Efficient Hamiltonian Parameter Estimation with Sequential Monte Carlo Technique

As quantum computers grow in size the task of calibrating them becomes more complex. There are many parameters to optimize for and the resulting performance of the qubits depends strongly on how well various quantities are measured. For example, knowledge of the coupling to neighboring qubits or resonators is necessary for two-qubit gates. In addition, it is preferable if qubits are located far away in frequency from harmful systems like Two-Level State defects.

By utilizing a frequency tunable qubit as a probe, we can measure the parameters of a resonance, namely the frequency and coupling strength, and use this information to calibrate gates or improve coherence times. Because these parameters fluctuate over time, this calibration should be repeated regularly. It is therefore important that it runs quickly. Using tunable transmon superconducting qubits, we demonstrate a method to efficiently measure the parameters of these resonant couplings. Initially, these parameters are estimated from a state of zero knowledge with a method based on Hamiltonian dynamics. Subsequently, a Sequential Monte Carlo technique can quickly refine and improve the accuracy of that estimate.