Bosonic qubits stored in microwave superconducting cavities are encoded in harmonic-oscillator modes to protect quantum information against dominant errors such as photon loss, dephasing, and residual nonlinearity. Engineers and theorists pursue encodings that trade off hardware simplicity, correctable error sets, and practical implementability under finite energy constraints.
Common bosonic encodings
The Gottesman-Kitaev-Preskill (GKP) code protects against small displacement errors by encoding logical states as lattice-like grid states in phase space. The GKP idea was introduced by Daniel Gottesman, Alexei Kitaev, and John Preskill; John Preskill at Caltech has written accessible expositions explaining the code’s ability to convert small continuous errors into discrete correctable shifts. GKP states are ideally unphysical because they require infinite energy, so realistic implementations use finite-energy approximations that balance protection and feasibility.
The cat code family encodes a qubit in superpositions of coherent states (Schrödinger cat states) and is inherently tailored to protect against single-photon loss when paired with engineered dissipation or two-photon drives. Stabilized cat qubits have been realized and tested in superconducting-cavity platforms by experimental groups led by Robert J. Schoelkopf and Michel H. Devoret at Yale, demonstrating extended lifetimes via active stabilization and error monitoring. Cat encodings exploit phase-space parity and can be made fault tolerant with bosonic error-correction routines.
Binomial and rotation-symmetric codes use finite superpositions of Fock states chosen so that dominant low-order error operators map logical states into orthogonal error spaces; these codes protect against a set of photon loss and dephasing errors with relatively low photon number. Each family offers different resource and syndrome-measurement costs, and some are easier to combine with transmon ancillas for syndrome extraction.
Practical performance, trade-offs, and context
Noise-resilient encodings differ in how they convert continuous physical errors into discrete logical faults and in how tolerant they are to realistic imperfections such as imperfect ancilla measurements, cavity nonlinearity, and finite-temperature effects. Experimentally, Yale’s cavity work shows that stabilization plus active correction can extend logical lifetimes but at the cost of increased control complexity and cryogenic infrastructure. From an environmental and territorial viewpoint, building and operating such dilution refrigerators concentrates research capabilities in well-funded labs, shaping where advances appear first and highlighting global inequality in access to quantum hardware. Choosing an encoding therefore depends on the dominant error model of the device, available control tools, and the broader institutional capacity to run long-duration, low-noise experiments.