Pith. sign in

REVIEW 2 major objections 4 minor 33 references

Coherent qubit noise is absent, not faint, from a single histogram, and a logarithmic set of extra measurement settings restores it.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-01 09:16 UTC pith:ZFPBGXBW

load-bearing objection The core impossibility and the logarithmic measurement cure are real and well proved for Z-diagonal inputs; the paper is worth refereeing, but the Z-diagonal assumption is load-bearing and the practical reach is narrower than the title suggests. the 2 major comments →

arxiv 2607.21663 v1 pith:ZFPBGXBW submitted 2026-07-23 quant-ph

Absent, Not Faint: Fisher-Information Limits and a Logarithmic Measurement-Design Cure for Passive Characterization of Coherent Qubit Noise

Yi Pan , Meng Hsiu Tsai , Weihang You , Hanqi Jiang , Junhao Chen , Wei Zhang , Isaac Lyngaas , Yingfeng Wang
show 1 more author
Tianming Liu
This is my paper
classification quant-ph MSC 81P6862B10 PACS 03.67.-a
keywords coherent over-rotationFisher information kernelsingular Cramér-Rao boundmeasurement designseparating codeconditioning floorqubit characterizationcoherence veil
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Small systematic gate over-rotations, a leading source of error in quantum processors, are usually assumed to be faint: weak but present in the data. This paper argues that when measured through the cheapest output a device provides—a single fixed-basis histogram—such a coherent error is absent, not faint: to first order it leaves the histogram unchanged, so the Fisher information is singular, the Cramér–Rao bound is infinite, and no estimator recovers it no matter how many shots are taken. The impossibility is exact at zero angle for commuting single- and two-qubit transverse over-rotations with known support on the canonical input, where each coherent angle lies in the Fisher kernel. The paper further proves that a fixed set of ⌈log2(n+1)⌉ extra measurement settings makes every such fault visible, that this count is minimal among product-Pauli augmentations, and that the finite-sample cost is governed by conditioning, not coverage, with a closed-form floor. A sympathetic reader cares because this moves the answer from 'collect more data or enrich the model' to 'design the measurement,' and gives an explicit design that attains the floor.

Core claim

At η=0, for commuting weight-1 and weight-2 transverse over-rotations with known support on a Z-diagonal input, every coherent angle lies exactly in the Fisher kernel of a single Z-basis histogram: its singular Cramér–Rao bound is infinite, and no finite-variance, locally unbiased, first-order estimator recovers it at any sample size. The kernel has a closed-form rank formula, and each veiled direction pairs a coherent move u with the compensating stochastic move (1/2)d⊙c(u). A generic nonzero angle partially lifts the veil; beyond four qubits the obstruction becomes conditioning rather than rank. The cure is measurement design: a fixed set of ⌈log2(n+1)⌉ product-Pauli settings restores iden

What carries the argument

The coherence veil: the kernel KZ of the classical Fisher information of the single Z-histogram, the subspace of parameter directions the histogram cannot resolve at any sample size. Its closed form is an incidence rank formula rank(JZ)=rank(BΩ)+rank((I−PB)H), with explicit exchange basis v(u)=(u, (1/2)d⊙c(u)): each veiled direction pairs a coherent over-rotation against the stochastic flip that exactly cancels its effect on the histogram. The argument runs on the extended singular Cramér–Rao bound, whose +∞ branch makes non-estimability a theorem rather than a heuristic. The cure is a separating code: assign each qubit a distinct nonzero binary codeword and read the corresponding X/Y settin

Load-bearing premise

The pre-noise probe is Z-diagonal (canonically |0^n⟩), so odd-Y Pauli expectations vanish; if the input carried Y-coherence, the single histogram would partially expose the coherent angle, and the 'absent' claim and the logarithmic cure would change.

What would settle it

Prepare |0^n⟩ with a single known over-rotation of angle η on one qubit and measure the Z-basis histogram at η=0 and at small η: the claim predicts the histogram's first-order derivative with respect to η is exactly zero at η=0 (the Fisher column vanishes), while a generic nonzero angle gives a small but nonzero slope. Equivalently, prepare the same over-rotation on an input with Y coherence (e.g., |+⟩ rotated by a small phase) and measure a single Z histogram: the paper's mechanism predicts the coherent angle immediately becomes visible from that single histogram, since the odd-Y expectation

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Adding a coherent parameter to a Z-only model cannot recover the angle at η=0; in simulation the enriched single-histogram estimator pays 12–67× the mean-squared error of the separating-code design at equal shots.
  • A fixed design of ⌈log2(n+1)⌉ extra product-Pauli settings lifts the veil for every known-support commuting weight-1/2 transverse family, and for the complete family this count is minimal at η=0 among product-Pauli augmentations of Z^n.
  • Once coverage holds, finite-sample recovery error is set by the design's smallest singular value, not by coverage: across 250 (n=3) and 120 (n=4) full-rank designs, RMSE tracks σmin (Spearman −0.63/−0.85) and worst-conditioned quintiles pay 3.1–4.1× the RMSE of the best at equal budget.
  • Below coverage the worst-direction sample complexity diverges as Ω(η^{−2}) as η→0, so the closer a device is to calibrated, the worse a blind single-basis measurement performs.
  • On superconducting hardware with injected errors, the single fixed-basis fit is 3–5× more biased than the separating-code fit, consistent with the predicted conditioning ordering.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same kernel-check diagnostic transfers to any fixed-measurement inverse problem: a target in the kernel of the averaged Fisher information is unrecoverable from passive data, and only a measurement change (or input change) can help. Applying this test to a classical linear inverse problem would be a direct, low-cost validation of the paper's transferable rule.
  • The ⌈log2(n+1)⌉ count is the group-testing bound; the paper leaves open whether entangled or adaptive measurements could lift the veil at η=0 with fewer settings. Answering that would sharpen the minimality statement beyond product-Pauli augmentations.
  • The closed-form floor suggests a crossover: at small n the fixed-setting code is the cheaper intervention, but the (1−2p)^n branch decays exponentially in n, so at scale an active twirling protocol that removes the coherent angle should become more shot-efficient. The paper's scaling remarks imply this; it is not a proved recommendation.
  • For non-commuting generators (always-on ZZ crosstalk), the first-order code leaves a residual second-order veil; quantifying how many additional settings a second-order Fisher analysis requires would be a natural follow-up, and the paper names this as an open problem.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 4 minor

Summary. This paper asks whether a fixed computational-basis (Z) histogram can identify small coherent over-rotation angles in a commuting weight-1/weight-2 transverse noise model, in the presence of stochastic Pauli flips, on the canonical |0^n> input. It proves that at zero coherent angle every such angle lies in the Fisher-information kernel, so its singular Cramér–Rao bound is infinite and no finite-variance locally unbiased estimator exists (Theorems 1–2). It then constructs a twirl-free product-Pauli separating code of ⌈log2(n+1)⌉ fixed settings that restores identifiability and is minimal for the complete family at η=0 (Theorem 3), and it analyzes finite-sample cost through a conditioning floor c_B = min{(1−2p)^n, √λ2} with an Ω(η^{-2}) sub-coverage divergence (Theorem 4). The paper includes exact density-matrix checks up to n=8, a design-population numerical study, a classical analogue, and a small hardware consistency check on IBM Heron.

Significance. The core contribution is a clean and useful example where a parameter is absent rather than faint from a passive measurement, and where the cure is measurement design rather than model enrichment. The central theorems are supported by explicit proofs in the appendices, the kernel and separating code are constructive, and the numerical checks match exact evolution to machine precision. The distinction between coverage and conditioning, with a closed-form floor, is a valuable conceptual and practical point. The results are, however, restricted to a narrow regime (Z-diagonal/canonical input, commuting weight-1/2 generators, known support), and the paper honestly states this.

major comments (2)
  1. [§IV, Theorem 4; Appendix E, Lemma 9] Theorem 4 advertises c_B = min{(1−2p)^n, √λ2} as a closed-form conditioning floor for the complete family, but λ2 is never actually defined or evaluated: the text refers only to 'the second eigenvalue of the readout-restricted information,' and the Z^n incidence Gram spectrum is not given. Without an explicit definition (or a formula/bound for λ2), the √λ2 branch of the floor cannot be verified, and the 'closed-form' claim is incomplete. Please define the readout-restricted information and state λ2 for the complete family, and indicate which branch dominates in the uniform-rate case.
  2. [§III/Appendix B; Abstract] The 'absent, not faint' claim is proved for Z-diagonal real inputs with vanishing odd-Y expectations, a scope the paper states in Table I and Section VIII. However, the title and the abstract's first paragraph present the absence as a property of the coherent fault itself. The paper's own numerical control with a local SH input (Appendix B) drops the kernel coherent weight from ≥0.99 to 0.55, showing that preparation errors producing Y-coherence partially lift the veil. Since the practical conclusion that calibration drives η to 0 inherits this fragility, please add an explicit scope caveat in the abstract and in Section VIII: for inputs carrying Y-coherence the coherent angle is faint rather than absent.
minor comments (4)
  1. [Appendix E / §IV] The constant '2 npmin' in the finite-sample bound should read 2^n p_min; as written (e.g., 'N·MSE≥2 npmin σ−2 min') it is inconsistent with the claim that the constant equals 1 at a uniform reference. Ensure exponents are typeset correctly throughout.
  2. [§VI.b / §VIII.b] Section VIII.b says the sub-coverage divergence is 'stated as a divergence without a precise exponent,' but Theorem 4 states the precise rate Ω(η^{-2}). Rephrase to avoid contradiction.
  3. [Abstract] 'beyond four qubits it clears entirely' should be qualified as 'at generic moderate angles'; at small angles the rank may still be deficient by the local analysis.
  4. [Data Availability] The data-availability statement says the code and data 'will be released publicly with the published version'; for review and reproducibility, please provide a repository link or a permanent availability statement.

Circularity Check

0 steps flagged

No circularity: the principal theorems are derived from explicit Pauli–Liouville computations and combinatorial counting; no fitted constant, self-citation chain, or imported ansatz is load-bearing.

full rationale

The central impossibility result (Theorem 2) is not circular: it follows from the exact column formulas of Lemma 1 and the identification ker F_Z = ker J_Z (Lemma 5). The vanishing of the coherent columns at eta=0 is a direct algebraic consequence of the stated Z-diagonal input assumption, quoted in Appendix A: 'The pre-noise state rho0 is real ... every Pauli expectation with an odd number of Y factors vanishes.' That is a disclosed modeling assumption delimiting the regime, not a hidden reuse of the conclusion. Theorem 3 (logarithmic separating code) is a constructive counting argument using distinct nonempty Y-signatures, giving n <= 2^{|B|} - 1; the count is also independently attributed to the external superimposed-code literature [8] and is not imported from the authors' prior work. Theorem 4's conditioning floor c_B = min{(1-2p)^n, sqrt(lambda2)} is derived from the block-diagonal singular-value structure of the augmented Jacobian, not fitted to data; no parameter is tuned to produce the predicted RMSE/conditioning ordering, and the hardware and simulation comparisons are presented as consistency checks rather than as inputs to the theorems. There are no load-bearing self-citations, no imported uniqueness theorems, and no ansatz smuggled in via citation. The weakest point, sensitivity to Y-coherence in the input, is explicitly scoped in Section VIII and Appendix F; it narrows the regime but does not make the derivation circular. The numerically checked (rather than fully proved) nonzero-angle coherent-fraction corollary is a validation gap, not a circular step.

Axiom & Free-Parameter Ledger

0 free parameters · 5 axioms · 0 invented entities

The central theory introduces no fitted parameters and no invented physical entities; it relies on standard Pauli/Fisher machinery plus explicit domain restrictions (Z-diagonal input, commuting known-support weight-1/2 generators, uniform p for the clean floor).

axioms (5)
  • domain assumption The pre-noise state is Z-diagonal (real; every odd-Y Pauli expectation vanishes) with canonical input |0^n>.
    Makes the coherent columns of the Z-histogram Jacobian vanish at η=0; this is the entire veil mechanism (Appendix A, Lemma 1). Non-Z-diagonal inputs would partially expose coherent angles.
  • domain assumption The generator set G is known a priori, pairwise commuting, with weights 1 and 2 only.
    Theorems 1–4 are stated and proved only in this commuting known-support regime; non-commuting generators and unknown support are explicitly left open in Section VIII.
  • standard math Score regularity for the finite-outcome Z-histogram: p_θ(x)>0 and C^1 in θ, so the extended singular CRB applies.
    Assumed in Appendix C before Lemma 6; standard for finite support, but required for the +∞ branch of Theorem 2.
  • standard math Analyticity of the Jacobian J(η,p) and a Lipschitz/Weyl perturbation bound to extend rank/conditioning statements off η=0.
    Used in the Theorem 4 proof (Remark 2); standard matrix perturbation theory.
  • domain assumption Uniform stochastic rate p for the complete-family floor; the (1−2p)^n branch assumes all flip rates equal.
    c_B = min{(1−2p)^n, √λ2} is proved for uniform p; heterogeneous rates get a star-dependent exponent, so the clean exponential floor is only uniform-p.

pith-pipeline@v1.3.0-alltime-deepseek · 31734 in / 12677 out tokens · 133609 ms · 2026-08-01T09:16:52.039479+00:00 · methodology

0 comments
read the original abstract

Calibrating a quantum processor means estimating error parameters, and estimation theory usually assumes a parameter hard to estimate is faint: its signal is weak but present, so more repetitions or a richer model will recover it. This assumption fails for a leading hardware fault. A coherent over-rotation is a small systematic gate miscalibration. Measured through the cheapest data a device returns--one fixed-basis histogram--it is not faint but absent: to first order it leaves the distribution unchanged, indistinguishable from a compensating stochastic error, exactly as two numbers cannot be separated from their sum. For commuting single- and two-qubit transverse over-rotations, with known support on the canonical input, the histogram's Fisher information is singular along the fault's direction at zero angle, its Cramer-Rao bound is infinite, and no finite-variance, locally unbiased estimator recovers it. At a generic nonzero angle the degeneracy partly lifts; beyond four qubits it clears entirely, leaving conditioning, not absence, as the obstruction. The cure is a richer measurement, not a richer model: a fixed, logarithmically small set of extra settings makes every such fault visible. Visibility alone is not enough. The sampling cost is set by conditioning, not coverage, through a floor whose complete-family closed form is exponentially small in the qubit count. We prove the impossibility and cure, confirm both in exact simulation, show conditioning predicts recovery error across hundreds of designs, and observe a 3-5x bias gap on IBM Heron hardware as a consistency check. Non-commuting faults and unknown support remain open.

Figures

Figures reproduced from arXiv: 2607.21663 by Hanqi Jiang, Isaac Lyngaas, Junhao Chen, Meng Hsiu Tsai, Tianming Liu, Weihang You, Wei Zhang, Yingfeng Wang, Yi Pan.

Figure 1
Figure 1. Figure 1: FIG. 1. Fisher geometry of the mechanism in the local parameter plane (schematic). [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2. Coverage versus conditioning at matched budget (50–60 seeds, 95% bootstrap CIs). [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3. Finite-sample recovery cost versus conditioning across the full design population. We plot, for [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4. Hardware consistency check on two IBM Heron processors ( [PITH_FULL_IMAGE:figures/full_fig_p021_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5. A classical analogue of the obstruction in overpa [PITH_FULL_IMAGE:figures/full_fig_p021_5.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

33 extracted references · 19 linked inside Pith

  1. [1]

    Hashimet al., A practical introduction to benchmark- ing and characterization of quantum computers, PRX Quantum6, 030202 (2025), arXiv:2408.12064

    A. Hashimet al., A practical introduction to benchmark- ing and characterization of quantum computers, PRX Quantum6, 030202 (2025), arXiv:2408.12064

    arXiv 2025

  2. [2]

    T. J. Rothenberg, Identification in parametric models, Econometrica39, 577 (1971)

    1971

  3. [3]

    Stoica and T

    P. Stoica and T. L. Marzetta, Parameter estimation prob- lems with singular information matrices, IEEE Transac- tions on Signal Processing49, 87 (2001)

    2001

  4. [4]

    Nielsen, J

    E. Nielsen, J. K. Gamble, K. Rudinger, T. Scholten, K. Young, and R. Blume-Kohout, Gate set tomography, Quantum5, 557 (2021), arXiv:2009.07301

    Pith/arXiv arXiv 2021

  5. [5]

    Proctor, K

    T. Proctor, K. Rudinger, K. Young, M. Sarovar, and R. Blume-Kohout, What randomized benchmarking ac- tually measures, Physical Review Letters119, 130502 (2017), arXiv:1702.01853

    Pith/arXiv arXiv 2017

  6. [6]

    Huang, S

    H.-Y. Huang, S. T. Flammia, and J. Preskill, Founda- tions for learning from noisy quantum experiments, arXiv preprint arXiv:2204.13691 (2022)

    Pith/arXiv arXiv 2022

  7. [7]

    Malik, Q

    Z. Malik, Q. Langfitt, and Z. Saleem, Coherence re- sponse in noisy quantum measurements, arXiv preprint arXiv:2512.13949 (2026)

    Pith/arXiv arXiv 2026

  8. [8]

    W. H. Kautz and R. C. Singleton, Nonrandom binary superimposed codes, IEEE Transactions on Information Theory10, 363 (1964)

    1964

  9. [9]

    S. Chen, Y. Liu, M. Otten, A. Seif, B. Fefferman, and L. Jiang, The learnability of pauli noise, Nature Com- munications14, 52 (2023), arXiv:2206.06362

    Pith/arXiv arXiv 2023

  10. [10]

    van den Berg, Z

    E. van den Berg, Z. K. Minev, A. Kandala, and K. Temme, Probabilistic error cancellation with sparse pauli–lindblad models on noisy quantum processors, Na- ture Physics19, 1116 (2023)

    2023

  11. [11]

    Harper, S

    R. Harper, S. T. Flammia, and J. J. Wallman, Effi- cient learning of quantum noise, Nature Physics16, 1184 (2020)

    2020

  12. [12]

    Erhard, J

    A. Erhard, J. J. Wallman, L. Postler, M. Meth, R. Stricker, E. A. Martinez, P. Schindler, T. Monz, J. Emerson, and R. Blatt, Characterizing large-scale quantum computers via cycle benchmarking, Nature Communications10, 5347 (2019)

    2019

  13. [13]

    Crupi, J

    M. Crupi, J. I. Cirac, and F. Baccari, Efficient charac- terization of coherent and correlated low-degree noise in layers of gates, PRX Quantum6, 040374 (2025), arXiv:2507.02030

    arXiv 2025

  14. [14]

    Kimmel, G

    S. Kimmel, G. H. Low, and T. J. Yoder, Robust cal- ibration of a universal single-qubit gate set via robust phase estimation, Physical Review A92, 062315 (2015), arXiv:1502.02677

    Pith/arXiv arXiv 2015

  15. [15]

    Blume-Kohout, J

    R. Blume-Kohout, J. K. Gamble, E. Nielsen, K. Rudinger, J. Mizrahi, K. Fortier, and P. Maunz, Demonstration of qubit operations below a rigorous fault tolerance threshold with gate set tomography, Nature Communications8, 14485 (2017)

    2017

  16. [16]

    Wiebe, C

    N. Wiebe, C. Granade, C. Ferrie, and D. G. Cory, Hamiltonian learning and certification using quantum resources, Physical Review Letters112, 190501 (2014), arXiv:1309.0876

    Pith/arXiv arXiv 2014

  17. [17]

    Frigerio and M

    M. Frigerio and M. G. A. Paris, Overcoming sloppi- ness for enhanced metrology in continuous-variable quan- tum statistical models, arXiv preprint arXiv:2410.02989 (2024)

    Pith/arXiv arXiv 2024

  18. [18]

    Mihailescu, S

    G. Mihailescu, S. Sarkar, A. Bayat, S. Campbell, and A. K. Mitchell, Metrological symmetries in singular quantum multi-parameter estimation, arXiv preprint arXiv:2503.05483 (2025)

    arXiv 2025

  19. [19]

    Tsang, Quantum noise spectroscopy as an incoherent imaging problem, Physical Review A107, 012611 (2023)

    M. Tsang, Quantum noise spectroscopy as an incoherent imaging problem, Physical Review A107, 012611 (2023)

    2023

  20. [20]

    J. J. Wallman, C. Granade, R. Harper, and S. T. Flam- mia, Estimating the coherence of noise, New Journal of Physics17, 113020 (2015), arXiv:1503.07865

    Pith/arXiv arXiv 2015

  21. [21]

    Li and P.-C

    Y.-H. Li and P.-C. Yeh, An interpretation of the moore– penrose generalized inverse of a singular fisher informa- tion matrix, IEEE Transactions on Signal Processing60, 5532 (2012), arXiv:1107.1944

    Pith/arXiv arXiv 2012

  22. [22]

    Chaloner and I

    K. Chaloner and I. Verdinelli, Bayesian experimental de- sign: A review, Statistical Science10, 273 (1995)

    1995

  23. [23]

    W. H.-C. Huang, Spectral thresholds for identifiability and stability: Finite-sample phase transitions in high- dimensional learning, arXiv preprint arXiv:2510.03809 (2025). 14

    arXiv 2025

  24. [24]

    Lang and J

    Q. Lang and J. Lu, A unified blockwise measurement de- sign for learning quantum channels and lindbladians via low-rank matrix sensing (2025), arXiv:2501.14080 [quant- ph]

    Pith/arXiv arXiv 2025

  25. [25]

    C. E. Granade, C. Ferrie, N. Wiebe, and D. G. Cory, Ro- bust online hamiltonian learning, New Journal of Physics 14, 103013 (2012), arXiv:1207.1655

    Pith/arXiv arXiv 2012

  26. [26]

    Dobrynin, L

    D. Dobrynin, L. Cardarelli, M. M¨ uller, and A. Bermudez, Compressed-sensing lindbladian quantum tomography with trapped ions, Quantum Science and Technology10, 10.1088/2058-9565/ae0363 (2025), arXiv:2403.07462

    Pith/arXiv arXiv 2058

  27. [27]

    S. T. Merkel, J. M. Gambetta, J. A. Smolin, S. Poletto, A. D. C´ orcoles, B. R. Johnson, C. A. Ryan, and M. Stef- fen, Self-consistent quantum process tomography, Physi- cal Review A87, 062119 (2013), arXiv:1211.0322

    Pith/arXiv arXiv 2013

  28. [28]

    S. T. Flammia and J. J. Wallman, Efficient estimation of Pauli channels, ACM Transactions on Quantum Com- puting1, 3:1 (2020), arXiv:1907.12976

    Pith/arXiv arXiv 2020

  29. [29]

    Huang, R

    H.-Y. Huang, R. Kueng, and J. Preskill, Predict- ing many properties of a quantum system from very few measurements, Nature Physics16, 1050 (2020), arXiv:2002.08953

    Pith/arXiv arXiv 2020

  30. [30]

    H. Kwon, K. Tsubouchi, C.-T. Chu, and L. Jiang, Cri- teria for unbiased estimation: applications to noise- agnostic sensing and learnability of quantum channel, arXiv preprint arXiv:2503.17362 (2025)

    Pith/arXiv arXiv 2025

  31. [31]

    H. Kwon, S. H. Lie, and L. Jiang, Universal sample com- plexity bounds in quantum learning theory via fisher information matrix, arXiv preprint arXiv:2602.21510 (2026)

    Pith/arXiv arXiv 2026

  32. [32]

    Du, M.-H

    Y. Du, M.-H. Hsieh, and D. Tao, Efficient learning for lin- ear properties of bounded-gate quantum circuits, Nature Communications16, 3790 (2025), arXiv:2408.12199

    arXiv 2025

  33. [33]

    ∆u∈col(B)

    J. J. Wallman and J. Emerson, Noise tailoring for scalable quantum computation via randomized compiling, Physi- cal Review A94, 052325 (2016), arXiv:1512.01098. Appendix A: Notation and standing conventions We collect the conventions used throughout the proofs: the coherent–stochastic noise model, theZ-histogram data map, and the Pauli–Liouville facts tha...

    Pith/arXiv arXiv 2016