{"id":"9b336aee-b605-4743-8ca3-5648f5a71166","arxiv_id":"2501.05270","paper_version":4,"verdict":"REJECT","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"The paper derives identifiability criteria for time-independent GKSL open quantum systems by mapping their measurement dynamics onto classical linear and bilinear dynamical systems and inverting the parameter map.","lead":"Open quantum systems are hard to identify: the paper connects their measurement dynamics to classical linear and bilinear system identification, giving rank and invertibility conditions under which the master equation can be reconstructed. If the conditions hold, experimenters get a formal test for when their sampling, pulses, and measurements are sufficient to recover Hamiltonian and noise parameters.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The identification theorems recover parameters only in the original coherence-vector basis, but LDS/BDS identification determines the realization only up to similarity; the paper never fixes this gauge, so the reconstructed theta and gamma need not be the physical parameters.","rationale":"The reader's weakest assumption identifies precisely the similarity-gauge gap: identification determines the realization only up to T, while the parameter reconstruction theorems require the specific coherence-vector-basis representative. This is the load-bearing issue because both main results, Corollary 17 and Theorem 19, rely on the second step that converts an identified LDS/BDS matrix into master-equation parameters. The paper's own proof of Corollary 17 acknowledges the similarity ambiguity but asserts it does not affect the recovered GKSL dynamics; that assertion is exactly what is unsupported, since a generic similar matrix leaves the image of phi. In the BDS case the issue is compounded because the matrices N_j are asserted to be known structure constants, yet they too transform under similarity. The proposed numerical test on the paper's own two-qubit example would settle the concern: if a random similar realization gives nonzero projection residuals, the central claim fails as written. Since the reader already rejected the paper on this basis and our analysis confirms the gap without finding an independent reason to alter the verdict, the reader's REJECT stands unchanged.","tokens_in":20,"tokens_out":10915,"duration_ms":298178,"concrete_test":"Use the Section VI two-qubit example. Construct A=phi(theta,gamma) from the given Hamiltonian and Kossakowski parameters using (90)/(101) and the published GitHub code. Choose a generic invertible 15x15 T (e.g., random with moderate condition number), compute A' = T^{-1} A T, and test the image conditions of Theorem 15: form A_sym=(A'+A'^T)/2 and A_anti=(A'-A'^T)/2, then compute the minimum-norm residuals of vec(A_anti) against Im T1 and vec(A_sym) against Im T3 using pseudo-inverses. If either residual substantially exceeds machine precision times the norm of A, then the similarity-transformed representative is not in the image of phi, so Algorithm 5 (Theorem 15) cannot recover theta,gamma from the identified matrix, refuting Corollary 17 and Theorem 19 as stated. Repeat for several random T to confirm genericity.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Identifiability of the LDS/BDS is established only up to similarity: Lemma 8 yields A only up to T^{-1}AT, and Proposition 9 with (79) gives the same for (A,N_j,b,C). Yet Theorem 13 and Theorem 15 invert the parameter map phi(theta,gamma)=A on matrices expressed in the original structure-constant basis. The proof of Corollary 17 states 'the matrix of the LDS is recovered up to similarity transform, which does not effect the dynamics,' but this does not imply that the particular representative A' produced by the identification step lies in Im phi. For n>3 a generic A'=T^{-1}AT will have an antisymmetric part outside Im T1 and a symmetric part outside Im T3, so the reconstruction equations (C3)/(C4) have no solution; no theta,gamma are recovered, and the asserted 'uniquely constructed GKSL equation with identical dynamics' is unsupported. The same problem is worse for BDS: the N_j matrices, asserted to be known structure constants, are also transformed to T^{-1}N_jT, so they are no longer known unless T lies in the centralizer of all N_j. Although known C, b, or N_j could in principle constrain T, the paper provides no gauge-fixing procedure, so the advertised claim is not established.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper connects time-independent GKSL master equations for open quantum systems to linear and bilinear dynamical systems (LDS and BDS) in the coherence-vector representation, and then imports classical system-identification results to give conditions under which the system matrices, and subsequently the Hamiltonian parameters theta and Kossakowski parameters gamma, can be recovered from discrete measurement data. For autonomous systems, the claimed result is Corollary 17: under full-rank controllability/observability and irrational sample-rate ratios, the LDS form is recoverable, and if the matrix M of Definition 12 is invertible, a GKSL equation with identical dynamics can be uniquely constructed. For controlled systems, Theorem 19 makes the analogous claim using full-rank bilinear controllability/observability matrices and pulses of the form (76). The paper also gives a symmetric-Kossakowski special case (Theorem 15), algorithms for the reconstruction procedure, and a two-qubit example. The derivations in Appendices A-C are coherent, and the identification theorems are quoted from standard sources, but the central reconstruction claim is not established as stated because the identified LDS and BDS realizations are only known up to similarity transformation, and the paper does not show that the particular representative returned by identification lies in the image of the parameter map phi.","tokens_in":44150,"tokens_out":8133,"duration_ms":91218,"significance":"If the main claims were established, the paper would provide practically useful sufficient conditions for deciding, from design choices such as sampling rates, control pulses, and measurement matrices, whether the parameters of a GKSL master equation can be recovered from discrete measurement data. The unification of autonomous and controlled open quantum systems within classical LDS/BDS identification is a useful conceptual contribution, and the explicit rank and invertibility criteria are concrete enough to be tested numerically. The paper also gives credit where it is due: it relies on external theorems for LDS/BDS identification, states its assumptions clearly, and provides a worked two-qubit example. However, the significance is conditional on resolving the similarity-transform ambiguity described below; without that resolution, the advertised reconstruction theorems do not follow from the identification theorems they invoke.","major_comments":[{"comment":"The proof of Corollary 17 acknowledges that 'the matrix of the LDS is recovered up to similarity transform, which does not effect the dynamics,' but this does not justify applying Theorems 13 and 15. Those theorems invert the parameter map phi(theta, gamma) = A^(l)(theta) + A^(d)(gamma) on matrices expressed in the original structure-constant basis, where the matrices T1 and T3 are defined. If identification returns A' = T^{-1} A T, then A' need not lie in the image of phi: for n > 3 the antisymmetric part of a generic A' lies outside Im T1 (equation (C3)), so the reconstruction equations have no solution and no theta or gamma is recovered. The similarity transform also changes the affine vector beta to T^{-1} beta, so the vectorized system (B2) is not the one being solved. To establish the claim, the paper must either prove that the identified realization can be chosen in the original basis (for example, by using a known full-column-rank C to enforce C T = C), or add a gauge-fixing procedure and show that the resulting A' satisfies the image conditions of Theorem 15 and the linear system of Theorem 13. As written, the statement 'uniquely construct a GKSL equation, which has identical dynamics to the original equation' is unsupported.","section":"V B, Corollary 17 proof; V A, Theorems 13 and 15"},{"comment":"For controlled systems the same ambiguity is worse. Proposition 9 and the similarity relations (79) show that BDS identification determines (A, N_j, b, C) only up to T^{-1} N_j T and C T. The paper repeatedly states that the bilinear coupling matrices N_j are known a priori because they are built from the antisymmetric structure constants, but this is only true in the original coherence-vector basis. Under the identification step the matrices actually obtained are T^{-1} N_j T, which are no longer the known structure constants unless T centralizes every N_j. The paper provides no argument that the similarity transformation from the identification algorithm has this centralizing property, and no procedure for using the known C, b, or N_j to fix T. Consequently, Theorem 19's assertion that one can 'uniquely construct a GKSL equation' from the identified BDS is not established.","section":"III C, Proposition 9 and Eq. (79); V A, Theorem 19"}],"minor_comments":[{"comment":"In the symmetric-Kossakowski case, Theorem 19 says the invertibility condition on M is replaced by invertibility of T2, but Corollary 17 and Theorem 15 require invertibility of T3, the symmetric reduction defined in Definition 14. This is an inconsistency in a load-bearing condition and should be corrected.","section":"V B, Theorem 19"},{"comment":"The heading calls Theorem 16 'necessary conditions for discrete identifiability,' but the statement begins with 'A sufficient condition.' The heading and statement should be aligned.","section":"V B, Theorem 16"},{"comment":"The corollary is titled 'on the identification of an autonomous open quantum systems' but the statement refers to the BDS form of a closed quantum system; the title appears to be a copy-paste error.","section":"V B, Corollary 20"},{"comment":"The two-qubit example assumes the system matrix A is known exactly in the original basis and verifies the rank/invertibility conditions on T1 and T3. It does not exercise the identification step, so it does not address the similarity-transform ambiguity raised above; this should be stated explicitly or the example should be extended.","section":"VI, Example"},{"comment":"The definition of the non-uniform partition uses both t_l = T and then t_{l+1}, and the notation is not fully consistent with the subsequent use of t_i in Definition 7; the indexing should be cleaned up.","section":"III B, Definition 5"},{"comment":"There are several typographical errors, such as 'pulses of the from (76)' in Theorem 19 and 'the structure constants' in Remark 4; these do not affect the mathematics but should be corrected in a revision.","section":"Various"}],"recommendation":"major_revision","confidential_remarks":"The similarity-transform gap is localized and, in principle, fixable by adding a gauge-fixing lemma and adjusting the statements of Corollary 17 and Theorem 19. The T2/T3 inconsistency in Theorem 19 suggests that the symmetric-case statements need a careful pass. I do not see grounds for rejection on grounds of circularity or bad faith; the issue is a genuine missing proof step in the central claim."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The genuinely new piece here is the linear-algebra criterion in Theorems 13 and 15: given the system matrix A in the original coherence-vector basis, invertibility of M or T3 lets you recover Hamiltonian and Kossakowski parameters from A and beta. That is a real contribution, and the paper does a competent job of assembling the classical identification background, connecting it to GKSL dynamics, and proving a nice sparsity property of su(2^N) structure constants in Appendix D. The writing is clear and the survey portions are solid.\n\nThe soft spot is load-bearing. LDS and BDS identification determine the realization only up to similarity, as the paper itself acknowledges in Lemma 8, Proposition 9, and the proof of Corollary 17. But Theorems 13 and 15 invert the parameter map phi on matrices expressed in the original structure-constant basis. A generic similar realization A' = T^{-1}AT will not lie in the image of phi: its antisymmetric part may fall outside Im T1 and its symmetric part outside Im T3, so the reconstruction equations have no solution. The proof of Corollary 17 says the similarity transform \"does not affect the dynamics,\" which is true for predictions but does not imply that the identified representative lies in Im phi. For BDS the problem is worse, because the N_j matrices also transform to T^{-1}N_j T, and they are no longer known structure constants unless T lies in their joint centralizer. The paper offers no gauge-fixing procedure, so the advertised unique construction of the original GKSL equation is not established.\n\nThere is a separate issue in Corollary 17: Rank B = n gives controllability, but observability of the LDS requires a condition on C and A, not just that C is \"full rank by construction.\" The example also outsources the key T3 invertibility check to a GitHub tutorial instead of proving or demonstrating it. Minor typo: Theorem 19 says T2 where it should be T3 in the symmetric case.\n\nFor all that, the paper is worth taking seriously. The reconstruction criterion is plausibly correct once the gauge issue is addressed, and the synthesis of classical identification results for quantum systems is useful. I would send it to peer review, but the authors need to fix the similarity ambiguity—for instance, by using known C, b, and N_j to constrain T, or by showing a canonical realization that preserves the structure-constant basis—or they need to reframe the claim as recovery up to dynamical equivalence. As is, the central claim is unsupported.","headline":"A useful import of classical LDS/BDS identifiability into open quantum systems with a new reconstruction criterion, but the similarity-transform ambiguity leaves the advertised parameter recovery unproven.","tokens_in":44703,"tokens_out":4740,"would_cite":false,"duration_ms":49441,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81Q93","93B30","93B05","93B07","81S22"],"pacs":["03.65.Yz","03.67.-a"],"model":"deepseek-v4-flash","headline":"Given full-rank observability and controllability plus suitable sampling or pulses, the parameters of a GKSL master equation can be uniquely reconstructed from discrete measurement data.","keywords":["open quantum systems","master equation identification","GKSL equation","bilinear dynamical systems","system identifiability","quantum state tomography","coherence vector","Kossakowski matrix"],"falsifier":"Simulate a single-qubit GKSL system with known $\\theta$ and $\\gamma$, sample at two rates with irrational ratio, identify the continuous-time LDS matrix $A$ with a standard subspace method, apply a random invertible similarity transformation $T$, and test whether $T^{-1}AT$ still satisfies the image conditions (103) and yields the original parameters through Theorem 15; if a generic such $T$ fails the test, the claim that the master equation is uniquely reconstructible from any identified realization is refuted.","tokens_in":43695,"feed_emoji":"⚛️","tokens_out":12814,"duration_ms":97628,"temperature":0.7,"pith_summary":"This paper asks when the parameters of a Markovian open quantum system — the Hamiltonian coefficients and the decoherence (Kossakowski) rates in a time-independent GKSL master equation — can be recovered from discrete-time measurement data. The authors show that, in the coherence-vector picture, the measurement dynamics of an autonomous open system is a linear dynamical system (LDS) and that of a driven open system is a bilinear dynamical system (BDS). They then prove sufficient conditions for identifiability: full-rank observability and controllability matrices, together with irrational sample-rate ratios for autonomous systems and non-constant probing pulses for driven ones. If an additional matrix $M$ built from the Lie-algebra structure constants is invertible (or its symmetric reduction $T_3$ in the symmetric-Kossakowski case), the GKSL parameters can be uniquely reconstructed from the identified system matrix. The results give experimentalists a pre-experiment design test and a post-estimation validity check for quantum state tomography and noise characterization.","feed_headline":"Rank and pulse tests decide when open quantum systems are identifiable","feed_subtitle":"Autonomous systems need irrational sample rates; driven ones need non-constant pulses. Then parameters are recoverable.","key_machinery":"The carrying object is the coherence-vector representation of the density matrix, $\\rho = \\frac{1}{\\sqrt{N}}\\mathbb{1} + \\sum_j x_j F_j$, in an orthonormal basis $\\{F_j\\}$ of $su(N)$ with structure constants $[F_j,F_k] = \\mathrm{i}\\sum_l f_{jkl}F_l$ and $\\{F_j,F_k\\} = \\frac{2}{N}\\delta_{jk}\\mathbb{1} + \\sum_l g_{jkl}F_l$. Substituting this into the GKSL equation turns the measurement dynamics into $\\dot{x} = (A^{(l)}+A^{(d)})x + \\beta + \\sum_j N_j u_j(t)x$, with $A^{(l)}$, $A^{(d)}$, $\\beta$, and $N_j$ all expressed through the structure constants as in equations (90). The identification results then hinge on two pieces of classical machinery: the non-uniform sampling reconstruction of continuous LDS from discrete data (Lemmas 6 and 8, requiring irrational sample-rate ratios), and the input-class characterization for BDS identifiability (the pulse family (76)). Finally, the parameter reconstruction theorems are carried by the matrices $T_1$, $T_2$ (and its symmetric reduction $T_3$) and the block matrix $M = \\begin{pmatrix} T_1 & T_2 \\\\ 0 & \\frac{\\mathrm{i}}{N}T_1^\\top \\end{pmatrix}$ (Definition 12), whose invertibility decides whether $\\theta$ and $\\gamma$ are uniquely recoverable from $A$ and $\\beta$.","core_discovery":"The central claim is that the identifiability of an open quantum system governed by a time-independent GKSL equation reduces to two classical system-identification questions: whether the induced LDS/BDS is minimal, and whether the sampling or input design is persistently exciting. Concretely, Corollary 17 states that an autonomous system has a recoverable LDS form when the rank of $B$ is $n$ and the sampling intervals have pairwise irrational ratios; Theorem 19 states that a controlled system has a recoverable BDS form when the bilinear observability and controllability matrices $OM^{(bi)}$ and $CM^{(bi)}$ are full rank and the probes are pulses of the form (76). Given the identified matrices, reconstruction of the physical parameters is governed by Theorems 13 and 15: writing the system matrix as $A = A^{(l)}(\\theta) + A^{(d)}(\\gamma)$, the parameter-to-matrix map is invertible exactly when the structure-constant matrix $M$ (Definition 12) is invertible, and in the symmetric-Kossakowski case when $T_1$ and $T_3$ have full rank and the identified symmetric and antisymmetric parts lie in their images. Since the $N_j$ coupling matrices are fixed structure constants, only $A$ and $\\beta$ need identification, and all reconstructions obtained from equivalent realizations have identical measurement dynamics.","pith_inferences":["The similarity-gauge gap suggests a concrete repair: restrict the identification output to realizations whose $A$ lies in the image of the parameter map $\\phi$, or solve for the similarity transform together with the parameters; until then, \"unique construction of a GKSL equation\" should be read as uniqueness within a fixed gauge.","Because the $N_j$ matrices are pure structure constants, the whole identification problem reduces to finding $A$ and $\\beta$; this suggests the framework transfers to any simple compact Lie algebra with a basis where each commutator has at most one nonzero coefficient, a property the paper proves for $su(2^N)$ and notes fails for $su(3)$ in the Gell-Mann basis.","A testable extension is to check numerically whether $M$ is generically invertible for random Kossakowski matrices in $su(2^N)$; if invertibility is generic, the practical barrier to parameter recovery is the rank and image conditions rather than the invertibility of $M$."],"forward_implications":["Before running an experiment, a practitioner can compute the observability and controllability ranks and check the sampling or pulse conditions to decide whether the master equation parameters are recoverable at all.","When $M$ is invertible, the recovered GKSL equation has measurement dynamics identical to the original, so the identifiability test doubles as a certificate that a quantum process tomography estimate is the true evolution.","Closed quantum systems are covered as a degenerate limit: Corollaries 18 and 20 give Hamiltonian identification criteria under the same rank, sampling, and pulse conditions.","For systems with a symmetric Kossakowski matrix (including the two-qubit example in Section VI), the number of unknown decoherence parameters drops from $n^2$ to $n(n+1)/2$ and the reconstruction condition reduces to a single full-rank check on $T_3$."],"supporting_citations":[{"why":"Supplies the class of non-constant pulses $V_\\alpha$ and the minimality-based result that bilinear systems identifiable under those pulses are globally input-output identifiable, used in Lemma 10 and Proposition 9.","marker":"[118]"},{"why":"Provides the non-uniform sampling framework and the Lemmas 6 and 8 that guarantee retention of observability and controllability and reconstruction of the continuous-time LDS matrix $A$.","marker":"[124]"},{"why":"Supplies the method of reconstructing a continuous-time system from multiple single-rate discrete systems, used in the LDS reconstruction outline of Section III B.","marker":"[123]"},{"why":"The similarity-transformation approach to quantum Hamiltonian identifiability that the paper extends to open systems with discrete sampling and to parameter reconstruction.","marker":"[23]"}],"fun_headline_variants":["Rank and pulse tests unlock quantum state recovery","Irrational sampling and nonconstant pulses: key to quantum ID","Quantum identifiability: minimality and excitation conditions","Full master equation recovery via rank and probe design","Autonomous vs controlled: rank and pulse conditions for quantum ID"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the identified LDS/BDS system matrix $A$ comes in the original coherence-vector basis where the bilinear coupling matrices $N_j$ are the known structure constants; classical identification only delivers $A$ up to an arbitrary similarity transformation, and the paper does not prove that a similarity-transformed realization still lies in the image of the parameter map $\\phi$.","fun_headline_variants_meta":{"raw":{"variants":["Rank and pulse tests unlock quantum state recovery","Irrational sampling and nonconstant pulses: key to quantum ID","Quantum identifiability: minimality and excitation conditions","Full master equation recovery via rank and probe design","Autonomous vs controlled: rank and pulse conditions for quantum ID"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001557,"raw_usage":{"total_tokens":6235,"prompt_tokens":973,"completion_tokens":5262,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":589,"completion_tokens_details":{"reasoning_tokens":5184}},"tokens_in":589,"tokens_out":5262,"duration_ms":39149,"temperature":1.0,"reasoning_tokens":5184,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T21:13:38.544523+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate a single-qubit GKSL system with known $\\theta$ and $\\gamma$, sample at two rates with irrational ratio, identify the continuous-time LDS matrix $A$ with a standard subspace method, apply a random invertible similarity transformation $T$, and test whether $T^{-1}AT$ still satisfies the image conditions (103) and yields the original parameters through Theorem 15; if a generic such $T$ fails the test, the claim that the master equation is uniquely reconstructible from any identified realization is refuted.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the non-uniform sampling framework and the Lemmas 6 and 8 that guarantee retention of observability and controllability and reconstruction of the continuous-time LDS matrix $A$."},{"cited_title":"Moiseyev,Non-Hermitian Quantum Mechanics (Cambridge University Press, 2011)","cited_arxiv_id":null,"evidence_quote":"Supplies the method of reconstructing a continuous-time system from multiple single-rate discrete systems, used in the LDS reconstruction outline of Section III B."}],"review_version":1}