Direct Randomized Benchmarking for Multiqubit Devices.

Area of Science:

• Quantum Information Science
• Quantum Computing
• Quantum Error Characterization

Background:

• Assessing quantum processor performance requires scalable benchmarking methods.
• Clifford randomized benchmarking (RB) is the industry standard but struggles to scale beyond a few qubits due to compilation inefficiencies.

Purpose of the Study:

• To develop and experimentally validate a direct randomized benchmarking (RB) protocol for efficiently benchmarking multiqubit quantum processors.
• To provide a more informative and flexible error rate compared to traditional Clifford RB.

Main Methods:

• Proposed a direct RB protocol using random native gate circuits seeded by Clifford-like randomization, minimizing compilation overhead.
• Experimentally demonstrated the protocol on a 2- to 5-qubit system using the ibmqx5 quantum processor.
• Performed simulations for 10+ qubit systems to show scalability.

Main Results:

• Successfully benchmarked quantum processors up to five qubits, a significant increase over previous Clifford RB implementations.
• The protocol successfully estimated an error rate from exponential decay, similar to Clifford RB.
• Demonstrated the protocol's flexibility by measuring separate error rates for distinct gate sets, including CNOT gates.

Conclusions:

• The direct RB protocol offers a practical and scalable solution for benchmarking larger quantum processors.
• This method provides a more direct and flexible error metric, crucial for advancing quantum hardware development.
• The protocol's ability to benchmark more qubits on accessible hardware marks a significant step in quantum processor assessment.