[2210.04330] Benchmarks for Analog Quantum Computers — I: Metrology Aspects

[Submitted on 9 Oct 2022]

Download PDF

Overview: Accurate and precise control of large-scale quantum systems is paramount to achieving practical advantages in quantum devices. Therefore, benchmarking hardware errors in quantum computers has received a lot of attention recently. Existing benchmarks for digital quantum computers involve global fidelity averaging over a large set of quantum circuits, making them unsuitable for the specific multi-qubit control operations used in analog quantum operations. . Moreover, average global fidelity is not the optimal figure of merit for some applications specific to analog devices, such as many-body physics studies, which often use local observables. In this two-part paper, he develops a new figure of merit suitable for analog/many-qubit quantum operations based on reduced Choi matrices of operations. In the first part, we develop an efficient and scalable protocol to fully characterize the reduced Choi matrix. We identify two sources of sampling error in the measurement of reduced Choi matrices and show that there are fundamental bounds on the rate of convergence of sampling errors, similar to the canonical quantum and Heisenberg bounds. The slow convergence of the sampling error means that many experimental shots are required. We develop a protocol using quantum information scrambling observed in chaotic systems. This is used, for example, to speed up the convergence of sampling errors during state preparation. Additionally, the sampling error in the measurements using the squeezed and entangled initial states. This reduces the metrology-enhanced process tomography protocol.