Use of Faulty States in Cat-Code Error Correction

1. Problem Statement

Bosonic codes, particularly cat codes (superpositions of coherent states with rotational symmetry), are promising candidates for fault-tolerant quantum computing and sensing. However, a major bottleneck in their implementation is the fault-tolerant preparation and maintenance of high-fidelity ancillary states required for error correction.

• The Challenge: Standard fault-tolerant schemes (like "tele-correction") typically require ancillary states that are perfect logical states (e.g., $|+\rangle$ states) within the code space. Preparing these states often requires complex multi-mode entanglement or strong non-linear interactions (Kerr non-linearities) that are difficult to realize experimentally.
• The Limitation: Existing proposals often rely on code concatenation (encoding a cat code within another code) to achieve fault tolerance, which is resource-intensive. Furthermore, generating high-fidelity multi-component cat states directly is hindered by the weakness of available non-linear interactions.
• The Gap: There is a need for a method to prepare ancillary states that relaxes the requirement for perfect code-space states, allowing for the use of "imperfect" or "faulty" states that can still facilitate syndrome extraction without compromising fault tolerance.

2. Methodology

The authors propose a teleportation-based error correction (tele-correction) scheme that utilizes generalized Yurke-Stoler (YS) states as ancillary inputs. These states are generated via Kerr non-linearities but possess relative phases between their coherent components that differ from standard cat code states.

Key Methodological Steps:

1. Generalized Yurke-Stoler States:
   • Instead of preparing perfect cat code states, the authors utilize $N$-component circular superpositions of coherent states generated by self-Kerr interactions ($H \propto \hat{n}^2$).
   • These states have a specific "incorrect" relative phase structure (a linear chirp signal) compared to the ideal cat code states.
2. Tele-correction Circuit Modification:
   • The standard tele-correction circuit (involving controlled rotations and heterodyne measurements) is adapted to accept these YS states.
   • The authors demonstrate that the "incorrect" phases of the YS states do not destroy the logical information but instead propagate through the circuit in a predictable manner.
3. Phase Correction Mechanisms:
   • Modular Photon Number Measurement: The authors show that the relative phases can be corrected by applying specific self-Kerr interactions (quadratic phase corrections) after the measurement.
   • Cross-Kerr Corrections: For the central rail of the tele-correction circuit, a combination of self-Kerr and cross-Kerr interactions is derived to remove the quadratic phase dependence introduced by the YS states, effectively "distilling" the correct logical state.
4. Noise Analysis:
   • The paper rigorously models four noise sources: dephasing, non-linear over/under-rotations, photon loss, and displacement noise (input state imperfections).
   • Crucially, the authors analyze loss during the non-linear interaction (mid-evolution loss), showing that while uniform loss is catastrophic, specific pulse shapes (e.g., Gaussian) can concentrate loss events in a way that preserves fault tolerance.

3. Key Contributions

• Broadening Ancillary State Acceptance: The paper proves that the class of acceptable ancillary states for cat-code error correction is broader than previously thought. States outside the code space (like YS states) can be used effectively if their phase errors are tracked and corrected.
• Bridge States for Weak Non-linearities: The proposal allows for the use of many-component states (large $N$) to simulate modular photon number measurements for smaller code distances ($K$). Since the required interaction strength scales as $(KN)^{-1}$, using large $N$ YS states reduces the demand on experimental non-linearities.
• Fault-Tolerant Phase Tracking: The authors derive explicit correction circuits (quadratic phase gates) that account for the deterministic phase errors introduced by YS states, ensuring the final logical state is recovered with high fidelity.
• Analysis of Mid-Evolution Loss: A significant contribution is the detailed analysis of photon loss occurring during the non-linear generation of the ancillary state. The authors show that while loss during interaction is generally dangerous, specific pulse shaping can mitigate the resulting phase errors, offering a path to fault tolerance even with imperfect materials.

4. Results

The authors provide quantitative estimates and numerical simulations to validate their proposal:

• Fidelity: Numerical simulations show that using YS states as inputs to the tele-correction circuit yields output state fidelities > 0.99 (specifically $\mu \approx 0.989$ with variance $\sigma^2 = 0.0043$) even in the presence of realistic noise.
• Error Suppression:
  • Dephasing & Rotation Errors: The measurement error rate is exponentially suppressed as the coherent state amplitude ($|\alpha|$) increases. The system tolerates relative phase rotation errors ($\theta$) up to $\sim 10^{-3}$ for 4-component states.
  • Displacement Noise: The protocol is robust against input displacement noise up to a scaling factor of $\sigma' \sim 10^{-3}$. Fidelity drops sharply only when noise exceeds this threshold.
• Loss Tolerance:
  • Loss occurring before or after the non-linear gate is manageable (compensated by amplitude reduction).
  • Loss during the gate is the critical failure point. However, for Gaussian pulse shapes, the induced phase errors are not uniformly distributed, allowing the code to retain protection.
• Material Feasibility: The appendix surveys non-linear materials. It concludes that standard materials (Silicon, GaAs) have $\chi^{(3)}$ values too low for direct implementation without massive $N$. However, near-resonant effects in coherent ensembles (e.g., hot rubidium vapors) or specific thermal/non-linear regimes might offer the necessary interaction strengths.

5. Significance

This work represents a significant step toward the practical realization of bosonic quantum error correction:

1. Resource Efficiency: By removing the strict requirement for perfect code-space ancillae, the proposal reduces the complexity of state preparation, potentially eliminating the need for code concatenation.
2. Experimental Viability: It offers a pathway to implement cat-code error correction with current or near-future non-linear optical technologies by trading off the number of coherent state components ($N$) against interaction strength.
3. Robustness: The detailed noise modeling, particularly regarding mid-evolution loss, provides a realistic framework for assessing fault tolerance in continuous-variable systems, moving beyond idealized discrete error models.
4. Theoretical Framework: The introduction of "bridge states" and the formalism for tracking their phase errors in tele-correction circuits provides a new theoretical toolset for designing fault-tolerant architectures for rotational symmetric codes.

In summary, Hanks et al. demonstrate that imperfect, non-code-space states can serve as effective resources for fault-tolerant cat-code error correction, provided that the specific phase errors they introduce are understood, tracked, and corrected. This relaxes the stringent hardware requirements for generating high-fidelity cat states, making the architecture more accessible to experimental realization.