{"id":"d5f07cc7-9f39-4e50-b22e-4cfe558884af","arxiv_id":"2607.24129","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A four-copy fixed-core architecture with five local SU(3)⊗SU(3) layers is shown to be locally universal for two-qutrit gates, via an explicit Clifford core that makes the differential an exact isometry.","lead":"This paper proves that a minimal circuit with four copies of a fixed two-qutrit gate and five adjustable layers can be locally universal, giving an explicit Clifford core that certifies full-rank reachability. It also shows that complex-symmetric fixed cores can never get such a certificate at the identity point, and it presents a numerical superconducting-style core that synthesizes 1000 random targets with high fidelity.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The Clifford-core certificate rests on Lemma 1's hand-checked 80-label partition; a single typo in the printed blocks or in S_KCl would break the rank-80 claim, and no machine check or code is provided.","rationale":"The central claim is Theorem 2: a specific Clifford core gives local surjectivity at the identity local point. The proof has a clean logical decomposition. The inverse-function-theorem step (Theorem 1) is standard: full rank of the right-trivialized differential at L* implies an open image neighborhood of Φ(L*). The nontrivial input is the computation that rank A_KCl(I)=80, which is established by the Pauli-label splitting criterion and Lemma 1. I checked representative entries in the printed blocks and the first transported block S(Λloc); the criterion is mathematically sufficient because disjoint label blocks give Hilbert-Schmidt-orthogonal real subspaces of the correct dimension. The remaining weak point is purely the lack of an independent, machine-executable verification of the finite partition. Since the manuscript prints all 80 labels, a referee can check this quickly, so this is a verifiability condition rather than a demonstrated mathematical error. I agree with the reader's weakest_assumption. One minor presentation slip: Proposition 6's proof writes τ(Y):=Y^T=-Y, but the subsequent eigenspace dimensions show the intended involution is τ(Y)=Y^T; the bound argument goes through with that reading, so this does not affect the central certificate. The numerical sections are explicitly finite-precision and do not bear the same weight. Recommendation: keep the CONDITIONAL verdict; the condition is exactly to publish or verify the finite-field check, and ideally the data archives. If the check passes, ACCEPT would be appropriate.","tokens_in":35793,"tokens_out":19156,"duration_ms":177422,"concrete_test":"Run a 20-line script (SageMath/GAP or plain Python over F_3) that takes S_KCl from Eq. (18) and Λloc from Eq. (14), computes Λ_t=S_KCl^t(Λloc) for t=0..4, and asserts: |Λ_t|=16 for each t; Λ_i∩Λ_j=∅ for i≠j; ∪_t Λ_t = F_3^4\\{0}; S_KCl^5=I_4; and S_KCl^T Ω S_KCl=Ω for the standard symplectic form. Separately compare the generated blocks against the printed lists (A1)-(A5) to rule out transcription errors. If all assertions pass, the finite certificate behind Theorem 2 is confirmed.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing step is Appendix A, Lemma 1. Theorem 2 reduces rank A_KCl(I)=80 exactly to the assertion that the five blocks Λ_t=S^t_KCl(Λloc), t=0,...,4, form a disjoint partition of V\\{0}=F_3^4\\{0}. The proof is a \"direct finite-field check\" of the printed lists (A1)-(A5), and no accompanying code or machine-readable verification is shipped; the data-availability statement offers archives only \"upon reasonable request.\" A single duplicated or missing label in those lists, or a transcription error in S_KCl in Eq. (18), would make the five transported local Pauli subspaces fail to be mutually orthogonal and dimension-additive, so the rank-80 certificate and the exact isometry of Corollary 1 would collapse. The surrounding geometric argument (Proposition 1, Theorem 1, Corollary 1) is standard and does not share this fragility; the finite partition is the one unmechanized, load-bearing computation. The check is small enough that it is likely correct, but it is exactly the step that would settle the claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the four-core fixed two-qutrit ansatz Φ_K(L1,...,L5) = L5 K L4 K L3 K L2 K L1, with K ∈ SU(9) fixed and five local layers in L = SU(3)⊗SU(3). Since 5 dim L = 80 = dim SU(9), this is the shortest fixed-core architecture not ruled out by parameter counting. The authors formulate a right-trivialized differential rank certificate for local universality, construct an explicit Clifford-word core K_Cl, and prove via an explicit five-block Pauli-label splitting that the identity-point differential has full rank 80 and is in fact an exact isometry. They classify 2304 symplectic actions realizing the splitting, prove that every complex-symmetric core has identity-point differential rank at most 78, and numerically study a superconducting asymmetric-drive core, reporting full-rank sampled Jacobians, 1000/1000 Haar-random target synthesis at F_avg ≥ 0.999 under a stated restart protocol, and additional robustness and structured-target diagnostics. The paper carefully separates exact statements from finite-precision numerical evidence and explicitly notes that global surjectivity remains open.","tokens_in":36000,"tokens_out":11058,"duration_ms":106834,"significance":"Assuming the finite certificate and a corrected proof of Proposition 6, this is a clean and nontrivial answer to a natural minimal-depth question: at exactly saturated parameter count, local universality can be certified for an explicit Clifford core, while a structurally important symmetric family is provably obstructed at the identity local point. The constructive Clifford certificate, the classification of good symplectic actions, the exact reachability of the local subgroup, and the clear separation of exact versus numerical claims are genuine strengths. The paper is also unusually explicit in the supporting appendices. The main weaknesses are that the central finite-field partition is not machine-verified and that the symmetric-core proof contains algebraic misstatements as printed; these are load-bearing but appear repairable.","major_comments":[{"comment":"The proof as printed contains two false algebraic statements. On su(9), the map τ(Y) := Y^T is not equal to −Y: for Y = iR with R real symmetric traceless, Y^T = Y, whereas for real antisymmetric Y, Y^T = −Y. In addition, a complex-symmetric unitary need not be Hermitian, so the sentence \"If K = K^T, then K^† = K\" is false. The subsequent identity τ(Ad_K X) = Ad_{K^{-1}}(τX) does follow from K = K^T alone, so the rank-78 bound appears salvageable, but the proof must be rewritten with τ defined as transpose and without the Hermiticity claim.","section":"Appendix C, Proposition 6"},{"comment":"The exact certificate in Theorem 2 is reduced entirely to the assertion that the five explicitly printed 16-label blocks (A1)–(A5) partition F_3^4 \\ {0}, and the proof is a \"direct finite-field check\" with no accompanying code or machine-readable table. A single duplicated or missing label, or a transcription error in S_KCl in Eq. (18), would invalidate the rank-80 and isometry claims. I am not asserting that the lists are wrong; I am asserting that the load-bearing finite arithmetic should be independently verifiable. Please include a small verification script or an electronic version of the tables as ancillary material, and report the output of the check explicitly.","section":"Appendix A, Lemma 1"}],"minor_comments":[{"comment":"Given that several numerical claims (the 1000/1000 synthesis benchmark, the regular-pair archive, and the robustness scans) are central to the experimental part of the paper, the statement that data are available \"upon reasonable request\" is too weak; an archival repository or DOI should be provided.","section":"Data availability statement"},{"comment":"The proof of Lemma 2 begins by assuming that every 16-element sign-closed block disjoint from B_0 has the form (W\\{0}) ⊔ (W^⊥\\{0}) for a nonsingular plane W; since \"block\" is not defined before this point, please clarify whether this is part of the definition of a block in this context and, if not, justify the structural claim.","section":"Appendix B, Lemma 2"},{"comment":"The phrase \"numerical rank floor\" is used without stating the singular-value tolerance; please state the threshold used when reporting that no sampled rank collapse is observed.","section":"Section III C and Appendix D"},{"comment":"The remark on degree theory is useful, but the sentence \"Φ_I has image contained in L and hence has degree zero\" deserves a one-sentence explanation that L is a proper closed subgroup, so the map is not surjective on homology in degree 80.","section":"Appendix C"}],"recommendation":"major_revision","confidential_remarks":"The Proposition 6 proof errors are local and easily corrected, and the finite-check concern can be addressed by adding a small verification code or machine-readable table. If the authors supply a corrected proof and an electronic verification of Lemma 1, I expect the paper to be publishable; I do not see grounds for rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a real result, not a parameter-counting fantasy. The authors prove that a four-copy fixed-core ansatz for two-qutrit synthesis—five local SU(3)⊗SU(3) layers with a fixed core—is locally universal at the identity point for an explicit Clifford-word core, with an exact isometry certificate and condition number one. They classify all 2304 symplectic actions with the splitting and prove a genuine obstruction: complex-symmetric cores (including real-Hamiltonian cores) have rank at most 78 at the identity point. If the finite-field partition is right, Theorem 2 is a proof, not a numerical hint.\n\nThe exact part is good and mostly standard Lie theory plus finite geometry. Theorem 1 is the inverse function theorem; the Clifford construction reduces to a finite splitting of the 80 nonzero Pauli labels into five transported 16-label blocks. The appendices do a real job: the 2304 count, the local double-coset equivalence, and the explicit stabilizer sequence are sketched tightly enough to check. The rank-78 bound for K=K^T is elegant and correctly flags the odd-number-of-blocks mechanism. I also give credit for labeling the numerical sections as finite-precision diagnostics and not claiming global surjectivity.\n\nThe soft spot is exactly where the stress-test note points. Lemma 1 is the load-bearing step and its proof is a 'direct finite-field check' of printed lists. That is checkable by hand, and I did not find an error, but the resilience of the paper should not depend on the reader's patience. There is no machine-checkable verification or shipped code; data is only on request. For a result whose entire certificate is finite arithmetic, this is a reproducibility gap, and it is easy to fix: provide the script or the Lean/Coq proof, or at minimum the raw lists in machine-readable form. A second minor issue: in Proposition 6's proof, τ(Y)=Y^T=-Y should read τ(Y)=Y^T for the real-linear involution; the argument is otherwise clear. The numerical sections are honestly labeled; the 1000/1000 F≥0.999 result and restart curves are empirical, not claims of uniform success. I would not treat the superconducting core as anything more than a case study.\n\nWho should read it: anyone working on qudit synthesis, Clifford-based decompositions, or fixed-entangling-gate hardware protocols. It deserves a serious referee. My recommendation: send to peer review, and ask for the finite-field partition to be independently verified as a condition of acceptance.","headline":"Exact local-universality certificate at the parameter-counting-minimal four-core depth, with an honest separation of exact and numerical claims; the only load-bearing step that needs mechanization is the hand-checked 80-label finite-field partition.","tokens_in":36557,"tokens_out":2019,"would_cite":true,"duration_ms":24366,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["22E70","81P68","20G40"],"pacs":[],"model":"deepseek-v4-flash","headline":"Four fixed two-qutrit cores, interleaved with five adjustable local layers, are shown to be locally universal at the minimal parameter-saturating depth.","keywords":["two-qutrit gates","local universality","fixed-core ansatz","Clifford group","Pauli-label splitting","symplectic group over F3","differential rank certificate","quantum gate synthesis"],"falsifier":"Independently enumerate $\\Lambda_t = S_{K_{\\rm Cl}}^t(\\Lambda_{\\rm loc})$ for $t=0,\\dots,4$ using the displayed symplectic matrix and check that the 80 labels are pairwise disjoint and exhaustive; equivalently, compute all 80 singular values of the identity-point differential matrix $M_{K_{\\rm Cl}}(I)$ and look for any value that is not exactly 1. A repeated or missing label, or any singular value deviating from one at machine precision, would falsify the exact local-universality certificate.","tokens_in":35584,"feed_emoji":"⚛️","tokens_out":8485,"duration_ms":75345,"temperature":0.7,"pith_summary":"The paper asks how short a fixed-hardware circuit can be and still be universal for two-qutrit gates, where a qutrit is a three-level quantum system. It studies the tightest possible architecture: four copies of a fixed non-tunable two-qutrit core interleaved with five adjustable local layers, whose parameter count exactly balances the 80 real dimensions of the two-qutrit unitary group. The central claim is that this minimal depth is genuinely attainable: an explicit Clifford-word core makes the identity-point differential an exact isometry of rank 80, so the reachable set contains an open neighborhood of a Clifford gate. A companion theorem says that no complex-symmetric core can be certified at the identity point, since its rank there is at most 78, steering hardware designs toward asymmetric drives. A superconducting-motivated asymmetric core is then shown numerically to synthesize all tested random targets, with the exact Clifford certificate kept distinct from finite-precision evidence.","feed_headline":"Four fixed qutrit gates open every local direction","feed_subtitle":"A Clifford core gives an exact 80-of-80 rank certificate at the minimal four-core depth for two-qutrit synthesis.","key_machinery":"The load-bearing object is the right-trivialized differential $A_K(L)(X_1,\\ldots,X_5) = \\sum_{i=1}^5 \\mathrm{Ad}_{P_i(L)}(X_i)$ of the fixed-core map, which at the identity local point reduces to $\\sum_{i=1}^5 \\mathrm{Ad}_{K^{5-i}}(X_i)$. Full rank 80 of this linear map is equivalent, via the inverse function theorem, to local surjectivity of the four-core synthesis map. For Clifford cores, conjugation by the core acts as a symplectic linear map $S_K$ on the two-qutrit Pauli labels over $\\mathbb{F}_3^4$; the Pauli-label splitting criterion asks that the five transported copies of the local label block partition the 80 nonzero labels, which forces the corresponding tangent subspaces to be mutually orthogonal and makes every singular value equal to one. The companion obstruction uses the transpose involution $\\tau(Y) = Y^{\\mathsf T}$ on $\\mathfrak{su}(9)$, whose positive eigenspace has dimension 44 while the local subalgebra contributes only 10 dimensions, forcing the rank bound of 78 for complex-symmetric cores.","core_discovery":"In the paper's own terms, the discovery is Theorem 2 together with Proposition 6. For the Clifford-word core $K_{\\rm Cl}$, the four-core map $\\Phi_{K_{\\rm Cl}}(L_1,\\ldots,L_5) = L_5 K_{\\rm Cl} L_4 K_{\\rm Cl} L_3 K_{\\rm Cl} L_2 K_{\\rm Cl} L_1$ has full-rank differential at the identity local point: the five transported Pauli-label blocks $\\Lambda_t = S_{K_{\\rm Cl}}^t(\\Lambda_{\\rm loc})$ partition all 80 nonzero labels, so the five transported copies of the local Lie algebra span $\\mathfrak{su}(9)$ in mutually orthogonal Hilbert–Schmidt subspaces. Consequently $\\Phi_{K_{\\rm Cl}}$ is locally surjective near the identity local point, and the differential there is an exact isometry with all 80 singular values equal to one. In contrast, every complex-symmetric core $K = K^{\\mathsf T}$ has identity-point differential rank at most 78, so no such core admits an identity-point full-rank certificate; the bound also applies to any core locally conjugate to a complex-symmetric core. The paper further classifies exactly 2304 symplectic actions satisfying the same splitting criterion, shows they are all two-sided locally equivalent, and supplies a hardware-motivated asymmetric core that escapes the symmetric-core obstruction and passes the reported sampled numerical tests.","pith_inferences":["A design heuristic follows from the paper: for dimension-saturating fixed-core synthesis, one could optimize the core's Hamiltonian against the smallest singular value of the identity-point differential rather than against a single-target fidelity; the paper makes the geometry computable, so this optimization loop is directly testable on hardware models.","If the numerical regular-pair findings survive interval verification, every observed fold-like critical value is covered by a regular preimage, and the paper's own closed-image plus regular-preimage criterion would point toward global surjectivity of the Clifford-core map, a result the paper leaves open.","The transpose-involution obstruction likely has an analogue in other qudit systems: an odd number of local blocks flanking a transpose-invariant core loses at least two tangent dimensions at the identity point, so symmetry-breaking should be a general design requirement for minimal-depth fixed-core synthesis rather than a special feature of two qutrits.","The exact identity-point isometry suggests the Clifford core provides a well-conditioned coordinate chart near that point; one could test whether the channel capacity of the ansatz across random targets correlates with the conditioning of $A_K(I)$ over the whole 2304-core family, whereas the paper fixes a single representative."],"forward_implications":["Four non-tunable two-qutrit cores are enough in principle: the image of the Clifford-core map contains an open neighborhood of the Clifford gate $K_{\\rm Cl}^{-1}$, so the minimal depth allowed by parameter counting achieves local universality rather than merely being formally possible.","Identity-point regularity is generic: because one core has nonzero determinant for the differential matrix, Haar-almost every core has full rank at the identity local point, with only a measure-zero exceptional set of cores failing there.","Complex-symmetric cores, including every core generated by a constant real-symmetric Hamiltonian, cannot be certified at the identity point; breaking this symmetry, for example with a noncommuting time-dependent drive, is necessary for an identity-point certificate.","All 2304 good Clifford-word cores are equivalent up to multiplication by local gates before and after the core, so they define the same reachable set; the inversion symmetry additionally implies that every purely local two-qutrit gate is exactly reachable by the same four-core Clifford architecture.","A hardware-motivated asymmetric superconducting core escapes the symmetric-core obstruction and reaches average fidelity at least 0.999 for all 1000 Haar-random targets tested under the stated restart protocol, while structured exchange-like targets remain below the threshold under the same fixed budget."],"supporting_citations":[{"why":"Establishes the Clifford-group to symplectic-group correspondence for higher-dimensional systems, letting a Clifford core act by conjugation on Pauli labels over the finite field.","marker":"[34]"},{"why":"Supplies the qutrit Clifford and stabilizer conventions and the modular-arithmetic description used to set up the Pauli-label splitting criterion.","marker":"[35]"},{"why":"Supplies the finite projective geometry language used to analyze blocks of Pauli labels and nonsingular-pair spreads.","marker":"[38]"},{"why":"Provides the stabilizer theorem for nonsingular-pair spreads used to count the 2304 good symplectic actions and prove their two-sided local equivalence.","marker":"[39]"},{"why":"Supplies the inverse function theorem that converts a full-rank differential into local surjectivity, giving Theorem 1.","marker":"[40]"},{"why":"Provides the numerical solver used to generate the time-ordered propagator of the hardware-motivated superconducting core.","marker":"[37]"},{"why":"Defines the average gate fidelity measure used to score the numerical synthesis benchmarks.","marker":"[36]"}],"fun_headline_variants":["Qutrit Clifford core proves local universality at minimal depth","Exact 80-dim certificate for two-qutrit fixed-core synthesis","Asymmetric core escapes symmetry bound for qutrit universality","Four fixed qutrit cores: full-rank differential at identity","Minimal qutrit ansatz: Clifford core gives isometric Jacobian"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The exact Clifford certificate rests on Lemma 1's assertion that the five explicitly listed label blocks partition all 80 nonzero labels, and the proof is a direct finite-field check of the displayed lists rather than a machine-verified computation, so an unnoticed duplicate or omission would break the rank-80 isometry claim.","fun_headline_variants_meta":{"raw":{"variants":["Qutrit Clifford core proves local universality at minimal depth","Exact 80-dim certificate for two-qutrit fixed-core synthesis","Asymmetric core escapes symmetry bound for qutrit universality","Four fixed qutrit cores: full-rank differential at identity","Minimal qutrit ansatz: Clifford core gives isometric Jacobian"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000338,"raw_usage":{"total_tokens":1990,"prompt_tokens":1186,"completion_tokens":804,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":802,"completion_tokens_details":{"reasoning_tokens":713}},"tokens_in":802,"tokens_out":804,"duration_ms":7933,"temperature":1.0,"reasoning_tokens":713,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T15:27:36.851486+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Independently enumerate $\\Lambda_t = S_{K_{\\rm Cl}}^t(\\Lambda_{\\rm loc})$ for $t=0,\\dots,4$ using the displayed symplectic matrix and check that the 80 labels are pairwise disjoint and exhaustive; equivalently, compute all 80 singular values of the identity-point differential matrix $M_{K_{\\rm Cl}}(I)$ and look for any value that is not exactly 1. A repeated or missing label, or any singular value deviating from one at machine precision, would falsify the exact local-universality certificate.","supporting_citations":[{"cited_title":"Kjaergaard, M","cited_arxiv_id":null,"evidence_quote":"Supplies the inverse function theorem that converts a full-rank differential into local surjectivity, giving Theorem 1."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the average gate fidelity measure used to score the numerical synthesis benchmarks."}],"review_version":2}