Cat Qubits Bit Flip Suppression

Short answer: A cat qubit stores quantum information in superpositions of two coherent states of a harmonic oscillator, and engineered two-photon dissipation can make bit-flip errors exponentially rarer as the photon number grows. A 2020 Nature Physics experiment measured that exponential suppression in a superconducting resonator, and later experiments report bit-flip times of 0.1 to 0.3 seconds. These are single-qubit laboratory results, and phase-flip errors remain to be corrected.

The idea

The 2020 Nature Physics paper says quantum systems interact with their environment, however slightly, so quantum bits undergo errors that put difficult constraints on hardware. It describes a strategy of encoding a quantum bit non-locally across the phase space of a physical system. Because most sources of decoherence come from local fluctuations, the promise is to suppress errors exponentially by increasing a measure of the non-locality. Topological qubits are one example, where spatial extent measures non-locality.

The 2020 experiment

The authors encode a quantum bit in the field quadrature space of a superconducting resonator equipped with a mechanism that dissipates photons in pairs. This pins two computational states to separate locations in phase space. By increasing the separation, they measured an exponential decrease of the bit-flip rate while only linearly increasing the phase-flip rate. Since bit-flips are corrected autonomously, only phase-flips remain to be corrected, via a one-dimensional quantum error correction code. They conclude that resonators with nonlinear dissipation are promising building blocks with drastically reduced hardware overhead. That conclusion is the authors' view.

A 2024 circuit with a filtered buffer

A 2024 arXiv paper describes cat qubits stabilized by two-photon dissipation, in which pairs of photons are added and removed from a harmonic oscillator by a lossy buffer mode. It says this needs a large loss rate and strong nonlinearities in the buffer mode that must not degrade the oscillator. The authors colour the loss environment of the buffer with a multi-pole filter and optimize the circuit, and report near-ideal growth of bit-flip times with photon number, reaching over 0.1 seconds at a mean photon number of only 4. They report phase-flip times corresponding to an effective lifetime of about 70 microseconds, comparable with the bare oscillator lifetime, even with an ancilla transmon for readout. They also describe pulsed stabilization, in which stabilization stays off for a significant fraction, for example two thirds, of a 3 microsecond cycle without degrading bit-flip times.

An autoparametric circuit

A Physical Review X paper describes cat qubits whose logical states are coherent states of a harmonic mode, stabilized by exchanging photon pairs for single photons of the environment, so bit-flip time grows exponentially with photon number. The authors introduce and operate an autoparametric superconducting circuit coupling the cat-qubit mode to a lossy mode at twice its frequency. The coupling is passive and needs no parametric pump, reaching a two-photon dissipation rate of about 2 MHz divided by 2 pi. They report bit-flip errors prevented for a characteristic time up to 0.3 seconds with only a mild impact on phase-flip errors, and show that the phase of a superposition can be changed while the dissipation stays active. The paper says the remaining phase-flip errors would be corrected in a repetition code of cat qubits.

What this does not tell you

These papers demonstrate individual cat qubits, not an error-corrected processor. Phase-flip suppression and multi-qubit gates are separate challenges, and these abstracts do not report them at scale.

FAQ

What is a cat qubit?

A qubit whose logical states are coherent states of a harmonic oscillator, per the Physical Review X paper.

Why does it suppress bit flips?

Two-photon dissipation pins the states apart in phase space, and bit-flip rates fall exponentially as the separation grows, per the 2020 paper.

How long did bit-flip times get?

Over 0.1 seconds at a mean photon number of 4 in one paper, and up to 0.3 seconds in another.

What about phase flips?

They increase only linearly in the 2020 experiment and must be corrected, for example with a repetition code.

Is this a working quantum computer?

No. These are experiments on single cat qubits.