{"id":"12aec7e7-95b2-49ca-b8c9-3d79eb9a5b39","arxiv_id":"2602.09468","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"For MMM (Bell-diagonal) two-qubit states, the quantum-illumination advantage has entanglement as a sufficient-but-not-necessary resource and discord as a necessary-but-not-always-sufficient resource, with a linear discord–advantage law in the high-noise limit.","lead":"This paper studies which quantum resource—entanglement or discord—powers the advantage in quantum illumination for two-qubit states. It concludes that higher entanglement guarantees higher advantage while higher discord is required, but not always enough, for it.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The printed discord formula (Eq. 13) is off by a constant from standard discord, and no code/data shows which formula generated the heat maps; if the printed formula was used, QA=δ_enc and all discord-based resource claims fail.","rationale":"The reader's verdict is CONDITIONAL, with the weakest assumption being that the extremal monotonicity claims are read off 1/80-mesh heat maps without convergence or error analysis. That is a real concern. However, an even more basic correctness issue appears in the printed definition of the central resource: Eq. (13) as written evaluates to non-standard values for the completely mixed state and Bell states, and the Appendix A calculation silently uses a different (correct) formula. This makes the quantitative discord values, the QA=δ_enc identification, and the discord-based extremal analysis non-reproducible from the manuscript alone. The concern is concrete and testable: one can check the formula at known states and regenerate the key plots. If the corrected formula is the one actually used, the qualitative conclusions may survive, because a constant offset in initial discord does not change fixed-discord slices and the high-noise asymptotic ratio is already computed with the corrected expression. But the manuscript as written does not allow a reader to verify this, and the discrepancy affects the foundational equality QA=δ_enc. I therefore keep the verdict CONDITIONAL, but the condition should include not only mesh refinement/error analysis but also clarification and correction of Eq. (13) and disclosure of which formula generated the numerical results. I partially agree with the reader's identified weak point, but I think the formula inconsistency is more load-bearing because it strikes at the definition of the resource being characterized.","tokens_in":14011,"tokens_out":18155,"duration_ms":167387,"concrete_test":"Evaluate Eq. (13) analytically at c=(0,0,0) and at c=(1,−1,1), and check whether it returns 0 and 1 (standard discord) or 1 and 2. Then regenerate the QA-vs-δ_enc plot of Fig. 4(b) and the conditional-extremal heat maps of Figs. 5(b)–7 using both the printed formula and the corrected formula δ = 1 + Σλ_k log₂λ_k − C(ρ). If the printed formula was used in the original numerics, the QA=δ_enc line will not pass through the origin for p₀=1/2; if the corrected formula was used, the manuscript must state that correction explicitly.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The displayed discord formula in Eq. (13), δ = 2 + Σλ_k log₂λ_k − C(ρ), is inconsistent with standard discord. For the completely mixed state (c_i=0), it gives δ=1; for a Bell-diagonal Bell state such as c=(1,−1,1), it gives δ=2. Standard Ollivier-Zurek discord for these states is 0 and 1, respectively. The correct Bell-diagonal discord formula is δ = 1 + Σλ_k log₂λ_k − C(ρ), i.e., Eq. (13) should have a '1' rather than a '2'. The Appendix A expansion implicitly uses the correct formula: it drops the constant and obtains δ ≈ (c₂²+c₃²)/(2 ln 2) near the fully mixed state, whereas the printed Eq. (13) would give 1 + that quantity. Since no code or data is provided, it is impossible to determine whether Figs. 3–7 were generated with the printed formula or the corrected one. If the printed formula was used, then δ_enc = p₀δ(0)−δ(ρ̄) carries an extra (p₀−1) offset; for the standard prior p₀=1/2, the claimed equality QA=δ_enc in Fig. 4(b) cannot hold, and the discord-based conditional-extremal classification in Sec. IV is not established. This issue is independent of the mesh-resolution concern and must be settled before the central resource claims can be accepted.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies quantum illumination with two-qubit maximally mixed marginal (MMM) states as the initial probe resource. It claims that the quantum advantage (QA) equals the discord of encoding (δ_enc), and then performs a conditional extremal analysis on the set of MMM states to determine how initial entanglement and initial discord relate to the advantage. The central conclusions are that (i) for fixed initial discord, the maximum initial entanglement in the cluster increases with advantage, so higher entanglement is sufficient but not necessary for higher advantage; (ii) for fixed initial entanglement, the minimum initial discord in the cluster increases with advantage, so higher discord is necessary but not always sufficient; and (iii) in the high-noise regime, QA is linearly proportional to initial discord with coefficient p_0 η²(1−p_0). The paper also claims in the abstract that the analysis is repeated with relative entropy of entanglement, Bures measure, and geometric discord, but this is not carried out in the body.","tokens_in":14392,"tokens_out":9162,"duration_ms":78626,"significance":"If the conclusions are correct, the paper provides a refined, operational characterization of when entanglement and discord contribute to the quantum-illumination advantage, going beyond the known QA=δ_enc theorem. The conditional extremal approach is a sensible way to address the broadened advantage-correlation relations. However, the significance is heavily undermined by a likely error in the printed discord formula, by the absence of the promised other-measure analyses, and by the reliance on finite mesh heat maps without convergence or error estimates. These issues prevent the reader from trusting the central necessary/sufficient claims as stated.","major_comments":[{"comment":"The printed Bell-diagonal discord formula, δ = 2 + Σλ_k log₂λ_k − C(ρ), is incorrect. For Bell-diagonal states the standard formula is δ = 1 + Σλ_k log₂λ_k − C(ρ), where C(ρ) is defined as in Eq. (14). With the printed constant 2, the completely mixed state (c_i=0) has δ=1, whereas its true discord is 0; for a Bell state the printed formula gives δ=2 instead of 1. Appendix A, however, expands to δ ≈ (c₂²+c₃²)/(2 ln 2), which is consistent with the correct constant (1), not with the printed 2. Since no code or data is provided, it is impossible to determine which formula generated Figs. 3–7. If the printed Eq. (13) was used, then δ_enc = p₀δ(ρ⁰)−δ(ρ̄) carries an extra (p₀−1) offset; for the typical p₀=1/2 the claimed equality QA=δ_enc in Fig. 4(b) cannot hold. This issue is load-bearing for the resource classification in Sec. IV and must be resolved.","section":"Sec. III, Eq. (13)"},{"comment":"The abstract states: 'We also repeat our analysis with other measures of quantum correlation. In particular, we show that relative entropy of entanglement, Bures measure of entanglement and geometric discord lead to the same conclusion...' The full manuscript contains no such analysis; these measures are not defined, computed, or even mentioned after the abstract. This is a significant discrepancy between the claimed scope and the actual content, and it misrepresents the paper's contributions.","section":"Abstract versus body"},{"comment":"The central necessary/sufficient conclusions are read off finite heat maps with mesh precision 1/80. The claims that E_max(δ_in, A) increases with A for fixed δ_in and that δ_min(E_in, A) increases with A for fixed E_in are inferred from grid extrema. No convergence analysis, error bars, or analytic proof is provided. The grid may miss the true continuous extrema, especially near the separable/entangled transition (the δ_in≈0.33 horn), the α/Werner crossover, and the entangled-state boundary. Without either an analytic proof or a convergence study, the classification 'higher entanglement sufficient, higher discord necessary' is not established.","section":"Sec. IV, Figs. 5–7"},{"comment":"The quantum advantage QA is defined as the difference of Holevo informations, but the explicit expression for QA for general MMM states is never derived or displayed. The text says 'it is a simple task' and plots Fig. 4(a). Since the equality QA=δ_enc and all subsequent extremal analysis depend on the values of QA, the reader cannot verify the calculations. The paper relies on the prior theorem of [33] for the qualitative equality, but the numerical heat maps require the actual formula. Provide the explicit expression for QA (or at least for χ_q and χ_c) in terms of c_i, η, and p₀.","section":"Sec. IV, Eqs. (16)–(19)"}],"minor_comments":[{"comment":"There are many typographical errors, e.g., 'quantum advanatge' (Sec. V), 'preicisely' (Sec. IV), 'refelecivity' (Fig. 7 inset), and inconsistent notation (ε vs. η in Eq. (2)). The manuscript needs careful proofreading.","section":"Throughout"},{"comment":"Eq. (2) uses η for reflectivity, but the text near it uses ε. Please standardize the notation.","section":"Sec. II, Eq. (2)"},{"comment":"The equality QA=δ_enc is shown as a scatter plot without numeric verification or fitting. A quantitative statement (e.g., maximum deviation) would strengthen the claim.","section":"Fig. 4(b)"},{"comment":"The expansion is performed only for the case max(|c_i|)=|c_1|, and single-axis paths (e.g., c₂=c₃=0) are excluded. This is stated, but the resulting limit in Eq. (24) is then presented as the generic high-noise behavior. Please clarify the domain of validity and whether the single-axis exception affects the conclusions.","section":"Appendix A"}],"recommendation":"major_revision","confidential_remarks":"The most serious issue is the discord formula in Eq. (13). If it is a typo and the figures were generated with the correct formula, the paper is still not acceptable in its current form because the abstract contains claims (other-measure analyses) that are entirely absent from the body, and the central extremal claims are not backed by any proof or convergence study. As an editor, I would ask the authors to (1) correct the discord formula and state which formula was used, (2) either add the promised other-measure analysis or remove it from the abstract, and (3) provide a derivation or at least a reproducible calculation of QA for MMM states, along with the extremal analysis on a converged mesh. Without these, the necessary/sufficient resource classification is not established."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"I took a close look, and the stress-test note about Eq. (13) does not hold up. The formula is the standard Ollivier-Zurek discord for Bell-diagonal states: 2 + Σλ log λ − C(ρ), where C(ρ) is the classical correlation (the (1±ζ)/2 log(1±ζ) terms). For the fully mixed state it gives 0, for a Bell state it gives 1. The Appendix A expansion is consistent with that. So the worry about a constant offset is unfounded.\n\nWhat the paper actually does: for two-qubit MMM (Bell-diagonal) states in a discrete-variable quantum illumination setup, it numerically confirms the QA=δ_enc equality (known from [33]) and then asks whether initial discord or entanglement is the better predictor of advantage. The new content is the conditional extremal classification: at fixed initial discord, the maximum entanglement in each (QA, δ) cluster grows with QA; at fixed initial entanglement, the minimum discord grows with QA. They interpret this as entanglement sufficient, discord necessary. That is a neat, plausible refinement, and the identification of α-, β-, and Werner states as the bounds is useful. The high-noise linear relation QA = p0 η²(1−p0) δ_in is derived analytically and looks correct under the stated generic-path assumption.\n\nThe soft spots are real but addressable. First, the extremal claims are inferred from finite-mesh heat maps (1/80) with no code, data, or convergence/error analysis. The boundaries (separable/entangled, the horn near δ_in≈0.33) are exactly where grid artifacts would matter. Second, the abstract says they repeat the analysis with relative entropy of entanglement, Bures measure, and geometric discord; that material is absent from the text. Either add it or cut the claim. Third, \"rigorously show\" for QA=δ_enc is too strong; the paper plots it, does not prove it (the proof is in [33]). None of this is fatal; all are fixable with more careful claims or additional numerics.\n\nWho it is for: people working on discrete-variable quantum illumination and on discord-versus-entanglement resource questions. It is a modest extension of an established program, not a paradigm shift, but if the numerical claims get hardened it would be a solid reference for the MMM-state resource boundary. I would send it to peer review; a good referee can ask for the missing data/code and tighten the claims. Reading group: maybe, if someone in the group cares about this specific question.","headline":"A focused numerical study with a plausible resource classification, but the central monotonicity claims rest on 1/80-mesh heat maps without error analysis, and the abstract promises a multi-measure robustness check that the paper does not contain.","tokens_in":14814,"tokens_out":7789,"would_cite":true,"duration_ms":60376,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81P40","81P45"],"pacs":["03.67.Mn"],"model":"deepseek-v4-flash","headline":"In quantum illumination with two-qubit mixed probes, the achievable advantage is set by an interplay: initial discord is a necessary resource and initial entanglement a sufficient one, with the advantage exactly equal to the discord consume","keywords":["quantum illumination","quantum discord","entanglement of formation","maximally mixed marginal states","discord of encoding","quantum advantage","Holevo information","high-noise regime"],"falsifier":"Compute the same conditional extrema on a finer mesh (or analytically) and find either a pair of MMM states with equal initial discord where the state with lower advantage has greater maximum entanglement than a higher-advantage state, or a pair with equal initial entanglement where a higher-advantage state has lower minimum discord than a lower-advantage state. Any such pair would falsify the sufficient/necessary claims. The natural test regions are the separable/entangled boundary near δ_in ≈ 0.33 and the α/Werner crossover where the upper/lower bounds switch.","tokens_in":13891,"feed_emoji":"🎯","tokens_out":8237,"duration_ms":62815,"temperature":0.7,"pith_summary":"The paper sets out to determine which quantum correlation actually powers the advantage in quantum illumination when the probe is a maximally mixed marginal (MMM) two-qubit state. It first shows that the quantum advantage equals the 'discord of encoding' — the discord consumed to encode the presence of the object — and then asks how initial discord and initial entanglement each constrain that advantage. By grouping states that share the same discord and advantage and examining the distribution of entanglement within each group, and by doing the symmetric exercise with entanglement and discord swapped, the authors conclude that higher entanglement is sufficient but not necessary for higher advantage, while higher discord is necessary but not always sufficient. In the high-noise regime the advantage becomes a linear function of initial discord, which they read as evidence that discord is the resource that survives noise.","feed_headline":"Discord necessary, entanglement sufficient for illumination gain","feed_subtitle":"For two-qubit mixed probes, advantage tracks consumed discord; in high noise it stays linear in initial discord","key_machinery":"The central object is the family of maximally mixed marginal (MMM) two-qubit states, ρ = (1 + Σ c_i σ_i⊗σ_i)/4, with |c_i|≤1. Two identities carry the argument: (i) for these states the state conditioned on target presence and the noise state commute, so the accessible information saturates the Holevo bound and the quantum advantage can be computed as a difference of Holevo informations; and (ii) the resulting advantage equals the discord of encoding, δ_enc = p0 δ(ρ0) − δ(ρ̄). The conditional extremal analysis is the method that converts these identities into resource statements: states are clustered by equal (advantage, discord) or (advantage, entanglement), and the max/min of the other cor","core_discovery":"For MMM states — two-qubit states whose reduced states are completely mixed, described by a correlation vector (c1,c2,c3) — the quantum advantage A in illumination is exactly the discord of encoding δ_enc. A conditional extremal analysis then shows that, among states with a fixed initial discord, the maximum initial entanglement in a cluster sharing the same advantage increases monotonically with A, while the minimum entanglement does not; hence entanglement is sufficient but not necessary for higher advantage. Conversely, for fixed initial entanglement, the minimum initial discord in each advantage-cluster increases monotonically with A, while the maximum discord increases only in the low-a","pith_inferences":["The monotonicity claims are read off 1/80-mesh heat maps without convergence or error analysis; an analytic proof, or a finer-mesh check near the separable/entangled boundary (the horn at δ_in ≈ 0.33) and the α/Werner crossover, would convert the sufficient/necessary classification from a numerical inference into a theorem.","The linear high-noise relation suggests a direct experimental probe: with near-maximally-mixed probes, measuring the advantage at two reflectivities η would extract the slope p0 η² (1−p0) and test whether discord, not entanglement, is the noise-resilient resource in practice.","The conditional-extremal clustering could be applied to other families of two-qubit states or to other correlation measures; if the asymmetry (entanglement sufficient, discord necessary) persists outside MMM states, it would be a generic feature of discrete-variable quantum illumination."],"forward_implications":["For every MMM state the quantum advantage equals the discord of encoding, so the advantage can be computed from the discord formula without a full POVM optimization.","Among states with fixed initial discord, the maximum entanglement at a given advantage rises monotonically with advantage while the minimum entanglement does not — entanglement is a sufficient resource, not a necessary one.","Among states with fixed initial entanglement, the minimum discord at a given advantage rises monotonically with advantage while the maximum discord rises only in the low-advantage regime — discord is necessary but not always sufficient.","In the high-noise limit the advantage becomes linear in initial discord, A = p0 η² (1−p0) δ_in, so discord persists as the resource when noise pushes the probe toward the completely mixed state.","The same necessary/sufficient pattern is reported for relative entropy of entanglement, Bures measure of entanglement, and geometric discord, indicating the result is not tied to one particular quantifier."],"fun_headline_variants":["Discord is the engine, entanglement the turbo in illumination","For mixed-state probes, discord decides, entanglement extends","Quantum illumination: discord required, entanglement for extra","Illumination advantage: discord necessary, entanglement sufficient"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The classification rests on the assumption that the monotonic trends seen in the 1/80-mesh heat maps of conditional extrema — maximum entanglement for fixed discord, minimum discord for fixed entanglement — are the true trends of the continuous state space; no analytic proof or error analysis is given for these monotonicities.","fun_headline_variants_meta":{"raw":{"variants":["Discord is the engine, entanglement the turbo in illumination","For mixed-state probes, discord decides, entanglement extends","Quantum illumination: discord required, entanglement for extra","Illumination advantage: discord necessary, entanglement sufficient"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000741,"raw_usage":{"total_tokens":3166,"prompt_tokens":785,"completion_tokens":2381,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":529,"completion_tokens_details":{"reasoning_tokens":2330}},"tokens_in":529,"tokens_out":2381,"duration_ms":15796,"temperature":1.0,"reasoning_tokens":2330,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T02:47:09.905090+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the same conditional extrema on a finer mesh (or analytically) and find either a pair of MMM states with equal initial discord where the state with lower advantage has greater maximum entanglement than a higher-advantage state, or a pair with equal initial entanglement where a higher-advantage state has lower minimum discord than a lower-advantage state. Any such pair would falsify the sufficient/necessary claims. The natural test regions are the separable/entangled boundary near δ_in ≈ 0.33 and the α/Werner crossover where the upper/lower bounds switch.","supporting_citations":[],"review_version":1}