{"id":"46da1e94-d24d-40ad-ac3d-95b0f6dd8e07","arxiv_id":"2607.28521","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Ironed two-qubit gadgets with a=5/9 match the iSWAP second-moment spectral gap on K_n (n≥5) because the largest negative eigenvalue of A_n always sits in the highest-spin sector.","lead":"Every ironed two-qubit gadget with KAK parameter a=5/9 has the same second-moment spectral gap as iSWAP on the complete graph for n≥5. The proof localizes the decisive eigenvalue to the highest-spin SU(2) sector and settles a Kong–Li–Liu conjecture on random-circuit designs.","discovery_kind":"extension","skeptic_critique":{"model":"grok-4.5","headline":"No significant objection identified","rationale":"The reader correctly isolates the only non-self-contained step (the cited complementary-subspace reduction) while judging the novel localization argument sound. Independent re-derivation of the Sturm ratio identity and spot-checks of the variational determinant sign and the positive-coefficient polynomials in s=n-9 reveal no algebraic fracture. The finite certificates are fully explicit and therefore auditable; the c-independence argument once localization is granted is immediate from B_n≤0 on lower spins. No stronger internal concern surfaces, so the ACCEPT / HIGH verdict is unchanged.","tokens_in":15279,"tokens_out":534,"duration_ms":57085,"concrete_test":"Recompute the four leading-principal-minor sequences of L_{n,1}-\theta_n I and the Sturm sequences of L_{n,0}-\theta_n I in Appendix A (n=5…8) by an independent computer-algebra session (exact rationals); confirm every listed entry and that the sign patterns remain exactly two changes (L_{n,0}) and all-positive (L_{n,1}). Any mismatch would falsify the finite-n half of Thm 4.5.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central localization (Thm 1.2/4.5) and the ensuing c-independent gap formula (Thm 1.1/5.4) hold up under scrutiny. The three new pieces—local rank-two PSD floor separating r≥2 (Prop 3.3), exact 2-D Rayleigh certificate for H°_{n,0} (Lem 4.1), and inductive Sturm/Sylvester argument plus rational certificates for H_{n,1} (Lem 4.3–4.4, App A)—are internally consistent; the key algebraic identity behind the Sturm ratio induction checks out exactly, and the threshold (3n-5)/N_n is compatibly tuned to all three regimes. Residual dependence on the external Kong–Li–Liu complementary-subspace reduction (Prop 5.3) is real but ordinary: the paper correctly verifies the numerical hypothesis via Lem 2.2 (λ_min(T^{IG}_2)≥-1/3 ⇒ n≥3 suffices), so the full-operator claim stands or falls with that prior lemma, not with a gap in the new argument.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.5","summary":"The paper proves that every ironed two-qubit gadget with KAK parameter a=5/9 has the same second-moment spectral gap as the iSWAP gadget on the complete graph K_n for all n≥5, thereby settling Conjecture 2 of Kong, Li, and Liu. The argument proceeds by Schur–Weyl reduction of the second-moment operator to spin blocks of an S_n-invariant operator A_n, followed by a localization theorem (Theorem 1.2/4.5): the largest strictly negative eigenvalue of A_n lies in the highest-spin SU(2) summand. A local rank-two PSD decomposition separates all sectors with r≥2; the remaining comparison between the highest-spin and next-to-highest-spin Jacobi matrices is obtained by a two-dimensional Rayleigh–Ritz upper bound after projecting out the zero mode, matched against an inductive Sturm/Sylvester lower bound (with exact rational certificates for 5≤n≤8). Once localization is established, the admissible range 0≤c≤1/3 and the sign of B_n make the gap independent of c, and a cited complementary-subspace reduction identifies it with the gap of the full moment operator.","tokens_in":15405,"tokens_out":896,"duration_ms":26218,"significance":"The result cleanly resolves a concrete conjecture in the spectral theory of random quantum circuits and ironed gadgets. The technical core—localization of the decisive eigenvalue to the highest-spin block—is of independent interest and is aligned with Aldous-type phenomena. Strengths include fully explicit Jacobi matrices, a parameter-free common threshold (3n−5)/N_n tuned simultaneously to the variational, Sturm, and r≥2 regimes, and machine-checkable exact rational Sturm certificates in Appendix A for the finite cases. The c-independence argument is sharp and identifies CNOT (c=0) as the least favourable case. Residual reliance on the Kong–Li–Liu complementary-subspace lemma is ordinary and is correctly reduced to a verified numerical hypothesis (λ_min(T^{IG}_2)≥−1/3).","major_comments":[],"minor_comments":[{"comment":"Throughout the manuscript (including the title and abstract) the string “iSW AP” appears with an internal space; likewise “theironed gadgetmodel” and similar broken compounds. These should be normalized to “iSWAP”, “ironed gadget model”, etc., before publication.","section":"Title, Abstract, passim"},{"comment":"In Corollary 5.5 the B-gate is written “Bgate” without a hyphen or space; match the earlier “B-gate” usage of Section 1 and [18, Table 3].","section":"Corollary 5.5"},{"comment":"Lemma 2.2 records the unordered pair of non-unit eigenvalues as −1/9 and 1/3−2c. A one-line remark that these are exactly the eigenvalues appearing in the local matrix (2.5) after setting a=5/9, b=2/9 would help readers who skip the determinant calculation.","section":"Lemma 2.2"},{"comment":"The reference [19] is listed as “in preparation, 2026”. If a preprint identifier becomes available before final proofs, it should be added; otherwise the citation is fine as is.","section":"References"},{"comment":"In the display of the Sturm sequences in Appendix A, the n=5 sequence for E^{(n)}_k mixes fractions with large numerators; adding a brief note that all entries are exact rationals (no floating-point) would reinforce the certificate character of the appendix.","section":"Appendix A"}],"recommendation":"accept","confidential_remarks":"The paper is a solid, self-contained resolution of a stated conjecture, with the new analytic work cleanly separated from the cited complementary-subspace reduction. Fit for a technical quant-ph / mathematical-physics venue is good. I see no novelty or citation-pattern concerns."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"The one thing to know: they actually prove the missing highest-spin localization for A_n itself, not just after adding the iSWAP B_n term. That is what Kong–Li–Liu needed and could not get. Once you have Theorem 1.2, c drops out for free on the complete graph and the conjecture falls.\n\nWhat is new is the three-piece comparison. Local rank-two PSD floor kills every r≥2 sector at 3/n. A two-dimensional Rayleigh–Ritz after projecting out the zero mode gives a strict upper bound on the first positive eigenvalue of the highest-spin Jacobi block. An inductive Sturm/Sylvester argument (plus four exact rational certificates in the appendix) puts the r=1 block strictly above the same rational threshold. The algebra checks; the threshold (3n−5)/N_n is chosen so the first pivots and the coarse floor all line up, not fitted to numerics. That is honest finite-dimensional spectral work.\n\nSoft spots are ordinary, not load-bearing. The upgrade from the V^{⊗n} restriction to the full moment operator still rides on Kong–Li–Liu’s complementary-subspace lemma; the authors only verify the numerical hypothesis via λ_min ≥ −1/3. If that prior reduction is wrong the full-operator claim fails, but the new localization stands on its own. Scope is narrow: complete graph, second moment, a=5/9 only. It does not reorganize design theory; it closes one concrete comparison among gate families.\n\nCitations look standard for the subfield. Math is written out, small-n cases are exact, no free parameters. This is for people who already care about second-moment gaps of ironed gadgets or Schur–Weyl reductions of all-to-all circuits. I would send it to a serious referee without hesitation. Worth reading if you work on random-circuit gaps; skip if you only want architecture-level depth bounds.","headline":"Clean proof that settles the Kong–Li–Liu conjecture: localization of the A_n gap to highest spin, hence c-independent iSWAP gap for all a=5/9 gadgets on K_n.","tokens_in":16157,"tokens_out":513,"would_cite":true,"duration_ms":16387,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81P45","05C50","15A18","60J10"],"pacs":[],"model":"grok-4.5","headline":"Every ironed two-qubit gadget with KAK parameter a=5/9 has the same second-moment spectral gap as iSWAP on the complete graph for n≥5.","keywords":["unitary designs","spectral gaps","Schur–Weyl duality","Jacobi matrices","Sturm sequences","ironed gadgets","iSWAP","second-moment operators"],"falsifier":"For a fixed n≥5, compute the second-largest eigenvalue of the full second-moment operator for CNOT (c=0) and for iSWAP (c=1/3) on Kn; if those eigenvalues differ, or if either differs from 1−(4/9)λ⁺_min of the explicit highest-spin Jacobi matrix, the central claim is false.","tokens_in":16033,"feed_emoji":"⚛️","tokens_out":1071,"duration_ms":19116,"temperature":0.7,"pith_summary":"This paper proves a conjecture about how fast certain random quantum circuits mix at the second moment. An ironed two-qubit gadget is a fixed two-qubit gate sandwiched between random one-qubit gates; its second-moment behavior is controlled by a few KAK parameters, and the special value a=5/9 covers iSWAP, the B-gate, and CNOT. On the complete graph with at least five qubits, the authors show that every such gadget has exactly the same second-moment spectral gap as iSWAP, independent of the remaining free parameter c. The proof works by localizing the decisive eigenvalue of an associated symmetric-group-invariant operator to the highest-spin SU(2) sector, then comparing the two highest spin blocks with variational and Sturm arguments while a local positive-semidefinite bound rules out all lower spins. A sympathetic reader cares because this pins down which gates set the mixing rate for all-to-all second-moment designs and shows that a whole family collapses to one gap.","feed_headline":"iSWAP gap is shared by every a=5/9 ironed gadget","feed_subtitle":"On complete graphs with n≥5, CNOT and B-gate match iSWAP at the second moment, independent of c.","key_machinery":"A representation-theoretic localisation theorem: after Schur–Weyl decomposition, the largest strictly negative eigenvalue of the Sn-invariant operator An always sits in the highest-spin SU(2) summand. A local PSD decomposition of the two-site term separates all spin sectors with r≥2; the r=0 and r=1 Jacobi blocks are then split across a common rational threshold by a two-dimensional Rayleigh–Ritz upper bound and a Sturm/Sylvester lower bound.","core_discovery":"For every n≥5 and every ironed two-qubit gadget with KAK parameter a=5/9, the second-moment spectral gap on the complete graph Kn equals the iSWAP gap. Equivalently, that gap equals (4/9) times the first positive eigenvalue of an explicit highest-spin Jacobi matrix and does not depend on the admissible parameter c in [0,1/3].","pith_inferences":["The same localisation strategy may decide whether other fixed a-values produce c-independent gaps, or whether a=5/9 is special to the complete-graph architecture.","Because the gap reduces to one explicit tridiagonal eigenvalue, asymptotic large-n expansions of that Jacobi ground state would give concrete depth formulas for all-to-all second-moment designs in this family.","If the complementary-subspace reduction extends below n=5 or to other graphs, the equality of gaps might hold in smaller or sparser architectures without new spin analysis."],"forward_implications":["iSWAP, B-gate, and CNOT ironed gadgets have identical second-moment spectral gaps on every complete graph Kn with n≥5.","At a=5/9 the gap is independent of c, so the least favorable case c=0 already determines the rate for the whole family.","The second-largest eigenvalue of the full moment operator is attained in the highest-spin sector and equals 1−(4/9)λ⁺_min(H°_{n,0}).","Lower-spin competitors are strictly farther from 1 once the localisation theorem holds, so only the highest-spin Jacobi matrix needs to be tracked for the gap."],"fun_headline_variants":["a=5/9 ironed gadgets match iSWAP second-moment gap on Kn n≥5","Every ironed two-qubit gadget with a=5/9 shares the iSWAP gap","Ironed a=5/9 gadgets equal iSWAP spectral gap for n≥5","Localisation shows a=5/9 gadgets match iSWAP gap on complete graphs","Second-moment gaps coincide for a=5/9 ironed gadgets and iSWAP"],"cache_read_input_tokens":128,"weakest_assumption_plain":"The claim that the full moment-operator gap equals the gap on the reduced space V⊗n rests on a prior complementary-subspace reduction that only applies once the local two-qubit operator is not too negative, which the paper checks via a lower bound of −1/3 and n≥5.","fun_headline_variants_meta":{"raw":{"variants":["a=5/9 ironed gadgets match iSWAP second-moment gap on Kn n≥5","Every ironed two-qubit gadget with a=5/9 shares the iSWAP gap","Ironed a=5/9 gadgets equal iSWAP spectral gap for n≥5","Localisation shows a=5/9 gadgets match iSWAP gap on complete graphs","Second-moment gaps coincide for a=5/9 ironed gadgets and iSWAP"]},"model":"grok-4.5","effort":"low","cost_usd":0.005514,"raw_usage":{"total_tokens":1402,"prompt_tokens":670,"num_sources_used":0,"completion_tokens":97,"cost_in_usd_ticks":55144000,"prompt_tokens_details":{"text_tokens":670,"audio_tokens":0,"image_tokens":0,"cached_tokens":128},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":635,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":670,"tokens_out":97,"duration_ms":10275,"temperature":1.0,"reasoning_tokens":635,"cache_read_input_tokens":128,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-31T04:43:49.820358+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"For a fixed n≥5, compute the second-largest eigenvalue of the full second-moment operator for CNOT (c=0) and for iSWAP (c=1/3) on Kn; if those eigenvalues differ, or if either differs from 1−(4/9)λ⁺_min of the explicit highest-spin Jacobi matrix, the central claim is false.","supporting_citations":[],"review_version":1}