Fig. 3: Process reconstruction for unitary quantum circuits containing single-qubit and two-qubit quantum gates.
From: Quantum process tomography with unsupervised learning and tensor networks
a Reconstruction fidelity during the LPDO training for a circuits with N = 4 qubits containing a single layer of Hadamard gates. Different curves corresponds to an increasing size M of the data set. b Scaling of the minimum number of samples M* as a function of N to reach a reconstruction infidelity of ε = 0.025 (i.e., the sample complexity) for a circuit with Hadamard gates (red) and a circuit with random single-qubit rotations R(φj) (blue). c Reconstruction fidelity for a circuit with N = 4 qubits containing 4 layers of controlled-not (CX) gates, for various data set sizes M. d Sample complexity for quantum circuits with different depths D containing layers of CX gates. For the sample complexity plots, the value M* is obtained by sequentially increasing M until the threshold in accuracy is met. Error bars are given by the step-size in M, and dashed lines are linear fits.
