Big cats: entanglement in 120 qubits and beyond

Basic Information

• Paper ID: 2510.09520
• Title: Big cats: entanglement in 120 qubits and beyond
• Authors: Ali Javadi-Abhari, Simon Martiel, Alireza Seif, Maika Takita, Ken X. Wei (IBM Quantum)
• Classification: quant-ph (Quantum Physics)
• Publication Date: October 10, 2025
• Paper Link: https://arxiv.org/abs/2510.09520v1

Abstract

This paper reports the largest Greenberger-Horne-Zeilinger (GHZ) state preparation experiment to date, successfully demonstrating many-body quantum entanglement on 120 superconducting qubits. The research team created this record-breaking quantum entangled state on a fixed-frequency, tunable-coupler superconducting processor through a combination of optimized compilation, low-overhead error detection, and temporary uncomputation techniques. The measured GHZ fidelity is 0.56(3) with a post-selection rate of 28%. The paper validates GHZ state fidelity using multiple methods and demonstrates their equivalence.

Research Background and Motivation

Problem Definition

1. Core Challenge: Preparing large-scale many-body quantum entangled states is an important benchmark for quantum computing hardware quality, but GHZ states are extremely sensitive to noise and difficult to prepare
2. Significance: GHZ states are critical resources for quantum algorithms with important applications in quantum secret sharing, quantum metrology, gate teleportation, and error correction
3. Existing Limitations: Traditional circuit compilation methods (constructing circuits before mapping to hardware) often fail in large-scale circuits, particularly when allocating auxiliary qubits for error detection

Research Motivation

• Hardware Benchmarking: Success in GHZ state preparation directly reflects the control quality and noise suppression capability of quantum computers
• Scale Breakthrough: Extending the previous 75-qubit record to 120 qubits represents a 45-qubit advancement
• Technical Innovation: Developing an adaptive compilation framework that handles arbitrary qubit connectivity constraints and varying error rates

Core Contributions

1. Record-Breaking Quantum Entanglement: Achieved 120-qubit GHZ state, the largest-scale many-body quantum entanglement to date
2. Adaptive Compilation Framework: Proposed novel compilation techniques based on graph theory, stabilizer groups, and circuit uncomputation
3. Optimized Error Detection: Maximized detection coverage through ZZ backpropagation
4. Multiple Verification Methods: Established equivalence between parity oscillation and direct fidelity estimation (DFE) methods
5. Cross-Platform Validation: Verified method reproducibility across multiple IBM quantum processors

Methodology Details

Task Definition

Preparation of n-qubit GHZ state: $|G_n\rangle = \frac{1}{\sqrt{2}}(|0\rangle^{\otimes n} + |1\rangle^{\otimes n})$

This state possesses qubit permutation invariance and allows low-weight stabilizers of the form $Z_iZ_j$.

Core Technical Architecture

1. Adaptive Tree Construction

• Minimum Eccentricity Root Selection: Select the vertex with minimum eccentricity $e(u) = \max\{d(u,v)|v \in V\}$ as the root node
• Breadth-First Search: Perform BFS from the root node, using CNOT gates to create entanglement along graph edges
• Circuit Depth: $D_{GHZ} \propto e(v_0) = \min_{u \in V} e(u) \leq radius(G)$

2. ZZ Backpropagation Detection Region

Lemma 1 (Detection Region for GHZ Parity Detection): Let $S_{i,j} := path(i, lca(i,j)) \cup path(j, lca(i,j))$, where $lca(i,j)$ is the lowest common ancestor of leaf nodes $i,j$. A single-qubit X or Y error at spacetime position $(q,t)$ is detected by the $Z_iZ_j$ measurement if and only if its corresponding edge lies in $S_{i,j}$.

3. Temporary Uncomputation

• ASAP Scheduling: Disentangle the earliest qubits as soon as possible, returning them to the ground state
• ALAP Recomputation: Re-entangle qubits as late as possible
• Purpose: Reduce decoherence effects during long idle periods

Technical Innovations

1. Dynamic Detection Optimization: Rather than fixing detection positions, dynamically allocate them during circuit preparation to maximize coverage
2. Randomized Perturbation Algorithm: Create additional auxiliary qubit space by randomly blocking certain nodes in BFS
3. Coverage-Driven Design: Optimize detection coverage rather than solely circuit depth or detection count

Experimental Setup

Hardware Platform

• Primary Experiment: IBM Aachen processor (120 qubits + 8 auxiliary qubits)
• Comparative Validation: IBM Kobe, Kingston, Fez processors (100-qubit experiments)
• Architecture: Fixed-frequency, tunable-coupler superconducting qubits

Evaluation Metrics

1. Fidelity: $F = \langle G_n|\rho|G_n\rangle$
2. Post-selection Rate: Success rate through parity detection
3. Many-Body Entanglement Threshold: Fidelity > 0.5 indicates genuine many-body entanglement

Verification Methods

1. Parity Oscillation: Measure population P and coherence χ, $F = \frac{1}{2}(P + \chi)$
2. Direct Fidelity Estimation (DFE): $F = \frac{1}{2^N}\sum_i Tr(\rho S_i)$, through uniform stabilizer sampling

Experimental Results

Primary Results

120-Qubit GHZ State (IBM Aachen)

• Fidelity: 0.56(3)
• Post-selection Rate: 28%
• Circuit Depth: 18 CNOT layers
• Detection Configuration: 8 auxiliary qubits, 1 uncomputation qubit

100-Qubit Cross-Platform Validation

Method Equivalence Verification

In 100-qubit experiments on IBM Kingston:

• Parity Oscillation Method: F = 0.536(8)
  • Coherence: 0.539(5)
  • Population: 0.54(1)
• DFE Method: F = 0.55(3)

Both methods agree within error margins, validating theoretical equivalence.

Ablation Studies

Dynamical Decoupling Effects

• Without DD: Non-diagonal stabilizers significantly degraded
• With DD: Non-diagonal stabilizers improved from ~0.4 to 0.8+
• Diagonal Stabilizers: DD impact minimal, slight improvement observed

Error Mitigation Effects

Through parity detection and readout error mitigation:

• Baseline Results: ~0.4
• After Parity Detection: ~0.5
• After Readout Mitigation: 0.55-0.8 (processor-dependent)

Related Work

Historical Development

• Ion Trap Systems: First demonstrated by Sackett et al. (2000), reached 14 qubits by Monz et al. (2011)
• Superconducting Systems: Initiated by Song et al. (2017), MQC protocol established by Wei et al. (2020)
• Neutral Atoms: Breakthrough work by Omran et al. (2019) and Song et al. (2019)
• Photonic Systems: Related progress by Wang et al. (2018)

Technical Comparison

• Traditional MQC Methods: Require double circuit depth, introduce non-Clifford operations
• This Work: Uses Clifford gates, compatible with error detection, more suitable for large-scale experiments

Conclusions and Discussion

Main Conclusions

1. Scale Breakthrough: Successfully prepared 120-qubit GHZ state, setting a new world record
2. Method Effectiveness: Adaptive compilation framework performs well across multiple processors
3. Verification Consistency: Multiple fidelity estimation methods yield consistent results
4. Practical Value: Provides large-scale entanglement resources for quantum algorithms

Limitations

1. Fidelity Constraints: While 0.56 fidelity exceeds the entanglement threshold, improvement is needed
2. Post-selection Overhead: 28% success rate necessitates multiple experimental runs
3. Hardware Dependence: Performance limited by qubit quality and connectivity
4. Scalability: Current methods constrained by available qubit count

Future Directions

1. Larger Scale: Extend to GHZ states with more qubits
2. Other Stabilizer States: Generalize techniques to other stabilizer state types
3. Gate Teleportation Integration: Combine with gate teleportation for scalable resource state injection
4. Non-Markovian Errors: Develop specialized post-selection strategies for specific noise models

In-Depth Evaluation

Strengths

1. Technical Innovation: Adaptive compilation framework addresses practical challenges in large-scale circuits
2. Experimental Rigor: Multiple verification methods and cross-platform testing enhance result credibility
3. Theoretical Contribution: Establishes theoretical connections between different fidelity estimation methods
4. Practical Value: Provides new benchmarking standards for quantum computing

Weaknesses

1. Theoretical Analysis: Noise model and error propagation analysis could be more comprehensive
2. Algorithm Complexity: Complexity analysis of random circuit optimization search lacks detail
3. Hardware Specificity: Method's dependence on specific hardware architecture requires further discussion

Impact

1. Academic Significance: Provides new technical approaches for large-scale quantum entanglement preparation
2. Industrial Value: Establishes new benchmarks for quantum processor performance evaluation
3. Reproducibility: Open code and data facilitate reproduction and improvement

Applicable Scenarios

1. Quantum Benchmarking: Evaluate quantum processor entanglement capabilities
2. Quantum Algorithms: Support algorithms requiring large-scale entanglement resources
3. Error Detection Research: Provide experimental platform for quantum error detection and correction

References

• 1 Greenberger, D. M., et al. Bell's theorem, quantum theory and conceptions of the universe. Springer, 1989.
• 11 Wei, K. X., et al. Verifying multipartite entangled GHZ states via multiple quantum coherences. Physical Review A 101, 032343 (2020).
• 21 Flammia, S. T. & Liu, Y.-K. Direct fidelity estimation from few Pauli measurements. Physical Review Letters 106, 230501 (2011).

Summary: This is an experimentally significant paper in quantum computing that not only sets a new record for quantum entanglement scale but, more importantly, proposes systematic technical solutions laying the foundation for large-scale quantum computing development. The paper achieves high standards in both technical innovation and experimental validation, representing an important milestone in quantum computing hardware advancement.