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
–
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