Gregory Baimetov

The main bottleneck for distributed quantum computing is the rate at which entanglement is produced between quantum processing units (QPUs). In this work, we prove that multiple QPUs connected through slow interconnects can outperform a monolithic architecture made with a single QPU. We present a distributed version of Clifford noise reduction (CliNR), a partial error correction scheme, and show that it outperforms a monolithic version of CliNR. Distributed CliNR has lower depth and lower logical error rates than monolithic CliNR even when the interconnects are slow. In simulations we show that the advantage persists with interconnects that are five times slower than two qubit gates. We also prove a sufficient condition for distributed CliNR to outperform monolithic CliNR. In addition, we present two methods for improving CliNR, allowing lower logical error rates and efficient performance for arbitrary length Clifford circuits. Finally we present results from an experimental implementation of a variant of CliNR on an ion trap quantum computer. This work is based on three papers that are available on arXiv (see extended abstract).