{"id":"3d4adc80-c6b5-462d-b120-1361659a329d","arxiv_id":"2411.09088","paper_version":3,"verdict":"REJECT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"The paper claims multidimensional KUR and TUR bounds for open quantum systems, with a Fisher-information off-diagonal term that vanishes for classical dynamics.","lead":"This paper claims new precision bounds for multiple currents in open quantum systems, derived from multiparameter estimation theory. If correct, the bounds would show that current fluctuations in a quantum machine can be constrained more tightly than single-current bounds suggest.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The D-optimality step in Eq. (11) replaces the full Jacobian determinant by the diagonal product (1+φ1)^2(1+φ2)^2; in the classical two-state process the Jacobian is singular, so Eq. (23) does not follow from the multiparameter CRB and is violated.","rationale":"The reader's weakest assumption is exactly the load-bearing flaw identified here. The derivation of Eq. (23) from Eq. (11) hinges on treating the Jacobian as diagonal, but the cross derivatives ∂_{φ_1}⟨Φ_2⟩ and ∂_{φ_2}⟨Φ_1⟩ are generically nonzero because perturbing one set of jump operators changes the state and therefore all currents. In the classical two-state limit the Jacobian is rank-one, so the correct D-optimality bound gives no lower bound on det(Ξ), and an explicit counterexample shows the claimed inequality is violated. Because both Eq. (23) and Eq. (29) are derived by the same step, the central claims of the paper are unsupported. The numerical examples do not resolve this: they are finite-time quantum simulations whose Jacobians are not singular, so they cannot detect the failure in the singular classical limit. The paper's idea remains interesting, and a corrected multidimensional bound with the full Jacobian might be possible, but the current derivation is not valid. No change to the reader's REJECT verdict is needed.","tokens_in":23034,"tokens_out":8467,"duration_ms":73782,"concrete_test":"Take the classical two-state chain with rates a (S_1: 0→1) and b (S_2: 1→0), steady state, and long time τ. Compute the full 2×2 Jacobian J and verify det(J)=0, while the product (1+φ_1)^2(1+φ_2)^2 = a^2b^2/(a+b)^4. Then evaluate the exact covariance determinant from the alternating renewal statistics: since N_1 and N_2 differ by at most 1, Var(N_1)Var(N_2)−Cov(N_1,N_2)^2 = O(τ), so the left side of Eq. (23) is O(τ^{−3}) whereas the paper's RHS is 1/[τ^2(a+b)^2]; this shows Eq. (23) is violated for large τ and that the diagonal-Jacobian replacement, not numerical error, is responsible. Optionally recompute with the full det(J)^2/det(F) bound to confirm it reduces to det(Ξ) ≥ 0.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing step is the passage from the matrix Cramér-Rao bound (9) to the scalar bound (11). The correct determinant consequence is det(Ξ) ≥ det(J)^2/det(F) with [J]_ij = ∂_{φ_j}⟨Φ_i⟩. Equation (11) instead uses the product of diagonal Jacobian elements, (∂_{φ_1}⟨Φ_1⟩)^2(∂_{φ_2}⟨Φ_2⟩)^2, which is only valid if J is diagonal. Appendix A computes only ∂_{φ_1}⟨Φ_1⟩ = (1+φ_1)⟨Φ_1⟩ and ∂_{φ_2}⟨Φ_2⟩ = (1+φ_2)⟨Φ_2⟩; it never computes the cross derivatives ∂_{φ_1}⟨Φ_2⟩ and ∂_{φ_2}⟨Φ_1⟩, which are generically nonzero because changing one set of jump operators alters the stationary state and hence both currents. For the classical two-state process with S_1 the 0→1 channel (rate a) and S_2 the 1→0 channel (rate b), both mean currents equal τab/(a+b), and both derivatives with respect to φ_1 equal τab^2/(a+b)^2 while both derivatives with respect to φ_2 equal τa^2b/(a+b)^2. Hence J has two identical rows and det(J)=0, so the correct CRB consequence is only det(Ξ) ≥ 0. The paper's formula gives RHS = (1+φ_1)^2(1+φ_2)^2/(A_1A_2) = 1/[τ^2(a+b)^2], since F_12=0 and Q_α=0 classically. In this process N_1 and N_2 differ by at most 1, so Var(N_1)Var(N_2)−Cov(N_1,N_2)^2 = O(τ), making the left side of Eq. (23) O(τ^{−3}), which is smaller than the O(τ^{−2}) RHS for large τ. Thus Eq. (23) is violated. A separate issue is that Eq. (A6) omits the cross terms ⟨N_1 ∂_{φ_2} ln q⟩ + ⟨N_2 ∂_{φ_1} ln q⟩, but the Jacobian replacement is already fatal. The same Jacobian issue affects the TUR (29) via Eq. (B25).","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper derives multidimensional kinetic and thermodynamic uncertainty relations for pairs of counting observables in Markovian open quantum systems. The authors imprint two perturbation parameters on the Hamiltonian and on the jump operators, apply a scalar D-optimal version of the multiparameter Cramér-Rao bound, and express the resulting bounds in terms of dynamical activities, entropy production, and the off-diagonal Fisher information. They illustrate the claimed KUR on a coherently driven qubit and a three-level maser by numerical trajectory simulation, arguing that the new bounds are tighter than products of single-observable bounds because they incorporate correlations, and that the off-diagonal Fisher information is a genuinely quantum signature that vanishes for classical rate equations.","tokens_in":23465,"tokens_out":12863,"duration_ms":121575,"significance":"The intended contribution is valuable if correct: a multiparameter KUR/TUR for open quantum systems that is tighter than products of single-observable bounds, with a quantum marker encoded in the off-diagonal Fisher information, would extend a substantial literature and give a practical tool for precision thermometry and current estimation. The paper also usefully connects the monitoring-operator formalism to precision bounds and gives a clean classical Fisher-matrix analysis in Appendix C. However, the central inequality is not a valid consequence of the multiparameter Cramér-Rao bound as stated, and a simple classical two-state process violates the claimed bound. The examples and numerical simulations cannot compensate for this, because the claimed universal bound is false within the paper's own stated domain.","major_comments":[{"comment":"The scalar D-optimal bound is stated incorrectly. Taking the determinant of the matrix inequality (9) gives det(J)^2/det(Ξ) ≤ det(F), i.e. det(Ξ) ≥ det(J)^2/det(F), where J_ij = ∂_{φ_j}⟨Φ_i⟩. Equation (11) instead uses [∂_{φ1}⟨Φ1⟩]^2 [∂_{φ2}⟨Φ2⟩]^2 in place of det(J)^2. These two quantities agree only when J is diagonal. For the parameter imprinting of Eqs. (21)-(22), each parameter changes the Liouvillian and therefore the stationary state and both mean currents, so the cross derivatives ∂_{φ2}⟨Φ1⟩ and ∂_{φ1}⟨Φ2⟩ are generically nonzero. The manuscript gives no argument that J is diagonal. Consequently, Eqs. (23) and (29) do not follow from Eq. (9).","section":"Section III, Eq. (11)"},{"comment":"The derivation computes only the diagonal derivatives ∂_{φ1}⟨Φ1⟩ and ∂_{φ2}⟨Φ2⟩ and then applies Eq. (11) as if J were diagonal. The cross derivatives are never computed. This is not a minor omission: in the classical two-state process with S1 = {0→1} (rate a) and S2 = {1→0} (rate b), one finds ⟨Φ1⟩ = ⟨Φ2⟩ = τab/(a+b) and J has two identical rows, so det(J) = 0. The correct Cramér-Rao consequence is only det(Ξ) ≥ 0. The paper's Eq. (23), with F12 = 0, Qα = 0, and φ1 = φ2 = 0, gives the positive right-hand side (a+b)^2/(a^2 b^2 τ^2). But since N1 and N2 differ by at most 1 in this process, Var(N1)Var(N2) − Cov(N1,N2)^2 = O(τ), so the left side of Eq. (23) is O(τ^{−3}) and is smaller than the O(τ^{−2}) right-hand side for sufficiently large τ. Thus Eq. (23) is violated in exactly the classically Markovian setting that the paper claims to cover.","section":"Appendix A, Eqs. (A12)-(A25)"},{"comment":"The expression for the off-diagonal Fisher information omits two terms. Since ln p = N1 ln(1+φ1) + N2 ln(1+φ2) + ln q, the product ∂_{φ1} ln p ∂_{φ2} ln p at φ = 0 contains N1 ∂_{φ2} ln q + N2 ∂_{φ1} ln q in addition to N1N2 and ∂_{φ1} ln q ∂_{φ2} ln q. Equation (A6) drops these two cross terms. They are not generally zero, and in the classical two-state example the full F12 must vanish (as Appendix C states), which requires the omitted terms to cancel the remaining contributions. The truncated expression in Eq. (A6) therefore does not give the correct F12, and any bound whose denominator uses this F12 is not justified.","section":"Appendix A, Eq. (A6)"},{"comment":"The numerical factors in the multidimensional TUR do not follow from the preceding equations. With Fαα = (1/2)Σα + Q′α, the determinant relation gives det(Ξ)/(⟨Θ1⟩^2⟨Θ2⟩^2) ≥ 4(1+ϑ1)^2(1+ϑ2)^2 / [(Σ1+2Q′1)(Σ2+2Q′2) − 4F12^2], not the factor 2 and 2F12^2 appearing in Eq. (29). Since the derivation uses an inequality in Eq. (B9) in the direction Fαα ≤ Σα/2 + Q′α, the bound would be loosened, not tightened, relative to this expression. Unless an additional inequality is invoked that is not stated, Eq. (29) is not a consequence of Eq. (11) and Eq. (B7). This is independent of the Jacobian issue raised above.","section":"Appendix B, Eqs. (B7) and (29)"}],"minor_comments":[{"comment":"The notation is inconsistent: Eq. (12) uses weights w_k, while Eq. (20) introduces ω_k and writes 'ωkj = ωk'; these should use a single symbol throughout.","section":"Section V, after Eq. (20)"},{"comment":"The sentence 'the diagonal elements of the Fisher information matrix are now lower bounded by the components of the entropy production, Fαα ≤ Σα + Q′α' contains two typographical errors: the inequality direction should be 'upper bounded', and the correct bound from Appendix B is Fαα ≤ Σα/2 + Q′α, not Σα + Q′α.","section":"Section V.B, text after Eq. (30)"},{"comment":"The definition 'ϑ1 ≡ ⟨Θ1⟩/⟨Θ⋆1⟩' is inverted relative to the main-text definition ϑ1 = ⟨Θ⋆1⟩/⟨Θ1⟩; as written it would give ∂θ1⟨Θ1⟩ = (1 + 1/ϑ1)⟨Θ1⟩, contradicting Eq. (B17).","section":"Appendix B, after Eq. (B17)"},{"comment":"The inequality in Eq. (A25) moves the covariance term to the right-hand side; this is equivalent to Eq. (23), but the sign of the covariance term changes between the two displays, which may confuse readers who do not track the rearrangement carefully.","section":"Appendix A, Eq. (A25)"}],"recommendation":"reject","confidential_remarks":"I recommend rejection. The central bound is not a consequence of the multiparameter Cramér-Rao bound, and a simple classical counterexample within the paper's stated scope violates Eq. (23). This is a load-bearing mathematical error that cannot be fixed by a local correction: the derivation would need to include the full Jacobian determinant and the omitted Fisher-information cross terms, and the resulting bound would generally be weaker or trivial. The examples and numerical plots therefore do not establish the claimed result. I saw no indication of any issue beyond the technical mathematics of the derivative and determinant manipulations."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this paper is not ready. The core inequalities (23) and (29) do not follow from the multiparameter Cramér-Rao bound as written. The problem is the step from (9) to (11): the correct determinant bound is det(Ξ) ≥ det(J)^2/det(F), but the paper replaces det(J) by the product of diagonal elements (∂φ1⟨Φ1⟩)^2(∂φ2⟨Φ2⟩)^2. That is only valid if the response matrix is diagonal. In the classical two-state process with S1={0→1} and S2={1→0}, both mean currents are equal and the Jacobian has two identical rows, so det(J)=0. The Cramér-Rao bound then gives only det(Ξ)≥0, while their formula gives a positive RHS of order τ^{-2}, and the actual LHS is O(τ^{-3}), so Eq. (23) is violated for large τ. This is a load-bearing error, not a gap. The same issue enters the TUR via Eq. (B25).\n\nNow the good parts. The basic idea is a natural extension of the single-observable quantum KUR/TUR: use multiparameter estimation with two perturbations and let the off-diagonal Fisher information appear as a correlation term. The observation that F12 vanishes for classical rate equations (Appendix C) is clean and potentially useful as a quantum signature. The numerical examples are honest, and the maser saturation is a nice illustration.\n\nTwo secondary problems: Eq. (A6) for F12 is incomplete—it drops the cross terms ⟨N1∂φ2 ln q⟩ + ⟨N2∂φ1 ln q⟩—and the numerical simulations (τ=10 for the qubit and maser) never approach the singular classical limit where the error shows. So the numerics do not test the claim.\n\nWho is this for? People working on quantum TUR/KUR will care about the question, and the F12 idea might survive a fix. But the submitted result as stated is false, and a corrected multidimensional bound needs a proper treatment of the full Jacobian, not just diagonal responses.\n\nRecommendation: I would not send this to a serious referee as is. It should go back to the authors with the Jacobian problem identified. If they fix it and the bound survives in a restricted form, it would be worth a round of review.","headline":"The multidimensional KUR/TUR idea is natural and the F12 quantum signature is interesting, but the central D-optimality step misuses the Cramér-Rao bound and the claimed bound fails in a simple classical two-state process.","tokens_in":24103,"tokens_out":4813,"would_cite":false,"duration_ms":48215,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":false},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Using two-parameter Fisher information, this paper derives multidimensional quantum KUR and TUR for pairs of currents that are tighter than single-observable bounds and can saturate.","keywords":["open quantum systems","kinetic uncertainty relation","thermodynamic uncertainty relation","quantum trajectories","Fisher information","parameter estimation","current fluctuations","Markovian dynamics"],"falsifier":"Evaluate both sides of Eq. (23) for a minimal two-state model in which the two jump channels are simultaneously perturbed and the state is driven by both parameters; if the left-hand side falls below the right-hand side while the covariance matrix remains positive semidefinite, the diagonal-response assumption behind the bound is violated.","tokens_in":1699,"feed_emoji":"⚛","tokens_out":3253,"duration_ms":96230,"temperature":0.7,"pith_summary":"The paper sets out to extend the quantum kinetic and thermodynamic uncertainty relations from one counting observable to two observables measured simultaneously along the trajectory of a Markovian open quantum system. The authors show that the product of the two relative fluctuations, minus the squared correlation between the currents, is bounded from below by a quantity built from each channel's dynamical activity, a quantum correction, and the off-diagonal element of the Fisher information matrix. Because that off-diagonal Fisher information vanishes for classical rate equations, the tightened bound is a genuinely quantum effect and is tighter than multiplying two single-observable bounds. The same construction yields a multidimensional TUR for heat currents under local detailed balance. If correct, these bounds state a fundamental trade-off: squeezing the fluctuations of several currents at once costs activity or entropy production, and the cost is exactly modified by how correlated the currents are.","feed_headline":"Correlated currents tighten quantum precision bounds","feed_subtitle":"Two-observable KUR and TUR beat products of single-observable bounds via the off-diagonal Fisher information.","key_machinery":"The load-bearing machinery is the multiparameter Cramér-Rao bound with its D-optimality determinant scalarisation, applied to the measurement record of a continuously monitored quantum system. Two control parameters $\\phi_1$ and $\\phi_2$ are imprinted on the dynamics by scaling the Hamiltonian with $1+\\phi_1+\\phi_2$ and each jump operator in channel subset $S_\\alpha$ by $\\sqrt{1+\\phi_\\alpha}$; the resulting Fisher information matrix has diagonal elements $A_\\alpha+Q_\\alpha$ and off-diagonal element $F_{12}$, which measures how the two parameter biases are correlated in the likelihood of trajectories. The determinant inequality converts the covariance matrix of the two estimators into a single scalar bound. The monitoring-operator formalism supplies the numerical route to the Fisher information on individual trajectories.","core_discovery":"The central claim is that for any Markovian open quantum system with two counting observables $\\Phi_1$ and $\\Phi_2$, the scalar multiparameter Cramér-Rao bound gives\n$$\\frac{\\det(\\Xi)}{\\langle\\Phi_1\\$rangle^{2}$\\langle\\Phi_2\\$rangle^{2}$}=\\frac{\\mathrm{Var}(\\Phi_1)\\mathrm{Var}(\\Phi_2)}{\\langle\\Phi_1\\$rangle^{2}$\\langle\\Phi_2\\$rangle^{2}$}-\\frac{\\mathrm{Cov}(\\Phi_1,\\Phi_2)^2}{\\langle\\Phi_1\\$rangle^{2}$\\langle\\Phi_2\\$rangle^{2}$}\\ge \\frac{(1+\\varphi_1)^2(1+\\varphi_2)^2}{(A_1+Q_1)(A_2+Q_2)-F_{12}^2},$$\nwith an analogous inequality for heat currents. The new term is $F_{12}$, the off-diagonal element of the Fisher information matrix, which enters in the denominator and always makes the bound larger; the authors prove that $F_{12}$ vanishes for classical stochastic dynamics, so a nonzero value is a quantum signature. The paper further demonstrates that the multidimensional KUR can be essentially saturated in the three-level maser, where the two heat currents are proportional to the same underlying stochastic variable.","pith_inferences":["Because only $F_{12}^2$ enters the denominator, the sign of the quantum correlation is lost; a natural follow-up is to seek a signed witness built from the off-diagonal Fisher information or the covariance itself, possibly connected to measurement invasiveness.","An implicit assumption in the derivation is that the average of each current responds only to its own perturbation parameter; if cross-responses are substantial, the determinant of the full response matrix should replace the product of diagonal entries, which is a testable modification.","The same multiparameter Cramér-Rao logic with different scalarisation criteria, such as A-optimality or E-optimality, would produce alternative scalar bounds in which $F_{12}$ appears in a matrix combination rather than only as a squared term, potentially giving different tightness for unequal-precision currents.","One concrete experimental direction is to extract $F_{12}$ from the monitored record of a driven qubit and check that the bound tightens precisely in the parameter regime where the two counting channels are most correlated."],"forward_implications":["For any pair of counting observables in a Markovian open quantum system, the product of relative fluctuations corrected by the covariance term is constrained by Eq. (23), and the heat-current version by Eq. (29).","The bound is tighter than the product of the two single-observable quantum KURs whenever $F_{12}\\neq 0$; for classical rate processes $F_{12}=0$, so the classical analogue of this tightening is absent.","The multidimensional KUR can be saturated when the two currents are perfectly correlated, as in the three-level maser example, meaning the underlying multiparameter Cramér-Rao bound is asymptotically attainable in that setting.","The same approach extends to three or more observables, as the authors state, at the cost of more complicated expressions; pairwise bounds can therefore be combined into multi-current trade-offs."],"supporting_citations":[{"why":"It supplies the single-observable quantum KUR and TUR that this paper generalises to multiple observables, including the parameter imprinting and the quantum corrections $Q$ and $Q'$.","marker":"[34]"},{"why":"It provides the multiparameter Cramér-Rao bound and the determinant scalarisation that forms the starting point of the derivations of Eqs. (23) and (29).","marker":"[16]"},{"why":"It defines the original thermodynamic uncertainty relation that the new quantum TUR extends to multiple currents.","marker":"[13]"},{"why":"It defines the kinetic uncertainty relation and the dynamical activity $A$ that appear in the multidimensional KUR.","marker":"[4]"},{"why":"It supplies the monitoring-operator formalism and the Fisher-Gillespie algorithm used in the paper to compute Fisher information from quantum trajectories.","marker":"[52]"},{"why":"It provides the three-level maser model whose two heat currents are proportional, giving the example in which the multidimensional KUR is essentially saturated.","marker":"[30]"},{"why":"It gives the complementary near-equilibrium multidimensional precision bounds that the present far-from-equilibrium Markovian bounds complete.","marker":"[32]"}],"fun_headline_variants":["Tighter bounds for multiple currents in open quantum systems","Quantum current correlations improve uncertainty bounds","Off-diagonal Fisher term strengthens quantum current bounds","Multidimensional TUR and KUR beat single-observable limits"],"cache_read_input_tokens":25856,"weakest_assumption_plain":"The derivation assumes that the average of each current responds only to its own perturbation parameter, rather than also responding to the other parameter through the changed steady state.","fun_headline_variants_meta":{"raw":{"variants":["Tighter bounds for multiple currents in open quantum systems","Quantum current correlations improve uncertainty bounds","Off-diagonal Fisher term strengthens quantum current bounds","Multidimensional TUR and KUR beat single-observable limits"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000211,"raw_usage":{"total_tokens":1447,"prompt_tokens":1008,"completion_tokens":439,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":624,"completion_tokens_details":{"reasoning_tokens":387}},"tokens_in":624,"tokens_out":439,"duration_ms":4102,"temperature":1.0,"reasoning_tokens":387,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T21:07:48.541414+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Evaluate both sides of Eq. (23) for a minimal two-state model in which the two jump channels are simultaneously perturbed and the state is driven by both parameters; if the left-hand side falls below the right-hand side while the covariance matrix remains positive semidefinite, the diagonal-response assumption behind the bound is violated.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It supplies the single-observable quantum KUR and TUR that this paper generalises to multiple observables, including the parameter imprinting and the quantum corrections $Q$ and $Q'$."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It defines the kinetic uncertainty relation and the dynamical activity $A$ that appear in the multidimensional KUR."},{"cited_title":"Manzano and R","cited_arxiv_id":null,"evidence_quote":"It supplies the monitoring-operator formalism and the Fisher-Gillespie algorithm used in the paper to compute Fisher information from quantum trajectories."},{"cited_title":"Ptaszyński, Coherence-enhanced constancy of a quan- tum thermoelectric generator, Phys","cited_arxiv_id":null,"evidence_quote":"It provides the three-level maser model whose two heat currents are proportional, giving the example in which the multidimensional KUR is essentially saturated."}],"review_version":1}