Random Party Distillation on a Superconducting Processor

This paper presents a qubit-based implementation of a random party distillation protocol on the IBM Aachen superconducting processor, demonstrating up to four rounds of local operations and classical communications to achieve a record-high distillation rate of approximately 0.85 EPR pairs per W state.

The Gist

The Big Idea: Turning a "Group Hug" into "Handshakes"

Imagine you have three friends: Alice, Bob, and Charlie. In the quantum world, they are holding hands in a special, invisible way called entanglement. Specifically, they are in a "W state."

Think of the W state like a group hug where all three friends are holding onto each other at once. It's a strong connection, but it's a bit messy because it involves everyone.

The goal of this experiment is to turn that messy group hug into a clean, direct handshake (an EPR pair) between just two of them. This is useful because many quantum tasks, like sending secret messages, work best when two people are directly connected.

The Problem: The "All-or-Nothing" Mistake

In the past, if you wanted to get a handshake between two people from a group hug, you had to ask one person to let go immediately.

• The Old Way: You ask Alice to let go. If she lets go at the right moment, Bob and Charlie get a handshake. But if she lets go at the wrong moment, the whole connection breaks, and Bob and Charlie are left holding nothing.
• The Success Rate: This old method worked about 66% of the time. It was a "one-shot" deal. If it failed, you had to start over with a new group hug.

The New Solution: The "Gentle Tap" Game

The researchers in this paper invented a smarter way called Random Party Distillation. Instead of asking someone to let go immediately, they play a game of "gentle taps."

1. The Setup: Alice, Bob, and Charlie are still in their group hug.
2. The Gentle Tap: Instead of a hard pull, they gently "tap" their connection to a helper (an "ancilla" qubit). This tap is so light that it doesn't break the hug; it just checks if the hug is still there.
3. The Decision:
   • If the tap shows the hug is still intact, they do it again. They keep tapping gently, round after round.
   • If the tap shows that the connection has naturally shifted, two of them suddenly find themselves holding hands directly, while the third is let go.
   • If the tap shows the connection is broken, the game ends in failure.

The Magic: By doing this "gentle tap" multiple times (up to 4 rounds in this experiment), they increase the odds that the group hug will naturally transform into a handshake without breaking.

The Experiment: The Superconducting Computer

The team tested this on a real quantum computer called ibm_aachen (a superconducting processor).

• The Challenge: Quantum computers are noisy. While the friends are waiting for their turn to "tap," the connection can get shaky (decoherence) or the computer might misread the result (measurement error).
• The Fix: To keep the connection steady while waiting, they used a technique called Dynamical Decoupling. Imagine this like a metronome or a gentle vibration that keeps the friends' hands steady so they don't slip while waiting for the next tap. They also used a special math trick (M3) to fix any mistakes the computer made when reading the results.

The Results: A New Record

The team ran this game for up to four rounds. Here is what they found:

• Success Rate: They managed to turn the group hug into a handshake 85% of the time.
• Comparison: This is much better than the old "one-shot" method (66%) and better than previous experiments that only managed 75%.
• The Trade-off: As they played more rounds, the quality of the handshake got slightly weaker (because the group hug got a bit shaky over time), but the chance of getting a handshake at all went up significantly.

Why This Matters (According to the Paper)

The paper claims this is the first time anyone has successfully run this "multi-round" game on real hardware. It proves that:

1. Patience pays off: You can get better results by using multiple rounds of gentle checks rather than one big, risky move.
2. Noise is manageable: Even on a noisy computer, you can use tricks (like the metronome vibration) to keep the quantum connection alive long enough to succeed.

In short, they showed that you can reliably turn a three-way quantum connection into a two-way connection much more often than before, paving the way for better quantum networks in the future.

Technical Summary

Technical Summary: Random Party Distillation on a Superconducting Processor

Problem Statement
Entanglement distribution is a fundamental subroutine for quantum communication and distributed computing. While bipartite entanglement (EPR pairs) is well-understood, converting multipartite entangled states (specifically W states) into bipartite entanglement via Local Operations and Classical Communication (LOCC) remains a challenging experimental regime. Theoretical frameworks, such as "random party distillation" proposed by Fortescue and Lo, suggest that by allowing randomness in which pair of parties shares the resulting EPR pair, one can asymptotically approach a success probability of unity for a single copy of a W state. However, experimental realizations have been limited to single-round implementations, restricting achievable success probabilities. Furthermore, multi-round LOCC protocols are rare in experimental settings due to the complexity of mid-circuit measurements, decoherence during idle periods, and the difficulty of maintaining high-fidelity multipartite states.

Methodology
The authors implement a multi-round random party distillation protocol on the 156-qubit superconducting processor ibm_aachen (Heron r3 system). The protocol involves three parties (Alice, Bob, and Charlie) sharing a W state:
$$|W\rangle_{abc} = \frac{1}{\sqrt{3}} (|011\rangle + |101\rangle + |110\rangle)$$

The core mechanism utilizes weak measurements rather than strong projective measurements to preserve entanglement across multiple rounds.

1. Weak Measurement Implementation: Instead of coupling to higher-dimensional Hilbert spaces (as in the original proposal), the authors use ancillary qubits coupled to the system qubits via controlled-rotation gates ($CRY(\theta)$). The rotation angle $\theta$ is related to the measurement strength $\epsilon$ by $\theta = 2 \arcsin(\epsilon)$.
2. Protocol Flow: In each round, parties measure their ancillae in the Z-basis.
   • If two parties measure "0" and one measures "1", the two "0" parties share an EPR pair (Success).
   • If all three measure "0", the system remains in a W state, and the protocol repeats.
   • If two or more measure "1", the protocol fails.
3. Dynamic Circuits: The experiment employs dynamic circuits where subsequent operations depend on mid-circuit measurement outcomes. The protocol continues only if all parties measure "0".
4. Error Mitigation: To counteract decoherence during the long idle periods (~836 ns) required for mid-circuit measurements, the authors apply XY4 dynamical decoupling to idle qubits. Additionally, they utilize Matrix-free Measurement Error Mitigation (M3) to correct readout errors.
5. Final Step: If the protocol completes $N$ rounds without success (all "0" outcomes), a final strong measurement is performed on one qubit to attempt distillation with a 2/3 probability, a modification proposed in prior work to boost success rates in finite-round scenarios.

Key Contributions

• First Multi-Round Experimental Demonstration: This work presents the first experimental realization of random party distillation exceeding a single round, successfully executing up to four rounds on superconducting hardware.
• Generalized Protocol for Qubits: The authors adapt the Fortescue-Lo protocol for transmon qubits without requiring access to higher energy levels, utilizing ancillary qubits and controlled rotations to simulate weak measurements.
• Performance Benchmarking: The study compares the performance of random party distillation against "specific party distillation" (where a specific party is chosen to measure) and previous experimental limits.
• Error Mitigation Strategy: The paper demonstrates the efficacy of combining dynamical decoupling and M3 to extend the viable number of protocol rounds on noisy intermediate-scale quantum (NISQ) devices.

Results

• Success Probability: The experiment achieved a record success probability of ~85% for a single copy of a W state after four rounds. This surpasses the previous experimental limit of 75% (single-round) and the theoretical 66% limit of specific party distillation.
• Fidelity Degradation: While the success rate increased with rounds, the fidelity of the W state degraded over time due to natural dephasing on the hardware. The initial W state fidelity was 0.96, compared to 0.866 in previous work [18], but the fidelity decreased as the number of rounds increased.
• Expected Entanglement: The authors calculated the expected entanglement of formation ($\langle E \rangle$), weighting success probabilities by the quality of the distilled pairs.
  • Single-round implementation: $\langle E \rangle \geq 0.6379$ pairs/W state.
  • Four-round implementation: $\langle E \rangle \geq 0.5404$ pairs/W state.
  • Specific party distillation (same hardware): $\langle E \rangle \geq 0.5993$.
  • The results demonstrate a clear advantage in the distilled entanglement rate per copy for random party distillation over specific party distillation, even with the degradation observed in multi-round execution.
• Error Analysis: The deviation between experimental results and theoretical predictions after the third round was attributed to the degradation of W state fidelity at the start of each round, primarily caused by dephasing during idle periods.

Significance and Claims
The paper claims to provide the first physical demonstration of random party distillation that exceeds one round of execution, establishing a new record for distillation rates. The work serves as a proof of concept for utilizing weak measurements in the "few-copy" regime for bipartite entanglement distribution.

The authors modestly frame the significance of their results as follows:

• They highlight the utility of multi-round LOCC protocols in an experimental regime that has remained comparatively unexplored.
• They demonstrate that random party distillation can outperform specific party distillation and single-round implementations in terms of entanglement yield per state copy.
• They identify that the primary limitations for future improvements are hardware-specific: specifically, the need for shorter readout periods to minimize decoherence during mid-circuit measurements and the development of better decoupling techniques (e.g., staggered dynamical decoupling) to handle cross-talk.
• The work suggests that such few-copy protocols are a necessary step for generating high-fidelity entanglement in future large-scale distributed quantum networks.

The authors do not claim to have solved the problem of perfect distillation on NISQ hardware; rather, they show that with current error mitigation techniques, multi-round protocols are feasible and offer a tangible advantage over existing methods, despite the inevitable trade-off with state fidelity over time.