{"id":"ebc9cf3d-6354-4d3e-a522-d3c6b36243cb","arxiv_id":"1908.03404","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Quantum dynamics of finite-dimensional systems is represented via pseudostochastic matrices acting on SIC-POVM probability vectors, with new measures of nonclassicality and non-Markovianity.","lead":"This paper rewrites quantum dynamics in a probability language based on symmetric informationally complete measurements (SIC-POVMs), where quantum states become ordinary probability vectors and quantum processes become matrices that can have negative entries. The authors derive dynamical equations in this picture, propose new measures of nonclassicality and non-Markovianity, and demonstrate them on an IBM quantum processor.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1 is sound; the load-bearing gap is the unproven SIC-POVM existence for general d, making the universal claim conditional.","rationale":"The reader's weakest assumption is exactly the SIC existence condition, and I agree that it is the load-bearing external constraint. I checked Theorem 1's proof: the forward direction uses a product limit of nonnegative stochastic matrices with column sums preserved, and the converse follows from the first-order term in exp(Lt); both are sound. The reconstruction formulas, H properties, and L derivation are consistent with the stated SIC axioms. The concern does not undermine the conditional mathematical results or the d = 2 experimental demonstration; it narrows the scope of the abstract's universal claim. The computational projectors Punit, PMark, and PCPTP in Secs. III-IV are introduced without full proof of well-definedness or uniqueness, but they are not needed for Theorem 1 and are secondary to the main claim. Verdict remains CONDITIONAL, with the SIC-existence caveat made explicit.","tokens_in":21863,"tokens_out":36979,"duration_ms":395250,"concrete_test":"Verify the framework for a non-qubit dimension with a known SIC, e.g., the d = 3 Hesse SIC: construct K via Eq. (7), compute L for a random GKSL generator via Eq. (44), and directly test Theorem 1's iff condition by comparing the off-diagonal signs of L with negativity of exp(Lt) for small t. If all identities hold, the central theorem is confirmed for every dimension that admits an explicit SIC, isolating the open existence conjecture as the only remaining limitation of the universal claim.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central construction is conditional on an unproven conjecture. Section II states 'we assume that it is possible to find a set of d^2 normalized states ...' satisfying Eq. (1), and the introduction concedes that analytic SIC existence proofs exist only in some cases. Every subsequent object -- the reconstruction matrix K (Eq. 7), the unitary generator H (Eqs. 24-26), the GKSL generator L (Eq. 44), and Theorem 1 -- is defined through a SIC-POVM. For a dimension not in the cited list (e.g., d = 152), no SIC is currently known, so the representation is undefined and the abstract's 'any finite-dimensional quantum dynamics' is not established. This is a genuine scope limitation rather than an internal inconsistency: the derivations are algebraically correct given Eq. (1), and the experimental section uses d = 2 with an explicit SIC (Eq. 62), so the IBM QX4 conclusions do not depend on the conjecture. The paper should state the conditional scope explicitly in the abstract and conclusion.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a probability representation of finite-dimensional quantum dynamics in which states are represented by SIC-POVM probability vectors and quantum channels act by pseudostochastic matrices. It derives explicit generator equations for unitary von Neumann dynamics (Eqs. (24)-(26)) and for GKSL dissipative dynamics (Eqs. (43)-(45)), and it proves Theorem 1: for a time-independent Markovian generator L, the map e^{Lt} is stochastic for every t>0 if and only if all off-diagonal elements of L are nonnegative. On this basis the paper defines measures of nonclassicality and non-Markovianity and applies them to a single-qubit S gate and a SIC-POVM measurement carried out on the IBM QX4 processor. The presentation is careful, and the algebraic core is self-contained, but the universal scope of the results is conditional on SIC-POVM existence and on unproved properties of the numerical projectors used in the measures.","tokens_in":22106,"tokens_out":9591,"duration_ms":112596,"significance":"The main theoretical contribution is a clean and largely self-contained translation of standard quantum dynamical equations into a pseudostochastic language, together with a simple and apparently correct necessary-and-sufficient criterion for stochastic (classically simulable) Markovian evolution in a fixed SIC-POVM representation. Theorem 1 gives a concrete, falsifiable condition, and the proposed nonclassicality and non-Markovianity measures have clear practical appeal. The experimental demonstration on IBM QX4 is a useful stress test and shows that the framework can be applied to real tomographic data. The derivation does not appear to fit free parameters to force the main theorem. However, the construction is conditional on the SIC-POVM existence conjecture for general dimension, and the two 'projectors' PMark and PCPTP are defined through nonconvex optimizations whose global optimality and idempotence are not established; these points must be addressed before the central claims can be accepted in full generality.","major_comments":[{"comment":"The entire construction is conditional on the existence of a SIC-POVM for the Hilbert space dimension d, which remains an open conjecture for general d. The manuscript states in Sec. II that 'we assume that it is possible to find out a set of d^2 normalized states' satisfying Eq. (1), and it lists only specific dimensions for which SIC-POVMs are known, yet the abstract and conclusion present the representation for 'finite-dimensional quantum systems' without this qualification. Since the matrices K, H, L, and Theorem 1 are all defined through Eq. (1), the scope of every result should be restricted to dimensions admitting a SIC-POVM, and the conditional status should be stated explicitly in the abstract and conclusion. This is a scope limitation rather than an internal inconsistency; the d=2 experimental part is unaffected because a concrete SIC is given in Eq. (62).","section":"Sec. II, Eq. (1); Abstract"},{"comment":"The maps PMark and PCPTP are introduced as 'projectors' but are defined by argmin over the matrix V of an objective that is nonconvex in V; no proof is given that the global minimum is attained, that the argmin is unique, or that the resulting map is idempotent. Without these properties the non-Markovianity measure in Eq. (60) and the physical projection used in the experimental reconstruction are not well-defined functions, and the equality in Eq. (53) is not a verified necessary-and-sufficient condition. Since D(V) and S(V) are linear in X = VV†, the optimization can be reformulated as a convex problem over the positive semidefinite cone; the authors should provide this reformulation or otherwise certify global optimality and idempotence, at least for the reported experimental instances.","section":"Sec. III.C, Eq. (52); Sec. V, Eq. (68)"},{"comment":"The definition L := log S presupposes a logarithm of a real pseudostochastic matrix, but no domain or branch specification is given. A pseudostochastic matrix, for example a unitary evolution operator in this representation, can have negative eigenvalues, in which case a real matrix logarithm does not exist and the expression exp[Punit(L)+PMark(L-Punit(L))] is either undefined or dependent on a choice of complex branch. The non-Markovianity measure therefore is not defined for all channels to which the paper applies it. Please specify the class of S for which the logarithm is taken and how non-unique branches are handled.","section":"Sec. IV.B, Eqs. (59)-(60)"}],"minor_comments":[{"comment":"The sentence describing δnMark as an 'average difference between elements' is imprecise: the quantity is a normalized Frobenius norm of the difference matrix, not an elementwise average. Please reword.","section":"Sec. IV.B, Eq. (60)"},{"comment":"The error estimate δ in Appendix D is an upper bound for the per-element standard deviation of the reconstructed pseudostochastic matrix, but the separation of Sdec and SU uses S_dec^{-1}, which can amplify errors; the text states that the error 'remained the same' without accounting for this additional propagation. Please add a comment or a more detailed estimate.","section":"Appendix D and Sec. V"},{"comment":"The reconstructed matrices are reported to four decimal places without confidence intervals. Given the estimated per-element error of about 0.031, some indication of the uncertainty in the reported nonclassicality and non-Markovianity values would improve the experimental discussion.","section":"Sec. V, Table II"},{"comment":"There is a typo in the sentence before Eq. (56): 'sone' should be 'some'. Please also check the notation in Eq. (A6), where the last displayed matrix element appears to contain an unbalanced parenthesis.","section":"Sec. IV.A, Eq. (56)"}],"recommendation":"major_revision","confidential_remarks":"The SIC-POVM conjecture is widely recognized, and the authors do state the assumption explicitly, so the conditional scope can likely be fixed with clear language in the abstract and conclusion. The more substantive issue is the definition of PMark and PCPTP as projectors without global-optimality or idempotence guarantees; since these operators enter both the theoretical measures and the experimental pipeline, the revision should either prove the needed properties or explicitly reframe the optimization as a heuristic and validate it numerically. I believe the paper is salvageable and recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper derives the von Neumann and GKSL equations in the SIC-POVM probability representation, which I believe is new, and it packages the result as a practical tool for nonclassicality and non-Markovianity witnesses. Theorem 1 is correct: for a time-independent Markovian generator, e^{Lt} is stochastic for all t>0 iff every off-diagonal element of L is nonnegative. The proof is clean, though it is basically the classical Kolmogorov criterion for stochastic semigroups translated into this representation. The real value is the explicit generator formulas and the experimental demonstration on IBM QX4, where they reconstruct the S gate and show its nonclassicality is robust against basis changes while the measurement noise is classically simulable.\n\nThe derivations in Sections II and III are straightforward linear algebra and check out. The representation of PTP and CPTP maps as pseudostochastic matrices is clearly written, and the experimental protocol is a sensible way to extract the pseudostochastic matrix. The error analysis is crude — a simple upper bound — but it supports the qualitative conclusions: the non-Markovianity measures are below the error, so the observed processes are well described as Markovian.\n\nThe soft spots are two. First, the whole framework assumes a SIC-POVM exists in the relevant dimension. That is conjectural in general; analytic existence proofs cover only a finite list of dimensions. The abstract's 'any finite-dimensional quantum dynamics' is too strong. The authors do state the assumption in Section II, but they should make the conditional scope explicit in the abstract and conclusion. For the d=2 experiment this is irrelevant because the SIC is written out, but readers should not take the universal claim at face value. Second, the 'projectors' PMark and PCPTP are defined as argmins over nonconvex optimization problems. The paper does not prove they are globally optimal, idempotent, or even true projections onto the intended sets. The non-Markovianity measure depends on SMark being the best Markovian approximation; without a proof, that interpretation is shaky. This is not a fatal flaw, but it is a real gap that should be addressed in revision.\n\nThe citation pattern is reasonable, with appropriate credit to the SIC and qplex literature. Self-citation is not an issue here. Who should read this? People working on probability representations of quantum mechanics, quantum tomography, and measures of nonclassicality or non-Markovianity. It deserves a serious referee and likely publication after revision, provided the projectors are either justified or softened, and the scope is stated honestly.","headline":"Conditional on SIC existence the derivations are sound and the experiment is a nice proof of concept, but the abstract overreaches and the optimization-based 'projectors' are not fully justified.","tokens_in":22600,"tokens_out":2291,"would_cite":true,"duration_ms":26372,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that finite-dimensional quantum dynamics can be written as linear equations on real SIC-POVM probability vectors, with pseudostochastic generators, and that a time-independent Markovian generator yields a classically…","keywords":["probability representation","SIC-POVM","pseudostochastic maps","quantum dynamics","Markovian master equation","nonclassicality","non-Markovianity","quantum process tomography"],"falsifier":"Search for a time-independent Markovian generator L in the SIC-POVM representation with at least one off-diagonal entry L_{ij} < 0 whose matrix exponential $e^{{Lt}}$ has all entries nonnegative for every t > 0; a single such example would disprove the only-if direction of Theorem 1.","tokens_in":21741,"feed_emoji":"⚛️","tokens_out":6339,"duration_ms":68231,"temperature":0.7,"pith_summary":"The paper develops a representation in which quantum states are described by true probability vectors obtained from SIC-POVM measurements, and quantum dynamics becomes a linear differential equation for those vectors. In this representation the generators of evolution are pseudostochastic matrices: their columns sum to one like conditional probabilities, but some entries are negative. The central result is a theorem: for a time-independent Markovian generator, the evolved map is stochastic for every positive time if and only if all off-diagonal elements of the generator are nonnegative. A direct corollary is that any nonzero unitary evolution necessarily produces negative entries at some times, so purely unitary quantum dynamics cannot be simulated by classical random processes of this kind. The paper then turns this condition into quantitative measures of nonclassicality and non-Markovianity and tests them on experiments with a superconducting quantum processor.","feed_headline":"Quantum evolution is classical only when noise kills all negative rates","feed_subtitle":"A theorem pins when quantum dynamics can be simulated by random processes, with measures tested on IBM's quantum processor.","key_machinery":"The central object is the SIC-POVM probability vector p, whose components are p_i = Tr(ρΠ_i)/d for d² symmetric projectors Π_i, together with the linear reconstruction map ρ = Kp where K_i = (d+1)Π_i − 1. This K bridges the density-matrix picture and the probability picture, and it is used to derive the generators H = −$iK^{{-1}}$(H ⊗ 1 − 1 ⊗ H^*)K and L = $K^{{-1}}$ΛK from the von Neumann and GKSL equations. Theorem 1 is carried by the semigroup representation $e^{{Lt}}$ = lim_{n→∞}(I + Lt/n)^n, which shows that nonnegative off-diagonal entries preserve stochasticity, and by a small-t expansion that exposes any negative off-diagonal entry as a negative matrix element of $e^{{Lt}}$. The additional machinery consists of projectors Punit, PMark, and PCPTP, which map arbitrary matrices onto the nearest physical Hamiltonian, Markovian, or completely-positive generator, and these projectors make the nonclassicality and non-Markovianity measures computable from tomographic data.","core_discovery":"For any Hilbert space dimension d in which a SIC-POVM exists, the von Neumann equation and every Markovian GKSL master equation can be recast as linear equations for a d²-component SIC-POVM probability vector p: unitary evolution becomes p-dot = Hp with H a real antisymmetric matrix, and dissipative evolution becomes p-dot = Lp with L a pseudo-Kolmogorov generator. Theorem 1 states that for a time-independent Markovian generator L, the evolution map S(t) = $e^{{Lt}}$ is stochastic for all t > 0 if and only if every off-diagonal element of L is nonnegative. Since any nonzero antisymmetric H has negative off-diagonal entries, Corollary 1 concludes that no nontrivial unitary evolution is classically simulable in this sense; decoherence is required to make the pseudostochastic map classical. The paper also constructs projectors onto the spaces of physical Hamiltonian generators, Markovian generators, and completely positive maps, defines a nonclassicality measure δquant and a non-Markovianity measure δnMark, and applies them to experimental reconstruction of a single-qubit gate and an imperfect SIC-POVM measurement, obtaining δquant = 0.781 for the gate and δquant = 0 for the measurement.","pith_inferences":["If SIC-POVMs continue to be found in higher dimensions, the same construction yields canonical pseudostochastic generators for those dimensions; if the SIC conjecture fails in some dimension, the explicit formulas for H and L would need to be replaced by those of a non-symmetric informationally complete measurement.","The theorem's if-and-only-if structure suggests a direct experimental test: engineer Markovian generators with small negative off-diagonal entries and look for the first time at which negative entries appear in the reconstructed evolution matrix.","Because Theorem 1 is stated for time-independent generators, an analogous criterion for time-ordered exponentials of time-dependent generators is a natural extension, and it can be checked numerically before attempting any experimental demonstration.","The non-Markovianity measure here is tied to the specific Markovian projection; comparing it with established trace-distance or divisibility-based measures on the same experimental data would clarify which aspects of non-Markovian behavior it actually captures."],"forward_implications":["Any nontrivial unitary evolution of a d-level system must show negative pseudostochastic entries at some times, so it cannot be reproduced by sampling ordinary classical conditional probabilities.","A quantum gate with measurable nonclassicality has δquant > 0, while a process that can be made completely classical by a change of basis has δquant = 0.","A process can be certified as Markovian, within the tomographic precision, when its non-Markovianity measure δnMark falls below the statistical error of the reconstructed pseudostochastic matrix.","The explicit pseudostochastic forms derived for transposition, the reduction map, and Rabi oscillations provide ready benchmarks for comparing classical and nonclassical behavior in qubit systems."],"supporting_citations":[{"why":"Introduces informationally complete and symmetric informationally complete POVMs, the measurement framework on which the probability vector is built.","marker":"[26–28]"},{"why":"Reviews the SIC existence problem and lists dimensions where SIC-POVMs are known, supporting the framework's main premise.","marker":"[29]"},{"why":"Defines the qplex and shows SIC-POVM probability vectors form a proper subset of classical probability vectors, which the paper uses to distinguish classical from quantum maps.","marker":"[38]"},{"why":"Provides the functorial embedding of quantum channels into quasi-stochastic matrices that motivates the pseudostochastic representation.","marker":"[40]"},{"why":"Introduces pseudo-stochastic matrices and pseudo-positive maps, supplying the terminology and prior basis for the maps used here.","marker":"[41, 42]"},{"why":"Supplies the non-Markovianity-measure approach that the paper's projection-based measure generalizes.","marker":"[43]"},{"why":"Gives the GKSL form of Markovian master equations whose generators are recast as pseudo-Kolmogorov matrices.","marker":"[56, 57]"},{"why":"Identifies the IBM QX4 cloud quantum processor used for the experimental gate and measurement reconstruction.","marker":"[55]"}],"fun_headline_variants":["Unitaries never classical: SIC-POVM dynamics proof","Noise makes quantum dynamics classically simulable","Negative rates block classical simulation in SIC-POVM","Pseudostochastic maps show unitaries stay quantum","Classical simulation of quantum needs nonnegative rates"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire representation assumes that a SIC-POVM exists in the Hilbert space dimension d under study, which is known for many small dimensions but remains an open mathematical conjecture in general.","fun_headline_variants_meta":{"raw":{"variants":["Unitaries never classical: SIC-POVM dynamics proof","Noise makes quantum dynamics classically simulable","Negative rates block classical simulation in SIC-POVM","Pseudostochastic maps show unitaries stay quantum","Classical simulation of quantum needs nonnegative rates"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000233,"raw_usage":{"total_tokens":1492,"prompt_tokens":946,"completion_tokens":546,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":562,"completion_tokens_details":{"reasoning_tokens":470}},"tokens_in":562,"tokens_out":546,"duration_ms":6312,"temperature":1.0,"reasoning_tokens":470,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:18:12.691167+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Search for a time-independent Markovian generator L in the SIC-POVM representation with at least one off-diagonal entry L_{ij} < 0 whose matrix exponential $e^{{Lt}}$ has all entries nonnegative for every t > 0; a single such example would disprove the only-if direction of Theorem 1.","supporting_citations":[{"cited_title":"Avanesov and Man’ko, Statistical properties of qutrit in probability representation of quantum mechanics, Phys","cited_arxiv_id":null,"evidence_quote":"Reviews the SIC existence problem and lists dimensions where SIC-POVMs are known, supporting the framework's main premise."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the qplex and shows SIC-POVM probability vectors form a proper subset of classical probability vectors, which the paper uses to distinguish classical from quantum maps."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the functorial embedding of quantum channels into quasi-stochastic matrices that motivates the pseudostochastic representation."},{"cited_title":"Appleby, C.A","cited_arxiv_id":null,"evidence_quote":"Supplies the non-Markovianity-measure approach that the paper's projection-based measure generalizes."},{"cited_title":"Bylicka, D","cited_arxiv_id":null,"evidence_quote":"Identifies the IBM QX4 cloud quantum processor used for the experimental gate and measurement reconstruction."}],"review_version":1}