Qubit Mapping is a pivotal stage in quantum compilation flow. Its goal is to convert logical circuits into physical circuits so that a quantum algorithm can be executed on realworld non-fully connected quantum devices. Qubit Mapping techniques nowadays still lack the key to quantum advantage, scalability. Several studies have proved that at least thousands of logical qubits are required to achieve quantum computational advantage. However, to our best knowledge, there is no previous research with the ability to solve the qubit mapping problem with the necessary number of qubits for quantum advantage in a reasonable time. In this work, we provide the first qubit mapping framework with the scalability to achieve quantum advantage while accomplishing a fairly good performance. The framework also boasts its flexibility for quantum circuits of different characteristics. Experimental results show that the proposed mapping method outperforms the state-of-the-art methods on quantum circuit benchmarks by improving over 5% of the cost complexity in one-tenth of the program running time. Moreover, we demonstrate the scalability of our method by accomplishing mapping of an 11,969-qubit Quantum Fourier Transform within five hours.

No abstract

Recently, successful applications of reinforcement learning to chip placement have emerged. Pretrained models are necessary to improve efficiency and effectiveness. Currently, the weights of objective metrics (e.g., wirelength, congestion, and timing) are fixed during pretraining. However, fixed-weighed models cannot generate the diversity of placements required for engineers to accommodate changing requirements as they arise. This paper proposes flexible multiple-objective reinforcement learning (MORL) to support objective functions with inference-time variable weights using just a single pretrained model. Our macro placement results show that MORL can generate the Pareto frontier of multiple objectives effectively.