[Paper Review] Demonstration of quantum volume 64 on a superconducting quantum computing system

Petar Jurcevic, Ali Javadi-Abhari+29 more|arXiv (Cornell University)|Nov 29, 2022

Quantum Computing Algorithms and ArchitectureComputer Science55 references421 citations

TL;DR

This paper demonstrates a quantum volume of 64 on IBM's 27-qubit superconducting quantum processor, ibmq montreal, by integrating compiler optimizations, shorter two-qubit gates, dynamical decoupling for idle qubits, and excited state promoted (ESP) readout. The combination of hardware-aware software and pulse-level control improvements achieved a heavy output probability of 0.701 ± 0.031, exceeding the 2/3 threshold with 98.744% confidence, marking a significant advance in fault-tolerant quantum computing readiness.

ABSTRACT

We improve the quality of quantum circuits on superconducting quantum computing systems, as measured by the quantum volume, with a combination of dynamical decoupling, compiler optimizations, shorter two-qubit gates, and excited state promoted readout. This result shows that the path to larger quantum volume systems requires the simultaneous increase of coherence, control gate fidelities, measurement fidelities, and smarter software which takes into account hardware details, thereby demonstrating the need to continue to co-design the software and hardware stack for the foreseeable future.

Motivation & Objective

• To achieve a quantum volume of 64 on a superconducting quantum processor, advancing toward fault-tolerant quantum computing.
• To address the limitations of gate fidelities, measurement errors, and decoherence in noisy intermediate-scale quantum (NISQ) systems.
• To demonstrate that quantum volume improvements require co-design of software and hardware, including low-level pulse control and compiler optimization.
• To validate that holistic system metrics like quantum volume are essential for measuring progress beyond isolated gate fidelities.

Proposed method

• Implemented compiler optimizations in Qiskit to reduce circuit depth and gate count, minimizing error accumulation.
• Reduced two-qubit gate duration to 199–309 ns using a direct CNOT gate with rotary echo pulses and optimized pulse shaping.
• Applied dynamical decoupling sequences to idle qubits to suppress decoherence from low-frequency noise.
• Introduced excited state promoted (ESP) readout by applying a π-pulse to transfer population from |1⟩ to |f⟩ before measurement, enhancing signal separation and reducing readout error.
• Used linear discriminant analysis (LDA) for state discrimination in the I-Q plane to improve assignment fidelity.
• Calibrated and integrated all improvements within the Qiskit runtime stack, enabling low-level pulse control and hardware-aware compilation.

Experimental results

Research questions

• RQ1 Can quantum volume be increased beyond 32 on a superconducting quantum processor through integrated hardware and software improvements?
• RQ2 To what extent do dynamical decoupling, shorter two-qubit gates, and ESP readout individually and collectively improve quantum circuit fidelity?
• RQ3 How does hardware-aware compiler optimization impact circuit depth and error rates in real-world quantum systems?
• RQ4 Can pulse-level control and measurement enhancements like ESP readout significantly reduce SPAM errors without increasing gate complexity?
• RQ5 What is the combined effect of multiple fidelity-improving techniques on the holistic quantum volume metric?

Key findings

• The IBM Quantum Falcon processor ibmq montreal achieved a quantum volume of 64, confirmed by a heavy output probability of 0.701 ± 0.031, exceeding the 2/3 threshold with 98.744% confidence (z = 2.25).
• The combination of compiler optimization, shorter two-qubit gates (199–309 ns), dynamical decoupling on idle qubits, and ESP readout enabled the QV64 result.
• Excited state promoted (ESP) readout reduced the total assignment error from 0.10 (standard procedure) to 3.5 × 10⁻², with a reset error of 3.7 × 10⁻².
• The average T1 and T2 coherence times were 113 µs and 122 µs, respectively, supporting longer gate and circuit execution windows.
• Two-qubit gate infidelity was reduced to 6.4 × 10⁻³, and single-qubit gate error to 3.8 × 10⁻⁴, indicating high-fidelity gate operations.
• The result confirms that quantum volume is a sensitive, holistic benchmark requiring simultaneous improvements across gate fidelity, measurement fidelity, coherence, and compiler efficiency.