Ground State Misidentification in Superconducting Qubits

To achieve fault tolerant quantum computation, it is necessary to maximize measurement fidelity and minimize readout-induced measurement errors. A new protocol was developed to measure $P_1(|g>)$, the probability of measuring the excited state without exciting the qubit, while not including stray tunneling present in superconducting phase qubits. We have confirmed the expected trend in $P_1(|g>)$ with device temperature. We then compared $P_1(|g>)$ for phase qubits with different readout mechanisms and found that it is $\sim$3\% for our dispersive readout scheme and $\sim$1.5\% for our prior SQUID-based readout scheme. We have further applied microwave power to the flux bias and microwave drive lines to understand the source of this difference.