{"id":"faa77a7b-7132-4180-8de7-2dcfcd89537d","arxiv_id":"2412.01850","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Using an all-to-all product of ZZ gates as a deterministic global unitary, classical shadow tomography achieves ~1.8^k sample complexity for dense k-local Pauli strings.","lead":"Classical shadow tomography estimates quantum state properties from few measurements; this paper designs a deterministic global circuit, the contractive unitary, that cuts the required samples for dense k-qubit Pauli observables from roughly 2^k to roughly 1.8^k. The protocol is tailored to atom-array processors with all-to-all connectivity.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Unknown-location sliding trick only works for operators contained in a length-k window; non-successive size-k operators (span > k) are never aligned, so the claimed k×1.8^k scaling does not cover them.","rationale":"The reader's weakest assumption was operator density (q identities inside a k-qubit chosen window). I read that as less load-bearing because the paper defines operator size as the number of non-identity factors, so the main theorem's 'size precisely equal to k' has q = 0 by definition; the q > 0 case is an explicitly discussed extension rather than a hidden assumption. The genuinely unprotected claim is the sliding trick's universality for non-successive operators. The known-location protocol (U_ct applied to the exact support) handles non-successive geometry because U_ct is all-to-all within the support. But when the location is unknown, the construction only searches consecutive length-k windows. Since a size-k non-successive operator has span > k, it can never be aligned, contradicting the stated 1/k probability. This does not invalidate the core 1.8^k scaling for known, fully supported operators, but it means the abstract's unqualified 'any non-successive local operators' and Table I's unknown-location entry need a scope caveat. A concrete simulation with an alternating-site operator would settle whether the sliding ensemble's shadow norm actually follows k×1.8^k; analytically it cannot. I therefore keep the reader's CONDITIONAL verdict, but for a different reason than the density concern.","tokens_in":16,"tokens_out":50933,"duration_ms":690370,"concrete_test":"For N = 2k with periodic boundary, take O = Z_1 Z_3 Z_5 ... Z_{2k-1} (support size k, span 2k-1). Compute the Pauli weight of the sliding-trick ensemble exactly and note that no shift aligns O with a length-k window, so the aligned 1/k term is absent and the operator is split across at least three windows. Then simulate N_snap = 10^6 snapshots as in Fig. 3 for k = 6, ..., 15 and compare the empirical variance with (32/19)k×1.8^k. If the variance is not consistent with that scaling, the unknown-location 'any string operator' claim is refuted. A simpler analytical check: enumerate the k shifts and count how many place O inside one window; for this O the count is 0, not 1.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The core known-location result (Eq. 4, shadow norm ~ 2×1.8^k) is internally consistent: for a Pauli string whose support is exactly the k-qubit subsystem, q = 0 and the derivation goes through. The load-bearing gap is in the generalization to unknown locations that the abstract's 'any non-successive local operators' invites. In Section 'Extensition to Situations without Knowing Operator Locations', the sliding trick partitions the chain into length-k windows and states 'any given string operator can become compatible with the circuit structure with a probability of 1/k'. This is only true for operators whose support is contained in some length-k consecutive window. A non-successive size-k operator, e.g., Z on sites 1, 3, 5, ..., 2k-1, has span 2k-1 > k and is contained in no such window; the aligned probability is 0, not 1/k. The ensemble then splits the operator across multiple independently scrambled windows, and the identity sites inside each window can be converted to Z by the contractive unitary, so the Pauli weight is not the sliding-trick expression Eq. (14) and the shadow norm is not bounded by ~k×1.8^k. Table I's 'Unknown: k×1.8^k' therefore overstates the scope for non-successive operators. This is a scope/implementation gap in the advertised universality, not a contradiction in the known-location derivation.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes replacing the random global Clifford unitary in classical shadow tomography with a fixed 'contractive unitary' U_ct = ∏_{i<j} exp(iπ/4 Z_i Z_j), sandwiched between random single-qubit Clifford rotations. For a Pauli string of size k with no identity factors on a known k-qubit subsystem, the authors derive the Pauli weight w(O)_ct in Eq. (4) and hence the shadow norm ∥O∥²_ct ≈ 2×1.8^k, improving on the random Clifford 2^k scaling. They extend the idea to unknown operator locations using a sliding-trick ensemble of k-qubit blocks, claiming k×1.8^k scaling, and they benchmark both protocols on GHZ and ZXZ states, finding agreement with the predicted scalings.","tokens_in":21914,"tokens_out":11045,"duration_ms":109264,"significance":"The known-location result is a clean, parameter-free analytical construction with independent numerical verification on stabilizer states; if it stands, it demonstrates a new principle: a deterministic global unitary can outperform fully random scrambling for dense Pauli observables, and it is implementable on atom-array platforms. The two-qubit optimality argument in the Supplementary provides a non-circular motivation for the unitary, and the derivation contains no fitted parameters. The significance is reduced, however, by the fact that the advertised universality ('any non-successive local operators') is not fully established: the density condition and the consecutive-window restriction in the unknown-location extension materially limit the claim.","major_comments":[{"comment":"The sliding-trick argument only aligns an operator with a circuit structure when its support is contained in a single length-k consecutive window. A non-successive size-k operator such as Z_1 Z_3 Z_5 ... Z_{2k-1} has span 2k-1 and is contained in no such window for any of the k shifts, so its alignment probability is 0, not 1/k. Consequently Eq. (14) and the k×1.8^k entry in Table I do not cover non-successive operators in the unknown-location setting, and the abstract's 'any non-successive local operators' overstates the proven scope. The numerical example in Fig. 3 uses a consecutive string, so it does not test the non-successive case. Please restrict the unknown-location claim to operators contained in a length-k window, or supply a distinct construction and analysis for general non-successive operators.","section":"Section 'Extensition to Situations without Knowing Operator Locations' and Table I"},{"comment":"The abstract's phrase 'any non-successive local operators with a size ∼k' is not qualified by the density condition derived in Supplementary Eq. (13). When the operator contains q identity factors, the shadow norm acquires a prefactor (5/3)^q, and for q=γk the contractive protocol beats random Clifford only for γ<0.206 (non-identity fraction above about 80%). A sparse size-k operator can therefore have sample complexity exceeding 2^k. The main text acknowledges the O(1) case, but the abstract's unqualified 'any' is too strong. Please state the density assumption explicitly in the abstract and the known-location summary, or revise the claim to 'dense' operators.","section":"Abstract and Supplementary Eq. (13)"}],"minor_comments":[{"comment":"The section heading 'Extensition to Situations without Knowing Operator Locations' contains a typo; it should read 'Extension'.","section":"Section heading"},{"comment":"The text contains the typo 'probalility' instead of 'probability' in the sentence explaining the sliding-trick sampling.","section":"Section 'Extensition to Situations without Knowing Operator Locations'"},{"comment":"There is an inconsistency in the claimed prefactor: the text first says ∼k×1.8^k, then says the sample complexity is bounded by 2k×1.8^k, while Table I lists k×1.8^k and the exact calculation gives (32/19)k×1.8^k; these should be reconciled.","section":"Section 'Extensition to Situations without Knowing Operator Locations'"},{"comment":"The notation (-1)^k/9^k in Eq. (4) and (-1/9)^k in Supplementary Eq. (12) should be unified for consistency.","section":"Eq. (4) and Supplementary Eq. (12)"}],"recommendation":"major_revision","confidential_remarks":"The known-location result is sound and novel, and the numerical benchmarks are convincing. My recommendation of major revision is driven by the gap between the abstract's broad claim ('any non-successive local operators') and the actual scope: the unknown-location sliding trick only covers operators contained in a length-k window, and the density condition from Supplementary Eq. (13) is omitted from the abstract. These are fixable by rewording and qualification; I do not see evidence of circularity or fitted parameters. I would be willing to accept after the claims are made precise and the scope is stated accurately."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The core result is sound. For a Pauli string of size k supported on a known, consecutive block with no identity insertions, the contractive unitary U_ct = ∏_{i<j} exp(iπ/4 Z_i Z_j) gives a shadow norm ~2×1.8^k, beating random Clifford's ~2^k. The Pauli-weight derivation (Eq. 4) is clean, the stabilizer-state numerics match the predicted scaling, and the idea of using a deterministic global unitary to contract operator size is a genuinely new design principle. I would use and cite the known-location protocol.\n\nThe soft spots are in the scope claims. The abstract says \"any non-successive local operators,\" but the advantage only holds when the operator is dense within its support. If a fraction γ of the k sites are identity operators, the shadow norm acquires a (5/3)^{γk} prefactor, and the crossover against random Clifford is at γ≈0.206. This caveat is in the supplementary, not the headline. The bigger issue is the sliding trick for unknown locations. The main text claims any given string operator matches the circuit structure with probability 1/k, but that is only true when the operator's support fits inside a consecutive length-k window. A non-successive size-k operator like Z on sites 1,3,5,…,2k-1 has span 2k-1 and is contained in no such window; the alignment probability is zero, the operator splits across multiple independently scrambled blocks, and the Eq. (14) calculation no longer applies. So Table I's \"unknown: k×1.8^k\" overstates the scope for non-consecutive operators.\n\nThese are scope/implementation gaps rather than contradictions in the known-location derivation. The authors are honest enough to put the identity-operator caveat and the two-block split analysis in the supplementary, but the main text \"any given string operator\" is factually wrong for non-consecutive supports. The fix is straightforward: restate the claims for dense operators on known or unknown consecutive blocks, and discuss the sparse and span>k cases explicitly.\n\nThis paper is for researchers using classical shadows on atom-array hardware, where all-to-all commuting ZZ gates are natural. The protocol is a meaningful contribution to that subfield. It deserves a serious referee: the core math is reproducible, the numerics are real, and the necessary corrections are exactly what refereeing is for. I would recommend conditional acceptance, with the abstract and Table I tightened and the unknown-location section amended to state the consecutive-support condition.","headline":"Known-location 1.8^k shadow-norm improvement is real and well-verified, but the abstract's 'any non-successive operators' and the unknown-location sliding trick overclaim: sparse operators and non-consecutive supports break the advertised scaling.","tokens_in":22573,"tokens_out":4669,"would_cite":true,"duration_ms":37983,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"By sandwiching a deterministic product of commuting ZZ rotations between random single-qubit gates, classical shadow tomography can estimate dense k-qubit Pauli strings with sample complexity scaling as about 1.8^k instead of 2^k.","keywords":["classical shadow tomography","contractive unitary","sample complexity","Pauli weight","operator size distribution","locally scrambled unitary ensemble","random Clifford","atom array quantum processor"],"falsifier":"Run the contractive-unitary shadow protocol on a known $k$-qubit support for a fully dense Pauli string with $k=20$ and compare the empirical shot variance with the random-Clifford protocol; the theory predicts a variance ratio of roughly $(2/1.8)^{20}\\approx 9.5$ in favor of the contractive protocol. Alternatively, insert $q=\\gamma k$ identity factors with $\\gamma=0.2$ into the operator and check whether the variance crosses the random-Clifford value at $\\gamma\\approx 0.206$ as predicted by Eq. (13); observing the crossover at a substantially different $\\gamma$, or a variance that scales as $3^k$ for dense operators, would refute the central claim.","tokens_in":21458,"feed_emoji":"⚛️","tokens_out":17013,"duration_ms":124406,"temperature":0.7,"pith_summary":"This paper aims to show that classical shadow tomography can break the $2^k$ sample-complexity barrier for estimating $k$-qubit Pauli observables, provided the operator's location is known and its support is dense. It introduces a deterministic global unitary, the contractive unitary $U_{ct}=\\prod_{i<j}\\exp(i\\pi Z_i Z_j/4)$, which contracts the size of a Pauli string whenever the string contains an odd number of $X$ or $Y$ factors, shifting the operator-size distribution toward smaller sizes. This yields a Pauli weight $w(O)=\\frac{1}{2}[\\frac{1}{3^k}+\\frac{(-1)^k}{9^k}]+\\frac{1}{2}[(\\frac{5}{9})^k-\\frac{1}{9^k}]$ and a shadow norm $\\sim 2\\times 1.8^k$, beating the random-Clifford $\\sim 2^k$ scaling. A sliding-trick variant extends the advantage to the case of unknown operator location with sample complexity $k\\times 1.8^k$. If true, the result matters because it shows that hybrid random-deterministic measurement strategies can outperform fully random ensembles in quantum state characterization.","feed_headline":"A deterministic circuit cuts quantum shadow sampling to 1.8^k","feed_subtitle":"Using commuting ZZ gates as the global unitary beats the 2^k random-Clifford scaling for k-qubit Pauli strings.","key_machinery":"The key machinery is the contractive unitary, a global all-to-all product of commuting two-qubit ZZ rotations $U_{ct}=\\prod_{i<j}\\exp(i\\pi Z_i Z_j/4)$, sandwiched between independent random single-qubit Clifford layers. Its action on a Pauli string is controlled by the parity of $N_{XY}$, the number of $X$ and $Y$ factors: for odd $N_{XY}$ each $Z$ becomes $I$ and each $I$ becomes $Z$, contracting the operator size from $k$ to $N_{XY}$, while for even $N_{XY}$ the string commutes with $U_{ct}$ and stays at size $k$. The Pauli weight $w(O)=\\sum_m \\pi(m)/3^m$, derived from the resulting size distribution, converts directly to the shadow norm via $\\|O\\|^2 = w(O)^{-1}$, and the broad size peak at $2k/3$ plus the minority delta peak at $k$ yields the $1.8^k$ scaling.","core_discovery":"The central discovery is that the deterministic all-to-all product of commuting ZZ rotations, $U_{ct}=\\prod_{i<j}\\exp(i\\pi Z_i Z_j/4)$, when used as the global unitary between two layers of random single-qubit Cliffords, reduces the shadow norm for a size-$k$ Pauli string from the random-Clifford value $\\sim 2^k$ to $\\sim 2\\times 1.8^k$. The mechanism is a parity-dependent size contraction: a Pauli string with $N_{XY}$ X/Y factors either collapses to size $N_{XY}$ when $N_{XY}$ is odd or stays at size $k$ when $N_{XY}$ is even. Averaging the resulting size distribution with the Pauli-weight factor $3^{-m}$ gives $w(O)=\\frac{1}{2}[\\frac{1}{3^k}+\\frac{(-1)^k}{9^k}]+\\frac{1}{2}[(\\frac{5}{9})^k-\\frac{1}{9^k}]$, and the shadow norm is its inverse. The same construction with a sliding trick handles operators whose location is unknown, yielding $k\\times 1.8^k$ scaling. The contractive unitary is readily implemented on atom-array platforms because all two-qubit gates commute and have identical form.","pith_inferences":["Other deterministic global unitaries, such as partial products of ZZ rotations acting on selected qubit pairs, might interpolate between random Clifford and the full contractive unitary, trading sample complexity for shallower circuits.","The parity-dependent contraction mechanism suggests that unitaries biased toward flipping the parity of $X/Y$ counts could push the sample-complexity exponent below $\\log_2(1.8)\\approx 0.848$, approaching the per-contracted-site bound $3^{-m}$ in the Pauli-weight sum.","The sliding-trick prefactor $(32/19)^k \\approx 1.684^k$ shows how boundary-crossing operators degrade the variance; a recursive or hierarchical sliding scheme might recover the pure $1.8^k$ scaling even for completely unknown operator locations.","Because the contractive unitary is a product of commuting CZ-like gates, the same size-distribution calculation could be adapted to fermionic shadow encodings or photon-number-resolving measurements, where Pauli weights obey different formulas."],"forward_implications":["For a known, dense $k$-qubit Pauli string, the contractive-unitary protocol reduces the number of measurement shots by a factor of $(2/1.8)^k \\approx (10/9)^k$ compared to random Clifford shadows.","When the operator location is unknown, the sliding trick gives sample complexity $k \\times 1.8^k$, which for sufficiently large $k$ beats the shallow-circuit protocol's $\\sim 2^k$ bound.","Because all two-qubit gates in the contractive unitary commute and are identical, the unitary can be implemented in at most $k-1$ parallel steps on a reconfigurable atom-array processor.","The protocol demonstrates a general principle: a deterministic global unitary tailored to contract operator size can outperform fully random ensembles in classical shadow tomography.","With $q \\le O(1)$ identity factors in the operator support, the $1.8^k$ scaling survives with only a prefactor $(5/3)^q$, so a small number of identity defects does not erase the advantage over random Clifford."],"supporting_citations":[{"why":"Defines classical shadow tomography, the random Clifford protocol, and the shadow norm, establishing the 2^k baseline this paper aims to beat.","marker":"[14]"},{"why":"Provides the locally scrambled unitary ensemble formalism that sandwiches the global unitary between random single-qubit rotations.","marker":"[26]"},{"why":"Derives the relation between shadow norm and Pauli weight (Eq. 1) and supplies the shallow-circuit protocol used as the unknown-location benchmark.","marker":"[32]"},{"why":"Extends the shadow-norm framework to Pauli-invariant unitary ensembles, underpinning the Pauli-weight calculation for the contractive ensemble.","marker":"[33]"},{"why":"Defines the operator size distribution whose parity-dependent contraction is the mechanism behind the contractive unitary's advantage.","marker":"[34]"}],"fun_headline_variants":["Contractive unitary cuts shadow tomography to ~1.8^k","Deterministic ZZ gates beat 2^k random Clifford shadow limit","Shadow tomography hits 1.8^k with contractive unitary","Random-deterministic mix improves shadow tomography to 1.8^k","Commuting ZZ circuit reduces shadow tomography to 1.8^k"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The advantage over random Clifford rests on the operator being dense in its known support: if a significant fraction of the $k$ sites carry identity operators, the parity-driven size contraction weakens and the $1.8^k$ scaling is lost.","fun_headline_variants_meta":{"raw":{"variants":["Contractive unitary cuts shadow tomography to ~1.8^k","Deterministic ZZ gates beat 2^k random Clifford shadow limit","Shadow tomography hits 1.8^k with contractive unitary","Random-deterministic mix improves shadow tomography to 1.8^k","Commuting ZZ circuit reduces shadow tomography to 1.8^k"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000632,"raw_usage":{"total_tokens":2963,"prompt_tokens":1034,"completion_tokens":1929,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":650,"completion_tokens_details":{"reasoning_tokens":1839}},"tokens_in":650,"tokens_out":1929,"duration_ms":14805,"temperature":1.0,"reasoning_tokens":1839,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T10:18:03.100093+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the contractive-unitary shadow protocol on a known $k$-qubit support for a fully dense Pauli string with $k=20$ and compare the empirical shot variance with the random-Clifford protocol; the theory predicts a variance ratio of roughly $(2/1.8)^{20}\\approx 9.5$ in favor of the contractive protocol. Alternatively, insert $q=\\gamma k$ identity factors with $\\gamma=0.2$ into the operator and check whether the variance crosses the random-Clifford value at $\\gamma\\approx 0.206$ as predicted by Eq. (13); observing the crossover at a substantially different $\\gamma$, or a variance that scales as $3^k$ for dense operators, would refute the central claim.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines classical shadow tomography, the random Clifford protocol, and the shadow norm, establishing the 2^k baseline this paper aims to beat."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the locally scrambled unitary ensemble formalism that sandwiches the global unitary between random single-qubit rotations."},{"cited_title":"Zhang, X","cited_arxiv_id":null,"evidence_quote":"Extends the shadow-norm framework to Pauli-invariant unitary ensembles, underpinning the Pauli-weight calculation for the contractive ensemble."},{"cited_title":"Bertoni, J","cited_arxiv_id":null,"evidence_quote":"Defines the operator size distribution whose parity-dependent contraction is the mechanism behind the contractive unitary's advantage."}],"review_version":1}