Pith. sign in

REVIEW 3 major objections 4 minor 75 references

Any nonzero depolarizing readout noise eliminates adaptive quantum state tomography's quadratic scaling advantage, reverting pure-state infidelity from 1/N to 1/√N.

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

2026-08-03 12:09 UTC pith:4WRIUS4E

load-bearing objection Good, useful paper with a real gap between the headline claim and the analytic proof at exactly the pure-state limit — but the numerics and mechanism make me think the conclusion is right. the 3 major comments →

arxiv 2601.04020 v2 pith:4WRIUS4E submitted 2026-01-07 quant-ph

Limitations for adaptive quantum state tomography in the presence of detector noise

Adrian Skasberg Aasen , Martin G\"arttner This is my paper
classification quant-ph
keywords adaptive quantum state tomographyreadout noisereadout-error mitigationFisher informationdepolarizing channelBayesian inferenceinfidelity scalingdetector tomography
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.

Adaptive measurement strategies can turn the hard problem of reconstructing pure quantum states from 1/√N scaling into 1/N scaling in the absence of noise. The paper asks whether that quadratic advantage survives real detectors, which are always noisy. It argues it does not: any nonzero depolarizing readout noise, even with perfect readout-error mitigation, restores the slower 1/√N scaling for pure-state targets estimated by mixed-state estimators. The reason is quantitative: depolarizing noise puts a floor under every measurement-outcome probability, so the adaptive trick of amplifying near-zero Fisher-information directions of an over-parameterized estimator can no longer work. Numerically, a constant-factor accuracy gain over non-adaptive strategies persists, and it grows as the noise weakens.

Core claim

The central claim is that for pure or close-to-pure target states, any nonzero depolarizing readout noise makes the asymptotic reconstruction infidelity of adaptive tomography no better than that of non-adaptive schemes: 1/√N instead of 1/N. Even when the noisy POVM is known exactly and used in the likelihood, the depolarized measurement operators give every outcome a probability at least τ_γ p / 2^n. This lower bound prevents the 1/p_γ prefactor in the Fisher information matrix from amplifying the rows and columns suppressed by the mismatch between a full-rank mixed-state estimator and a rank-deficient pure target state, so the boundary contribution that causes 1/√N scaling cannot be remove

What carries the argument

The Fisher information matrix for the readout-error-mitigated likelihood, written as I(Ψ) = N Σ_γ (4/p_γ) A_γ |Ψ⟩⟨Ψ| A_γ, whose singular values control the average infidelity. For pure targets the matrix is effectively rank-deficient; adaptive strategies become optimal by choosing measurement operators orthogonal to the large-eigenvalue eigenvectors, which makes p_γ small and lets 1/p_γ amplify the suppressed directions. The paper's result comes from the lower bound p_γ ≥ τ_γ p / 2^n, which holds for any POVM element depolarized with strength p: the amplification factor is capped, the Fisher matrix stays rank-deficient, and the 1/√N boundary term survives.

Load-bearing premise

The analytic result assumes that suppressed Fisher-information directions generically contribute 1/√N to infidelity and that depolarizing noise has the same strength for every measurement setting; if some estimator or schedule restores information in those directions by another route, or if device-specific noise allows rare high-information outcomes, the scaling advantage could survive.

What would settle it

Run an adaptive state-tomography experiment or simulation with depolarizing readout of known strength p>0, exact readout-error mitigation, pure target states, and a mixed-state estimator, and fit the infidelity at large N. The paper predicts the power-law exponent approaches -0.5; an exponent that approaches -1, or a Fisher information matrix that becomes full-rank as N grows, would refute the claim.

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

Share X LinkedIn Reddit HN

If this is right

  • In any experiment whose readout has depolarizing noise, adaptive tomography's asymptotic sample-complexity advantage disappears; doubling the number of measurements improves infidelity only by the 1/√N law.
  • Adaptive strategies still give a constant-factor accuracy gain, and the gain increases as readout noise decreases, so they remain attractive in well-calibrated systems.
  • For mixed target states the 1/N scaling is unaffected by readout noise; the limitation is specific to pure and near-boundary states.
  • When detector calibration uses finite copies, both adaptive and non-adaptive estimators acquire bias; the simulations indicate at least five times more copies for detector tomography than for a single state reconstruction.
  • Readout-error mitigation trades bias for variance in adaptive protocols just as it does in static ones, shifting the infidelity curves without restoring optimal scaling.

Where Pith is reading between the lines

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

  • The strongest version of the conclusion depends on depolarizing noise acting with a single, setting-independent strength. A structured or noise-aware measurement design could in principle leave high-information outcomes rare enough to restore amplification; the paper itself leaves this open.
  • A direct experimental test is available: measure the infidelity scaling exponent in a noisy adaptive experiment with exact readout-error mitigation. The paper predicts it approaches -0.5 rather than -1; observing the latter would refute the bound.
  • Because the depolarizing lower bound scales as p/2^n, the constant-factor window of usefulness shrinks exponentially with qubit number for fixed p, so the practical pain is most acute in multi-qubit devices.
  • Overestimated readout noise pushes the likelihood outside the physical domain and can improve boundary estimation, suggesting that noise mis-modeling interacts nontrivially with adaptivity and is not purely a nuisance.

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

3 major / 4 minor

Summary. The paper analyzes adaptive quantum state tomography (QST) with readout-error mitigation (REM) under depolarizing readout noise. Its central claim is that any nonzero depolarizing noise destroys the asymptotic scaling advantage of adaptive strategies for pure and close-to-pure states reconstructed with mixed-state estimators: the infidelity reverts from the adaptive 1/N scaling to the non-adaptive 1/sqrt(N) scaling, while a constant-factor gain may survive. The analytical argument is built on a Fisher-information optimality criterion imported from Bogdanov and Struchalin et al., with the key new ingredient being a lower bound on outcome probabilities, Eq. (22), p_gamma >= tau_gamma p / 2^n, which prevents the 1/p_gamma prefactor in the Fisher information from amplifying the rows and columns suppressed by the estimator's rank deficiency. The paper supports this with large-scale Bayesian-mean-estimation simulations for one- and two-qubit systems, and studies realistic finite-detector-tomography effects. The authors make their code and data publicly available.

Significance. If the central no-go statement were rigorously established, the result would be practically important: it would show that the asymptotic advantage of adaptive QST, a widely discussed feature, is fragile under realistic readout noise and that only finite-sample constant-factor gains remain. The paper has clear strengths: a transparent analytical bound, extensive numerical simulations with reproducible code and data, and a useful practical discussion of detector-tomography calibration. However, the analytical proof as written does not rigorously cover the exactly-pure-state case that is the paper's headline claim, and the numerical fits, while suggestive, do not by themselves prove an asymptotic scaling law. The result is likely correct in its physical essence, but the gap between the derived inequalities and the universal conclusion needs to be closed or the claim appropriately qualified.

major comments (3)
  1. [Sec. IV, App. B.5, Eqs. (B27)-(B30)] The derivation is performed for close-to-pure states with lambda_i = lambda > 0 and measurement operators satisfying alpha_1 = 0. It shows that the per-measurement Fisher information in the suppressed directions is O(lambda). For any fixed lambda > 0, the total information after N measurements is O(N lambda), so the infidelity contribution is O(1/(N lambda)) — that is, still 1/N asymptotically, with an enhanced prefactor. The claimed 1/sqrt(N) behavior appears only in the singular limit lambda -> 0 (exactly pure states), where the calculation is not defined: the orthogonal measurement has zero outcome probability in the noiseless case, and the true-state Fisher matrix is rank-deficient regardless of noise. The noiseless adaptive advantage for pure states is produced by data-dependent misalignment of the measurement basis, which is not modeled in this fixed-state Fisher-information analys
  2. [Sec. IV, Eq. (22)] The lower bound p_gamma >= tau_gamma p / 2^n is true but too weak to support the 'any nonzero readout noise' conclusion for all close-to-pure states. For lambda >> p, the outcome probability of the optimally chosen orthogonal measurement is approximately tau_gamma (1-p) lambda S, which is dominated by the eigenvalue term, not by the noise floor. In that regime the noiseless cancellation still operates and the per-measurement Fisher information is O(1), so the asymptotic scaling remains 1/N. The proof needs to state and analyze the threshold lambda <~ p (or N lambda <~ 1) where the noise floor actually dominates. Without this, the analytical result is not a proof of loss of 1/N scaling for the parameter regime claimed in the abstract.
  3. [Sec. III.C, Eq. (15), and App. B] The paper's argument relies on the imported criterion that optimal adaptive scaling requires all true-state Fisher singular values to be non-vanishing. For exactly pure states, this criterion cannot be satisfied by any measurement, even in the noiseless case, yet adaptive protocols are known to achieve 1/N. The connection between the true-state Fisher matrix and the asymptotic infidelity of a data-dependent adaptive protocol is therefore not automatic. The manuscript needs either a direct derivation for the pure-state case that accounts for the estimator's uncertainty and the resulting measurement misalignment, or an explicit limiting argument showing how the close-to-pure analysis controls the pure-state limit. Without this, the universal no-go claim is not established.
minor comments (4)
  1. [Abstract] The abstract says 'any nonzero readout noise eliminates the asymptotic quadratic scaling advantage,' but the body and the Conclusion restrict the result to depolarizing noise and explicitly allow device-specific noise models that might preserve better scaling. The abstract should be qualified to 'any nonzero depolarizing readout noise' to match the proven statement.
  2. [Sec. II.C, Eq. (5)] The display equation contains a stray 'n=1' after the first line; this appears to be a typesetting artifact and should be removed.
  3. [Fig. 3 and Sec. V.1] The power-law fits between 10^4 and 10^5 measurements yield exponents around -0.57, -0.54, -0.55 for the noisy cases, not -0.50. The text says the curves eventually settle into 1/sqrt(N) scaling, but the numerical results show only a gradual approach. It would be more precise to state that the fitted exponents are decreasing toward -0.5 and are consistent with, but not conclusive proof of, the asymptotic claim.
  4. [Sec. VI] The recommendation to use at least five times as many state copies for detector tomography as for state tomography is based on a specific simulation setup with depolarizing noise. It would be helpful to state more explicitly that this is an empirical rule for the studied noise model, not a general bound.

Circularity Check

0 steps flagged

No circular reduction in the central claim; self-citations are minor and not load-bearing.

full rationale

Walking the derivation: Eq. (15), the singular-value expression for averaged infidelity, and the 1/sqrt(N) boundary contribution are imported from Refs. [9,10] (Bogdanov 2009; Struchalin et al. 2018), which are external to the authors and are re-derived in Appendix B. The new noise mechanism is derived, not assumed: with depolarizing noise, Eq. (B26) yields p_gamma = tau_gamma p/2^n + tau_gamma(1-p) sum_i lambda_i |<psi_gamma|lambda_i>|^2 >= tau_gamma p/2^n, i.e. Eq. (22), and the conclusion that the prefactor 1/p_gamma in Eq. (B28)/(16) can no longer cancel small-eigenvalue suppression follows algebraically. No parameter is fitted and then renamed a prediction: the numerical scaling exponents in Figs. 2-3 are diagnostics of independent Bayesian mean-estimation simulations (Appendix C), not inputs to the analytical claim. The only self-citations, [31] and [37], provide the readout-error-mitigation protocol and a static-noise benchmark; they are not invoked as a uniqueness theorem or as evidence forbidding alternative adaptive schedules, so they are not load-bearing. A scope gap exists (the analytic proof keeps small eigenvalues lambda>0 and does not bridge the singular lambda=0 pure-state limit; the Conclusion also concedes device-specific noise models may preserve scaling), but that is a correctness/scope concern, not circularity: the claimed loss of scaling is not equivalent to its assumptions by construction.

Axiom & Free-Parameter Ledger

2 free parameters · 5 axioms · 0 invented entities

The analytic central claim is parameter-free except for the noise model itself; the two free parameters listed are swept or implementation-level choices. The paper imports the Fisher-optimality framework (Refs [9,10]) rather than proving it, and assumes a uniform depolarizing noise channel; it introduces no new physical entities.

free parameters (2)
  • Readout noise strength p = 0, 0.05, 0.15, 0.25 (swept, not fitted)
    Model input swept across the simulations; the theorem itself must hold for all p>0, so no single value is fitted to make the claim work.
  • BME hyperparameters (n_part, tau, k, n_MH) = 1 qubit: (200, 0.1, 0.15, 100); 2 qubits: (1000, 0.05, 0.4, 200); Sec. VI: n_part=2000
    Chosen by hand (Table I) to ensure particle-swarm convergence in the Bayesian mean estimator; they affect the finite-N transient curves and the detector-tomography saturation levels, but not the analytic asymptotic bound.
axioms (5)
  • domain assumption Haar-averaged reconstruction infidelity equals the sum of inverse non-vanishing Fisher information singular values; suppressed (rank-deficient) singular values contribute a generic 1/sqrt(N) term (Eq. 15, App. B.3).
    The central bound in Sec. IV reduces the loss-of-scaling to this characterization from Refs [9,10]; it is imported, not proven, and the paper's own numerics are consistent with it.
  • domain assumption Depolarizing readout noise acts setting-independently on every POVM element as E(M) = (1-p)M + p tau 1/2^n (Eqs. 5, 21, 25).
    Required for the floor p_gamma >= tau_gamma p / 2^n (Eq. 22). The paper motivates depolarizing noise via typicality and Pauli twirling, and concedes in the Conclusion that device-specific noise models might preserve better scaling.
  • domain assumption Bayesian mean estimation with the particle-swarm sampler asymptotically attains the Fisher-information scaling of the infidelity.
    The claim that the observed numerical exponents are the true asymptotics presumes BME's convergence; Appendix E checks estimator uncertainty, but the two-qubit noisy-case uncertainties grow logarithmically with N.
  • domain assumption Averaging the infidelity over Haar-random pure target states is the correct proxy for the worst-case / average scaling claims.
    All scaling claims are for Haar-averaged infidelity; the paper does not establish state-by-state worst-case bounds, relying on the standard average-case treatment in the tomography literature.
  • standard math Standard quantum-mechanics machinery: POVMs, Born rule, quantum fidelity, depolarizing channel, multinomial likelihood.
    Background formalism used throughout; no controversy.

reviewed 2026-08-03 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Limitations for adaptive quantum state tomography in the presence of detector noise." pith.science (2026). https://pith.science/paper/4WRIUS4E

@misc{pith2026260104020,
  author       = {Pith},
  title        = {Pith review of: Limitations for adaptive quantum state tomography in the presence of detector noise},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4WRIUS4E}},
  note         = {Machine review of arXiv:2601.04020}
}
Share X LinkedIn Reddit HN
read the original abstract

Assumption-free reconstruction of quantum states from measurements is essential for benchmarking and certifying quantum devices, but it remains difficult due to the extensive measurement statistics and experimental resources it demands. An approach to alleviating these demands is provided by adaptive measurement strategies, which can yield up to a quadratic improvement in reconstruction accuracy for pure states by dynamically optimizing measurement settings during data acquisition. A key open question is whether these asymptotic advantages remain in realistic experiments, where readout is inevitably noisy. In this work, we analyze the impact of readout noise on adaptive quantum state tomography with readout-error mitigation, focusing on the challenging regime of reconstructing pure states using mixed-state estimators. Using analytical arguments based on Fisher information optimization and extensive numerical simulations using Bayesian inference, we show that any nonzero readout noise eliminates the asymptotic quadratic scaling advantage of adaptive strategies. We numerically investigate the behavior for finite measurement statistics for single- and two-qubit systems with exact readout-error mitigation and find a gradual transition from ideal to sub-optimal scaling. We furthermore investigate realistic scenarios where detector tomography is performed with a limited number of state copies for calibration, showing that insufficient detector characterization leads to estimator bias and limited reconstruction accuracy. Although our result imposes an upper bound on the reconstruction accuracy that can be achieved with adaptive strategies, we nevertheless observe numerically a constant-factor gain in reconstruction accuracy, which becomes larger as the readout noise decreases. This indicates potential practical benefits in using adaptive measurement strategies in well-calibrated experiments.

Figures

Figures reproduced from arXiv: 2601.04020 by Adrian Skasberg Aasen, Martin G\"arttner.

Figure 1
Figure 1. Figure 1: FIG. 1. Simulated single-qubit state reconstruction with and [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2. Comparison of mean infidelity scaling between adap [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3. Single- and two-qubit mean infidelities with different amount of depolarizing readout noise. A power-law [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4 [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 6
Figure 6. Figure 6: FIG. 6. Mean infidelity curves with different number of [PITH_FULL_IMAGE:figures/full_fig_p009_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: FIG. 7. Mean reconstruction infidelity averaged over 1500 [PITH_FULL_IMAGE:figures/full_fig_p015_7.png] view at source ↗
Figure 3
Figure 3. Figure 3: In the single-qubit case the uncertainty is approx [PITH_FULL_IMAGE:figures/full_fig_p015_3.png] view at source ↗
Figure 10
Figure 10. Figure 10: FIG. 10. Single- and two-qubit estimator uncertainty. The [PITH_FULL_IMAGE:figures/full_fig_p016_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: FIG. 11. Estimator uncertainty from Fig. 6 where a fi [PITH_FULL_IMAGE:figures/full_fig_p016_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: FIG. 12. Mean infidelity curves with different depolariz [PITH_FULL_IMAGE:figures/full_fig_p017_12.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

75 extracted references · 8 canonical work pages

  1. [1]

    If the target state is known in ad- vance, the best strategy is to perform a projective mea- surement in the eigenbasis of the state

    Two-step strategies Single update methods are the more economical of the two approaches, since updating measurement settings can be very slow compared to continuing measurements in the same setting. If the target state is known in ad- vance, the best strategy is to perform a projective mea- surement in the eigenbasis of the state. This measure- ment schem...

  2. [2]

    Perform an initial state estimateρ init based on an informationally complete measurement, such as the MPauli-6 measurements

  3. [3]

    0⟩=∣λ 1⟩

    Find the eigenstate∣λ 1⟩with the largest eigen- valueλ 1 ofρ init, and perform the remaining measurements using the generalized measurement UM Pauli-6U †, whereUis a unitary rotation matrix that rotates the all-zero qubit state to this eigen- state ofρ init,U∣0. . .0⟩=∣λ 1⟩. The best way to divide the total measurement budget between the two steps has bee...

  4. [4]

    Multi-step strategies Multi-step strategies are a form of adaptive experimen- tal design, where the estimator continuously updates the measurement strategy based on prior measurement out- comesD. A prominent example is Bayesian adaptive tomography, where Bayesian inference is used to incor- porate the measurement outcomes into a continuously updated poste...

  5. [5]

    3 we show the mean infidelity curves for multi- step adaptive Bayesian mean estimates under different strengths of depolarizing readout noise

    Single- and two-qubit transient infidelity In Fig. 3 we show the mean infidelity curves for multi- step adaptive Bayesian mean estimates under different strengths of depolarizing readout noise. In both the single- and two-qubit case, the noise-free scenario ex- hibits the expected optimal≈1/Nscaling, while even small amounts of depolarizing readout noise ...

  6. [6]

    5, we present the binned distribution of the infidelity reached for a fixed number of measurements, which corresponds to a vertical cross sections of all curves in Fig

    Infidelity distribution at a given number of measurements In Fig. 5, we present the binned distribution of the infidelity reached for a fixed number of measurements, which corresponds to a vertical cross sections of all curves in Fig. 3. We see that the adaptive distributions are slightly narrower than their nonadaptive counterparts. The noiseless adaptiv...

    2000

  7. [7]

    The target statesρare pure or close-to- pure in ad-dimensional Hilbert space

    Defining notation To begin with, we need to define the notation for our specific problem. The target statesρare pure or close-to- pure in ad-dimensional Hilbert space. The target state can be written in terms of an eigendecomposition (keep- ing the small eigenvalues for later convenience), ρ= d ∑ k=1 λk ∣λk⟩⟨λ k∣,(B2) where{∣λ k⟩}forms a basis in thed-dim...

  8. [8]

    Fisher information matrix for a multinomial distribution Our measurement follows the multinomial distribution, and the likelihood functions can be written in the purified notation as L(v;n γ)∝Π γ(vT Oγv)nγ ,(B6) where to simplify the notation, we limit ourselves to one measurement setting, which leads us to omit the indexα that was previously used in Sec....

  9. [9]

    (B13) and Eq

    Rank-deficient estimator From the Fisher information matrix in Eq. (B13) and Eq. (B12), it is clear that the singular valuesσ i are gen- erally proportional toN. However, this is not the case when the target state has fewer relevant degrees of free- dom compared to the estimated state. When discussing the rank of the Fisher information matrix, we refer to...

  10. [10]

    This effectively amplifies the suppressed rows and columns by a factor 1/p γ

    Optimal adaptive measurement strategies Adaptive strategies recover optimal asymptotic scal- ing by amplifying the suppressed columns and rows of I(∣Ψ⟩)by finding measurement operatorsM γ that sat- isfyM γ ∣λk⟩=0 for state vectors∣λ k⟩with non-vanishing eigenvalues. This effectively amplifies the suppressed rows and columns by a factor 1/p γ. To see why t...

  11. [11]

    (B23) explains how using multiple measurement settings can recover opti- mal asymptotic infidelity scaling for rank-deficient esti- mators

    Optimal adaptive measurement strategies with noise The condition described in Eq. (B23) explains how using multiple measurement settings can recover opti- mal asymptotic infidelity scaling for rank-deficient esti- mators. We can now investigate how readout noise in- fluences the optimality of adaptive strategies. We will consider the depolarizing channel ...

  12. [12]

    The Bayesian update cycle The prior distributionπ 0(ρ)is initialized by draw- ingn part random particle statesρ i from the Hilbert- Schmidt distribution of mixed states [59] and the weights equalizedw i =1/n part. The data is integrated one by one through the Bayesian update rule, expressed in dis- cretized form as wn+1 i = wn i Tr(ρiMon+1) ∑j wn j Tr(ρjM...

  13. [13]

    The key idea is to perturb existing particles so that they clus- ter more densely near regions of high posterior probabil- ity

    Metropolis-Hastings resampling Resampling refines the particle swarm to better ap- proximate the posterior distribution during updates. The key idea is to perturb existing particles so that they clus- ter more densely near regions of high posterior probabil- ity. The resampling follows the procedure from Appendix C of Ref. [12], summarized here. We random...

    2000

  14. [14]

    7 we show the reconstruction infidelity for mixed target states with varying readout noise strength

    Mixed target states In Fig. 7 we show the reconstruction infidelity for mixed target states with varying readout noise strength. The target states are drawn from the Hilbert-Schmidt distribution [59]. Since mixed states do not suffer from the quadratic accuracy penalty of pure states, even non- adaptive methods achieve the ideal 1/Nscaling. Per- forming a...

  15. [15]

    8 we show the interquartile ranges of the single- and two-qubit curves with depolarizing readout strength p=0.05 from Fig

    Interquartile ranges In Fig. 8 we show the interquartile ranges of the single- and two-qubit curves with depolarizing readout strength p=0.05 from Fig. 3. The interquartile ranges are com- puted by sorting the infidelity at each measurement num- 15 102 103 104 105 Number of measurements 10 4 10 3 10 2 Mean Infidelity One qubit Readout noise strength p = 0...

  16. [16]

    9, we show the binned distribution of the num- ber of measurements required to reach a specified infi- delity, corresponding to a vertical slice of Fig

    Number of measurements to reach specified infidelity In Fig. 9, we show the binned distribution of the num- ber of measurements required to reach a specified infi- delity, corresponding to a vertical slice of Fig. 3. The results are qualitatively similar to the binned distribu- tion at a constant number of measurements in Fig. 5, which instead corresponds...

    2000

  17. [17]

    Massar and S

    S. Massar and S. Popescu, Optimal extraction of infor- mation from finite quantum ensembles, Physical Review Letters74, 1259–1263 (1995)

    1995

  18. [18]

    Bagan, M

    E. Bagan, M. A. Ballester, R. D. Gill, A. Monras, and R. Mu˜ noz-Tapia, Optimal full estimation of qubit mixed states, Physical Review A73, 10.1103/phys- reva.73.032301 (2006)

    doi:10.1103/phys- 2006

  19. [19]

    J. Haah, A. W. Harrow, Z. Ji, X. Wu, and N. Yu, Sample- optimal tomography of quantum states, inProceedings of the Forty-Eighth Annual ACM Symposium on Theory of Computing, STOC ’16 (Association for Computing Ma- chinery, New York, NY, USA, 2016) p. 913–925

    2016

  20. [20]

    Pelecanos, J

    A. Pelecanos, J. Spilecki, E. Tang, and J. Wright, Mixed state tomography reduces to pure state tomography (2025), arXiv:2511.15806 [quant-ph]

    arXiv 2025

  21. [21]

    W. K. Wootters and B. D. Fields, Optimal state- determination by mutually unbiased measurements, An- nals of Physics191, 363–381 (1989)

    1989

  22. [22]

    D. H. Mahler, L. A. Rozema, A. Darabi, C. Ferrie, R. Blume-Kohout, and A. M. Steinberg, Adaptive quan- tum state tomography improves accuracy quadratically, Phys. Rev. Lett.111, 183601 (2013)

    2013

  23. [23]

    R. B. A. Adamson and A. M. Steinberg, Improving quantum state estimation with mutually unbiased bases, Phys. Rev. Lett.105, 030406 (2010)

    2010

  24. [24]

    Kahn and M

    J. Kahn and M. Gut˘ a, Local asymptotic normality for finite dimensional quantum systems, Communications in Mathematical Physics289, 597–652 (2009)

    2009

  25. [25]

    G. I. Struchalin, E. V. Kovlakov, S. S. Straupe, and S. P. Kulik, Adaptive quantum tomography of high- dimensional bipartite systems, Physical Review A98, 10.1103/physreva.98.032330 (2018)

    doi:10.1103/physreva.98.032330 2018

  26. [26]

    Y. I. Bogdanov, Unified statistical method for recon- structing quantum states by purification, Journal of Ex- perimental and Theoretical Physics108, 928–935 (2009)

    2009

  27. [27]

    Husz´ ar and N

    F. Husz´ ar and N. M. T. Houlsby, Adaptive bayesian quantum tomography, Physical Review A85, 10.1103/physreva.85.052120 (2012)

    work page doi:10.1103/physreva.85.052120 2012

  28. [28]

    G. I. Struchalin, I. A. Pogorelov, S. S. Straupe, K. S. Kravtsov, I. V. Radchenko, and S. P. Kulik, Experimen- tal adaptive quantum tomography of two-qubit states, Phys. Rev. A93, 012103 (2016)

    2016

  29. [29]

    S. S. Straupe, Adaptive quantum tomography, JETP Let- ters104, 510–522 (2016)

    2016

  30. [30]

    Sugiyama, P

    T. Sugiyama, P. S. Turner, and M. Murao, Adaptive ex- perimental design for one-qubit state estimation with fi- nite data based on a statistical update criterion, Phys. Rev. A85, 052107 (2012)

    2012

  31. [31]

    Ferrie, Self-guided quantum tomography, Phys

    C. Ferrie, Self-guided quantum tomography, Phys. Rev. Lett.113, 190404 (2014)

    2014

  32. [32]

    Granade, C

    C. Granade, C. Ferrie, and S. T. Flammia, Practical adaptive quantum tomography, New Journal of Physics 19, 113017 (2017)

    2017

  33. [33]

    Y. Quek, S. Fort, and H. K. Ng, Adaptive quantum state tomography with neural networks, npj Quantum Infor- mation7, 10.1038/s41534-021-00436-9 (2021)

    work page doi:10.1038/s41534-021-00436-9 2021

  34. [34]

    B. Qi, Z. Hou, Y. Wang, D. Dong, H.-S. Zhong, L. Li, G.-Y. Xiang, H. M. Wiseman, C.-F. Li, and G.-C. Guo, Adaptive quantum state tomography via linear re- gression estimation: Theory and two-qubit experiment, npj Quantum Information3, 10.1038/s41534-017-0016-4 (2017)

    work page doi:10.1038/s41534-017-0016-4 2017

  35. [35]

    O. E. Barndorff-Nielsen and R. D. Gill, Fisher informa- tion in quantum statistics, Journal of Physics A: Mathe- 18 matical and General33, 4481–4490 (2000)

    2000

  36. [36]

    Kalev and I

    A. Kalev and I. Hen, Fidelity-optimized quantum state estimation, New Journal of Physics17, 093008 (2015)

    2015

  37. [37]

    Zambrano, L

    L. Zambrano, L. Pereira, S. Niklitschek, and A. Delgado, Estimation of pure quantum states in high dimension at the limit of quantum accuracy through complex opti- mization and statistical inference, Scientific Reports10, 10.1038/s41598-020-69646-z (2020)

    work page doi:10.1038/s41598-020-69646-z 2020

  38. [38]

    S. T. Ahmad, A. Farooq, and H. Shin, Self-guided quan- tum state tomography for limited resources, Scientific Re- ports12, 10.1038/s41598-022-09143-7 (2022)

    work page doi:10.1038/s41598-022-09143-7 2022

  39. [39]

    Vargas, L

    C. Vargas, L. Pereira, and A. Delgado, Near-optimal pure state estimation with adaptive fisher-symmetric measure- ments (2024), arXiv:2412.04555 [quant-ph]

    Pith/arXiv arXiv 2024

  40. [40]

    S. M. Kazim, A. Farooq, J. ur Rehman, and H. Shin, Adaptive quantum state tomography with iterative par- ticle filtering, Quantum Information Processing20, 10.1007/s11128-021-03267-x (2021)

    work page doi:10.1007/s11128-021-03267-x 2021

  41. [41]

    Mondal and A

    S. Mondal and A. K. Dutta, A bayesian quantum state tomography along with adaptive frameworks based on linear minimum mean square error criterion, New Journal of Physics25, 123001 (2023)

    2023

  42. [42]

    Farooq, M

    A. Farooq, M. A. Ullah, J. u. Rehman, K. Lee, and H. Shin, Self-guided quantum state learning for mixed states, Quantum Information Processing21, 10.1007/s11128-022-03585-8 (2022)

    work page doi:10.1007/s11128-022-03585-8 2022

  43. [43]

    K. S. Kravtsov, S. S. Straupe, I. V. Radchenko, N. M. T. Houlsby, F. Husz´ ar, and S. P. Kulik, Experimental adap- tive bayesian tomography, Phys. Rev. A87, 062122 (2013)

    2013

  44. [44]

    Okamoto, M

    R. Okamoto, M. Iefuji, S. Oyama, K. Yamagata, H. Imai, A. Fujiwara, and S. Takeuchi, Experimental demonstra- tion of adaptive quantum state estimation, Phys. Rev. Lett.109, 130404 (2012)

    2012

  45. [45]

    Z. Hou, H. Zhu, G.-Y. Xiang, C.-F. Li, and G.-C. Guo, Achieving quantum precision limit in adaptive qubit state tomography, npj Quantum Information2, 10.1038/npjqi.2016.1 (2016)

    doi:10.1038/npjqi.2016.1 2016

  46. [46]

    Hwang, J

    H. Hwang, J. Choi, and E. Kim, Adaptive quantum tomography in an indistinct measurement system with superconducting circuits, Phys. Rev. Appl.20, 064007 (2023)

    2023

  47. [47]

    A. S. Aasen, A. Di Giovanni, H. Rotzinger, A. V. Ustinov, and M. G¨ arttner, Readout error mitigated quantum state tomography tested on superconducting qubits, Commu- nications Physics7, (2024)

    2024

  48. [48]

    Arrasmith, A

    A. Arrasmith, A. Patterson, A. Boughton, and M. Paini, Development and demonstration of an efficient readout error mitigation technique for use in nisq algorithms (2023), arXiv:2303.17741 [quant-ph]

    Pith/arXiv arXiv 2023

  49. [49]

    F. B. Maciejewski, Z. Zimbor´ as, and M. Oszmaniec, Mit- igation of readout noise in near-term quantum devices by classical post-processing based on detector tomography, Quantum4, 257 (2020)

    2020

  50. [50]

    Luis and L

    A. Luis and L. L. S´ anchez-Soto, Complete characteriza- tion of arbitrary quantum measurement processes, Phys. Rev. Lett.83, 3573 (1999)

    1999

  51. [51]

    J. S. Lundeen, A. Feito, H. Coldenstrodt-Ronge, K. L. Pregnell, C. Silberhorn, T. C. Ralph, J. Eisert, M. B. Plenio, and I. A. Walmsley, Tomography of quantum de- tectors, Nature Physics5, 27–30 (2008)

    2008

  52. [52]

    Z. Cai, R. Babbush, S. C. Benjamin, S. Endo, W. J. Hug- gins, Y. Li, J. R. McClean, and T. E. O’Brien, Quan- tum error mitigation, Reviews of Modern Physics95, 10.1103/revmodphys.95.045005 (2023)

    doi:10.1103/revmodphys.95.045005 2023

  53. [53]

    Di Giovanni, A

    A. Di Giovanni, A. S. Aasen, J. Lisenfeld, M. G¨ arttner, H. Rotzinger, and A. V. Ustinov, Benchmarking the qual- ity of multiplexed qubit readout beyond assignment fi- delity, Physical Review Applied24, 044043 (2025)

    2025

  54. [54]

    Y. I. Bogdanov, B. I. Bantysh, N. A. Bogdanova, A. B. Kvasnyy, and V. F. Lukichev, Quantum states tomog- raphy with noisy measurement channels, International Conference on Micro- and Nano-Electronics 201610224, 102242O (2016)

    2016

  55. [55]

    Y. I. Bogdanov, N. A. Bogdanova, Y. A. Kuznetsov, K. B. Koksharov, and V. F. Lukichev, Precise tomography of qudits, Russian Microelectronics52, 135–143 (2023)

    2023

  56. [56]

    V. N. Ivanova-Rohling, N. Rohling, and G. Burkard, Optimal quantum state tomography with noisy gates, EPJ Quantum Technology10, 10.1140/epjqt/s40507- 023-00181-2 (2023)

    doi:10.1140/epjqt/s40507- 2023

  57. [57]

    Demkowicz-Dobrza´ nski, J

    R. Demkowicz-Dobrza´ nski, J. Czajkowski, and P. Sekatski, Adaptive quantum metrology under general markovian noise, Phys. Rev. X7, 041009 (2017)

    2017

  58. [58]

    Oshio, Y

    K. Oshio, Y. Suzuki, K. Wada, K. Hisanaga, S. Uno, and N. Yamamoto, Adaptive measurement strategy for noisy quantum amplitude estimation with variational quantum circuits, Phys. Rev. A110, 062423 (2024)

    2024

  59. [59]

    Fiur´ aˇ sek, Maximum-likelihood estimation of quantum measurement, Physical Review A64, (2001)

    J. Fiur´ aˇ sek, Maximum-likelihood estimation of quantum measurement, Physical Review A64, (2001)

    2001

  60. [60]

    Blume-Kohout, Optimal, reliable estimation of quan- tum states, New Journal of Physics12, 043034 (2010)

    R. Blume-Kohout, Optimal, reliable estimation of quan- tum states, New Journal of Physics12, 043034 (2010)

    2010

  61. [61]

    A. I. Lvovsky, Iterative maximum-likelihood reconstruc- tion in quantum homodyne tomography, Journal of Op- tics B: Quantum and Semiclassical Optics6, S556–S559 (2004)

    2004

  62. [62]

    Gorini, A

    V. Gorini, A. Kossakowski, and E. C. G. Sudarshan, Completely positive dynamical semigroups of n-level sys- tems, Journal of Mathematical Physics17, 821 (1976)

    1976

  63. [63]

    A. S. Holevo and V. Giovannetti, Quantum channels and their entropic characteristics, Reports on Progress in Physics75, 046001 (2012)

    2012

  64. [64]

    Magesan, Depolarizing behavior of quantum channels in higher dimensions, Quantum Information and Compu- tation11, 466–484 (2011)

    E. Magesan, Depolarizing behavior of quantum channels in higher dimensions, Quantum Information and Compu- tation11, 466–484 (2011)

    2011

  65. [65]

    C. A. Fuchs and C. M. Caves, Ensemble-dependent bounds for accessible information in quantum mechan- ics, Phys. Rev. Lett.73, 3047 (1994)

    1994

  66. [66]

    R. D. Gill and S. Massar, State estimation for large en- sembles, Phys. Rev. A61, 042312 (2000)

    2000

  67. [67]

    M. K. Patra, Quantum state determination: estimates for information gain and some exact calculations, Jour- nal of Physics A: Mathematical and Theoretical40, 10887–10902 (2007)

    2007

  68. [68]

    A. W. R. Smith, K. E. Khosla, C. N. Self, and M. S. Kim, Qubit readout error mitigation with bit- flip averaging, Science Advances7, eabi8009 (2021), https://www.science.org/doi/pdf/10.1126/sciadv.abi8009

    doi:10.1126/sciadv.abi8009 2021

  69. [69]

    Aasen, CREMST (2025)

    A. Aasen, CREMST (2025)

    2025

  70. [70]

    Limitations for adaptive quan- tum state tomography in the presence of detector noise

    A. Aasen, Dataset for “Limitations for adaptive quan- tum state tomography in the presence of detector noise”, 10.5281/zenodo.18076123 (2025)

    work page doi:10.5281/zenodo.18076123 2025

  71. [71]

    C. R. Harris, K. J. Millman, S. J. van der Walt, R. Gom- mers, P. Virtanen, D. Cournapeau, E. Wieser, J. Tay- lor, S. Berg, N. J. Smith, R. Kern, M. Picus, S. Hoyer, M. H. van Kerkwijk, M. Brett, A. Haldane, J. F. del R ´ ıo, M. Wiebe, P. Peterson, P. G´ erard-Marchant, K. Shep- pard, T. Reddy, W. Weckesser, H. Abbasi, C. Gohlke, 19 and T. E. Oliphant, Ar...

    2020

  72. [72]

    Virtanen, R

    P. Virtanen, R. Gommers, T. E. Oliphant, M. Haber- land, T. Reddy, D. Cournapeau, E. Burovski, P. Pe- terson, W. Weckesser, J. Bright, S. J. van der Walt, M. Brett, J. Wilson, K. J. Millman, N. Mayorov, A. R. J. Nelson, E. Jones, R. Kern, E. Larson, C. J. Carey, I. Polat, Y. Feng, E. W. Moore, J. VanderPlas, D. Laxalde, J. Perktold, R. Cimrman, I. Henriks...

    2020

  73. [73]

    Joblib Development Team, Joblib: running python func- tions as pipeline jobs (2021)

    2021

  74. [74]

    J. D. Hunter, Matplotlib: A 2d graphics environment, Computing in Science & Engineering9, 90 (2007)

    2007

  75. [75]

    ˙Zyczkowski, K

    K. ˙Zyczkowski, K. A. Penson, I. Nechita, and B. Collins, Generating random density matrices, Journal of Mathe- matical Physics52, 062201 (2011)

    2011

This paper was first reviewed by deepseek-v4-flash on August 3, 2026.