Quantum bits are prone to two types of errors due to uncontrolled interactions with their environment. To build a quantum computer, these errors must be reduced while maintaining the ability to control the qubit. In my PhD, I work with cat qubits, superconducting qubits whose bits are encoded in coherent states of a harmonic oscillator with opposite phases. To stabilize such states, we use a driven-dissipative scheme, where photons are exchanged in pairs with the environment. This kind of qubit is very interesting as it can be in a regime where one of the two types of errors is suppressed (noise bias qubit). To remove the other errors, the strategy is to encode the information redundantly across several qubits and regularly measure them to detect and correct the errors (quantum error correction).

To perform error correction, one needs a gate between two qubits which is called the CNOT. Depending on the state of the first qubit (control), a “NOT” operation is performed or not on the second qubit (target), corresponding to flipping the 0 to 1 and 1 to 0. The goal of my PhD is to experimentally implement such a gate in a way that doesn’t break the noise bias of the cat qubit. In other words, the CNOT gate should not reintroduce the error against which cat qubits are inherently protected.

​

This project is done at LPENS under the supervision of Zaki Leghtas, with Adrien Bocquet and in partnership with Alice & Bob (CIFRE PhD).