A score for comparing superconducting qubit chips of any design

Researchers from The University of Osaka and QuEL, Inc. have developed a new benchmark, percolating entanglement quality (PEQ), that assesses how well quantum entanglement spreads across a processor. The metric can inform the design of scalable chips.

Percolating Entanglement Quality (PEQ) as a Benchmark

The assessment of quantum processors now extends beyond simple qubit counts with the introduction of a new metric, percolating entanglement quality (PEQ), which evaluates how effectively entanglement spreads across a chip’s architecture. This benchmark normalizes for topology, addressing a limitation of previous methods that focused solely on the size of entanglement clusters without considering the quality of individual connections. The development of PEQ allows for meaningful comparisons between processors built with differing designs, as this capability is important as the field diversifies.

PEQ is based on measurements of conditional two-qubit negativity and a topology-specific bond-percolation threshold, providing a single, chip-wide score that doesn’t rely on identifying a subset of best-performing qubits. The team demonstrated PEQ by preparing a graph state on a 64-qubit superconducting processor, identifying a 52-qubit entanglement cluster, and then evaluating the processor’s overall entanglement quality.

This approach moves beyond simply verifying the presence of entanglement to quantifying its robustness and distribution across the chip’s network of qubits. A key advantage of PEQ lies in its ability to compare processors with fundamentally different coupling topologies, which dictate how many connections must remain intact for an entanglement cluster to persist. This demonstrates that PEQ can discern performance nuances beyond overall cluster size, offering a more granular assessment of processor capabilities.

“PEQ reflects the quality of entanglement generated during chip-wide operation,” the authors write, noting that this quality is affected by factors like qubit coherence, gate imperfections and crosstalk. The team’s analysis highlights how PEQ complements existing quantum benchmarks, each of which focuses on different aspects of processor performance. While metrics like error per logical gate (EPLG) assess the accuracy of individual operations, PEQ evaluates the entanglement of the prepared state across the entire chip.

The researchers explain that defining a percolating gate fidelity, ranking two-qubit gate fidelities and applying the percolation threshold, can further characterize gate quality across the full chip. This multi-faceted approach allows for a more comprehensive understanding of a processor’s strengths and weaknesses. PEQ is not limited to specific lattice structures like square or heavy-hex designs.

The methodology can be adapted to any coupling topology by determining the bond-percolation threshold for the corresponding infinite lattice. This adaptability positions PEQ as a versatile benchmark for evaluating a wide range of quantum processor architectures, potentially accelerating the development of scalable quantum computing.

Graph States Measure Superconducting Processor Entanglement

Unlike previous benchmarks focused solely on qubit count, PEQ evaluates how effectively entanglement, a core quantum phenomenon, spreads across the processor, providing a more subtle understanding of performance. The metric relies on conditional two-qubit negativity, a measure of entanglement between connected qubits, and a topology-specific bond-percolation threshold, allowing for comparisons even between processors with fundamentally different connection layouts. This experiment involved coordinated single- and two-qubit operations, revealing the combined performance of both the processor hardware and the control systems during simultaneous operation.

The team then analyzed the resulting entanglement, reconstructing the two-qubit state between coupled qubits based on measurement outcomes of their neighbors, and evaluating the negativity of these states to create a spatially resolved entanglement map. This map not only revealed the overall coupling quality but also pinpointed the largest entanglement cluster present on the chip.

PEQ moves beyond simply identifying the largest cluster, as cluster size alone does not fully represent the quality of entanglement within individual connections. Processors can achieve similar cluster sizes with varying degrees of entanglement strength, making size an incomplete metric. All 68 measured couplings had negativity estimates exceeding twice their resampling standard deviations, indicating a robust level of entanglement.

PEQ Evaluation on 64-Qubit Square-Lattice Processor

The ability to objectively compare quantum processors with differing architectures received a boost with the development of a new metric yielding a PEQ of 0.42 on a 64-qubit square-lattice processor. This score, termed percolating entanglement quality, or PEQ, offers a single value reflecting the overall entanglement quality of a chip, regardless of its specific coupling topology. The research demonstrates that PEQ can differentiate between processors that appear similar based on cluster size alone.

While several heavy-hex-lattice processors have demonstrated entanglement clusters spanning their entire chips, their PEQ values varied substantially, ranging from 0.20 to 0.74. This highlights the limitations of relying solely on cluster size as a benchmark. This suggests that the arrangement of connections, and their individual fidelity, is important for overall processor performance.

The methodology behind PEQ focuses on evaluating entanglement at a specific point, the percolation threshold, where the network of connections begins to break down. “PEQ provides a common basis for comparing different chip designs,” the researchers write, emphasizing the metric’s potential to standardize evaluation across the rapidly diversifying field of quantum computing hardware. This process probes the combined performance of the processor and its control system during chip-wide operation, offering a detailed view of its capabilities.

The researchers note that PEQ complements existing benchmarks, which often focus on verifying specific aspects of quantum computation. Unlike circuit fidelity measurements, which require computationally intensive classical simulations and become impractical with increasing qubit counts, PEQ provides a single chip-wide score without requiring a selection of best-performing qubit subsets or paths. This makes it particularly valuable for assessing the scalability of different coupling topologies. By providing a topology-normalized score, PEQ facilitates a more meaningful comparison of different chip designs and can inform the development of more scalable and efficient quantum computing architectures.

PEQ Differentiates Chip-Wide Entanglement Performance

Applying the percolating entanglement quality (PEQ) metric to existing data reveals substantial differences in entanglement performance among quantum processors with similarly sized entanglement clusters, demonstrating that qubit count alone is an incomplete measure of capability. This variation emphasises how PEQ offers a more detailed assessment than simply identifying the largest entanglement cluster, a metric that can be misleading when comparing differing architectures.

This detailed evaluation is important because PEQ characterizes entanglement quality at the percolation threshold specific to each processor’s coupling topology, a factor that influences how effectively entanglement spreads across the chip. The team’s work on a 64-qubit square-lattice processor with defects yielded a PEQ of 0.42, alongside the identification of a 52-qubit entanglement cluster, establishing a baseline for comparison with other designs.

The researchers also introduced the concept of entanglement yield (EY), a normalization factor that accounts for chip size, allowing for comparisons between processors with varying qubit numbers. EY is calculated by dividing the number of qubits within the largest connected entanglement cluster by the total number of qubits on the chip, providing a measure of how efficiently the processor uses its resources.

For the 64-qubit square-lattice processor, EY(0), the yield at zero cutoff, where all couplings with positive negativity are retained, was found to be 0.81, indicating that 81% of the qubits participate in a single connected entanglement cluster. Combining EY with PEQ, the team provides two complementary measures that describe both the extent and quality of chip-wide entanglement, offering a comprehensive basis for comparison across processors. The resulting PEQ value of 0.42 ± 0.04 characterizes the entanglement quality supporting percolation on the square lattice.

Topology-Normalized PEQ Accounts for Coupling Quality

The new metric, percolating entanglement quality (PEQ), delivers a single, normalized score for assessing the entanglement capabilities of any superconducting processor, moving beyond reliance on qubit count as the primary measure of performance. Unlike previous methods, PEQ accounts for the specific coupling topology of each chip, a critical factor in sustaining entanglement across the entire system. These maps not only reveal the extent of chip-wide entanglement but also pinpoint the quality of individual couplings, identifying areas of strength and weakness.

The utility of PEQ extends beyond simply quantifying entanglement; the metric distinguishes between processors with similar cluster sizes but differing coupling quality. For instance, a square-lattice processor achieved an entanglement yield (EY) of 0.81, indicating 81% of qubits participate in a single connected cluster, yet its PEQ of 0.42 contrasted sharply with the 0.20 PEQ of ibmq_rochester, despite the latter’s perfect cluster coverage.

The team’s work establishes PEQ as a complementary measure to existing benchmarks, providing a detailed assessment of chip-wide entanglement performance. PEQ’s ability to generate a single chip-wide score without requiring selection of a best-performing qubit subset or path further streamlines the evaluation process.

By focusing on the percolation threshold, the minimum number of couplings needed to maintain connectivity, the metric offers a topology-specific benchmark that accurately reflects the inherent limitations and strengths of each processor’s design. This detailed analysis, summarized in a zero-to-one scale, provides a clear and concise measure of entanglement quality, facilitating objective comparisons and accelerating progress in quantum computing hardware development.

👉 More information
🗞 Benchmarking superconducting quantum processors with percolating entanglement quality
✍️ Akinori Machino, Ryo Matsuda, Nilton F. G. Filho and Makoto Negoro (Graduate School of Engineering Science, The University of Osaka, Toyonaka, Osaka, Japan); Kazuhisa Ogawa, Takefumi Miyoshi, Hidehisa Shiomi, Shinichi Morisaka and Takafumi Miyanaga (Center for Quantum Information and Quantum Biology, The University of Osaka, Toyonaka, Osaka, Japan); Koichiro Ban (Graduate School of Information Science and Technology, The University of Osaka, Suita, Osaka, Japan); Keisuke Koike (e-trees.Japan, Inc., Hachioji, Tokyo, Japan); Ryutaro Ohira, Toshi Sumida, Yoshinori Kurimoto, Yuuya Sugita and Yosuke Ito (QuEL, Inc., Hachioji, Tokyo, Japan); Yasunari Suzuki, Peter A. Spring, Shiyu Wang, Arvind Mamgain, Shuhei Tamate, Yutaka Tabuchi and Yasunobu Nakamura (RIKEN Center for Quantum Computing, Wako, Saitama, Japan); Hiroto Mukai (Graduate School of Engineering, The University of Tokyo, Tokyo, Japan)
🧠 ArXiv: https://arxiv.org/abs/2611.17941v1

Stay current

See today’s quantum computing news on Quantum Zeitgeist for the latest breakthroughs in qubits, hardware, algorithms, and industry deals.

Avatar of Dr. Donovan

Latest Posts by Dr. Donovan: