Towards High-performance Distributed Quantum Computing with Qubit Placement and Provisioning

In Distributed Quantum Computing (DQC), quantum bits (qubits) used in a task may be distributed on multiple Quantum Computers (QCs) connected by a Quantum Data Network (QDN). When we have to perform a quantum gate operation involving two qubits on different QCs, an Entanglement Connection (EC) has to be established between these two QCs. Since quantum gate operations can be performed at the data speed, the completion time of a DQC task is dominated by the time to establish ECs.To minimize the DQC task completion time, we propose QuPEP to jointly optimize data qubit (i.e., dbit) placement (that determines the number of ECs we have to establish between each pair of QCs) and entangled qubit (i.e., ebit) provisioning (that minimizes the time to establish each EC). QuPEP has an offline algorithm to optimize the dbit placement based on Genetic Simulated Annealing (GSA) and an online algorithm to optimize ebit provisioning based on Lagrange’s relaxation and the stochastic gradient descent method. By setting the fitness of each chromosome in GSA as the minimum DQC task completion time that can be achieved by the proposed online ebit provisioning scheme, QuPEP joints dbit placement and ebit provisioning. Extensive simulations show that compared with only optimizing dbit placement or ebit provisioning, QuPEP can reduce the DQC task completion time by up to 38% and 95%, respectively.