Dual-Rail Encoding

• Dual-rail encoding is a method that represents logical bits or qubits using two orthogonal physical modes, mapping each logical state to a unique single excitation pattern.
• It enables robust error detection by converting physical errors such as photon loss and amplitude damping into identifiable erasure events across superconducting, optical, and ion-trapped platforms.
• This encoding underpins scalable quantum error correction, high-fidelity cluster-state generation, and reliable asynchronous digital logic in classical systems.

Dual-rail encoding is a paradigm for representing and manipulating quantum or classical information using two orthogonal physical modes per logical data unit (qubit or bit). The defining property is that logical values are mapped to a single excitation shared between two distinct modes, such that logical “0” and “1” are associated with mutually exclusive, physically distinguishable occupancy patterns. This redundancy provides robust error detection capability by mapping dominant physical errors (such as photon loss, amplitude damping, or wire faults) to easily identifiable erasure events. Dual-rail encodings are realized in a variety of platforms, including superconducting circuits, linear optical and nanophotonic systems, vibrational modes of trapped ions, and asynchronous delay-insensitive classical electronics. This encoding underpins advances in scalable quantum error correction, high-fidelity cluster-state generation, and hardware-robust asynchronous computation.

1. Dual-Rail Encoding: Formal Definition and Physical Realizations

In dual-rail encoding, a logical bit or qubit is represented by the joint state of two physical modes, ensuring at most one mode is excited at a time. For quantum systems, this typically means exactly one bosonic excitation (e.g., photon, phonon) shared between two modes:

• Quantum basis:

$$|0\rangle_L = a_1^\dagger|vac\rangle\,,\quad |1\rangle_L = a_2^\dagger|vac\rangle$$

or, equivalently, $$|0\rangle_L \equiv |10\rangle,\, |1\rangle_L \equiv |01\rangle$$, with the indices denoting occupation numbers in the two modes (Wang et al., 14 Aug 2025, Teoh et al., 2022, Østfeldt et al., 2021).

• Classical basis (asynchronous circuits):

$$\text{Bit 0:}\ (x_0,\,x_1) = (1,\,0)\ ;\quad \text{Bit 1:}\ (x_0,\,x_1) = (0,\,1)$$

with $(0,0)$ representing the idle or “spacer” state, and $(1,1)$ forbidden (Balasubramanian et al., 2017, Balasubramanian, 2018).

Physical realizations span:

• Superconducting microwave cavities and transmons: logical states correspond to single-photon Fock states distributed over two bosonic modes (Teoh et al., 2022, Chou et al., 2023, Koottandavida et al., 2023, Wills et al., 18 Jun 2025).
• Optical photonics: spatial or polarization modes of photons in linear optics or guided waveguide systems (Østfeldt et al., 2021, Dhara et al., 2021, Drahi et al., 2019).
• Trapped ion vibrational modes: each logical qubit uses exactly one phonon shared between two normal modes (Kang et al., 19 May 2025).
• Asynchronous digital electronics: two wires per logical bit, manipulating pulse transitions with four-phase handshaking (Balasubramanian et al., 2016, Balasubramanian et al., 2017, Balasubramanian, 2018).

2. Error Detection and Erasure Conversion

Dual-rail encoding inherently maps dominant physical errors to heralded erasure events. In quantum settings, photon loss, amplitude damping, or energy decay events remove population from the one-excitation manifold, sending the system into the “vacuum” or higher-excitation subspaces outside the logical code space. Such leakage is immediately identified and can be flagged as an erasure (Teoh et al., 2022, Chou et al., 2023, Wills et al., 18 Jun 2025, Huang et al., 16 Apr 2025, Koottandavida et al., 2023):

• Superconducting dual-rail cavity and transmon systems achieve $>99\%$ efficiency in detecting photon-loss events as erasures (Chou et al., 2023, Huang et al., 16 Apr 2025).
• In frequency-bin dual-rail photonic qubits, absence of an excitation in both frequency modes immediately signals erasure (Wang et al., 14 Aug 2025).
• Asynchronous dual-rail logic in delay-insensitive circuits ensures that any missing or simultaneous transition on both rails is detected as illegal or “spacer” (Balasubramanian et al., 2017, Balasubramanian, 2018).
• In trapped-ion vibrational dual-rail encoding, phonon leakage outside the single-phonon subspace is detected by parity checks or QND measurements (Kang et al., 19 May 2025).

This erasure-dominance is quantified experimentally, e.g., in superconducting systems where the logical error hierarchy is $$p_\text{erasure} \sim 0.2\%/\mu\text{s} \gg p_\phi \sim 0.02\%/\mu\text{s} \gg p_X \sim 10^{-5}/\mu\text{s}$$ (Chou et al., 2023), and in multi-mode dimon devices where logical bit- and phase-flip rates are improved by over an order of magnitude relative to physical error rates (Wills et al., 18 Jun 2025).

3. Quantum Information Processing with Dual-Rail Encoding

Dual-rail encoding supports universal gate sets, high-fidelity cluster state generation, and scalable architectures:

• Single- and two-qubit gates are engineered to preserve the single-excitation (logical) subspace. This is achieved via beam-splitter pulses, frequency-converting couplers, or exchange interactions that conserve photon or phonon number:
  • In superconducting cQED, universal control is achieved with beamsplitter unitaries and controlled-parity or $ZZ(\theta)$ gates (Teoh et al., 2022, Huang et al., 16 Apr 2025, Guan et al., 15 Jan 2025).
  • Frequency-bin dual-rail cluster states are generated using time-frequency multiplexing of a transmon and resonator, with explicit emission protocols producing up to 11-qubit logical chains (Wang et al., 14 Aug 2025).
  • Trapped-ion systems use state-dependent beamsplitters and conditional sideband transitions to engineer arbitrary single- and multi-qubit unitaries within the dual-rail manifold (Kang et al., 19 May 2025).
• Cluster state preparation:
  • Frequency-bin encoded clusters: An $N$-qubit linear dual-rail cluster is prepared by sequential photon emission and circuit-level Hadamard and controlled-phase operations (Wang et al., 14 Aug 2025).
  • Chiral nanophotonic sources deterministically generate spatial dual-rail Bell pairs by engineering light-matter interactions to ensure one photon per rail per emission cycle (Østfeldt et al., 2021).
• Dense-encoding and channel capacity: Dual-rail encoding in linear optics, when lifted beyond the pure rail-qubit subspace to the full Fock space, increases the classical channel capacity, enabling protocols that transmit more information than is possible with strict qubit encoding (Smith et al., 2015).

4. State Characterization and Metrics

State fidelity, entanglement, and error metrics in dual-rail platforms are quantified by a range of experimental and theoretical tools:

• Quantum state tomography and heterodyne measurement are used for full density matrix reconstruction in multi-mode photonic and superconducting devices, often leveraging efficient matrix-product-operator (MPO) representations (Wang et al., 14 Aug 2025).
• Logical-subspace fidelity, after discarding or postselecting on erasure events, quantifies performance exclusive of hardware-induced leakage. For frequency-bin clusters, raw fidelities above 50% are sustained up to $N=5$, and logical-subspace fidelities remain above 50% up to $$|0\rangle_L \equiv |10\rangle,\, |1\rangle_L \equiv |01\rangle$$0 (Wang et al., 14 Aug 2025).
• Localizable entanglement, measured via concurrence after Pauli measurements, is used to verify multipartite entanglement; entanglement persists over extended dual-rail chains, up to 11 physical qubits in leading implementations (Wang et al., 14 Aug 2025).
• In hardware implementations, error rates for bit-flip, phase-flip, and erasure are extracted from decay and tomography experiments, defining error hierarchies at the $$|0\rangle_L \equiv |10\rangle,\, |1\rangle_L \equiv |01\rangle$$1 or lower level for Pauli errors, and percent-level for erasures (Chou et al., 2023, Koottandavida et al., 2023, Huang et al., 16 Apr 2025).

5. Applications: Quantum Error Correction, Photonic Computation, and Delay-Insensitive Classical Logic

The advantages of dual-rail encoding drive progress in resource-efficient quantum error correction, scalable photonic quantum computing, and robust classical asynchronous design.

• Quantum error correction:
  • Erasure-dominated noise models in surface codes have much higher thresholds—up to 50% for pure erasure channels, compared to $$|0\rangle_L \equiv |10\rangle,\, |1\rangle_L \equiv |01\rangle$$2 for depolarizing noise (Koottandavida et al., 2023). Dual-rail qubits enable this regime by converting amplitude damping to flagged erasures.
  • Concatenated erasure codes exploit the high-fidelity detection of erasures to reduce logical failure rates exponentially with code distance or level (Huang et al., 16 Apr 2025).
  • Bias-preserving gates in dual-rail cavity systems maintain low infidelity ($$|0\rangle_L \equiv |10\rangle,\, |1\rangle_L \equiv |01\rangle$$3) and low erasure rates ($$|0\rangle_L \equiv |10\rangle,\, |1\rangle_L \equiv |01\rangle$$4 per gate) while suppressing bit-flip errors below $$|0\rangle_L \equiv |10\rangle,\, |1\rangle_L \equiv |01\rangle$$5 (Mehta et al., 13 Mar 2025, Huang et al., 16 Apr 2025).
• Photonic and circuit-based cluster state generation:
  • Frequency-bin and spatial dual-rail encodings enable scalable, robust entangled-state generation with intrinsic loss detection, facilitating measurement-based computation or quantum networking (Wang et al., 14 Aug 2025, Østfeldt et al., 2021, Dhara et al., 2021).
• Asynchronous delay-insensitive digital logic:
  • Dual-rail codes (1-of-2) provide simple, robust data representation immune to wire delays and enable early-output, area-efficient arithmetic (e.g., ripple-carry adders), often outperforming higher-order codes in power, area, and robustness (Balasubramanian et al., 2016, Balasubramanian et al., 2017, Balasubramanian, 2018).

6. Scalability, Architectural Integration, and Outlook

Dual-rail encoding is highly scalable due to efficient error detection and compatibility with existing hardware platforms:

• Hardware integration: Dual-rail qubits can be implemented with minimal hardware overhead—each logical qubit requiring only two physical modes and a dispersively coupled ancilla for erasure detection (Teoh et al., 2022, Koottandavida et al., 2023). In trapped ions, full connectivity and all-to-all couplings are retained when dual-rail vibrational modes are used as qubit carriers (Kang et al., 19 May 2025).
• Frequency-bin encoding packs multiple logical modes into a single photonic channel, allowing for time-frequency multiplexed architectures and modular extension to 2D cluster-state generation (Wang et al., 14 Aug 2025).
• Fault-tolerance: Erasure conversion greatly raises error correction thresholds, reducing the physical overhead required to achieve a target logical failure rate. Hardware-efficient, mid-circuit erasure checks minimize time overhead and facilitate “soft” decoding strategies using analog information from physical measurements (Hung et al., 17 Apr 2026).
• Future directions include hybrid quantum networks exploiting dual-rail–single-rail–continuous-variable interconversion (Drahi et al., 2019), multiplexed deterministic photonic Bell-pair sources (Dhara et al., 2021), and concatenated erasure codes with sub-millisecond-level logical coherence (Huang et al., 16 Apr 2025).

---

In summary, dual-rail encoding unifies robust error-detectable information representation across quantum and classical domains. Its principles—excitation-parity-based logical coding, efficient erasure conversion, and operational compatibility with universal gate sets—are foundational for current and emerging platforms in quantum error correction, scalable cluster-state computation, high-fidelity photonic entanglement sources, and asynchronous classical circuits (Wang et al., 14 Aug 2025, Teoh et al., 2022, Chou et al., 2023, Huang et al., 16 Apr 2025, Koottandavida et al., 2023, Balasubramanian et al., 2017).