(英) |
In order to precisely design quantum operations and quantum analog simulations in quantum computers,it is essential to know the effective Hamiltonian of qubits. Quantum process tomography has been the mainstreammethod for estimating the effective Hamiltonian, but while it can estimate the effective Hamiltonian comprehensively,it cannot provide accurate evaluation due to the large estimation error. In addition, in order to obtain the matrixelements of the Kraus operator, we must evaluate $16^n$ matrix elements fornqubits, which requires a huge memorycost. To solve this problem, we propose an effective Hamiltonian estimation of superconducting qubits systems bycross entropy minimization. In this method, the memory cost can be reduced to $4^n$ even with the introduction ofnoise, and the effective Hamiltonian can be estimated in a realistic time even for a relatively large number of qubits.Numerical simulations assuming a superconducting qubit system are performed to estimate the computation time withrespect to the number of parameters, and it is shown that a relatively large number of qubits can be handled by usingparallel computation to speed up the computation time. |