{"id":"4b030e3b-03be-4f0f-bfec-97708ef0c042","arxiv_id":"2505.10716","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"For graph states made of identical controlled-phase gates on |+> qubits, the Entanglement Distance per qubit equals 1 minus the average of cos(θ) to the power twice the vertex degree.","lead":"This paper shows that for quantum states built from a directed graph with identical controlled-phase connections on |+> qubits, the Entanglement Distance per qubit is set only by the number of connections at each vertex. Edge direction drops out, giving a simple design rule for entanglement in quantum network models.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Counterexample to Theorem 1: reciprocal edges (i,j) and (j,i) are allowed by the definitions but break Eq. (6); the proof's case iii assumes the outgoing and incoming neighbor sets are disjoint.","rationale":"The reader's verdict identified the homogeneous-coupling assumption and the incorrect phase in Eq. (24), but the most serious problem is a counterexample inside the stated definitions. The theorem's proof case iii silently assumes that the outgoing neighbor set and incoming neighbor set of a vertex are disjoint; the definition of directed graph states does not require this, and reciprocal pairs are allowed and satisfy the ordering-free commutativity condition. For such a pair the two controlled-phase gates act on the same two qubits, producing an effective two-qubit diagonal gate whose induced single-qubit Bloch vector length is |cos 2θ| rather than |cos θ|^{d}. This changes the ED at θ=π/4 from the predicted 1/2 (or 3/4 by arc count) to 1, so the central degree-only formula is false as stated. The fix would be to restrict to oriented graphs (no reciprocal arcs) and to repair the mixed-case derivation, but that is a nontrivial restriction not present in the paper. Hence the verdict should move from conditional to reject.","tokens_in":5756,"tokens_out":25140,"duration_ms":232702,"concrete_test":"Evaluate exactly the two-qubit case M=2, L={(1,2),(2,1)}, θ=π/4, ψ=0. State coefficients are (1, e^{iπ/4}, e^{iπ/4}, -i)/2 in the computational basis; the single-qubit reduced density matrix is I/2, giving E=1. Compare with Theorem 1: using d(1)=|Γ(1)|=1 it gives E=1−cos²(π/4)=1/2, and counting the two arcs gives E=1−cos⁴(π/4)=3/4. A 2×2 linear-algebra check (or any symbolic package) reproduces E=1 and disproves Eq. (6).","verdict_should_be":"REJECT","load_bearing_attack":"The claim (6) is derived only when Γ→(i) and Γ←(i) are disjoint. Definitions 1–2 allow both (i,j) and (j,i) in L, and the two diagonal gates commute, so such a graph is ordering-free. In proof case iii, Eq. (20) writes U_tot as a product over outgoing targets followed by incoming controls and then decomposes the state as |φ>^{⊗d→}|φ>^{⊗d←}|φ>^{(i)}; if a neighbor appears in both sets, that qubit is both a target and a control and the tensor-product decomposition is invalid. The failure is substantive, not a proof gap. For M=2 with L={(1,2),(2,1)}, the state is U_21 U_12 |++>. A direct calculation gives, for each qubit, ⟨σ_x⟩=cos(ψ+θ) cos 2θ, ⟨σ_y⟩=sin(ψ+θ) cos 2θ, ⟨σ_z⟩=0, so ||⟨σ⟩||²=cos²(2θ) and E_i=1−cos²(2θ). With the paper's degree d(i)=|Γ→∪Γ←|=1, Eq. (6) predicts 1−cos²θ; even counting the two arcs gives 1−cos⁴θ. At θ=π/4 these predictions are 1/2 and 3/4, while the exact E is 1. Thus the degree-only, direction-insensitive formula is false for an allowed configuration. The theorem requires an explicit no-reciprocal-edges assumption, and the phase in Eq. (24) is also incorrect for general d→,d←.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies a family of quantum states generated from directed graphs, where each directed edge applies the same two-qubit controlled-\\bar U operator of Eq. (5) to the initial state |+>^{⊗M}. The main result, Theorem 1, claims that the Entanglement Distance per qubit, E(θ;{d(i)}) = 1 − (1/M) Σ_i [cos θ]^{2d(i)}, depends only on the vertex degree d(i) and is independent of edge direction. The proof is organized into three cases according to whether vertex i has only outgoing, only incoming, or both kinds of incident edges. The paper concludes that the entanglement is determined by the degree distribution and that this underscores a topological nature of the measure.","tokens_in":6072,"tokens_out":3425,"duration_ms":35894,"significance":"The claimed formula is remarkably simple and, if valid, would give a closed-form, degree-only characterization of entanglement in a natural family of controlled-phase graph states. The proof strategy is direct and the statement is falsifiable, which are strengths. However, the central claim fails for configurations that are explicitly allowed by the paper's definitions: when reciprocal directed edges (i,j) and (j,i) are present, the tensor-product decomposition in Proof case iii is invalid and the degree-only formula is false. With an added no-reciprocal-edges assumption, the formula appears to be correct, and the corrected calculation would still be a useful contribution. The paper also overstates the result by calling the degree dependence topological.","major_comments":[{"comment":"The theorem is false for an allowed graph configuration. Definition 1 permits both (i,j) and (j,i) to belong to L, and if the two diagonal gates commute, such a graph is ordering-free. In Proof case iii, Eq. (20) writes U_tot as a product over outgoing targets followed by incoming controls and then decomposes the initial state as |φ>^{⊗d→}|φ>^{⊗d←}|φ>^{(i)}. When a neighbor appears in both Γ→(i) and Γ←(i), that neighbor is simultaneously a target and a control, so the tensor-product decomposition is invalid. For M=2 with L={(1,2),(2,1)}, a direct calculation gives ⟨σ_x⟩=cos(ψ+θ) cos 2θ, ⟨σ_y⟩=sin(ψ+θ) cos 2θ, ⟨σ_z⟩=0, hence E_i=1−cos²(2θ). With the paper's degree d(i)=|Γ→(i)∪Γ←(i)|=1, Eq. (6) predicts 1−cos²θ, and even counting two arcs predicts 1−cos⁴θ. At θ=π/4 these predictions are 1/2 and 3/4, while the exact value is 1. The theorem must be amended with an explicit no-reciprocal-edges assumption, or the definition of the graph must exclude reciprocal edges.","section":"Theorem 1 and Proof case iii (§2)"},{"comment":"The phase in Eq. (24) is incorrect for the mixed in/out case. Summing the binomial average in Eq. (22) gives the phase d→(i)ψ + d←(i)θ, not d←(i)(ψ+θ). The missing factor is e^{-i d←(i) θ} that arises from averaging e^{-2ikθ} over the binomial distribution. Although this phase error does not change the norm squared and hence does not alter the final Entanglement Distance value, Eq. (24) as written is a wrong identity and must be corrected.","section":"Eq. (24), Proof case iii"},{"comment":"The claim that Eq. (6) reveals a 'topological nature' of the measure is an overstatement. The degree sequence is an isomorphism invariant of a graph, but it is not a topological invariant in the standard sense, and Eq. (6) explicitly depends on the numerical degrees, so changing the degrees changes the entanglement. The conclusion should be rephrased to say that, within the restricted no-reciprocal-edge family, the entanglement depends only on the degree sequence and is insensitive to edge orientation.","section":"Remark 3 and Conclusion (§2, §3)"}],"minor_comments":[{"comment":"The definition L={(a,b) | a,b∈V} allows loops (a,a) as well as reciprocal edges; the paper should specify that the underlying graph is a simple directed graph with no loops, and state explicitly whether reciprocal edges are excluded.","section":"Definition 1"},{"comment":"The phrase 'all two-particle unitary operators Uab commutate' should read 'commute'.","section":"Definition 2"},{"comment":"The notation d→(i) appears in Eq. (10) and in the proof before it is defined; please define d→(i)=|Γ→(i)| and d←(i)=|Γ←(i)| explicitly before first use.","section":"Eq. (10) and surrounding text"},{"comment":"The notation 'c.c' in the displayed expectation value is unclear; writing the conjugate terms explicitly would improve readability.","section":"Eq. (13)"}],"recommendation":"major_revision","confidential_remarks":"The stress-test concern is valid and decisive: the counterexample with reciprocal edges is an allowed configuration under the paper's own definitions, and it directly falsifies Theorem 1 as stated. The fix is local—add a no-reciprocal-edges assumption—and the central formula survives for that restricted family. The paper deserves a major revision rather than rejection, provided the authors correct the theorem statement, fix the phase in Eq. (24), and temper the topological claim."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Theorem 1 as stated is false. The proof in case iii assumes the outgoing and incoming neighbor sets are disjoint, but Definitions 1–2 allow both (i,j) and (j,i) in L, and the gates commute so such graphs are ordering-free. For M=2 with L={(1,2),(2,1)}, the exact state gives for each qubit E=1−cos²2θ, while Eq. (6) predicts 1−cos²θ (or 1−cos⁴θ if you count arcs). So the degree-only formula breaks for reciprocal edges. This is not a proof gap; it is a counterexample to the claimed theorem.\n\nWhat the paper does well: for the disjoint case the calculation is correct. The formula E=1−(1/M)Σ_i cos^{2d(i)}θ is a clean closed form, and cases i and ii are derived cleanly. Expressing this particular entanglement measure in terms of the vertex degree distribution is a neat observation, and the connection to the authors' earlier ED framework is explicit.\n\nSoft spots: first, the missing no-reciprocal-edges assumption is load-bearing. Add it and Theorem 1 holds. Second, Eq. (24) has a phase error even in the disjoint case: the correct phase is d→ψ + d←θ, not d←(ψ+θ). The entanglement distance only depends on the norm squared, so the final E is unaffected, but the displayed expectation value is wrong. Third, the \"topological nature\" claim is overstatement; degree distribution is graph-isomorphism invariant, not topological in any standard sense. Fourth, the formula assumes a uniform coupling angle θ per edge; with different angles the degree-only dependence fails, which the paper does not state. Novelty is incremental: it is a short algebraic consequence of earlier ED work applied to standard graph-state purity calculations, and the directed-graph framing adds no new physics because orientation cancels in the disjoint case.\n\nWho is this for? Someone working with graph states and this specific ED measure might find the restricted formula useful, after the missing assumption is added. It does not resolve an open problem. I would not cite it in its current form. It does deserve a serious referee because the error is subtle and fixable, and the restricted result is correct. My recommendation: send to peer review, but require the no-reciprocal-edges assumption to be stated explicitly, correct Eq. (24), and soften the topological claim.","headline":"Theorem 1 is false as stated: reciprocal edges are allowed by the definitions but break the degree-only formula, though the restricted disjoint case is correct and fixable.","tokens_in":6560,"tokens_out":9199,"would_cite":false,"duration_ms":79553,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["03.67.Mn"],"model":"deepseek-v4-flash","headline":"Entanglement in qubit networks is set by each node's link count, not by link direction.","keywords":["entanglement distance","graph states","directed graphs","vertex degree distribution","Fubini–Study metric","multiqubit entanglement","quantum networks"],"falsifier":"Evaluate Eq. (3) numerically for a small ordering-free directed graph state — for example, the directed path $1\\to 2\\to 3\\to 4$ with $\\theta=\\pi/3$ — and compare each vertex's contribution with $1-[\\cos(\\pi/3)]^{2d(i)}$; any mismatch at any vertex would disprove the degree-only formula, while agreement across many random small digraphs would confirm the theorem.","tokens_in":5548,"feed_emoji":"🔗","tokens_out":10463,"duration_ms":95887,"temperature":0.7,"pith_summary":"The paper studies entanglement in multiqubit states built from directed graphs, where every oriented edge applies the same two-qubit controlled unitary to the two qubits it joins. It proves a closed formula for the Entanglement Distance per qubit: for a graph with $M$ vertices and initial state $|+\\rangle^{\\otimes M}$, the measure equals $1 - \\frac{1}{M}\\sum_i [\\cos\\theta]^{2d(i)}$, with $\\theta$ the rotation angle of the edge unitary and $d(i)$ the total number of edges incident to vertex $i$. The consequence is that entanglement is fully determined by the vertex degree distribution: edge orientation drops out, incoming and outgoing links contribute identically, and relabeling vertices does not change the measure. A reader would care because this turns a seemingly direction-sensitive network quantity into a purely topological one, making predictions and design for quantum networks built from graph states simpler.","feed_headline":"Entanglement in qubit networks is set by each node's link count","feed_subtitle":"Edge direction drops out; the entanglement formula depends only on vertex degree distribution.","key_machinery":"The central object is the Entanglement Distance per qubit, $E(|\\psi\\rangle) = 1 - \\frac{1}{M}\\sum_i \\|\\langle\\psi|\\boldsymbol{\\sigma}^{(i)}|\\psi\\rangle\\|^2$, an entanglement measure obtained from the Fubini–Study metric on projective Hilbert space. The mechanism that carries the proof is the controlled unitary $U_{ab} = \\Pi_0^{(a)} I^{(b)} + \\Pi_1^{(a)} \\bar{U}^{(b)}$ with $\\bar{U}^{(b)} = e^{-i\\psi}\\,\\mathrm{diag}(e^{i\\theta}, e^{-i\\theta})$, applied identically to every edge. Because all these unitaries commute and the initial state is $|+\\rangle^{\\otimes M}$, each incident edge contributes one factor of $\\cos\\theta$ to the transverse Bloch components of the control qubit while leaving the $z$-component unchanged; an induction on symmetric projectors $P_k^{(n)}$ handles the incoming-link case, and combining the two sets of links yields the final degree-only power $[\\cos\\theta]^{2d(i)}$.","core_discovery":"The central claim is Theorem 1: for an ordering-free directed graph state $|G\\rangle$ built from the product state $|+\\rangle^{\\otimes M}$ and the controlled unitary of Eq. (5), the Entanglement Distance per qubit is exactly $$E(\\$\\theta$;\\{d(i)\\}) = 1 - \\frac{1}{M}\\sum_{i\\in V} [\\cos\\$\\theta$]^{2d(i)},$$ where $d(i)$ is the degree of vertex $i$. The proof splits the contribution of each vertex into the cases of only outgoing links, only incoming links, and a mixture of both; in every case the expectation value of the Pauli vector on that qubit shrinks by a factor $[\\cos\\theta]^{d(i)}$, and squaring that length is what the Entanglement Distance subtracts from 1. Consequently the measure cannot tell an incoming edge from an outgoing edge, and depends on the graph only through its degree distribution.","pith_inferences":["If the equal-coupling assumption were relaxed, the same vertex-by-vertex calculation suggests the per-vertex entanglement would become $E_i = 1 - \\prod_{e\\ni i} \\cos^2\\theta_e$, a weighted-degree quantity rather than a function of the degree alone.\n","Because the formula sees only degrees, two non-isomorphic graphs with the same degree sequence — for instance a six-cycle and two disjoint triangles — would be assigned exactly the same Entanglement Distance, so the measure cannot distinguish all network topologies.\n","The degree-only law offers a direct experimental probe: prepare a small graph state, measure the three Pauli expectation values per qubit, and check the predicted $[\\cos\\theta]^{2d(i)}$ decay; systematic deviations would signal non-ideal or non-identical edge interactions.\n"],"forward_implications":["The Entanglement Distance of any ordering-free directed graph state can be computed directly from the degree sequence: $E = 1 - \\frac{1}{M}\\sum_i [\\cos\\theta]^{2d(i)}$.\n","Reversing any edge, or any subset of edges, leaves all vertex degrees unchanged, so the entanglement of the graph state is invariant under changing edge directions.\n","Vertex relabeling is a symmetry of the measure, exactly as stated in the paper's topological-invariance claim.\n","For a $k$-regular graph state, $E = 1 - [\\cos\\theta]^{2k}$, so the entanglement per qubit increases monotonically with both the degree $k$ and the coupling angle $\\theta$.\n","An isolated vertex ($d(i)=0$) contributes zero to the entanglement sum, consistent with a fully separable qubit.\n"],"supporting_citations":[{"why":"Supplies the definition of Entanglement Distance per qubit, Eq. (3), which the theorem evaluates for graph states.","marker":"[6–8]"},{"why":"Introduces Entanglement Distance as a Fubini–Study-derived entanglement measure, motivating Eq. (3).","marker":"[3–5]"},{"why":"Provides the graph-state construction and prior entanglement analyses that the paper extends to directed, ordering-free graphs.","marker":"[9–13]"},{"why":"Gives the Fubini–Study metric underlying the Entanglement Distance.","marker":"[1,2]"}],"fun_headline_variants":["Entanglement ignores edge direction in directed graph states","Graph-state entanglement: only node degree counts","Directed graph entanglement: direction is irrelevant","Entanglement distance depends only on vertex degrees"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"Every edge carries exactly the same two-qubit interaction, with a single coupling angle $\\theta$ (and phase $\\psi$); if different edges had different coupling angles, the contribution of a vertex would involve a product of cosines over its incident edges rather than a power of one cosine, and the degree-only formula would fail.","fun_headline_variants_meta":{"raw":{"variants":["Entanglement ignores edge direction in directed graph states","Graph-state entanglement: only node degree counts","Directed graph entanglement: direction is irrelevant","Entanglement distance depends only on vertex degrees"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000792,"raw_usage":{"total_tokens":3435,"prompt_tokens":837,"completion_tokens":2598,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":453,"completion_tokens_details":{"reasoning_tokens":2543}},"tokens_in":453,"tokens_out":2598,"duration_ms":17784,"temperature":1.0,"reasoning_tokens":2543,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T21:07:01.445989+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Evaluate Eq. (3) numerically for a small ordering-free directed graph state — for example, the directed path $1\\to 2\\to 3\\to 4$ with $\\theta=\\pi/3$ — and compare each vertex's contribution with $1-[\\cos(\\pi/3)]^{2d(i)}$; any mismatch at any vertex would disprove the degree-only formula, while agreement across many random small digraphs would confirm the theorem.","supporting_citations":[],"review_version":1}