Pith. sign in

REVIEW 2 major objections 5 minor 1 cited by

Quantum observables over time for information recovery

T0 review · 2 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read This paper claims that a two-time quantum observable over time can define recovery maps that restore a chosen observable's noiseless expectation value, with optimal sampling overhead in the two worked examples.

desk verdict A genuinely new observable-dual of QSOT with correct optimal examples, but the lambda-regularization for traceless observables is unproven and the Appendix F trace constraint has a missing factor of 2. read the letter →

arxiv 2412.11659 v1 pith:X4KRERRD submitted 2024-12-16 quant-ph

Gabriele Bressanini , Farhan Hanif , Hyukjoon Kwon , M. S. Kim This is my paper · ORCID
classification quant-ph
keywords quantumobservablesovertimestaterecoverymapstime-reversalsymmetryerrormitigationprobabilisticcancellationJordanproductgeneralizedamplitudedamping
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 observables over time (QOOT) are operators that jointly describe an observable and its Heisenberg-evolved image across two times, the observable analogue of the quantum state over time. This paper claims that a QOOT exists exactly when the evolution preserves the trace of that observable, and that using the Jordan-product form of the QOOT turns a time-reversal symmetry equation into a linear system whose solution is a recovery map restoring the observable's noiseless expectation value for every state. Because such recovery maps are usually non-physical, the paper decomposes them into physically realizable channels, yielding unbiased estimators of the noiseless expectation value. In the two examples worked out in detail, generalized amplitude damping with reference Pauli X and stochastic Pauli noise with reference Pauli Z, the sampling overhead equals the previously established lower bound and beats probabilistic error cancellation based on the channel inverse.

What carries the argument

The load-bearing object is the Jordan product quantum observable over time, $E^\dagger \star O = \tfrac{1}{2}\{O\otimes I, D[E^\dagger]\}$, built from the anti-commutator of the reference observable with the Jamiołkowski state $D[E^\dagger]=\sum_{ij} |i\rangle\langle j| \otimes E^\dagger(|j\rangle\langle i|)$. This object converts the abstract time-reversal symmetry equation into solvable linear systems whose solutions are the pre- and post-processing recovery maps, and it supplies the well-definedness condition $\{O, E(I)\}=2O$, which at the general QOOT level is equivalent to the trace condition $\mathrm{Tr}[E^\dagger(O)]=\mathrm{Tr}[O]$. The time-reversal map $\tau$, defined as the conjugate-linear extension of $\tau(B\otimes A)=A^\dagger\otimes B^\dagger$, puts the forward and backward joint observables on equal footing. For traceless reference observables the linear systems become singular, and the paper works around this by regularizing $O$ to $O+\lambda I$ and taking $\lambda\to 0$ after solving.

What would settle it

Compute the recovery map for Pauli X under generalized amplitude damping using two different regularization paths: $O_\lambda = X+\lambda I$ and $O_\lambda = X+\lambda I+\mu Z$, sending $\mu\to 0$ before or after $\lambda\to 0$. If the resulting maps, or their sampling costs $\gamma$, disagree along different paths, the claimed systematic construction is not well defined for traceless reference observables.

Watch

Extended reading notes

Core claim

The paper's central discovery is a systematic route from a joint object, a quantum observable over time, to a recovery map that protects a chosen observable from a known noise channel. The Jordan-product QOOT, $E^\dagger \star O = \tfrac{1}{2}\{O\otimes I, D[E^\dagger]\}$, is defined by anti-commuting the reference observable with the Jamiołkowski state of the adjoint channel; it automatically yields $E^\dagger(O)$ as one of its marginals and reproduces $O$ as the other whenever the consistency condition $\{O, E(I)\}=2O$ holds. Imposing the time-reversal symmetry $E^\dagger \star O = \tau(P^\dagger \star E^\dagger(O))$ for pre-processing, or $R^\dagger \star O = \tau(E^\dagger \star R^\dagger(O))$ for post-processing, gives explicit linear equations for $P$ and $R$, and any solution automatically satisfies the recovery property $P^\dagger(E^\dagger(O))=O$ or $E^\dagger(R^\dagger(O))=O$. For the GAD channel with reference $X$ and the stochastic Pauli channel with reference $Z$, the resulting recovery maps are stochastic Pauli maps with negative coefficients, and their sampling costs $\gamma=1/\sqrt{1-\epsilon}$ and $\gamma=1/|p_0-p_1-p_2+p_3|$ coincide with the lower bounds reported in Ref. [23].

Load-bearing premise

The results for traceless reference observables hinge on the limit of a regularization parameter, $O \to O+\lambda I$ followed by $\lambda \to 0$, being independent of the regularization scheme, and the paper does not prove that uniqueness.

Editorial extensions

If this is right

  • Whenever the consistency condition is met, the recovery map is obtained by solving a linear system defined by the noise channel and the reference observable, with no optimization over candidate maps.
  • Because $P^\dagger(E^\dagger(O))=O$ or $E^\dagger(R^\dagger(O))=O$, the protocol returns the exact noiseless expectation value of the reference observable for every input state, not only for a reference state.
  • For generalized amplitude damping protecting $X$, the sampling overhead $\gamma=1/\sqrt{1-\epsilon}$ is smaller than the inverse-channel PEC cost for all $p$ and $\epsilon$.
  • For stochastic Pauli noise protecting $Z$, $\gamma=1/|p_0-p_1-p_2+p_3|$, and both examples match the optimal lower bound of Ref. [23].
  • The construction is not universal: a QOOT exists only when $\mathrm{Tr}[E^\dagger(O)]=\mathrm{Tr}[O]$, and when the noise is unitary the recovery map reduces to the channel inverse.

Reading between the lines

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

  • The $\lambda\to 0$ regularization for traceless observables deserves a direct test: if two different regularizations, say $O+\lambda I$ and $O+\lambda I+\mu Z$, give different recovery maps in the limit, the systematic construction is not canonical for traceless references.
  • The same machinery should extend to multi-qubit Pauli observables under twirled Pauli noise, with the sampling cost controlled by the spectrum of $E^\dagger(O)$; this is a natural next test of the optimality pattern.
  • Because the uncorrelated QOOT already satisfies the recovery property but does not select a unique map, the formalism suggests that choosing an error-mitigation strategy is equivalent to choosing a temporal joint observable, an interpretation the paper does not develop.
  • The observed optimality in two examples may hint at a general theorem relating QOOT-derived recovery maps to information-recoverability lower bounds; the paper explicitly leaves this as an open question.
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

2 major / 5 minor

Summary. Bressanini et al. introduce the quantum observable over time (QOOT), a bipartite operator on HA⊗HB whose partial traces recover a reference observable O and its Heisenberg-evolved image E†(O). They characterize when such a QOOT exists (Tr[O]=Tr[E†(O)]) and, for their Jordan-product instance, when the additional condition {O,E(I)}=2O holds. They then define pre- and post-processing recovery maps P and R through time-reversal equations involving QOOTs, obtaining P†(E†(O))=O or E†(R†(O))=O. The maps are solved explicitly for qubits, and for generalized amplitude damping with reference X (pre-processing) and stochastic Pauli noise with reference Z (post-processing), the paper provides Choi decompositions into Pauli channels, sampling costs γ, and compares these with lower bounds from Ref. [23], finding that the costs match the optimal values and are lower than the PEC cost of implementing the channel inverse.

Significance. The paper makes a useful conceptual contribution by extending the state-over-time formalism to observables and linking it to observable-specific error mitigation. It provides fully explicit recovery maps for two standard noise models, with costs that reproduce the lower bounds of Ref. [23] and improve on standard PEC; the examples are worked out in enough detail to be checked, and their internal algebra is consistent. The main weakness is that the examples with traceless Pauli observables require a λI regularization of a singular equation, and the paper does not prove that the λ→0 limit is scheme-independent; for the GAD-X case the regularized observable even violates the well-definedness condition of the very QOOT used in the time-reversal equation. If this gap is closed, the framework would indeed provide a systematic closed-form route to optimal observable-specific recovery maps.

major comments (2)
  1. [VI A, Eqs. (49)-(52) and (16)] The GAD-X pre-processing map is obtained by applying Eq. (48) to the regularized observable Oλ=λI+X and taking λ→0. This step is not justified within the QOOT framework. For 2p≠1 the noise is non-unital, E(I)=I+ε(2p−1)Z, so {Oλ,E(I)} = 2(λI+X)+2λε(2p−1)Z = 2Oλ+2λε(2p−1)Z, which violates Eq. (16), the condition for the Jordan QOOT to satisfy the marginal property (2). The time-reversal equation (19) that defines the recovery map is therefore applied to a pair (E,Oλ) for which the QOOT is not well defined. Moreover, even setting this aside, the formula (48) is singular at λ=0 because the eigenvalues of E†(X) sum to zero, and the paper gives no argument that the limiting map is independent of the regularization scheme (e.g., X+λZ would also lift the degeneracy when E†(Z) has a trace component). Since the claimed optimality is for the specific map obtained from the λI regularization, scheme dependence would invalidate the claim that the QOOT framework systematically produces the optimal recovery map. The authors should either prove uniqueness of the limit or provide a derivation that avoids applying the QOOT equation outside its domain.
  2. [VI B, Eqs. (34)-(38) and (66)-(67)] The stochastic Pauli-Z example uses the same λI regularization for the degenerate eigenvalues of Z. Although the noise is unital and the QOOT well-definedness condition (16) is satisfied, the core equation (34) is singular at λ=0, and the text states only that solving the linear system and taking λ→0 yields Eq. (70). No derivation or uniqueness proof is given for this limiting solution. The recovery property alone fixes only R†(Z); the action on X and Y is determined by the time-reversal equation, so one needs to know that the λ→0 limit is independent of the regularization to conclude that the QOOT construction uniquely defines the optimal map. Please provide a proof of limit uniqueness or an explicit argument that the limiting solution is the unique HPTP extension satisfying the time-reversal equation.
minor comments (5)
  1. [IV B and Appendix F] The trace constraint for the post-processing map is misstated: Appendix F's calculation gives (1/2)Σ_{kℓ} Tr[X_{kℓ}]|ω_k><ω_ℓ| = O, so the condition {O,R(I)}=2O is equivalent to Tr[X_{kℓ}]=2q_kδ_{kℓ}, not q_kδ_{kℓ} as stated in Sec. IV B and in the final sentence of Appendix F. The examples satisfy the factor-2 version, so this is a typo rather than a substantive error, but it should be corrected.
  2. [IV, Eq. (17) and Eq. (23)] The symbol O is reused for both the reference observable and the operator that anticommutes with it in the condition E(I)=I+O; this is confusing and the second O should be a different symbol, e.g., Δ with {O,Δ}=0.
  3. [II] Calling the trace-condition result a 'no-go theorem' is misleading: Eq. (5) provides a universal construction whenever Tr[O]=Tr[E†(O)], so the result is a characterization of when a QOOT exists, not a no-go statement in the usual sense.
  4. [VI B] The sentence 'The solution to these equations, upon taking λ→0, yields the following recovery map' would benefit from a brief derivation or a reference to an appendix; currently the reader must trust the algebra, especially because the solution is central to the optimality claim.
  5. [IV A] There is a typo in the sentence preceding Eq. (49): 'this condition condition is not satisfied' should read 'this condition is not satisfied'.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the recovery maps are derived from an explicit time-reversal equation and benchmarked against external lower bounds, not fitted to the claimed outputs.

full rationale

The paper's central derivation is self-contained. The Jordan-product QOOT is defined in Eq. (13), and substituting it into the time-reversal equations (7) and (11) yields algebraic equations (22) and (33); solving those equations gives explicit recovery maps, e.g. Eq. (26) for pre-processing and the qubit post-processing solution in Eqs. (70). The recovery property E†(R†(O)) = O follows from the partial-trace condition on the QOOTs, so it is a designed consequence of the construction rather than an empirical prediction or a fitted parameter. The claimed optimality of the examples is checked against an external lower bound from Ref. [23], which optimizes over all HPTS recovery maps; the paper does not tune any parameter to match that bound. The λ-regularization used when q_k + q_l = 0 is a technical device to handle a singular eigen-decomposition; concerns about scheme-dependence of the limit would be a mathematical well-posedness issue, not circularity. The self-citations that appear (e.g., Refs. [7] and [30]) are background references and are not load-bearing for the recovery-map derivation or the optimality claims. No step in the derivation reduces to its own input by construction, and no fitted quantity is relabeled as a prediction.

Assumptions & free parameters 1 free parameters · 5 assumptions · 0 invented entities

The central derivations rest on standard linear algebra and Choi-Jamiolkowski identities, plus several modeling choices specific to this paper: the choice of the Jordan product QOOT, the imposition of the time-reversal equations as the definition of recovery, the use of a lambda-regularization for traceless observables, and the external optimality bounds of Ref. [23]. No free parameters are fitted to data; lambda is a regulator sent to zero.

free parameters (1)
  • regularization parameter lambda = lambda -> 0 (limit)
    Ad hoc parameter introduced in Eqs. (49) and (66) to make the recovery-map equations non-singular for traceless Pauli observables. The final maps are lambda-independent in the examples, but scheme-independence is not proven.
assumptions (5)
  • ad hoc to paper The Jordan product QOOT E† ⋆ O = 1/2 {O⊗I, D[E†]} is chosen as the instance of quantum observable over time; this selection is not unique and the recovery maps depend on it.
    Section IV introduces this choice due to its desirable properties; other QOOT definitions would yield different time-reversal equations and different recovery maps.
  • domain assumption A recovery map is defined by imposing the time-reversal symmetry equations (7) and (11); this is a postulate of the framework, not a theorem.
    Section III states these equations as the implicit definition of recovery maps. Alternative definitions such as the Petz map or SDP-optimized maps exist; the equation is chosen so that recovery follows from partial tracing.
  • ad hoc to paper For traceless reference observables, the recovery map is the lambda to zero limit of the regularized problem, and this limit exists and is independent of the regularization scheme.
    Used in Section VI A, Eq. (49), and Section VI B, Eq. (66). The regularized observable O+lambda I does not satisfy the well-definedness condition for non-unital noise at lambda > 0, and scheme-independence is unproven.
  • standard math HPTP maps can be decomposed into CPTP maps with real coefficients summing to one, with sampling cost gamma = sum of absolute coefficients, as per Ref. [21].
    Invoked in Section V, Eqs. (39)-(41); this is a known result in quantum error mitigation.
  • domain assumption The optimal sampling overhead lower bounds for the GAD-X and stochastic-Pauli-Z tasks are taken from Ref. [23] and are not re-derived in this paper.
    The optimality claims rest on the external SDP results of Ref. [23]; the paper does not reproduce those lower bounds.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Quantum observables over time for information recovery." pith.science (2026). https://pith.science/paper/X4KRERRD

@misc{pith2026241211659,
  author       = {Pith},
  title        = {Pith review of: Quantum observables over time for information recovery},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/X4KRERRD}},
  note         = {Machine review of arXiv:2412.11659}
}
read the original abstract

We introduce the concept of quantum observables over time (QOOT), an operator that jointly describes two observables at two distinct time points, as a dual of the quantum state over time formalism. We provide a full characterization of the conditions under which a QOOT can be properly defined, via a no-go theorem. We use QOOTs to establish a notion of time-reversal for generic quantum channels with respect to a reference observable, enabling the systematic construction of recovery maps that preserve the latter. These recovery maps, although generally non-physical, can be decomposed into realizable channels, enabling their application in noiseless expectation value estimation tasks. We provide explicit examples and compare our protocol with other error mitigation methods. We show that our protocol retrieves the noiseless expectation value of the reference observable and can achieve optimal sampling overhead, outperforming probabilistic error cancellation.

Figures

Figures reproduced from arXiv: 2412.11659 by the authors.

Figure 2
Figure 2. FIG. 2. Pictorial representation of the post-processing protocol. De [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Experimental Virtual Quantum Broadcasting

    quant-ph 2025-01 conditional novelty 5.0 of 10

    Researchers experimentally realized virtual quantum broadcasting on an NMR processor by combining a universal cloner with a universal antisymmetrizer via linear combination of unitaries and classical post-processing.

Reference graph

Works this paper leans on

31 extracted references · 17 canonical work pages · cited by 1 Pith paper

  1. [23]

    S. Endo, S. C. Benjamin, and Y . Li, Practical quantum error mitigation for near-future applications, Phys. Rev. X 8, 031027 (2018)

    work page 2018

  2. [1]

    M. A. Nielsen and I. L. Chuang, Quantum computation and quantum information (Cambridge university press, 2010)

    2010

  3. [2]

    Furthermore, since R† is manifestly TP, it implies that R is unital, thereby satisfying the consistency condition in Eq. (30). Next, we want to express the recovery map as a linear combination of physically realizable maps. To this end, following the procedure outlined in the previous example, we compute the Choi matrix of R†, diagonalize it, and unvec- t...

    work page 2020

  4. [3]

    M. M. Wilde, Quantum information theory (Cambridge univer- sity press, 2013)

    2013

  5. [4]

    Fullwood and A

    J. Fullwood and A. J. Parzygnat, On quantum states over time, Proceedings of the Royal Society A 478, 20220104 (2022)

    2022

  6. [5]

    A. J. Parzygnat and F. Buscemi, Axioms for retrodiction: achieving time-reversal symmetry with a prior, Quantum 7, 1013 (2023)

    2023

  7. [6]

    A. J. Parzygnat and J. Fullwood, From time-reversal symmetry to quantum bayes’ rules, PRX Quantum 4, 020334 (2023)

    2023

  8. [7]

    Petz, Sufficient subalgebras and the relative entropy of states of a von neumann algebra, Communications in mathematical physics 105, 123 (1986)

    D. Petz, Sufficient subalgebras and the relative entropy of states of a von neumann algebra, Communications in mathematical physics 105, 123 (1986)

    1986

Show all 31 references
  1. [8]

    H. Kwon, R. Mukherjee, and M. S. Kim, Reversing lindblad dynamics via continuous petz recovery map, Phys. Rev. Lett. 128, 020403 (2022)

    2022

  2. [9]

    Barnum and E

    H. Barnum and E. Knill, Reversing quantum dynamics with near-optimal quantum and classical fidelity, Journal of Math- ematical Physics 43, 2097 (2002)

    2002

  3. [10]

    M. M. Wilde, Recoverability in quantum information theory, Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences 471, 20150338 (2015)

    2015

  4. [11]

    Temme, S

    K. Temme, S. Bravyi, and J. M. Gambetta, Error mitigation for short-depth quantum circuits, Phys. Rev. Lett. 119, 180509 (2017)

    2017

  5. [12]

    M. S. Leifer and R. W. Spekkens, Towards a formulation of quantum theory as a causally neutral theory of bayesian in- ference, Physical Review A—Atomic, Molecular, and Optical Physics 88, 052130 (2013)

    2013

  6. [13]

    Horsman, C

    D. Horsman, C. Heunen, M. F. Pusey, J. Barrett, and R. W. Spekkens, Can a quantum state over time resemble a quan- tum state at a single time?, Proceedings of the Royal Soci- ety A: Mathematical, Physical and Engineering Sciences 473, 20170395 (2017)

    2017

  7. [14]

    Ohya, Note on quantum probability, Lettere al Nuovo Ci- mento (1971-1985) 38, 402 (1983)

    M. Ohya, Note on quantum probability, Lettere al Nuovo Ci- mento (1971-1985) 38, 402 (1983)

    1983

  8. [15]

    Ohya, On compound state and mutual information in quan- tum information theory (corresp.), IEEE Transactions on infor- mation theory 29, 770 (1983)

    M. Ohya, On compound state and mutual information in quan- tum information theory (corresp.), IEEE Transactions on infor- mation theory 29, 770 (1983). 10

    1983

  9. [16]

    Asorey, A

    M. Asorey, A. Kossakowski, G. Marmo, and E. G. Sudarshan, Relations between quantum maps and quantum states, arXiv preprint quant-ph/0602228 (2006)

    2006 arXiv

  10. [17]

    S. H. Lie and N. H. Y . Ng, Quantum state over time is unique, Phys. Rev. Res. 6, 033144 (2024)

    2024

  11. [18]

    J. F. Fitzsimons, J. A. Jones, and V . Vedral, Quantum correla- tions which imply causation, Scientific reports5, 18281 (2015)

    2015

  12. [19]

    M. S. Leifer, Quantum dynamics as an analog of conditional probability, Phys. Rev. A74, 042310 (2006)

    2006

  13. [20]

    M. S. Leifer and R. W. Spekkens, Towards a formulation of quantum theory as a causally neutral theory of bayesian infer- ence, Phys. Rev. A 88, 052130 (2013)

    2013

  14. [21]

    Buscemi, M

    F. Buscemi, M. Dall’Arno, M. Ozawa, and V . Vedral, Direct ob- servation of any two-point quantum correlation function, arXiv preprint arXiv:1312.4240 (2013)

    2013 arXiv

  15. [22]

    Jiang, K

    J. Jiang, K. Wang, and X. Wang, Physical implementability of linear maps and its application in error mitigation, Quantum 5, 600 (2021)

    2021

  16. [24]

    X. Zhao, B. Zhao, Z. Xia, and X. Wang, Information recover- ability of noisy quantum states, Quantum 7, 978 (2023)

    2023

  17. [25]

    Jamiołkowski, Linear transformations which preserve trace and positive semidefiniteness of operators, Reports on mathe- matical physics 3, 275 (1972)

    A. Jamiołkowski, Linear transformations which preserve trace and positive semidefiniteness of operators, Reports on mathe- matical physics 3, 275 (1972)

    1972

  18. [26]

    Hoeffding, Probability inequalities for sums of bounded ran- dom variables, The collected works of Wassily Hoeffding , 409 (1994)

    W. Hoeffding, Probability inequalities for sums of bounded ran- dom variables, The collected works of Wassily Hoeffding , 409 (1994)

    1994

  19. [27]

    X. Zhao, L. Zhang, B. Zhao, and X. Wang, Power of quan- tum measurement in simulating unphysical operations, arXiv preprint arXiv:2309.09963 (2023)

    2023 arXiv

  20. [28]

    Z. Cai, R. Babbush, S. C. Benjamin, S. Endo, W. J. Huggins, Y . Li, J. R. McClean, and T. E. O’Brien, Quantum error mitiga- tion, Rev. Mod. Phys. 95, 045005 (2023)

    2023

  21. [29]

    Dankert, R

    C. Dankert, R. Cleve, J. Emerson, and E. Livine, Exact and approximate unitary 2-designs and their application to fidelity estimation, Phys. Rev. A 80, 012304 (2009)

    2009

  22. [30]

    Khatri, K

    S. Khatri, K. Sharma, and M. M. Wilde, Information-theoretic aspects of the generalized amplitude-damping channel, Phys. Rev. A 102, 012401 (2020)

    2020

  23. [31]

    Taylor, G

    A. Taylor, G. Bressanini, H. Kwon, and M. S. Kim, Quantum error cancellation in photonic systems: Undoing photon losses, Phys. Rev. A 110, 022622 (2024). Appendix A: Useful identities In this Appendix we prove that, for a CPTP map E it holds true that τ (D[E †]) = D[E]. D[E] =...

    2024

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.