| bosonic codes | |
|---|---|
| Name | Bosonic codes |
| Type | Quantum error-correcting codes |
| Field | Quantum information science |
| Invented | 2000s–2010s |
| Inventors | Emanuel Knill; Peter Zoller; Sergio M. Girvin; John Preskill; David Gottesman; Barbara Terhal; others |
| Institutions | Caltech; Harvard University; Yale University; MIT; University of Innsbruck; Google Quantum AI; IBM Quantum |
| Notable works | Gottesman–Kitaev–Preskill code; cat codes; Binomial code |
bosonic codes
Bosonic codes are quantum error‑correcting codes that encode logical quantum information into bosonic degrees of freedom such as modes of the electromagnetic field or collective vibrational modes. They provide hardware‑efficient protection against common continuous‑variable errors (loss, dephasing, and thermal noise) encountered in implementations based on microwave cavities, optical modes, and mechanical resonators. Bosonic codes matter in Quantum information science and Quantum computing because they can reduce overhead relative to qubit‑based encodings and enable fault‑tolerant logical operations using native bosonic interactions in platforms developed by groups at Yale University, Caltech, IBM Quantum, and Google Quantum AI.
A bosonic mode is described by annihilation and creation operators a and a† and an infinite‑dimensional Hilbert space spanned by Fock states |n⟩. Continuous‑variable techniques developed in quantum optics and the theory of the harmonic oscillator provide the mathematical framework for bosonic encodings. Important mathematical constructs include the Wigner quasiprobability distribution, displacement operator, and squeezed states. Encodings leverage superpositions of Fock states, phase‑space lattices, or rotation‑symmetric subspaces to represent logical qubits or qudits, enabling error correction tailored to noise channels like photon loss described by the Lindblad equation.
Several families of bosonic codes have been proposed and studied:
- Cat codes: based on coherent state superpositions (Schrödinger cat states) and studied in works by S. M. Girvin and experimental groups at Yale University and University of Innsbruck. Cat codes protect against single‑photon loss using parity‑based stabilizers and have been implemented in circuit QED with superconducting qubit ancillas. - Binomial codes: introduced by researchers including Guido Burkard and Michael H. Devoret's collaborators; they use finite superpositions of Fock states with designed photon‑number moments to correct loss and dephasing up to a chosen order. - Gottesman–Kitaev–Preskill (GKP) code: proposed by Daniel Gottesman, Alexei Kitaev, and John Preskill; it encodes qubits into displacement‑periodic grid states in phase space and corrects small shift errors. GKP codes connect to modular variables and have motivated experimental efforts in trapped ions, optical frequency combs, and microwave cavities. - Rotation‑symmetric (binomial/number‑parity generalizations): codes exploiting discrete rotational symmetry in phase space (e.g., four‑fold symmetry) to protect against rotations and dephasing; theoretical development involves symmetry groups and stabilizer formalisms related to stabilizer codes.
Each family balances protection, resource cost, and ease of preparation/measurement, and typical analyses compare logical error rates, fault tolerance thresholds, and overhead.
Bosonic error models include photon loss (amplitude damping), dephasing (phase diffusion), thermal noise, and non‑Gaussian error processes such as Kerr nonlinearities and leakage via higher modes. Photon loss is commonly modeled by a quantum channel with jump operator a and described by the master equation in Lindblad form. Dephasing arises from stochastic phase shifts and is modeled by number‑operator coupling to baths. Non‑Markovian environments and correlated errors appear in multimode settings and in hybrid platforms coupling cavities to superconducting qubits or mechanical resonators. Accurate noise characterization uses techniques like quantum process tomography, randomized benchmarking adapted to continuous variables, and Wigner tomography.
Encoding and decoding of bosonic logical states employ unitary gates, engineered dissipation, and ancilla‑assisted operations. Techniques include autonomous stabilization via driven dissipative processes (reservoir engineering) pioneered in circuit QED experiments at Yale University and ETH Zurich, gate‑based encodings using controlled displacements and conditional phase gates with transmon ancillas, and state‑transfer protocols from trapped ions or optical modes. Syndrome extraction commonly uses parity measurements, modular quadrature measurements for GKP codes, and photon‑number selective rotations implemented with Josephson junction circuits. Error correction can be performed via active feedback or via continuous monitoring and reservoir‑engineered stabilization.
Bosonic codes can be concatenated with qubit‑level surface codes or Steane code type stabilizers to achieve fault tolerance with reduced overhead; logical bosonic qubits serve as lower‑level encodings that suppress dominant bosonic errors before qubit‑level correction. Fault‑tolerant gate constructions exploit bias‑preserving gates for cat codes, lattice‑surgery‑like protocols for GKP, and teleportation‑based logical operations. Thresholds and resource estimates have been derived in theoretical studies from groups including John Preskill and Emanuel Knill, showing that hybrid bosonic–qubit architectures can lower the number of physical qubits required for scalable quantum computation.
Experimental platforms demonstrating bosonic codes include microwave cavities coupled to transmon qubits in circuit QED (landmark experiments from Yale University and ETH Zurich), trapped ions implementing GKP‑like encoding, optical implementations using squeezed light and photonic integrated circuits, and mechanical resonators. Industry efforts by IBM Quantum and Google Quantum AI explore bosonic encodings for logical qubits in superconducting architectures. Key milestones include demonstrations of logical state preparation, syndrome extraction via ancilla qubits, and multi‑millisecond logical coherence times in protected modes. Ongoing challenges include high‑fidelity state preparation, scalable syndrome readout, and integrating bosonic modules into modular fault‑tolerant processors.
Category:Quantum error correction Category:Quantum optics Category:Quantum computing