Speeding up entangling gates for fixed-frequency superconducting qubits

Superconducting transmon qubits have a weak nonlinearity which leads to spectral crowding in multi-qubit circuits. This is one of the obstacles to fast and accurate entangling gates. Previously we introduced SWIPHT, an analytical protocol for designing fast, high-fidelity two-qubit gates, and we showed that SWIPHT exhibits high performance in a particular regime of static system parameters. Here, we show that this regime can be greatly extended by overcoming coherent errors and leakage through the use of the DRAG protocol and pulse shaping, respectively. This enables a reduction of SWIPHT CNOT gate times down to the order of 10 ns while maintaining fidelities above the 99% level using simple pulse shapes.