A dataset for quantum circuit mapping

Area of Science:

• Quantum Computing
• Machine Learning
• Computational Science

Background:

• Quantum computing offers advantages over classical computing but faces challenges in mapping algorithms to physical processors.
• Current deterministic mapping algorithms are time-consuming, hindering scalability for next-generation quantum computers with thousands of qubits.

Purpose of the Study:

• To address the quantum circuit mapping problem by proposing the first datasets suitable for a classification task.
• To facilitate the development of machine learning models for efficient quantum circuit mapping.

Main Methods:

• Creation of datasets comprising random quantum circuits mapped to specific IBM quantum processors.
• Each dataset instance includes device calibration data and quantum circuit features.
• Instances are labeled with optimal mappings derived from deterministic algorithms.

Main Results:

• The proposed datasets enable the training of machine learning models for quantum circuit mapping.
• These models can achieve mapping quality comparable to deterministic methods.

Conclusions:

• The developed datasets significantly reduce the time required for quantum circuit mapping.
• This research paves the way for more efficient quantum algorithm design and deployment on larger quantum systems.