{"id":"e62cffdb-3c98-4e55-bd67-01395b120ba7","arxiv_id":"2607.14222","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"At stabilizer states, magic curvature equals quantum Fisher information up to 1/ln2, and for Ising evolution on forests the magic is exactly (L-c(G)) times a single-qubit function, giving exact revivals with finite magic density and zero entanglement density.","lead":"The paper derives exact formulas for 'magic' — the non-Clifford resource needed for fault-tolerant quantum computing — in simple spin-chain dynamics, and shows that at special stabilizer points the initial growth of magic is exactly the quantum Fisher information. This gives a concrete way to detect magic using established metrology measurements, and exactly solvable examples where magic density is finite while entanglement density vanishes.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Central theorems survive scrutiny; residual concern is non-reproducible MPS verification and lengthy un-machine-checked algebra.","rationale":"The reader's CONDITIONAL verdict is reasonable, but the specific weakest_assumption identified by the reader---forest graph plus Clifford-aligned stabilizer input---is not, in my view, a load-bearing flaw: the paper explicitly and accurately delineates this scope in SM S4.E/S4.F and does not claim the formula for bare unrotated generators on arbitrary stabilizer states. The analytic arguments I checked are consistent: the tangent theorem's projector calculation, the CNOT pruning normal form, the one-qubit m_alpha(m2) building block, the open-chain Schmidt decomposition, and the perturbative coefficients A_L and B_L all hold together. The genuinely load-bearing residual issue is the absence of public code and data for the Pauli-MPS numerical verification, together with the lack of machine-checked confirmation of the lengthy S9/S10 algebra. This is a verification gap, not a detected error, so the verdict should remain CONDITIONAL rather than being upgraded or rejected. A single concrete reproducibility check would settle whether this gap actually hides an error.","tokens_in":42815,"tokens_out":43509,"duration_ms":403869,"concrete_test":"Request the authors' MPS code and raw data, then independently rerun: (i) the open-chain h=0 M2(t) curve at L=64 and compare with Eq. (13), and (ii) the revival-lifting scans for the quench and two-kick Floquet revivals, fitting the leading quadratic coefficients and comparing with Eqs. (17) and (18). If the MPS curves match the exact formulas to truncation error and the fitted A_L and B_L agree with the SM derivations, the conditional reservation is discharged.","verdict_should_be":"UNCHANGED","load_bearing_attack":"I cannot identify a load-bearing objection to the central mathematical claims. Theorem 1's proof is sound: the stabilizer projector identity correctly isolates the quadratic term, and the O(theta^3) remainder is legitimate because non-stabilizer Pauli strings enter only at order theta^4. I also checked the generalized tangent argument used in the perturbative sections: for any smooth unitary path through a stabilizer state, the M2 curvature is fixed solely by the tangent generator K, so the use of interaction-picture K_L and K_F is valid even though the paths are not simple exponentials. The forest theorem is internally consistent: the CNOT pruning mechanism works leaf-to-root, and the product formula gives (L-c)m_alpha for the declared representative/conjugated stabilizer inputs. The acknowledged restriction in SM S4.E/S4.F---arbitrary stabilizer inputs with bare Z_iZ_j gates need not obey Eq. (7), and cycles obstruct the product formula---is a scope condition, not a flaw, because the paper's claims are explicitly stated for the aligned family.\n\nThe only concrete concern is evidential rather than logical: the MPS calculations and the long S9/S10 coefficient evaluations (A_L, B_L) are not independently reproducible because no code or data are provided. The paper's confidence rests partly on these numerical benchmarks, so until they can be rerun, a fully unconditional acceptance is not warranted.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper derives two structural results: (i) a tangent theorem showing that at any stabilizer state the quadratic coefficient of the second stabilizer Rényi entropy along e^{-iθK}|s> equals F_Q/ln2; and (ii) a forest theorem evaluating all M_α for commuting Ising evolution on a forest as (L-c(G))m_α(θ), via a Clifford CNOT-pruning normal form. It applies these to the open Ising chain and a kicked-Ising Floquet chain, giving exact finite-size and thermodynamic magic densities, distinct magic/entanglement revival periods, Clifford points with zero magic but finite entanglement, and quadratic lifting of revivals under perturbations. Pauli-basis MPS checks are reported for finite sizes.","tokens_in":43090,"tokens_out":14456,"duration_ms":137103,"significance":"These are clean, parameter-free analytic results with supplied proofs. The one-qubit Pauli distribution behind m_α(θ) and the CNOT edge-parity identity behind the forest theorem are correct, and the leaf-to-root pruning argument is internally consistent. Theorem 1 gives a striking local bridge between a computational resource (magic) and a metrological quantity (QFI), with a concrete experimental protocol. The forest theorem provides exact many-body examples of extensive magic density with vanishing entanglement density, a useful separation result. No fitted constants appear anywhere. The main limitation is that the perturbative revival-lifting sections rely on an extension of the tangent theorem that is not proved in the manuscript, and the claimed numerical verification is not independently rerunnable from the text alone.","major_comments":[{"comment":"Theorem 1 is stated and proven only for paths of the form e^{-iθK}|s>. In the perturbative revival calculations, however, the state after stripping the Clifford factor is a time-ordered exponential Texp(-iη∫H_X(u)du)|+> (quench) or a product of two first-order expansions (Floquet), not e^{-iηK_L}|+> or e^{-iεK_F}|+>. The SM simply says 'Eq. (S35) then applies with θ=η and K=K_L'. This is a gap. What is needed is a lemma: any normalized smooth path through a stabilizer state |s> with first-order tangent -iK|s> has M_2(θ)=(θ²/ln2)F_Q(|s>,K)+O(θ³). The lemma is true — the projector identity plus normalization fixes the contribution of the second-order acceleration term — but it is not stated or proved. Without it, the claimed derivations of Eqs. (17) and (18) are not consequences of Theorem 1 as proven.","section":"SM §S9–S11; Eqs. (17)–(18)"},{"comment":"The paper states that 'large-scale Pauli-basis MPS calculations verify all predictions' and reports benchmarks for the QFI, the quench-lifting coefficient A_L, and the Floquet-lifting coefficient B_L. No code, raw data, or machine-readable tables are provided; 'available upon reasonable request' is not sufficient for independent verification. Since the central analytic results stand alone, this does not affect the theorems, but the numerical verification claim should be reproducible. Please provide a public repository with the code and data, or explicitly identify which numerical checks cannot be independently rerun.","section":"Data Availability; SM §S13"}],"minor_comments":[{"comment":"The typeset formula for m_α(θ) is ambiguous. It should read m_α(θ) = (log₂[1+cos^{2α}(2θ)+sin^{2α}(2θ)] − 1)/(1−α), not log₂[...] − 1/(1−α). Please place the numerator in parentheses.","section":"Eq. (6) and SM Eq. (S86)"},{"comment":"The α=1 limit is mentioned but not displayed. State explicitly m_1(θ) = −(1/2)[cos²(2θ)log₂cos²(2θ)+sin²(2θ)log₂sin²(2θ)], which is the Shannon limit used in the text.","section":"SM §S3.C"},{"comment":"The red thermodynamic curve m_2(ϑ) is used for both the quench and Floquet data with different abscissae (ϑ=Jt and ϑ=kθ). State this in the caption to avoid confusion.","section":"Fig. 1 caption"},{"comment":"The spectral convention for χ''_{KK}(ω) in Eq. (S75) is nonstandard; the sign convention matters. One sentence defining χ'' as the imaginary part of the retarded response with a fixed sign would make the 4/π prefactor unambiguous.","section":"SM §S2.C"}],"recommendation":"major_revision","confidential_remarks":"The two main theorems are sound and the paper is likely to be a valuable contribution after revision. The perturbative-lifting gap is fixable by adding the generalized tangent lemma; the lack of code/data is the main obstacle to full verification. No concerns about novelty or citation practice."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read this one. The two structural results are the real thing. The tangent theorem — at any stabilizer state, the M2 curvature under any Hermitian generator is exactly the QFI divided by ln2 — is clean, novel, and proven honestly in the SM. I checked the argument; the stabilizer-projector identity isolates the quadratic term correctly, and the order-theta^4 suppression of non-stabilizer strings is legitimate. The QFI bridge is a genuine contribution, not just a restatement.\n\nThe forest theorem is also as advertised. The CNOT pruning mechanism is elegant: for commuting Ising evolution on any forest, the whole L-qubit Pauli distribution reduces to L-c(G) identical one-qubit factors. That gives exact quench and Floquet magic, exact revivals, and the finite-magic-with-vanishing-entanglement-density family in the thermodynamic limit. The proofs are parameter-free, with no fitted constants, and the one- and two-qubit calculations I spot-checked match.\n\nThe soft spots are real but proportionate. The forest theorem applies only to the Clifford-aligned family — stabilizer states of the form C|+> with generators conjugated by the same C — and to acyclic graphs. The authors say this explicitly in SM S4.E/S4.F, so it is a scope condition, not a hidden flaw. The headline claims are all stated for that family, so the paper does not overreach. The more concrete concern is reproducibility: the MPS verification is described in detail, but no code or data are deposited, and the data-availability line is the generic 'upon reasonable request.' The long coefficient evaluations for A_L and B_L were also not machine-checked. That is enough to keep the verdict conditional, but it does not touch the central mathematics.\n\nI would engage with this paper, and I would cite it. The theoretical core is solid, the results are substantial, and the limitations are stated by the authors themselves. For peer review: yes, send it to a serious referee. The referee should push for code/data or at least a sharper numerical-error statement, but the paper deserves referee time.","headline":"The tangent and forest theorems are real, exact, and hold up under checking; the only real gap is the non-reproducible MPS numerics.","tokens_in":43583,"tokens_out":1594,"would_cite":true,"duration_ms":19268,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["03.67.-a","03.67.Mn"],"model":"deepseek-v4-flash","headline":"This paper proves that at any stabilizer state the initial growth of magic equals the quantum Fisher information, and that acyclic Ising dynamics yield exact revivals with finite magic but vanishing entanglement density.","keywords":["stabilizer Rényi entropy","nonstabilizerness (magic)","quantum Fisher information","Ising quench","Floquet dynamics","entanglement entropy","Clifford circuits","exact many-body revivals"],"falsifier":"Two concrete tests. (1) Prepare a stabilizer state, evolve briefly under e^{-iθK}, and measure M2 for small θ: if the quadratic coefficient deviates from F_Q(|s⟩,K)/ln2 — with F_Q fixed independently by the variance, the squared-fidelity curvature, or the dynamical susceptibility — the tangent theorem fails. (2) Compute M_α exactly for commuting Ising evolution on a graph containing a closed loop (e.g., a triangle or a square) at a non-Clifford angle such as θ=π/8: agreement with (L−c(G))m_α(θ) would refute the cycle obstruction, while any deviation confirms the forest condition is necessary;","tokens_in":42686,"feed_emoji":"⚛️","tokens_out":10914,"duration_ms":90297,"temperature":0.7,"pith_summary":"Magic is the quantum resource that measures how far a state lies from the stabilizer sector — the states that Clifford circuits can prepare and classically simulate. This paper establishes two exact structural principles about how magic is born and how it behaves in many-body dynamics. First, at every stabilizer state, the leading growth of the second stabilizer Rényi entropy under any Hermitian generator is exactly proportional to the quantum Fisher information, so the metrological sensitivity of a probe is the same object as its initial magic curvature. Second, for commuting Ising evolution on any forest graph, a Clifford pruning circuit collapses the dynamics to independent single-qubit rotations, giving the full magic family exactly for arbitrary size and spatial embedding. Applied to the open Ising chain — as a quench and as a kicked Floquet circuit — this yields exact magic revivals with period π/4, entanglement that revives on a different schedule and vanishes in density, and a precise account of how the revivals lift when the ideal dynamics are detuned.","feed_headline":"Magic growth equals Fisher information at stabilizer states","feed_subtitle":"Exact Ising-chain solutions show magic reviving every π/4 while entanglement density stays zero","key_machinery":"The two load-bearing objects are (1) the tangent identity M2(e^{-iθK}|s⟩) = (θ²/ln2)F_Q(|s⟩,K) + O(θ³), which identifies the quantum Fisher information — four times the variance of the generator, equivalently four times the squared speed in projective Hilbert space — as the exact curvature of magic at a stabilizer point, and (2) the CNOT pruning circuit C_G, a Clifford product of controlled-NOT gates along a leaf-to-root ordering of each tree, which conjugates every Ising edge gate e^{-iθZ_iZ_j} into an independent single-qubit rotation e^{-iθZ_v}. Together they reduce the entire 4^L-term Pauli distribution of the forest state to L−c(G) identical one-qubit factors, making the magic additive","core_discovery":"On its own terms, the paper claims two theorems. Theorem 1 (tangent geometry of magic): for any stabilizer state |s⟩ and any Hermitian generator K, the second stabilizer Rényi entropy along e^{-iθK}|s⟩ obeys M2 = (θ²/ln2) F_Q(|s⟩,K) + O(θ³), with F_Q = 4(⟨K²⟩_s − ⟨K⟩_s²) the quantum Fisher information; the quadratic term comes entirely from the 2^L stabilizer Pauli strings, while all other strings enter only at order θ⁴. Theorem 2 (forest normal form): for commuting Ising evolution exp(−iθΣZᵢZⱼ)|+⟩^L on a forest graph, M_α(ψ_G(θ)) = (L − c(G)) m_α(θ), where m_α is the one-qubit function, because a CNOT pruning circuit carries each edge parity to an independent Z-rotation. From these, the pap","pith_inferences":["Editorial inference: the tangent theorem suggests that near any stabilizer point, quantum metrology and magic are not separate optimizations — a probe that saturates the Fisher-information bound should also maximize early magic growth, which could be tested by comparing QFI extracted from M2 curvature with that from spin-squeezing or susceptibility data on the same device.","Editorial inference: the cycle obstruction is a clean algebraic fact (edge-parity vectors become linearly dependent on loops), so loop graphs can serve as a controlled starting point for perturbation theory: the first corrections to the forest formula should scale with the rank deficiency |E| − (L − c(G)).","Editorial inference: the distinct revival periods give a resource-resolved diagnostic — in a noisy simulator, the ratio of magic to entanglement revival rates could distinguish Clifford-type errors (which leave magic untouched but alter entanglement) from non-Clifford ones.","Editorial inference: conversely, under the tangent bridge, measurements of M2 curvature at short times give a thermodynamic-limit route to Fisher information that does not require accessing all Pauli correlators, which may be useful in large systems where the full distribution is inaccessible."],"forward_implications":["At any stabilizer reference state, measuring the quantum Fisher information through established susceptibility, interferometric, or randomized-measurement protocols directly fixes the leading magic curvature — magic detection without stabilizer tomography.","In these exact models magic and entanglement are dynamically decoupled: magic revives every π/4 of accumulated Ising angle, entanglement every π/2, so the two resources carry independent clocks in a single quantum simulator.","At θ=π/8, g=π/2, one kick reaches maximal magic density, two kicks return exactly to zero magic, four kicks restore the entanglement pattern, and eight kicks return the unitary to the identity — a complete, exact many-body stroboscopic cycle.","The ideal revivals are stable: detuning the field or the kick lifts the magic minimum quadratically with analytic size-dependent coefficients, and the thermodynamic-limit curvature (1/4 + π²/64 and 5/4) matches tensor-network numerics in the Pauli basis.","Because the forest theorem is purely graph-theoretic, the same exact formulas hold for any acyclic interaction graph in any spatial dimension and any embedding, not just one-dimensional chains."],"fun_headline_variants":["Magic curvature equals Fisher information at stabilizer states","Exact Ising-chain solution: magic revives at π/4, entanglement zero","Fisher information origin of magic revivals in Ising chains","Magic without entanglement: exact revivals from Fisher info","Stabilizer Rényi curvature: QFI bridges magic and metrology"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The exact thermodynamic results rest on two structural conditions the paper states explicitly: the interaction graph must be a forest (acyclic), and the state/generator pair must be Clifford-aligned, |s⟩ = C|+⟩^L with gates C ZᵢZⱼ C†; the text and the Supplemental Material (Secs. S4.E–S4.F) note that arbitrary stabilizer inputs driven by bare, unrotated ZᵢZⱼ gates need not satisfy the product formula, and that cycles break it generically, leaving only the weaker zero-magic re","fun_headline_variants_meta":{"raw":{"variants":["Magic curvature equals Fisher information at stabilizer states","Exact Ising-chain solution: magic revives at π/4, entanglement zero","Fisher information origin of magic revivals in Ising chains","Magic without entanglement: exact revivals from Fisher info","Stabilizer Rényi curvature: QFI bridges magic and metrology"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000228,"raw_usage":{"total_tokens":1356,"prompt_tokens":831,"completion_tokens":525,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":575,"completion_tokens_details":{"reasoning_tokens":437}},"tokens_in":575,"tokens_out":525,"duration_ms":5230,"temperature":1.0,"reasoning_tokens":437,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T02:46:03.872287+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Two concrete tests. (1) Prepare a stabilizer state, evolve briefly under e^{-iθK}, and measure M2 for small θ: if the quadratic coefficient deviates from F_Q(|s⟩,K)/ln2 — with F_Q fixed independently by the variance, the squared-fidelity curvature, or the dynamical susceptibility — the tangent theorem fails. (2) Compute M_α exactly for commuting Ising evolution on a graph containing a closed loop (e.g., a triangle or a square) at a non-Clifford angle such as θ=π/8: agreement with (L−c(G))m_α(θ) would refute the cycle obstruction, while any deviation confirms the forest condition is necessary;","supporting_citations":[],"review_version":1}