{"id":"739d7ac1-588b-4db7-aac7-96de7ef05065","arxiv_id":"2512.11055","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For Gaussian states, every correlated single mode has a unique partner (pure states) or separate correlation/entanglement partners (mixed states), constructed from the state's complex structure.","lead":"A new framework identifies exactly which other modes carry all correlations and entanglement of a chosen mode in Gaussian quantum systems. It gives explicit formulas for these \"partner\" modes and separates total correlations from entanglement in mixed states.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Paper does not prove that the projected vector defining Γ_Aep is non-null; if null, the entanglement partner is not a valid single-mode subsystem and the central formula fails.","rationale":"The reader's verdict (CONDITIONAL) centers on the imported 1×N PPT theorem and the restricted sign flip identity. I find both to be ultimately sound (the theorem is standard; the sign-flip identity is correct once the missing superscript on J^{T_b} is restored). However, the reader's broader point that the closed-form partner formula must be valid is correct, and I identify a specific unproven requirement: the projected vector must be non-null for Γ_Aep to be a legitimate symplectic subspace. The paper proves only that the projection is nonzero, not that it is non-degenerate. This gap is directly load-bearing for the central claim that the entanglement partner is exactly single-mode. Since this is a proof gap rather than a demonstrated counterexample, the verdict should remain CONDITIONAL, pending either a proof of non-degeneracy or a numerical falsification. I propose a concrete computational test that would settle the issue.","tokens_in":26921,"tokens_out":44801,"duration_ms":375338,"concrete_test":"Generate an ensemble of random Gaussian states (e.g., random 4- and 5-mode covariance matrices satisfying σ + iΩ ≥ 0). For each, choose a single-mode subsystem A, compute the partial transpose J^{T_a}, and if it has a symplectic eigenvalue ν < 1, find the eigenvector e^{T_a}_1. Compute v = Π⊥_a e^{T_a}_1 and check ⟨v,v⟩ ≠ 0. Also verify that the smallest symplectic eigenvalue of the partial transpose of the reduced state on A ⊕ span{v,v*} equals ν. If any state yields ⟨v,v⟩ = 0, the central claim is falsified; if all states pass, the gap is filled.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Proposition 4 defines the entanglement partner as Γ_aep = span[Π⊥_a e^{T_a}_1, (Π⊥_a e^{T_a}_1)*]. For this to be a single-mode subsystem, the vector v = Π⊥_a e^{T_a}_1 must have non-zero symplectic norm ⟨v,v⟩ ≠ 0; otherwise the span is totally isotropic and does not define a symplectic subspace. The proof only shows e^{T_a}_1 ∉ Γ_a, so v ≠ 0, but it does not rule out ⟨v,v⟩ = 0. Since e^{T_a}_1 has positive norm and decomposes as e_a + v, it is possible in principle that ⟨e_a,e_a⟩ > 0 and ⟨v,v⟩ = 0. Without proving ⟨v,v⟩ ≠ 0, the construction may fail to produce any physical subsystem. This is load-bearing: the central claim that a unique single-mode partner exists for every non-PPT state depends on the projected subspace being a genuine symplectic subsystem. The Lemma in Appendix H also contains a typo ('symplectically orthogonal to γ_{p_a}' should be 'γ_{x_a}'), but that is fixable; the non-degeneracy gap is substantive.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a phase-space, complex-structure framework for identifying the degrees of freedom that carry total correlations and entanglement with a chosen mode in bosonic Gaussian systems. For pure Gaussian states, Proposition 2 gives an explicit partner formula Γ_ap = Π⊥_a(JΓ_a), claimed to be the unique single-mode subsystem capturing all correlations. For mixed Gaussian states, the paper splits the notion into a correlation partner (Proposition 3) and an entanglement partner (Proposition 4), the latter constructed from the subunity eigenvector e^{T_a}_1 of the partially transposed restricted complex structure: Γ_aep = span[Π⊥_a e^{T_a}_1, (Π⊥_a e^{T_a}_1)*]. The construction is extended to multi-mode subsystems in Propositions 5–7. The main claims are constructive and parameter-free, with appendices covering the covariance metric, the restricted complex structure, PPT/negativity, and the number of subunity symplectic eigenvalues.","tokens_in":27197,"tokens_out":37493,"duration_ms":324878,"significance":"If correct, this provides a basis-independent, constructive answer to where correlations and entanglement of a chosen mode reside, extending earlier QIC/partner-mode ideas to mixed states and multi-mode subsystems. The formulas are explicit, coordinate-free, and do not rely on fitted parameters; the paper also gives numerous worked examples and proofs in appendices. The pure-state result is a clear geometric reformulation of known partner constructions, while the mixed-state entanglement-partner theorem is a substantive new claim. However, the central mixed-state result depends on a non-degeneracy property that is not proved, and one of the key identities used in the reciprocity proof is stated rather than fully derived. These gaps currently prevent the paper from being accepted as is.","major_comments":[{"comment":"The proof of Proposition 4 only shows that e^{T_a}_1 is not entirely contained in Γ_a, hence v = Π⊥_a e^{T_a}_1 is nonzero. It does not show that ⟨v,v⟩ ≠ 0. If ⟨v,v⟩ = 0, the two-dimensional complex span of v and v* is totally isotropic and does not define a symplectic single-mode subsystem. In that case Eq. (47) fails to produce a physical subsystem, and the claimed existence/uniqueness of the entanglement partner collapses. This is load-bearing for the main mixed-state claim, so a proof of non-degeneracy (or a counterexample) is required.","section":"§III B 2, Eq. (47)"},{"comment":"The 'restricted sign flip identity' is used to show that the entanglement partner of A_ep is A, which is essential for establishing that A_ep captures all the entanglement between A and the rest. The identity is stated with a terse chain of equalities, and the operator expression T_aT_b J T_b T_a T_b is ambiguous as written (missing parentheses or a typo in the ordering of T factors). Since this identity is load-bearing, a full derivation with clear definitions of Ω^{T_S} and J^{T_S}, and a step-by-step verification of the sign flips, should be provided.","section":"§III B 2, Eq. (58), proof of statement b)"},{"comment":"The proof asserts that Γ_b = ⊕_I(Π^+_I Γ_a ⊕ Π^-_I Γ_a) is 'the smallest subsystem satisfying points 1 and 2', but no argument for minimality is given. Since the correlation partner A_cp is defined as the smallest uncorrelated subsystem containing A, this minimality step is part of the definitional content of the result and should be proved explicitly, not asserted.","section":"§III B 1, Proposition 3"}],"minor_comments":[{"comment":"The text says that after flipping γ_{p_a}, T_a reduces to the identity in the subspace symplectically orthogonal to the direction γ_{p_a}. The fixed subspace of the momentum flip is actually the subspace symplectically orthogonal to γ_{x_a}. Please correct the wording.","section":"Appendix H"},{"comment":"The reported value ν^{T_a}_1 = 1/2(−35 + √1201) ≈ −0.17 is negative, which is impossible for a symplectic eigenvalue. Please check the expression; this appears to be a typo in the worked example.","section":"§III B 2, example"},{"comment":"In several places the correlation partner is denoted Γ_aep (e.g., the lines defining the correlation partner in items 3 and 4), which conflicts with the notation Γ_acp used in the main text. Please use consistent notation.","section":"Appendix F"},{"comment":"Equation (41) defines projectors Π^±_I; the signs in the normalization conventions are not fully explained, though they follow from ⟨e_I,e_I⟩>0 and ⟨e_I^*,e_I^*⟩<0. A sentence making this explicit would improve readability.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The paper is likely to be a solid contribution once the proof gaps are addressed. The most important issue is the missing non-degeneracy of the projected vector in Proposition 4; if a short proof exists, the paper could be upgraded, but as it stands the central mixed-state theorem is not fully established. The example typo in the symplectic eigenvalue also needs correction. I would not recommend rejection at this stage, but the authors need to supply the missing arguments."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this paper is worth refereeing. It gives a clean, basis-independent geometric construction of partner modes for Gaussian states, and the genuinely new part—the mixed-state entanglement partner, a single mode that captures all entanglement with a chosen mode—is proved in enough detail to be credible. The pure-state formula is honestly acknowledged as equivalent to earlier work, and the phase-space formalism with restricted complex structures is well chosen. The worked examples are concrete and useful.\n\nThe main soft spot is Proposition 4. The entanglement partner is defined as span[Π⊥_a e, (Π⊥_a e)*], where e is the subunity eigenvector of the partially transposed J. For that span to be a symplectic subspace, v = Π⊥_a e needs nonzero symplectic norm. The proof shows v ≠ 0 but never rules out ⟨v,v⟩ = 0. If v is null, the span is isotropic and there is no single-mode partner. That is a load-bearing gap in the central claim. I tried to find a concrete counterexample and couldn't, and I suspect a positivity argument using the partial-transpose quadratic form can close it, but it is not in the paper.\n\nThe second soft spot is the imported 1×N PPT theorem, which underlies the claim that the entanglement partner exists exactly when the state is non-PPT. That theorem is standard and the citation is appropriate, but the paper leans on it without proof. Proposition 7's multi-mode extension is also abbreviated—more sketch than full proof.\n\nOtherwise the logic is clean, the citation pattern looks right, and there is no fitted-parameter or circularity issue. The stress-test concern about the null projection is the only substantive mathematical complaint I see. This deserves a serious referee: send it out, and ask the authors to prove non-degeneracy of v and expand the Prop. 7 proof.","headline":"Worth refereeing: a clean geometric split between correlation and entanglement partners for Gaussian states, with one load-bearing gap in the entanglement-partner proof that is likely fixable.","tokens_in":27666,"tokens_out":15090,"would_cite":true,"duration_ms":142615,"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":"For Gaussian states, every entangled mode has a unique partner, and the paper constructs it explicitly.","keywords":["Gaussian states","entanglement partner","correlation partner","restricted complex structure","symplectic eigenvalues","positive partial transpose","logarithmic negativity","continuous-variable quantum information"],"falsifier":"For a concrete three-mode mixed Gaussian state with non-PPT A, construct A_ep from Eq. (47), then compute the logarithmic negativity of the reduced state on A⊕A_ep from its covariance matrix and compare it with the logarithmic negativity of the full bipartition A versus the rest. If the two numbers differ, or if applying the same construction to A_ep does not return a single mode equal to A, the claimed completeness and reciprocity of the entanglement partner fail.","tokens_in":26832,"feed_emoji":"🎯","tokens_out":5536,"duration_ms":56898,"temperature":0.7,"pith_summary":"The paper claims that in Gaussian (continuous-variable) quantum states, the correlations and entanglement of any chosen single mode with the rest of the system are concentrated in specific partner subsystems, and it gives closed-form constructions for those partners. For pure states, the partner is always another single mode and is obtained by applying the state's restricted complex structure to the original mode and projecting outside it. For mixed states, the notion splits: a correlation partner that may contain several modes and collects all correlations, and an entanglement partner that is always at most one mode and exists exactly when the state fails the positive-partial-transpose test. If true, this turns a many-body entanglement problem into an effective two-mode description and gives a constructive way to locate entanglement resources.","feed_headline":"One formula finds the mode that carries all entanglement","feed_subtitle":"In Gaussian quantum states, each entangled mode has a unique partner—the paper tells you exactly how to build it.","key_machinery":"The restricted complex structure J = -ℏ Ωσ (equivalently, the covariance metric σ) encodes the state in phase space; its eigenpairs define a symplectic-orthonormal basis. Symplectic projectors Π_a and Π^⊥_a provide a basis-independent way to separate a subsystem from its complement. Partial transposition T_a is a momentum flip on A, and the crucial object is the unique eigenvector e^{T_a}_1 of J^{T_a} whose symplectic eigenvalue ν < 1. The formulas Γ_ap = Π^⊥_a(JΓ_a) and Γ_aep = span(Π^⊥_a e^{T_a}_1, conjugate) carry the argument: they convert the abstract existence of partners into explicit subspaces.","core_discovery":"For a pure Gaussian state with restricted complex structure J, the partner of a correlated single-mode subsystem A is Γ_ap = Π^⊥_a(JΓ_a); A⊕A_p is uncorrelated with the rest and the reduced state on it is pure, so A_p captures all correlations, which for pure states are entanglement. For a mixed Gaussian state, the entanglement partner is Γ_aep = span(Π^⊥_a e^{T_a}_1, (Π^⊥_a e^{T_a}_1)^*), where e^{T_a}_1 is the unique eigenvector of the partially transposed J with symplectic eigenvalue below 1; this mode encodes all entanglement between A and the rest, and the partner of A_ep is A again. The correlation partner, by contrast, is the smallest subsystem that makes A⊕A_cp uncorrelated and gener","pith_inferences":["A practical consequence the paper leaves implicit: entanglement harvesting or teleportation protocols would need to address only A and its partner A_ep, since the formula identifies a minimal set of modes to act on.","Because the formulas are basis-independent, they should survive a continuum limit; applying them to lattice quantum-field-theory vacua would locate entanglement partners in space and could sharpen claims about entanglement structure in quantum fields.","The same restricted-complex-structure logic is sketched for fermionic pure states; a mixed fermionic analogue, if it exists, would likely follow the same pattern with the metric and symplectic roles interchanged.","The construction could be tested as an operational entanglement witness: for states where the reduced negativity computed from the partner formula deviates from the full negativity beyond numerical error, the imported PPT theorem or its Gaussian faithfulness would be the suspect."],"forward_implications":["Every correlated mode in a pure Gaussian state can be paired into an effective two-mode pure state, so many-body entanglement structure can be read off mode by mode.","In mixed states, all entanglement of a mode is concentrated in a single partner mode whenever the state is non-PPT, even though total correlations may be spread over many modes.","The entanglement-partner map is involutive in the single-mode case: the partner of A_ep is A, so the construction is symmetric and well defined.","For a multi-mode subsystem A in a pure state, the partner has exactly as many modes as there are symplectic eigenvalues of the reduced J larger than 1; uncorrelated modes inside A simply drop out.","For multi-mode A in a mixed state, the entanglement partner is at most N_a modes and captures the distillable entanglement, giving a finite-dimensional handle on distillability in Gaussian systems."],"fun_headline_variants":["Each entangled mode has one true partner","Pinpoint the mode that holds all entanglement","Gaussian entanglement: find the unique partner","Entanglement partner: always a single mode","Correlation vs entanglement partners in Gaussian states"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The mixed-state results depend on the imported theorem that for one mode versus many Gaussian modes, failing the positive-partial-transpose test is equivalent to being entangled and that logarithmic negativity faithfully measures the entanglement; the proof that the partner's partner is A also leans on the restricted sign-flip identity, Eq. (58).","fun_headline_variants_meta":{"raw":{"variants":["Each entangled mode has one true partner","Pinpoint the mode that holds all entanglement","Gaussian entanglement: find the unique partner","Entanglement partner: always a single mode","Correlation vs entanglement partners in Gaussian states"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000156,"raw_usage":{"total_tokens":1051,"prompt_tokens":735,"completion_tokens":316,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":479,"completion_tokens_details":{"reasoning_tokens":251}},"tokens_in":479,"tokens_out":316,"duration_ms":4013,"temperature":1.0,"reasoning_tokens":251,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T16:57:51.884687+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For a concrete three-mode mixed Gaussian state with non-PPT A, construct A_ep from Eq. (47), then compute the logarithmic negativity of the reduced state on A⊕A_ep from its covariance matrix and compare it with the logarithmic negativity of the full bipartition A versus the rest. If the two numbers differ, or if applying the same construction to A_ep does not return a single mode equal to A, the claimed completeness and reciprocity of the entanglement partner fail.","supporting_citations":[],"review_version":1}