient (seeing 2 out of every 3 signals) to be sure.
The Twist: If one camera is perfect (100% efficient) and the other is bad, the bad one only needs to be 50% efficient. It's like having one super-accurate referee and one slightly distracted one; the super-accurate one carries the team.

2. The Instrumental Scenario (The Messenger)

Imagine a game where Alice doesn't just flip a coin; she sends a message to Bob telling him how to flip his coin.

The Challenge: This setup is stricter. Because Alice is influencing Bob directly, it's harder to prove the "magic" is real if you miss data.
The Result: You need much better cameras here. To be absolutely sure, you need about 96% efficiency. If you use a specific type of math trick (absorbing the "missed" signals into a "wrong answer" category), you can get away with about 67%, similar to the classic test.

3. The Prepare-and-Measure Scenario (The Dimension Check)

Here, Alice sends a physical object (like a letter) to Bob. The goal isn't just to prove magic, but to prove the letter is written in a complex language (high dimension) rather than simple binary code (0s and 1s).

The Problem: If the letter gets lost in the mail (detection loss), Bob might think it was a simple letter that just got lost, rather than a complex one.
The Result: To prove the letter is complex, you need about 71% efficiency. Interestingly, if you have a mix of perfect and bad detectors, the bad one can be almost useless, and you can still prove the complexity, provided the setup is right.

4. The Bilocality Scenario (The Network Effect)

This is the most exciting part. Imagine a chain: Alice sends a message to Bob, and Bob sends a message to Charlie. But here's the kicker: Alice and Bob get their messages from Source 1, while Bob and Charlie get theirs from Source 2. The two sources are totally independent.

The Magic: Because there are two independent sources, the rules change. The "trickster" has a harder time faking the results because he has to coordinate two separate, unrelated conspiracies.
The Result: This network structure is a superpower! It relaxes the requirements. You can prove the quantum connection with lower efficiency (around 53% or even lower in some cases) than in the standard test. It's like having two independent witnesses; if they both agree on a story, you don't need to be as sure your camera caught every single frame.

The Big Picture: Why Does This Matter?

This paper is like a manual for building unhackable communication systems (Quantum Key Distribution) and future quantum computers.

Security: If you are building a quantum bank vault, you need to know exactly how good your sensors must be so a hacker can't sneak in by exploiting "missed" signals.
Feasibility: It tells engineers, "Hey, you don't need perfect 100% cameras to prove quantum mechanics works. If you build a network (like the Bilocality scenario), you can get away with cheaper, slightly worse equipment."

Summary Analogy

Think of the "Detection Loophole" as a leaky bucket.

In the Standard Test, the bucket leaks so much that you need a very tight seal (67% efficiency) to catch enough water to prove the hose is working.
In the Network Test (Bilocality), the bucket is designed with a double-walled structure. Even if it leaks a bit more, the water pressure from the two independent sources is so high that you can still prove the hose is working with a much looser seal (lower efficiency).

This review gathers all the math and history to tell us: "Don't give up on your quantum experiments just because your detectors aren't perfect. There are clever ways to set up the experiment so that even imperfect detectors can catch the magic."

1. Problem Statement

The central problem addressed is the detection loophole in quantum foundations and quantum information. In experiments testing non-classicality (e.g., Bell tests), imperfect detectors fail to register a fraction of events. If the detection efficiency ( $\eta$ ) falls below a critical threshold, a Local Hidden Variable (LHV) model can exploit these "no-click" events to mimic quantum correlations, thereby falsifying the claim of non-classicality.

While the detection loophole is well-studied in the canonical Bell scenario, this paper argues that efficiency requirements vary significantly across different causal structures used to test quantumness. The authors aim to provide a unified review and new results regarding the critical efficiency thresholds ( $\eta_{crit}$ ) required to close the detection loophole in:

Bell Scenarios (Standard and Multipartite).
Instrumental Scenarios (Asymmetric causal structures with communication).
Prepare-and-Measure (PAM) Scenarios (Semi-device-independent, dimension witnessing).
Bilocality Scenarios (Quantum networks with independent sources).

2. Methodology

The paper employs a combination of theoretical review, causal modeling, and numerical optimization.

Detector Modeling: The authors rigorously define two primary models for handling non-detection events:
1. Absorption Model (Model 1): No-click events are absorbed into an existing output (e.g., a "no-click" is registered as outcome $\hat{a}=1$ ). This is common in polarization experiments where a missed photon might be misidentified.
2. Extra-Outcome Model (Model 2): No-click events are treated as a distinct, explicit additional outcome (e.g., $\emptyset$ ), expanding the output alphabet.
Causal Frameworks: The analysis utilizes Directed Acyclic Graphs (DAGs) to define the causal assumptions (locality, independence, measurement independence) for each scenario.
Optimization: For scenarios where analytical bounds are complex, the authors perform numerical optimizations. They parametrize quantum states (e.g., non-maximally entangled states, qutrits) and local measurements to find the minimal $\eta$ required to violate specific inequalities (CHSH, Mermin, IJ, Dimension Witnesses).
Mapping: A key methodological contribution is the formal mapping between the Instrumental Scenario and the Bell Scenario, showing how instrumental inequalities can be derived from Bell inequalities under specific post-selection rules.

3. Key Contributions and Results

A. The Bell Scenario (Review and Extensions)

CHSH Inequality: Confirms the famous Eberhard bound of $\eta > 2/3 \approx 66.7\%$ for symmetric efficiencies using non-maximally entangled states.
Asymmetry: Demonstrates that if one detector is perfect ( $\eta_1=1$ ), the threshold for the other drops to 50%.
Multipartite Systems:
- Mermin Inequalities: Show that for $n$ parties, the critical efficiency can decrease as $n$ increases (scaling as $\approx n/2^n$ ), making multipartite systems potentially more robust to loss than bipartite ones.
- Svetlichny Inequalities: Conversely, for genuine multipartite nonlocality (GMN), the threshold rapidly approaches 100% as $n$ increases, making these tests extremely sensitive to loss.
- Qutrits: Numerical results show that using qutrits (3-level systems) with the CGLMP inequality requires $\eta \approx 81\%$ (symmetric) or $\eta \approx 68\%$ (asymmetric), which is higher than the qubit CHSH bound.

B. The Instrumental Scenario (Novel Results)

This section presents new findings (marked with $\dagger$ in the paper).

Equivalence to Bell: The authors prove that the instrumental scenario with binary variables is formally equivalent to the Bell scenario regarding efficiency bounds.
Thresholds:
- For the inequality $I_{222}$ (derived from Pearl's inequalities), the critical efficiency mirrors the CHSH results.
- Symmetric Case: $\eta_{crit} \approx 0.67$ (if absorbing non-detections into $\hat{a}=1, \hat{b}=0$ ) or $\eta_{crit} \approx 0.84$ (if absorbing into $\hat{a}=1, \hat{b}=1$ ).
- Asymmetric Case: If one detector is perfect, the other requires $\eta > 0.5$ .
Hybrid Polytopes: The paper explores combining observational and interventional data, showing that while this defines a tighter polytope, the efficiency thresholds remain comparable to standard Bell tests.

C. Prepare-and-Measure (PAM) Scenario

Dimension Witnesses: Analyzes how inefficiencies affect the certification of quantum system dimension (e.g., distinguishing a qubit from a classical bit).
Scaling: For the standard dimension witness $T_2$ (2-bit random access code), the threshold is $\eta > 1/\sqrt{2} \approx 0.707$ .
Robustness: The paper highlights that dimension witnesses can be more robust to loss in asymmetric setups. If one measurement basis is perfect, violations are possible even with arbitrarily low efficiency in the other basis.
Noise Channels: The authors model state preparation inefficiencies using Amplitude Damping and Depolarizing channels. They identify critical noise thresholds ( $t_c \approx 0.433$ for damping, $q_c \approx 0.216$ for depolarization) beyond which dimension certification fails.

D. The Bilocality Scenario

Network Advantage: This scenario involves two independent sources ( $\lambda_1, \lambda_2$ ). The assumption of source independence imposes stricter constraints on classical models than standard locality.
Relaxed Thresholds:
- The standard bilocality inequality ($IJ$) requires $\eta > 1/\sqrt{2} \approx 0.707$ for symmetric detection.
- However, by optimizing the entanglement of the sources, the authors show that non-bilocal correlations can be certified with efficiencies as low as $\eta \approx 0.529$ (symmetric) and even $\eta \approx 0.00001$ in highly asymmetric cases (one perfect detector).
Significance: This demonstrates that quantum networks can certify non-classicality with significantly lower detection efficiencies than standard Bell tests, provided the network topology (independent sources) is exploited.

4. Significance and Conclusion

Unified Perspective: The paper provides a comprehensive map of detection efficiency requirements across the landscape of device-independent and semi-device-independent protocols.
Experimental Guidance: It offers concrete targets for experimentalists. For instance, it suggests that bilocality tests are the most promising route for loophole-free demonstrations in systems with high loss (e.g., long-distance fiber networks), while multipartite Mermin tests offer a path for high-party systems.
Security Implications: In semi-device-independent protocols (like QKD), the paper clarifies that detection inefficiencies can be exploited to fake dimension violations, emphasizing the need for rigorous efficiency modeling in security proofs.
New Bounds: The specific numerical bounds derived for the Instrumental scenario and the detailed analysis of asymmetric bilocality thresholds represent novel contributions to the field, refining our understanding of the trade-off between causal structure and experimental feasibility.

In summary, the paper concludes that while the detection loophole remains a formidable challenge, the choice of causal scenario (Bell vs. Instrumental vs. Bilocal) and the optimization of quantum states (non-maximal entanglement) can drastically lower the required detection efficiency, enabling robust demonstrations of quantum non-classicality in realistic, lossy environments.