QTensor: Fast QAOA Quantum Simulator

ORAL

Abstract


We present a quantum circuit simulator* designed to execute large QAOA quantum circuits. The simulator is based on the tensor network representation of quantum circuits. To achieve this goal, we developed and implemented a number of techniques: lightcone optimization, custom gates to reduce tensor network complexity, and reduction of tensor dimensions through the utilization of diagonal structure of gates. As a result, we were able to execute large QAOA circuits to solve MaxCut. Using QTensor, we simulated quantum circuits with hundreds of qubits and large depths (p ≥ 7) within seconds and performed an extensive analysis of large QAOA circuits. We found that simulation complexity is non-monotonic with respect to the graph size due to the statistical properties of subgraphs of random-regular graphs.

*QTensor code is publicity available at https://github.com/danlkv/QTensor

*This research is supported by the U.S. Department of Energy, Office of Science, Basic Energy Sciences, Materials Sciences and Engineering Division, and by the Exascale Computing Project (17-SC-20-SC). This material is based upon work supported by the Defense Advanced Research Projects Agency (DARPA) under Contract No. HR001120C0068.

Presenters

  • Danylo Lykov

    • Argonne National Laboratory
    • Northern Illinois University

Authors

  • Danylo Lykov

    • Argonne National Laboratory
    • Northern Illinois University
  • Yuri Alexeev

    • Argonne National Laboratory
  • Cameron Ibrahim

    • Argonne National Laboratory
  • Alexey Galda

    • University of Chicago
    • James Franck Institute, University of Chicago