Pith. sign in

REVIEW 3 major objections 4 minor 26 references

Optimized four-qubit quantum error correcting code for amplitude damping channel

T0 review · 3 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Co-optimizing encoding and recovery yields a four-qubit amplitude-damping code with entanglement fidelity $1 - 1.09\gamma^2$, beating the previous $1 - 1.25\gamma^2$.

desk verdict Plausible numerical four-qubit AD code, but the analytical recovery section's printed equations don't support the claimed fidelity; needs correction and code release. read the letter →

arxiv 2411.12952 v2 pith:USNRPVXK submitted 2024-11-20 quant-ph

Xuanhui Mao , Qian Xu , Liang Jiang This is my paper · ORCID
classification quant-ph MSC 81P7090C22 PACS 03.67.Pp03.65.Yz
keywords amplitudedampingchannelfour-qubitcodebiconvexoptimizationsemidefiniteprogrammingentanglementfidelityapproximatequantumerrorcorrectionnoise-adapted
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Quantum error correction usually targets generic Pauli errors, but real noise like amplitude damping has a specific structure. This paper optimizes the encoding and the recovery together, using alternating semidefinite programming, and reports a four-qubit code tailored to amplitude damping. The optimized code reaches an entanglement fidelity of $1 - 1.09\gamma^2$ for small damping probability $\gamma$, beating the earlier Leung-Nielsen-Chuang-Yamamoto four-qubit code at $1 - 1.25\gamma^2$. An analytical recovery map is also constructed, giving $1 - 1.85\gamma^2$ versus $2.75\gamma^2$ for the earlier code with analytical recovery. If correct, the result shows that co-designing the code and its recovery around the actual noise can buy a concrete fidelity advantage with only four qubits.

What carries the argument

The load-bearing object is the biconvex optimization over the encoding Choi matrix and the combined recovery-decode Choi matrix, solved by alternating semidefinite programming: optimize the recovery for a fixed encoding, then optimize the encoding for that recovery, and iterate. The optimized codewords are extracted from the solution, and the recovery/decoding map is decomposed into Kraus operators. The analytical version is a set of eight recovery operators $R_1$ through $R_8$: $R_1$-$R_4$ correct single-qubit damping on each of the four qubits, $R_5$ and $R_6$ handle two specific two-qubit damping errors on $|0_L\rangle$, and $R_7$ and $R_8$ restore the logical states after the no-jump error, using fitted coefficients $\alpha$ and $\beta$. The approximate QEC criterion, which relaxes the perfect error-correction condition to hold up to $O(\gamma^2)$, is what licenses the shorter four-qubit block.

What would settle it

Compute the Kraus sum $\sum_{i=1}^{8} R_i^\dagger R_i$ for the analytical recovery using Eqs. (30)-(31) and evaluate the entanglement fidelity of the full map on the amplitude-damping channel at, say, $\gamma=0.01$; if the sum is not the identity or the fidelity deviates from $1 - 1.85\gamma^2$, the analytical claim fails.

Watch

Extended reading notes

Core claim

The paper's central claim is that there exists a four-qubit code, with codewords $|0_L\rangle = \sqrt{1 - \tfrac{1}{2}(1-\gamma)^2}\,|0000\rangle + \tfrac{1}{\sqrt{2}}(1-\gamma)\,|1111\rangle$ and $|1_L\rangle = \tfrac{1}{2}(|0011\rangle + |0101\rangle - |1010\rangle + |1100\rangle)$, that, combined with a recovery optimized by alternating semidefinite programming, achieves entanglement fidelity $F_{\mathrm{ent}} = 1 - 1.09\gamma^2 + O(\gamma^3)$ over the amplitude damping channel. This outperforms the Leung-Nielsen-Chuang-Yamamoto code, which reaches $1 - 1.25\gamma^2 + O(\gamma^3)$ with optimized recovery. The paper also constructs an analytical recovery map whose fidelity is $1 - 1.85\gamma^2 + O(\gamma^3)$, against $1 - 2.75\gamma^2$ for the earlier code's analytical recovery. The code is shown to satisfy an approximate quantum error correction criterion: errors from the no-damping and first-order damping subspaces are suppressed to second order in $\gamma$, with the maximal deviation from perfect QEC smaller than for the earlier code.

Load-bearing premise

The analytical outperformance claim stands on the assumption that the recovery operators $R_1$ through $R_8$ with the fitted coefficients $\alpha,\beta$ from Eq. (30) form a valid quantum recovery map that really achieves $F_{\mathrm{ent}} = 1 - 1.85\gamma^2 + O(\gamma^3)$.

Editorial extensions

If this is right

  • For small damping probabilities, the new code's leading error term $1.09\gamma^2$ is smaller than the earlier code's $1.25\gamma^2$, so in the $\gamma \lesssim 0.1$ regime it predicts a measurable fidelity advantage.
  • The alternating-SDP recipe provides a general template: fix a noise channel, co-optimize encoding and recovery, then read off a code and a recovery from the Choi matrices.
  • The analytical recovery operators are close to the numerical optimum, suggesting the scheme can be implemented without solving an SDP at runtime.
  • The approximate-QEC check shows the code corrects all single-qubit damping errors and the no-jump distortion to second order, which is the feature that explains the fidelity gain.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the fidelity gap persists at realistic finite $\gamma$, the code is a natural candidate for a logical memory experiment on superconducting qubits, where the benchmark would be the logical error rate per memory step as a function of $T_1$.
  • The same biconvex pipeline could be applied to dephasing or to a channel mixing amplitude damping and dephasing; the resulting code might interpolate between the amplitude-damping code and a stabilizer code.
  • Because the analytical recovery uses only a few fitted parameters, one could compile it into a concrete circuit and measure the actual fidelity, testing whether the $1 - 1.85\gamma^2$ prediction holds beyond the model.
  • The paper's reliance on the $\Delta M$ bound suggests a stronger statement may be true: that this four-qubit code is near the best possible four-qubit code for amplitude damping, but the paper only establishes the heuristic connection.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The paper proposes a four-qubit quantum error correcting code tailored to the amplitude damping channel, obtained by alternating semidefinite programming over the encoding and recovery channels. The central claims are that the optimized code achieves an entanglement fidelity Fent = 1 − 1.09γ² + O(γ³), beating the Leung–Nielsen–Chuang–Yamamoto code's Fent = 1 − 1.25γ² + O(γ³), and that an analytically constructed recovery achieves Fent = 1 − 1.85γ² versus 2.75γ² for the LNCY analytical recovery. The paper also argues that the code satisfies an approximate QEC criterion up to O(γ²) for no-damping and single-damping errors.

Significance. If correct, the proposed code would be a useful noise-adapted QEC scheme for a physically relevant error channel, and the comparison against the external LNCY benchmark is a genuine strength rather than a circular claim. The explicit codeword, the explicit recovery operators, and the use of alternating SDP are attractive features. However, the analytical recovery derivation contains concrete algebraic inconsistencies that leave the analytical outperformance claim unsupported as stated, and the numerical comparison is not backed by code, data, or error bars. The manuscript is therefore promising but requires substantial correction before the main claims can be accepted.

major comments (3)
  1. [§III C 2, Eqs. (28)–(31)] Equation (28) does not follow from the printed recovery operators. With t = 1 − γ, the codewords in Eqs. (14)–(15) give E0|0L⟩ = sqrt(1 − t²/2)|0000⟩ + t⁵/√2 |1111⟩ and E0|1L⟩ = t²|1L⟩. Applying R7 alone gives a |0L⟩ coefficient α sqrt(1 − t²/2) + β t⁵/√2 and a |1L⟩ coefficient t², whereas applying R7 + R8 gives (α+β) sqrt(1 − t²/2) + (β−α) t⁵/√2 for |0L⟩ and t² for |1L⟩. Equation (28) instead reports (β−α)(1−γ)/√2 and (1−γ)/2. The origin of the factor 1/2 and the replacement of t⁵ and t² by t is not explained by any convention in the paper. Since Eq. (29) optimizes α, β against this incorrect target, the quoted α in Eq. (30) and the claimed Fent = 1 − 1.85γ² are unsupported by the printed operators. The analytical half of the outperformance claim needs a corrected derivation or a corrected set of recovery operators.
  2. [Appendix, Eqs. (A.5) and (A.8)] The expansion for the leading diagonal entry of ⟨0L|Ea†Eb|0L⟩ is incorrect. From Eq. (14), ⟨0L|E0†E0|0L⟩ = 1 − t²/2 + t¹⁰/2 with t = 1 − γ, which expands to 1 − 4γ + 22γ² + O(γ³), not 1 − 2γ + O(γ²) as stated in Eq. (A.5). The corresponding |1L⟩ entry is t⁴ = 1 − 4γ + 6γ² + O(γ³), so the difference is still O(γ²) and Eq. (A.2) may survive, but the displayed expansion and the printed expression for A_{1,1} are not well formed. Because Eq. (A.8) uses these matrix elements to claim a maximal QEC-matrix deviation of 1/(2√2)γ + O(γ²), that bound needs to be rederived from corrected matrix elements.
  3. [§III A and Fig. 3] The numerical support for the central comparison is incomplete. The codeword ansatz in Eq. (14) is verified only over γ ∈ [0.01, 0.1], while the asymptotic claims in Eq. (18) and the analytical fidelity statement are small-γ statements. No code, SDP solver inputs, or numerical data are provided, and the fitted fidelity coefficients are quoted without error bars or residuals. The authors should specify the full γ range used in Fig. 2, make the data or code available, and verify the codeword over a wider interval so that the claimed γ² coefficients can be checked against contamination by higher-order terms.
minor comments (4)
  1. [Eqs. (23) and (26)] The coefficients in these equations are missing powers of (1−γ). A direct calculation gives E1000|0L⟩ = sqrt(γ(1−γ)⁴/2) |0111⟩, not sqrt(γ(1−γ)/2) |0111⟩, and E0110|0L⟩ = γ(1−γ)³/√2 |1001⟩, not γ/√2 |1001⟩. These errors do not change the logical correction action because the relevant states remain proportional to the same syndromes, but the displayed amplitudes should be corrected.
  2. [Fig. 2] The figure caption does not state what is plotted on the axes, whether the curves are raw numerical data or fits, or the γ range used. Adding axis labels, the fitting form, and the fitted coefficient values with uncertainties would greatly improve reproducibility.
  3. [References] References [18]–[22] are not cited in the body of the paper, and reference [21] appears to duplicate reference [9]. The reference list should be pruned or the citations should be added.
  4. [Introduction] There is a typo in the first paragraph: "tolerents" should be "tolerates." Also, the phrase "This expression of codeword works for γ ≪ 1" is vague; a quantitative statement of the validity range and the error incurred outside it would be preferable.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the central outperformance claim is benchmarked against the external Leung-Nielsen-Chuang-Yamamoto code, and the analytical recovery is an explicit construction, not a disguised input.

full rationale

The paper's derivation chain is a constructive design-and-evaluate procedure rather than a derivation that covertly re-inserts its conclusions. The codeword (Eqs. 14-15) is presented as an ansatz refined by biconvex optimization and verified against the numerical optimizer; this is an optimization output, not a prediction from its own assumptions. The primary fidelity comparison (Eq. 18 vs. F_LNCY = 1 - 1.25 gamma^2) is made against the independent Leung-Nielsen-Chuang-Yamamoto code, so the central claim has external content. The analytical recovery (Section III C 2) is explicitly constructed and its coefficients alpha and beta are fitted via Eq. (30); reporting the resulting fidelity is a property of the construction, not a renamed input. The only self-citation that appears in a supporting role is Ref. [17] (same-group authors) used to relate the QEC-matrix deviation to fidelity; this is heuristic and not load-bearing for the numerical or analytical fidelity values. There is no equation in the paper that reduces a claimed prediction to a defining relation or a fitted parameter renamed as a result. Internal inconsistencies in Eqs. (23)-(30) are correctness risks, but they do not constitute circularity under the specified definitions.

Assumptions & free parameters 2 free parameters · 6 assumptions · 0 invented entities

The core formalism (CPTP, Choi matrices, entanglement fidelity) is standard. The main added assumptions are the identification of the optimized encoder with a simple gamma-dependent ansatz (Eq 14), the use of a heuristic alternating SDP without global optimality guarantees, and the fitted alpha/beta coefficients in the analytical recovery. No new physical entities are introduced.

free parameters (2)
  • alpha (and beta = sqrt(1 - alpha^2)) in analytical recovery = alpha approx sqrt(1 - 1/2(1-gamma)^2 + 0.71 gamma + 0.76 gamma^2 + O(gamma^3))
    Chosen by minimizing the imbalance in the no-jump recovery coefficient (Eq 29). The constants 0.71 and 0.76 are unexplained numerical fits, not derived. The analytical fidelity Fent = 1 - 1.85 gamma^2 depends on this choice.
  • gamma-dependent entries of numerical recovery operators R1 and R6 = not given in closed form
    Section III C 1 states 'For the entries of these operators, we fit their gamma-dependence to capture their behavior accurately.' No formulas or residuals are provided.
assumptions (6)
  • standard math Choi's theorem and the CPTP formalism for channels
    Used in Section II A 2 to characterize encoding and recovery channels.
  • domain assumption Amplitude damping channel model: A0 = diag(1, sqrt(1-gamma)), A1 = [[0, sqrt(gamma)], [0, 0]], acting independently on each qubit
    The noise model is the single-qubit amplitude damping channel, assumed to act independently on each of four qubits (Section II A 3).
  • domain assumption Entanglement fidelity is the correct objective for QEC performance
    Used throughout as the figure of merit for comparing codes and recoveries.
  • ad hoc to paper The decoding operator is the inverse of the encoding
    Assumed in Section III C 1 when separating recovery from decoding; not proven and not generally required.
  • ad hoc to paper The codeword ansatz in Eq (14) matches the biconvex optimization solution
    Verified only numerically for gamma in [0.01,0.1] (Appendix Fig 3); no details of the comparison are given.
  • domain assumption Biconvex alternating SDP converges to a good (near-optimal) solution
    The optimization is heuristic; no global optimality guarantee is claimed, but the result's validity depends on the quality of the found solution.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Optimized four-qubit quantum error correcting code for amplitude damping channel." pith.science (2026). https://pith.science/paper/USNRPVXK

@misc{pith2026241112952,
  author       = {Pith},
  title        = {Pith review of: Optimized four-qubit quantum error correcting code for amplitude damping channel},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/USNRPVXK}},
  note         = {Machine review of arXiv:2411.12952}
}
read the original abstract

Quantum error correction (QEC) is essential for reliable quantum information processing. Targeting a particular error channel, both the encoding and the recovery channel can be optimized through a biconvex optimization to give a high-performance, noise-adapted QEC scheme. We solve the biconvex optimization by the technique of alternating semi-definite programming and identify a new four-qubit code for amplitude damping channel, one major noise in superconducting circuits and a good model for spontaneous emission and energy dissipation. We also construct analytical encoding and recovery channels that are close to the numerically optimized ones. We show that the new code notably outperforms the Leung-Nielsen-Chuang-Yamamoto four-qubit code in terms of the entanglement fidelity over an amplitude damping channel.

Figures

Figures reproduced from arXiv: 2411.12952 by the authors.

Figure 1
Figure 1. FIG. 1. flowchart for alternating semidefinite program. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2. Entanglement Fidelity Comparison Between Opti [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3. Comparison between Optimized Four-qubit Code [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

26 extracted references · 17 canonical work pages

  1. [1]

    In this space, a QEC scheme can be established as follows: Initially, a state |Φ⟩ is prepared in a noiseless local memory

    QEC Scheme Consider a d-dimensional Hilbert space H0. In this space, a QEC scheme can be established as follows: Initially, a state |Φ⟩ is prepared in a noiseless local memory. This state is encoded from H0 into a larger Hilbert space Hn (where n ≥ d) to introduce redun- dancy. This encoding is performed via a Completely Pos- itive and Trace-Preserving (C...

    arXiv 2025

  2. [2]

    Choi Matrix For a linear mapA that operates from L(H1) to L(H2), the Choi matrix ˆXA is an operator in L(H1) ⊗ L(H2), defined by: ( ˆXA)[ij],[i′j′] = ⟨jH2 |A(|iH1 ⟩⟨i′ H1 |)|j′ H2 ⟩. (4) Each element of ˆXA corresponds to the action of A on a specific basis element of H1 and H2, where |iH1 ⟩, |i′ H1 ⟩, |jH2 ⟩, and |j′ H2 ⟩ form the orthonormal bases of H1...

    work page

  3. [3]

    It is represented by the operators: A0 = 1 0 0 √1 − γ A1 = 0 √γ 0 0 (6) where γ denotes the probability of transition from the state |1⟩ to |0⟩, symbolizing the decay process

    Amplitude Damping Channel The amplitude damping channel describes the process of energy loss in a quantum system. It is represented by the operators: A0 = 1 0 0 √1 − γ A1 = 0 √γ 0 0 (6) where γ denotes the probability of transition from the state |1⟩ to |0⟩, symbolizing the decay process. In the context of a four-qubit system, the amplitude damping operat...

    work page

  4. [4]

    Optimized QER Using the optimized recovery results from convex op- timization, we achieve an entanglement fidelity of: Fent = 1 − 1.09γ2 + O(γ3) (18) which surpasses the Fent = 1 − 1.25γ2 + O(γ3) achieved by Leung et al.’s four-qubit code with optimized recovery (as illustrated by the solid lines in Fig. 2). We decompose the optimized recovery and decodin...

    work page

  5. [5]

    Analytical QER We also derive an analytical recovery map for the new codeword, which is described as follows: R1 = |0L⟩⟨0111| + 1√ 2 |1L⟩(−⟨0010| + ⟨0100|) R2 = |0L⟩⟨1011| + 1√ 2 |1L⟩(⟨0001| + ⟨1000|) R3 = |0L⟩⟨1101| + 1√ 2 |1L⟩(⟨0001| − ⟨1000|) R4 = |0L⟩⟨1110| + 1√ 2 |1L⟩(⟨0010| + ⟨0100|) R5 = |0L⟩⟨1001| R6 = |0L⟩⟨0110| R7 = |0L⟩(α⟨0000| + β⟨1111|) + |1L...

    work page 2020

  6. [6]

    C. H. Bennett, D. P. DiVincenzo, J. A. Smolin, and W. K. Wootters, 54 (1996), lANL e-print quant- ph/9604024

    arXiv 1996

  7. [7]

    Ekert and C

    A. Ekert and C. Macchiavello, Physical Review Letters 77, 2585 (1996)

    work page 1996

  8. [8]

    Knill and R

    E. Knill and R. Laflamme, Physical Review A (1996), lANL E-print quant-ph/9604034

    arXiv 1996

Show all 26 references
  1. [9]

    M. A. Nielsen, H. Barnum, C. M. Caves, and B. W. Schumacher, (1996), in preparation

    1996

  2. [10]

    A. R. Calderbank, E. M. Rains, P. W. Shor, and N. J. A. Sloane, (1996), lANL E-print quant-ph/9608006

    1996 arXiv

  3. [11]

    Schumacher, Phys

    B. Schumacher, Phys. Rev. A 54, 2614 (1996)

    1996

  4. [12]

    D. W. Leung, M. A. Nielsen, I. L. Chuang, and Y. Ya- mamoto, Physical Review A 56, 2567 (1997)

    1997

  5. [13]

    Dutta, P

    S. Dutta, P. Mandayam, and A. Jain, Smallest quantum codes for amplitude-damping noise, arXiv:2410.00155 [quant-ph] (2024)

    2024 arXiv

  6. [15]

    K. Noh, V. V. Albert, and L. Jiang, arXiv (2018), 1801.07271

    2018 arXiv

  7. [16]

    R. L. Kosut and D. A. Lidar, Quantum Information Pro- cessing 8, 443 (2009)

    2009

  8. [17]

    Reimpell and R

    M. Reimpell and R. F. Werner, Physical Review Letters 94, 080501 (2005)

    2005

  9. [18]

    Choi, Linear Algebra and its Applications 10, 285 (1975)

    M.-D. Choi, Linear Algebra and its Applications 10, 285 (1975)

    1975

  10. [19]

    C. A. Floudas and V. Visweswaran, Computers & Chem- ical Engineering 14, 1397 (1990)

    1990

  11. [20]

    Cvx: Matlab software for disci- plined convex programming, version 2.0,

    CVX Research, Inc., “Cvx: Matlab software for disci- plined convex programming, version 2.0,” (2012), [On- line; accessed 20-Jan-2024]

    2012

  12. [21]

    Grant and S

    M. Grant and S. Boyd, in Recent Advances in Learning and Control, edited by V. Blondel, S. Boyd, and H. Kimura (Springer-Verlag Limited, 2008) pp. 95–110

    2008

  13. [22]

    The near- optimal performance of quantum error correction codes,

    G. Zheng, W. He, G. Lee, and L. Jiang, “The near- optimal performance of quantum error correction codes,” (2024), arXiv:2401.02022 [quant-ph]

    2024 arXiv

  14. [23]

    Yamamoto, S

    N. Yamamoto, S. Hara, and K. Tsumura, Physical Re- view A 71, 022322 (2005)

    2005

  15. [24]

    Audenaert and B

    K. Audenaert and B. De Moor, Physical Review A 65, 030302 (2002), rapid Communications

    2002

  16. [25]

    Lecture note for quantum computation, chapter 3. foundations of quantum theory ii: Measure- ment and evolution,

    J. Preskill, “Lecture note for quantum computation, chapter 3. foundations of quantum theory ii: Measure- ment and evolution,” (2005), [Online; accessed 20-Jan- 2024]

    2005

  17. [26]

    A. S. Fletcher, P. W. Shor, and M. Z. Win, Physical Review A 75, 012338 (2007)

    2007

  18. [27]

    Preskill, (2018), based on a Keynote Address at Quan- tum Computing for Business, 5 December 2017

    J. Preskill, (2018), based on a Keynote Address at Quan- tum Computing for Business, 5 December 2017

    2018

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.