LNCY [[4,1,2]] Quantum Error Code

• The LNCY code is a four-qubit [[4,1,2]] error-detecting scheme that encodes one logical qubit using GHZ and pairwise flipped states to flag single-qubit errors.
• It uses a set of three commuting stabilizers and logical operators to enable syndrome extraction and post-selection recovery under amplitude-damping noise.
• Its integration in distributed quantum architectures offers enhanced fidelity and resource efficiency compared to simpler repetition codes and high-overhead distillation protocols.

The Leung-Nielsen-Chuang-Yamamoto (LNCY) code denotes a family of four-qubit quantum error-detecting and -correcting codes, with primary focus on the [[4,1,2]] code for amplitude damping and noise mitigation in near-term quantum information processing. These codes were first introduced as approximate quantum error-correcting codes capable of detecting and partially correcting single-qubit errors, and have since assumed central status in quantum error detection (QED) strategies for distributed quantum computing and as benchmarks for channel-adapted code optimization.

1. Logical Structure, Codeword Construction, and Stabilizers

The LNCY [[4,1,2]] code encodes one logical qubit into four physical qubits, engineered to detect any single-qubit error. Logical codewords in the computational basis are: $$\begin{aligned} |0_L\rangle &= \frac{1}{\sqrt{2}}\big(|0000\rangle + |1111\rangle\big), \ |1_L\rangle &= \frac{1}{\sqrt{2}}\big(|0011\rangle + |1100\rangle\big). \end{aligned}$$ Codeword $|0_L\rangle$ is a four-qubit GHZ state, while $|1_L\rangle$ results from pairwise flipping of middle qubits in $|0_L\rangle$ (Mao et al., 2024).

A convenient generating set of three commuting stabilizers is: $$\begin{aligned} S_1 &= Z_1Z_2Z_3Z_4, \ S_2 &= Z_1Z_3, \ S_3 &= X_1X_2X_3X_4, \ \end{aligned}$$ where all products are tensor products ($Z_1Z_3 = Z\otimes I \otimes Z\otimes I$). Logical operators can be defined as: $Z_L = Z_1Z_2,\quad X_L = X_1X_3,$ which satisfy the usual commutation relations with code stabilizers and mutually anticommute ($\{Z_L,X_L\}=0$) (Campbell et al., 21 Jan 2026). The code distance is 2, enabling single-error detection but not correction.

2. Encoding and Recovery Operations

Encoding is achieved by an isometry $U_{\text{enc}}$ such that: $$\begin{aligned} |0\rangle_{\text{in}} &\mapsto |0_L\rangle, \ |1\rangle_{\text{in}} &\mapsto |1_L\rangle. \end{aligned}$$ Two explicit encoding circuits are prevalent. The “4QED” encoder executes:

1. CNOT from $p_0$ (data) to $p_1$, $p_2$, $p_3$ (ancillae).
2. Hadamard on $p_0$.
3. CNOTs: $p_0\to p_3$ and $p_1\to p_3$.

The “SS” (short Shor) encoder is a gate-reduced alternative involving:

1. Hadamard on $p_1$.
2. CNOT $p_1\to p_3$, $p_1\to p_0$, $p_1\to p_2$.
3. Hadamard on $p_1$.

Decoding proceeds in two stages:

• Stabilizer extraction: Ancillas (prepared in $|+\rangle$ for $X$-type and $|0\rangle$ for $Z$-type checks) interact with data qubits through CNOTs to measure each stabilizer’s syndrome.
• Unitary decode: If and only if syndromes are trivial (+1, +1), the inverse of the encoder is run (Campbell et al., 21 Jan 2026).

3. Error Handling and Channel Adaptation

The code’s principal channel of application is the independent amplitude-damping process $\mathcal{N}_\gamma$ on each qubit: $A_0=\begin{pmatrix}1&0\0&\sqrt{1-\gamma}\end{pmatrix},\quad A_1=\begin{pmatrix}0&\sqrt{\gamma}\0&0\end{pmatrix},$ acting as

$$\mathcal{N}^{\otimes4}(\rho) = \sum_{k_1,\ldots,k_4}\left(A_{k_1}\otimes\cdots\otimes A_{k_4}\right)\rho \left(A_{k_1}\otimes\cdots\otimes A_{k_4}\right)^\dagger.$$

First-order correction is achieved via syndrome-subspace recovery. For each error type (zero or one amplitude-damping error), define recovery Kraus operators $\{R_0,R_1,\ldots,R_4\}$ that project the corresponding syndrome subspace back to the code space. This procedure restores logical amplitudes up to $O(\gamma^2)$ (Mao et al., 2024).

The code detects (but does not correct) all single-qubit errors:

• Any single $Z$ error anticommutes with at least one $X$-type stabilizer; any single $X$ error anticommutes with at least one $Z$-type stabilizer.
• All such events are flagged; runs with nontrivial syndrome are post-selected out (Campbell et al., 21 Jan 2026).

4. Integration in Distributed Quantum Architectures

In distributed settings such as quantum data centres, the LNCY code is incorporated within remote-gate execution protocols, notably the “partial-coding 1TP” (PC-1TP) variant. Here:

1. The control qubit is encoded in the code on QPU A.
2. Each physical qubit is teleported via parallel or sequential ebit channels to QPU B.
3. On QPU B, syndrome extraction via stabilizer measurement is performed; failures are rejected.
4. Successful events are unitary-decoded and the remote CNOT is completed on logical and target qubits.

A “happy-path” resource audit (assuming no errors during a run) is:

• QPU A: 4 processing qubits + 4 communication qubits.
• QPU B: 4 communication qubits + 2 ancillas + 1 output qubit = 7.
• Gates: ~$25$–$30$ two-qubit gates, $\sim12$ single-qubit gates, $\sim10$ measurements (Campbell et al., 21 Jan 2026).

5. Code Performance and Limitations

The probability of a successful (post-selected) attempt is given by

$$p_s = F_w^4 + 2F_w^2(1-F_w)^2 + \frac{(1-F_w)^4}{27},$$

where $F_w$ is the Werner-state ebit fidelity. The logical output fidelity $\overline{F}_{\text{out}}$ exhibits robust improvement relative to unencoded operations for $F_w<0.99$ under ideal local gates (Fig. 6(a) in (Campbell et al., 21 Jan 2026)). For hardware error rates in the range $\epsilon_{sg}=1.8{\times}10^{-5}$, $\epsilon_{tg}=9.7{\times}10^{-4}$, $\epsilon_m=2.33{\times}10^{-3}$, LNCY-encoded QED outperforms unencoded protocols for $F_w\lesssim0.98$. Two-round four-ebit DEJMPS distillation achieves higher fidelity for $F_w\lesssim0.95$ but at increased latency and qubit overhead.

Simulation reveals the three-qubit repetition code offers negligible or negative gain, primarily because it cannot detect phase errors, while the LNCY code consistently grants $\gtrsim5$–$10$\% fidelity improvement for $F_w\in[0.90,0.97]$ (Campbell et al., 21 Jan 2026). The average success rate of the LNCY-coded gate run is $10$–$30$\% below that of DEJMPS, reflecting the cost of post-selection.

In the amplitude-damping channel, explicit entanglement-fidelity calculation gives for the standard LNCY code: $$F_e(\gamma)= \begin{cases} 1-1.25\gamma^2 + O(\gamma^3), & \text{(numerical recovery)} \ 1-2.75\gamma^2 + O(\gamma^3), & \text{(analytical recovery)} \end{cases}$$ compared to $1-1.09\gamma^2$ (biconvex-optimized code, numerical) and $1-1.85\gamma^2$ (biconvex code, analytical) (Mao et al., 2024). This demonstrates LNCY’s competitiveness for moderate $\gamma$, while channel-adapted optimization is advantageous for small, but nonzero, damping probabilities.

6. Comparison with Alternative Schemes

Scheme	Correctable Error Type	Fidelity Gain	Latency/Overhead
LNCY [[4,1,2]]	Detects all single-qubit errors	$\sim$5–10%	4 ebits + post-selection
Three-qubit repetition	X (bit-flips) only	$<2\%$ or negative	Minimal, but no phase error detection
Two-ebit distillation	Purifies entanglement	$\lesssim$2%	High success rate, low error suppression
Four-ebit DEJMPS (2rnd)	Entanglement distillation	$>$ LNCY (for $F_w<0.95$)	Double ebit use, high latency

The LNCY code enables a nontrivial boost in noisy intermediate-scale quantum (NISQ) remote-gate fidelity with only moderate resource additions, while distillation-based approaches ultimately dominate for high fidelity at increased resource and time costs (Campbell et al., 21 Jan 2026).

7. Significance, Impact, and Frontier Developments

The LNCY [[4,1,2]] code remains the archetype for small error-detecting codes outside the bit-flip code class, and offers the minimal paradigm beyond the three-qubit repetition scheme for structural error detection in distributed architectures. It is particularly relevant for near-term quantum data centres where hardware constraints and noise models do not yet support full correction and/or nested distillation.

Modern SDP-based optimization techniques yield four-qubit codes with further reduced infidelity for the amplitude-damping channel, but at the expense of codeword symmetry and circuit simplicity, and with no practical advantage for generic mixed-error channels where detection of all single-qubit errors is required (Mao et al., 2024).

A plausible implication is that in architectures where error detection and fast post-selection are feasible and moderate resource overheads are tolerable, the LNCY code offers a balance of simplicity, efficacy, and hardware suitability, whereas in environments dominated by pure amplitude damping or with substantial optimization capacity, numerically adapted four-qubit codes are preferred.

• "Combatting noise in near-term quantum data centres" (Campbell et al., 21 Jan 2026)
• "Optimized four-qubit quantum error correcting code for amplitude damping channel" (Mao et al., 2024)