Cat code

Cat codes encode quantum information in superpositions of coherent states |alpha> and |-alpha>, often organized by even/odd photon parity or multicomponent rotational symmetry. Their central advantage is to turn dominant photon loss into detectable parity changes or strongly biased logical noise.

Two-component cats superpose opposite coherent amplitudes, while four-component and higher cats use discrete phase-space rotation orbits. A complete convention specifies logical basis, parity sector, alpha, and stabilization because papers may exchange the logical X/Z labels.

Single-photon loss changes parity and can be detected by repeated parity monitoring, while coherent-state overlap decreases exponentially with |alpha| squared and suppresses one logical Pauli component. Increasing alpha also raises photon number and loss-event rate, producing an engineering optimum.

A cat code often stores one logical qubit in one oscillator mode, but n=1 is not a sufficient resource statement. Report |alpha| squared, mean photon number, stabilization strength, cavity lifetime, parity-measurement cadence, and logical bit- and phase-flip rates.

Displacements, phase-space rotations, conditional phases, and engineered dissipation can implement bias-preserving logical gates. Operations that pass through small-alpha regions or convert loss into the unsuppressed Pauli component can destroy the bias; universality and bias preservation require separate verification.

Repeated parity records form a noisy time series that can be decoded with hidden-Markov, maximum-likelihood, or Pauli-frame tracking of loss events. Final parity alone cannot distinguish two losses from none and discards event timing.

Fault tolerance depends on ancilla backaction, parity-measurement propagation, pump-induced transitions, and leakage while preserving bias. A long-lived bosonic memory is not by itself a fault-tolerant set of gates, measurements, and resets.

High-Q microwave cavities with superconducting ancillas have demonstrated cat stabilization, repeated parity monitoring, and logical-lifetime gains; optical platforms prepare traveling cats. Cross-platform comparisons must normalize energy, preparation success, and postselection.

Cat, binomial, and GKP codes are all bosonic but use coherent-state separation, finite Fock support, and phase-space lattices respectively. They may share oscillator hardware while having different error sets, syndromes, and energy scaling.

“Cat code” covers distinct protocols including two-component, four-component, Kerr-cat, and dissipatively stabilized cats. External results transfer directly only when logical basis, component count, stabilizing Hamiltonian or dissipator, and noise model agree.