Entanglement distribution: To herald or not to herald

This paper compares the entanglement distribution rates of two heralded and one unheralded SPDC-based systems using spectral islands, finding that while unheralded operation generally outperforms heralded approaches under specific memory and efficiency constraints, the optimal choice depends on a trade-off between achievable heralding efficiencies and the differing equipment burdens of each system.

The Gist

Imagine you are trying to build a Quantum Internet. This isn't just a faster version of the internet we use today; it's a network that uses the weird rules of quantum physics to send information that is perfectly secure and can teleport data instantly.

The most important ingredient for this network is entanglement. Think of entangled particles as a pair of "magic dice." If you roll one in New York and the other in London, they will always land on matching numbers, no matter how far apart they are. To build the internet, we need to send these magic dice pairs to two people, let's call them Alice and Bob.

The problem? Sending these dice is incredibly hard. They are fragile, and most of them get lost or destroyed before they reach their destination. To make this work, we need to generate these pairs very quickly and send them efficiently.

This paper compares three different "factories" (or methods) for making and sending these magic dice pairs. The authors are trying to answer a simple question: "Should we use a 'Herald' (a warning signal) or not?"

Here is the breakdown of the three methods, explained with everyday analogies:

The Three Methods

1. The "Double-Factory" with a Perfect Matchmaker (ZALM)

• How it works: Imagine two factories (Sagnac sources) working side-by-side. They both make magic dice pairs. To make sure the dice are perfectly entangled, the factories send their "idler" dice (the second half of the pair) to a central "Matchmaker" station.
• The Herald: The Matchmaker checks the idler dice. If it sees a specific pattern (one red, one blue), it shouts, "Hey Alice and Bob! The dice you have are the real deal!" This shout is the Herald.
• The Catch: This system is very complex. It requires two perfectly identical factories and a massive, high-tech Matchmaker station with hundreds of super-sensitive detectors. It's like trying to synchronize two grand pianos perfectly while a conductor checks the notes in real-time.
• Pros: When the Matchmaker works well (high efficiency), this method is the fastest at sending pairs when you only have one "memory slot" (one place to store the dice) at Alice and Bob's end.

2. The "Single-Factory" with a Magic Eraser (Chahine et al.)

• How it works: Instead of two factories, we use just one. Inside this factory, the light travels in two directions (clockwise and counter-clockwise). The factory makes pairs in both directions.
• The Herald: The "idler" dice are sent to a detector, but the "signal" dice (the ones for Alice and Bob) are sent through a Magic Eraser (a beam splitter). This eraser wipes out the information about which direction the dice came from. Because we don't know the path, the dice become entangled.
• The Catch: It's simpler than the Double-Factory, but it's slightly less efficient at generating the "perfect" pairs compared to the first method.
• Pros: It's easier to build and requires fewer components.

3. The "Blind" Factory (Unheralded)

• How it works: This is the simplest approach. We just have a factory that spits out magic dice pairs as fast as it can. We don't check the idler dice. We don't have a Matchmaker. We don't shout "Herald!"
• The Catch: Because we aren't checking, we don't know which pairs are "good" and which are "bad" (or if we accidentally sent two pairs at once, which ruins the quantum magic). We have to send them blindly and hope they arrive.
• Pros: It's very simple and doesn't need complex detectors.
• Cons: It's usually slow because we have to keep the "production rate" very low to avoid sending too many pairs at once (which causes errors).

---

The Big Showdown: To Herald or Not to Herald?

The authors ran simulations to see which factory wins under different conditions. Here is what they found, using a Restaurant Analogy:

Scenario A: The Small Restaurant (One Table/Memory)
Imagine Alice and Bob only have one table (one quantum memory) to serve customers.

• The Winner: The Double-Factory with the Matchmaker (ZALM) wins.
• Why? Even though the Matchmaker is complicated, it filters out the bad orders so perfectly that Alice and Bob get a fresh, perfect meal every time the Matchmaker shouts. The "Blind" factory has to wait so long to avoid serving a bad meal that it ends up serving fewer customers overall.
• The Lesson: If you have limited storage space, the "Herald" (the Matchmaker) is worth the extra cost.

Scenario B: The Buffet (Unlimited Tables/Memories)
Imagine Alice and Bob have a massive buffet with hundreds of tables (many quantum memories).

• The Winner: The Blind Factory (Unheralded) wins.
• Why? With so many tables, the Blind Factory can just flood the room with food. Even if 90% of the food is slightly imperfect, there are so many tables that they can still find enough perfect meals to fill the room. The complex Matchmaker system is too slow to keep up with the sheer volume the Blind Factory can produce.
• The Lesson: If you have unlimited storage, the "Herald" becomes a bottleneck. It's better to just blast everything out and sort it later.

The "Real World" Problems

The paper also looked at two real-world annoyances:

1. Dark Counts (False Alarms): Imagine the Matchmaker's eyes are tired, and they sometimes see a flash of light that isn't actually there. The authors found that with modern, high-quality detectors, these false alarms are so rare they don't really matter.
2. Background Light (Sunlight): If you are sending these dice through the air (like from a satellite), sunlight can interfere. The authors found that as long as you filter out the "noise" (like using a narrow window to look at the sun), the system still works well during the day.

The Final Verdict

• If you are building the first generation of the Quantum Internet (where you have limited memory and need high quality): Use the "Herald" (ZALM). It's complex and expensive, but it gives you the best speed and quality for small-scale networks.
• If you are building the massive, future Quantum Internet (where you have unlimited memory): Drop the Herald. The "Blind" factory is actually faster and simpler when you scale up to massive numbers.

In short: The "Herald" is like a strict quality control inspector. You need them when you are small and can't afford mistakes. But when you are a giant factory, the inspector slows you down, and it's better to just produce as much as you can and sort it out later.

Technical Summary

Here is a detailed technical summary of the paper "Entanglement distribution: To herald or not to herald" by Jeffrey H. Shapiro, Clark Embleton, and J. Gabriel Richardson.

1. Problem Statement

The creation of a quantum internet requires high-rate, high-fidelity distribution of entangled photon pairs between remote nodes (Alice and Bob). Current preferred sources are Spontaneous Parametric Down-Converters (SPDCs), specifically those utilizing phase-matched spectral islands (discrete, spectrally factorable frequency bins) to enable single-mode temporal behavior.

However, SPDCs operate at low brightness to avoid multi-pair events, and long-distance transmission suffers from significant losses (10–20 dB). To achieve the high rates necessary for a quantum internet, multiplexing many sources is essential. The paper addresses the critical trade-off between two approaches:

1. Heralded Operation: Using detection of "idler" photons to signal ("herald") the presence of entangled "signal" photons, allowing for higher brightness but introducing complexity and potential losses in the heralding path.
2. Unheralded Operation: Sending signal photons without heralding, relying on simple multiplexing but suffering from multi-pair errors and lower fidelity at higher brightness.

The core question is: Which architecture (Heralded vs. Unheralded) maximizes the entanglement distribution rate under realistic constraints (efficiency, loss, and memory resources)?

2. Methodology

The authors compare three specific architectures based on islands-based SPDC sources (using PPLN crystals with $N_I$ spectral islands):

1. Islands-based Zero-Added-Loss Multiplexing (ZALM):

   • Configuration: Uses two Sagnac SPDC sources.
   • Mechanism: Idler photons from both sources are combined on a 50:50 beam splitter and undergo a partial Bell-State Measurement (BSM) using Polarizing Beam Splitters (PBS) and Dense Wavelength Division Multiplexing (DWDM) filters.
   • Heralding: Uses Same-Plus-Cross-Island (SPCI) heralding. A herald is declared if one H-polarized and one V-polarized idler are detected, regardless of whether they come from the same island or different islands.
   • Goal: Achieve zero-added-loss by using quantum state wavelength conversion at the receivers.

2. Chahine et al.'s Signal-Path Erasure Source:

   • Configuration: Uses a single Sagnac SPDC source.
   • Mechanism: The source generates H-polarized pairs in the clockwise (cw) path and V-polarized pairs in the counter-clockwise (ccw) path. Idlers are detected to identify the path, while the signal paths are combined on a 50:50 beam splitter to erase path information, creating entanglement.
   • Heralding: Also uses SPCI heralding.
   • Advantage: Simpler hardware (one crystal/source) compared to ZALM.

3. Unheralded Operation:

   • Configuration: A single Sagnac source with $N_M$ spectral islands.
   • Mechanism: No idler detection. Signal and idler photons are sent directly to Alice and Bob.
   • Multiplexing: Simple wavelength-division multiplexing where each island corresponds to a dedicated quantum memory.

Analytical Approach:

• The authors utilize anti-normally ordered characteristic functions to model the quantum states, accounting for losses, detector efficiencies, and multi-pair events.
• They derive performance metrics: Bell-state fraction ($B$), Bell-state fidelity ($F$), and the entanglement distribution rate ($R_E$).
• They analyze scenarios with single-memory ($N_M=1$) and multi-memory ($N_M > 1$) allocations per pump pulse.
• They evaluate non-idealities: dark counts in detectors and background light in free-space propagation.

3. Key Contributions

• Comparative Performance Analysis: The first rigorous comparison of ZALM, the Chahine et al. signal-path erasure scheme, and unheralded operation using the spectral island model, which accounts for multi-pair events of all orders.
• SPCI Heralding Optimization: Demonstrating that SPCI heralding (allowing cross-island detection) significantly reduces the number of required spectral islands compared to same-channel heralding to achieve high-fidelity entanglement.
• Trade-off Identification: Establishing that while heralded systems (ZALM and Chahine) suppress multi-pair errors better than unheralded systems, their advantage is heavily dependent on heralding efficiency ($\eta_T$).
• Scalability Insights: Showing that the optimal choice shifts based on the number of available quantum memories ($N_M$) and spectral islands ($N_I$).
• Robustness Analysis: Quantifying the impact of dark counts and background light, concluding that these non-idealities are manageable if the number of extraneous background modes is kept low.

4. Key Results

A. Single-Memory Operation ($N_M = 1$)

• ZALM vs. Chahine: ZALM generally outperforms the Chahine source in distribution rate because its dual-source configuration provides a higher heralding probability, which outweighs its slightly lower Bell-state fidelity.
• Heralded vs. Unheralded: When heralding efficiency is high ($\eta_T \ge 90\%$) and the number of islands is sufficient ($N_I \ge 10$), ZALM significantly outperforms unheralded operation.
  • Example: With $N_I=50$ and $\eta_T=0.9$, ZALM achieves a rate ~2.4$\times$ higher than a single-island unheralded source while maintaining $B \ge 0.999$ and $F \ge 0.99$.
  • ZALM can reach the quasideterministic threshold (entanglement generation probability $\approx 0.25$), whereas unheralded sources are limited to $\sim 0.01$ to avoid multi-pair errors.

B. Multi-Memory Operation ($N_M > 1$)

• The "Third-Generation" Regime ($N_M = N_I$): When the number of quantum memories matches the number of spectral islands (unlimited resources), unheralded operation becomes superior.
  • Unheralded operation scales linearly with $N_M$ without the overhead of heralding detection losses.
  • Example: With $N_M = N_I = 30$, unheralded operation offers a 2.0$\times$ rate advantage over ZALM.
• The "Second-Generation" Regime ($N_M < N_I$): In intermediate scenarios (e.g., $N_I=25, N_M=5$), heralded systems (specifically ZALM) still hold a modest advantage (~2.2$\times$) over unheralded systems with the same number of memories, due to the suppression of multi-pair errors.

C. Impact of Non-Idealities

• Heralding Efficiency: If $\eta_T$ drops (e.g., to 75%), the advantage of heralded systems diminishes rapidly. ZALM requires many more islands to maintain an advantage over unheralded systems.
• Dark Counts: With realistic SNSPD rates ($10^3$s$^{-1}$), dark counts have a negligible impact on performance metrics.
• Background Light: For free-space links, background light only significantly degrades performance if the number of collected background modes ($M_B$) is very high ($>100$). For realistic parameters (e.g., $M_B \approx 25$), the fidelity degradation is minimal (<1%).

5. Significance and Conclusion

The paper provides a definitive framework for designing quantum repeater nodes:

1. For Near-Term/Initial Deployments: Where quantum memory resources are limited ($N_M=1$) and high-fidelity entanglement is critical, Heralded ZALM is the superior choice, provided high-efficiency detectors and wavelength conversion can be achieved.
2. For Long-Term/Scalable Networks: As technology matures to support large numbers of quantum memories ($N_M \approx N_I$), Unheralded Operation becomes the most efficient architecture due to its linear scaling and lack of heralding overhead.
3. Implementation Burden: The authors note that ZALM requires two perfectly matched Sagnac sources and complex path-switching for multi-memory reception, whereas unheralded systems are simpler to implement but require more memories to achieve high rates.

Final Verdict: The choice "to herald or not to herald" depends on the specific resource constraints. Heralding is essential for high-rate, high-fidelity distribution when memory resources are scarce. However, unheralded multiplexing is the ultimate solution for scalable, high-memory networks.