{"id":"578c25e5-0deb-4bb2-bf6e-fe2cfa8f5a5a","arxiv_id":"2412.12533","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Genuine multipartite entanglement among neighboring spins peaks near the quantum critical point in 1d, 2d, and 3d Ising models, weakens with dimension, and vanishes for non-adjacent spins.","lead":"The authors map out how strongly small groups of neighboring spins in transverse-field Ising magnets are entangled in one, two, and three dimensions. The study provides a reference dataset for genuine multiparticle entanglement near quantum phase transitions and connects the most collective entanglement to a simple single-spin measurement.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claim that GME vanishes for all non-adjacent subregions in 2d/3d rests on finite geometry checks plus numerical biseparability certificates; untested disjoint geometries may still harbor weak GME, so the abstract overstates what is proven.","rationale":"The reader's weakest assumption is also the most load-bearing concern I can identify. Positive GME detections near the QCP are robust because I2>0 is a sufficient criterion for GME independent of the lower-bound theorem, and QMC and ED agree on the adjacent-site results. The global GMC formula has an explicit finite-N extrapolation caveat, but it is a smaller part of the paper. The §VIIC proof of 2I2≤CGME contains an unproved inequality and apparent typos, but even if that proof required repair, the physical detection claims would stand. By contrast, the universal non-adjacent vanishing claim is featured in the abstract, the bullets, and the discussion, and it is supported only by a finite set of geometries and numerical biseparability certificates. The 1d precedent in Ref. [25] demonstrates that weak GME can exist in disjoint subregions while I2 remains negative, so the same possibility in 2d/3d cannot be excluded without a direct test. A single counterexample would falsify the headline, making this the single most load-bearing assumption. The proposed SDP-witness check on an untested disjoint geometry would settle whether the concern lands. If the check finds no GME, the claim can be retained as a well-supported numerical finding; if it finds GME, the paper requires a softened statement. In either case the reader's conditional verdict is appropriate, so I recommend no change to the verdict.","tokens_in":23807,"tokens_out":6838,"duration_ms":61801,"concrete_test":"Using the RDMs deposited on GitHub [41], take the 24x24 QMC 3-site/4-site RDM at h=3.044 for the 2d lattice and evaluate the SDP-based GME witness of Ref. [53] on a disconnected geometry not listed in Table II, e.g., sites {1,2,4,6} (the 2x3 rectangle with site 5 omitted). If the witness is positive beyond Monte Carlo error, the universal vanishing claim is false. If it is non-positive, run the Ref. [48] biseparability algorithm to see whether a certificate is found. Repeat for one 3d disjoint geometry, e.g., {1,2,6} plus the site directly above site 6, using the 8x8x8 QMC RDM at h=5.2. This directly tests the only evidence gap separating the finite checks from the abstract's universal statement.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section IVB and Table II report I2 for a finite set of 3- and 4-site geometries. I2>0 certifies GME, but I2≤0 does not certify its absence. For many disjoint geometries the paper gives a biseparability certificate from the iterative algorithm of Ref. [48], but for others (e.g., subregion 126 in 2d) it only reports that an estimated CGME is 'consistent with a biseparable state.' No theorem rules out GME in untested geometries. The 1d case already shows the danger: Ref. [25] detected weak disjoint GME (e.g., 124, 1245) with an SDP witness even though I2 is negative there; the present paper acknowledges this in §IVB. Since 2d/3d may similarly have weak GME below the I2 detection threshold, the abstract's universal statement 'GME disappears for subregions consisting of non-adjacent sites in both 2d and 3d' goes beyond the evidence. This is the load-bearing overclaim: one positive witness for an untested disjoint geometry would falsify the headline, whereas the §VIIC proof gap, while real (unproved fijk≤0 and typos), leaves the central detections intact because I2 is an established sufficient criterion.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper applies entanglement microscopy to the transverse-field Ising model in d=1,2,3, obtaining reduced density matrices of up to four sites via quantum Monte Carlo, exact diagonalization, and the exact 1d solution. It evaluates bipartite negativity and several genuine multipartite entanglement criteria (CGME, I2, W1), reporting that 3- and 4-spin GME for adjacent sites is present over wide field ranges, peaks near the quantum critical point, and decreases rapidly with dimension. It also reports that GME vanishes for non-adjacent subregions in 2d and 3d, identifies logarithmic singular scaling of dI2/dh in 1d, derives a formula for the N-party GMC in terms of the single-site transverse magnetization, and claims completion of a proof that I2 lower-bounds the GMC.","tokens_in":23995,"tokens_out":10356,"duration_ms":95941,"significance":"If the main claims hold, this is a valuable systematic dataset on multipartite entanglement across archetypal quantum critical points in one, two, and three dimensions. The strengths are the convergence checks between QMC and ED, the use of the exact 1d solution, the public release of the sampled RDMs on GitHub, and the demonstration that I2 detects GME in 2d and 3d where the W1 criterion fails. The proposed global GMC formula and the symmetry-adapted I2 parametrizations are conceptually attractive and should be useful for future work. The universal vanishing claim and the completed-proof claim, however, currently outrun the evidence presented.","major_comments":[{"comment":"The abstract and Section IVB state that GME disappears for all non-adjacent subregions in 2d and 3d, but the evidence consists of the finite set of geometries in Table II plus, for some subregions, biseparability certificates from the iterative algorithm of Ref. [48]. For subregion 126 no biseparability certificate is given; the text only says that the estimated CGME is 'consistent with a biseparable state', and the geometric-distance upper bounds are not entanglement witnesses. Since I2≤0 does not prove biseparability, and since the 1d case already exhibits weak disjoint GME below the I2 threshold (Ref. [25], acknowledged in this section), the universal statement goes beyond the evidence. Please either prove a general statement or rephrase the claim to 'no GME is detected in the tested non-adjacent geometries'.","section":"§IVB and Abstract"},{"comment":"The claimed completion of the proof that 2I2≤CGME depends on the assertion 'fijk≤0 for (i,j,k)≠(1,1,1)', which is stated without proof. In addition, the displayed expression for I2[|ψ⟩⟨ψ|,|000xyz⟩] replaces |a000 Σ aijk xi yj zk| by a sum of absolute values, which is an upper bound rather than an equality; this step needs justification. The formula also appears to contain a typo, since 'a0j0a0j0' should presumably be 'a0j0 ai0k'. Until these steps are supplied, the claim of a complete proof is not established. This gap does not invalidate the positive I2 detections, because I2>0 is itself an established sufficient condition for GME.","section":"Appendix VIIC"},{"comment":"The global GMC formula assumes that the minimum over all bipartitions is attained by a single-site partition. The argument that adding sites to Ai always increases the concurrence is heuristic, and the numerical verification is reported only for lattices with N≤20, while Fig. 7 uses 25, 25, and 27 spins for 1d, 2d, and 3d respectively. Please provide a proof for site-transitive pure states or verify the assumption for the actual lattice sizes used; otherwise Eq. (12) should be presented as a conjectured lower/upper bound rather than an exact expression.","section":"§V, Eq. (12)"}],"minor_comments":[{"comment":"The definition of fijk contains a repeated 'a0j0a0j0' term; the intended term is likely 'a0j0 ai0k', and the surrounding derivation should be rechecked for further notational typos.","section":"Appendix VIIC"},{"comment":"For the (2,2) configuration in 2d, the ED and QMC sudden-death distances differ (ED gives r=2 with E(2)=0 on 6×5, while QMC gives a small value consistent with zero); the text should state explicitly whether this is a finite-size or convergence effect.","section":"Section IVB"},{"comment":"The caption of Fig. 11 reports a scaling exponent 0.89 for ⟨σx⟩, but the main text uses this to infer dSvN/dh∼|h−hc|−0.11; the sign convention should be stated clearly so that readers do not confuse the order-parameter exponent with the derivative exponent.","section":"Appendix VIID"},{"comment":"The GitHub repository link in Ref. [41] should be accompanied by a version identifier or retrieval date so that the released RDM data can be cited reproducibly.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The central positive detections of I2 for adjacent 3- and 4-spin subregions appear sound and are cross-checked by ED and QMC. The main issues are a universal vanishing claim that should be qualified, an incomplete proof step in Appendix VIIC, and an unverified assumption behind the global GMC formula for the specific lattice sizes used. All three are fixable within the scope of a major revision. I do not see grounds for rejection, but the manuscript should not be accepted before these load-bearing points are addressed."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my take on arXiv:2412.12533. The paper extends entanglement microscopy to the TFIM in 1d, 2d, and 3d and gives the first systematic dataset of multipartite entanglement across the phase diagram. The genuinely new pieces are the I2-based GME detection in 2d and 3d, the completion of the proof that I2 lower-bounds the GME concurrence for arbitrary product states, and the closed-form expression for the global GMC in terms of single-site magnetization. The shared RDMs on GitHub and the QMC/ED cross-validation are concrete strengths; the 3-spin adjacent results are converged and robust.\n\nThe soft spots are real but mostly fixable. The abstract says GME 'disappears' for non-adjacent subregions in 2d/3d. That overstates what is shown: I2 is negative for a finite set of disjoint geometries, and some of those states are certified biseparable by an iterative algorithm. That is evidence, not a proof for all disjoint subregions, and in 1d the same logic would miss the weak disjoint GME that Ref. [25] finds with an SDP witness. The claim should be softened to 'we find no GME in the non-adjacent subregions tested.'\n\nThe proof of the I2 bound in Appendix VIIC has a typo (the term -|a0j0 a0j0| should be -|a0j0 ai0k|). With that correction, the claim that fijk<=0 for all (i,j,k) != (1,1,1) is a simple case check, so this is a presentation gap, not a real hole. The authors should fix it and add the two lines.\n\nTwo smaller points: the 3d 4-site I2 signal is within QMC error bars, though ED on 3x3x3 supports a small positive value; and the global GMC formula assumes the single-site bipartition is optimal, verified only up to N<=20 but used for N=25/27. Both are worth a sentence in the revision.\n\nBottom line: this deserves a serious referee. The main detections are probably right, and the overclaims can be corrected without changing the core of the paper.","headline":"Useful dataset and a few new results on multipartite entanglement in the TFIM, but the abstract overstates the vanishing of GME for non-adjacent subregions.","tokens_in":24653,"tokens_out":7866,"would_cite":true,"duration_ms":61154,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Genuine multipartite entanglement between three or four neighbouring Ising spins peaks near the quantum critical point, weakens sharply with dimension, and disappears when the spins are not adjacent in 2d and 3d.","keywords":["genuine multipartite entanglement","transverse-field Ising model","quantum Ising model","entanglement microscopy","quantum Monte Carlo tomography","quantum critical point","biseparability","GME concurrence"],"falsifier":"Run the same tomography and $I_2$ maximization on a 24x24 square-lattice transverse-field Ising model at the critical field $h\\approx 3.044$ for a disjoint three-site geometry not covered by Table II, for example sites 1, 5 and 9 forming a diagonal chain two lattice steps long. If $I_2$ evaluates to a strictly positive number at a precision where the reduced-density-matrix errors are below $10^{-5}$, the claim that all non-adjacent subregions in 2d are free of GME is refuted; conversely, certified biseparability for every such geometry would support it.","tokens_in":23482,"feed_emoji":"🔬","tokens_out":11479,"duration_ms":96478,"temperature":0.7,"pith_summary":"The paper maps where genuine multipartite entanglement (GME) lives in the transverse-field Ising model in one, two, and three dimensions by reconstructing the full quantum state of small blocks of three or four neighbouring spins. It finds that these microscopic blocks share GME over broad portions of the phase diagram, that the entanglement peaks in the paramagnetic phase near the quantum critical point, and that its strength drops sharply as the spatial dimension grows. It also finds that once the spins in a block are not all adjacent, GME is no longer detected in 2d and 3d, with many non-adjacent geometries certified to be biseparable, meaning the entanglement is extremely short-ranged. On the formal side, the paper completes a previously incomplete proof that the computable criterion $I_2$ lower-bounds the genuine multipartite concurrence for every product state, and it shows how the lattice symmetries of the state reduce the optimization needed to evaluate $I_2$. If these results hold, GME becomes a practical, microscope-scale diagnostic of quantum criticality and dimensionality in equilibrium spin systems.","feed_headline":"Multipartite entanglement peaks at criticality, dies with distance","feed_subtitle":"Three- and four-spin blocks in 1d, 2d, and 3d show genuine entanglement only when adjacent","key_machinery":"The load-bearing object is the $I_2$ criterion, defined as a maximization over product states of a two-copy quantity built from swap operators on the doubled Hilbert space; a strictly positive value certifies genuine multipartite entanglement, and because it is a maximization any positive value found is already a certificate. The paper supplies the missing step in the proof that $2I_2$ bounds the genuine multipartite concurrence, showing that for any product state the maximum reduces without loss of generality to the |000111\\rangle product state, so the bound holds generally. The numerical machinery is the tomography of small reduced density matrices (up to four spins) from quantum Monte Carlo and exact diagonalization, with parity, $Z_2$, and lattice permutation symmetries used to reduce the product-state search space to a few angles. A second central identity is the reduction of the $N$-party GME concurrence for uniform states to $\\sqrt{1-\\langle \\sigma^x\\rangle^2-\\langle \\sigma^z\\rangle^2}$, turning the most collective entanglement into a local magnetization measurement.","core_discovery":"The central claim is that genuine multipartite entanglement in the transverse-field Ising model is a local, dimension-sensitive phenomenon that is strongest when three or four spins sit next to each other in one dimension, and is concentrated in the paramagnetic phase just before the quantum critical point. This is established by performing tomography of the reduced density matrices of at most four spins, using the exact solution in 1d, quantum Monte Carlo in 2d and 3d, and exact diagonalization as a check, then evaluating the genuinely multipartite concurrence, its $I_2$ lower bound, and a weaker $W_1$ witness. The paper proves that $2I_2$ is a lower bound on the genuine multipartite concurrence for any product state in the defining maximization, completing a gap left in the original proof; it also proves that $I_2$ is strictly stronger than $W_1$ because $W_1$ corresponds to restricting the optimization to local-unitary transforms of a fixed product state. For adjacent triples and quadruples, $I_2$ is positive across wide ranges of the transverse field in all dimensions and reaches a maximum near the critical point, while for non-adjacent subregions in 2d and 3d no GME is found and a finite list of geometries is certified biseparable. Finally, for all $N$ spins together, the GME concurrence of a uniform state reduces to a single-site observable, $\\sqrt{1-\\langle \\sigma^x\\rangle^2-\\langle \\sigma^z\\rangle^2}$, which in the ferromagnetic phase is large but fragile: it is killed by one local $\\sigma^z$ measurement, while the post-measurement GMC peaks near the critical point.","pith_inferences":["Inference: the short-range pattern of GME could serve as a sharper diagnostic of classical versus quantum correlations at criticality, since long-range order-parameter correlations would not be accompanied by long-range multipartite entanglement.","Inference: the completed $I_2$ proof applies beyond Ising spins, so any quantum Monte Carlo sampled reduced density matrix with enough symmetry to tame the optimization can be certified for GME at the same cost, opening fermionic or frustrated models to the same microscopy.","Inference: the post-measurement global concurrence is the more physical measure of collective entanglement in the ordered phase, since a single local projective measurement already destroys the cat-state GME; experiments should therefore compare pre- and post-selected entanglement when probing ordered phases.","Inference: a testable extension is to compute $I_2$ for the untested disjoint geometries in 3d with improved reduced-density-matrix precision below $10^{-5}$; a positive value in any such geometry would replace the 'all non-adjacent' statement with a list of short-ranged exceptions."],"forward_implications":["For neighbouring three- and four-spin blocks, GME is present over large portions of the phase diagram in 1d, 2d, and 3d, so genuine multipartite entanglement is an ordinary, measurable property of quantum Ising matter near criticality.","In 2d and 3d, non-adjacent blocks show no GME, and the listed geometries are certified biseparable; if the pattern is general, entangled correlations at criticality remain effectively short-ranged even when two-point correlations are long-ranged.","The global concurrence formula means the collective GME of any uniform pure state can be read off from single-site magnetization, without solving the exponentially large partition optimization.","The completed $I_2$ lower bound makes GME certification robust: any optimization attempt that lands on a positive $I_2$ value is already a proof, not a heuristic, for 3- and 4-spin blocks.","The observed singular scaling of $dI_2/dh$ near the critical point extends the known critical scaling of bipartite entanglement measures to a genuine multipartite measure."],"supporting_citations":[{"why":"Supplies the entanglement microscopy protocol for sampling reduced density matrices of small subregions.","marker":"[1]"},{"why":"Defines the GME concurrence and proves its lower bound only for a restricted product-state family, a gap the present paper closes.","marker":"[35]"},{"why":"Introduces the $I_2$ criterion as a maximization over product states on two copies of the state.","marker":"[36]"},{"why":"Certifies the biseparability of disjoint-geometry subregions, the load-bearing step for the claim that GME vanishes for non-adjacent spins.","marker":"[48]"},{"why":"Prior GME study of the XY model whose $I_2$ values are compared and explained as effectively the $W_1$ criterion.","marker":"[24]"},{"why":"Prior scaling analysis of genuine multiparticle entanglement in the 1d Ising chain that reports weak GME in some disjoint subregions, providing the comparison this paper re-examines.","marker":"[25]"},{"why":"Exact solution of the 1d transverse-field Ising chain used to obtain the thermodynamic-limit RDMs and the small-field asymptotic behavior.","marker":"[28]"},{"why":"Supplies the 2d critical field $h_c\\approx 3.044$ used to position the QMC data.","marker":"[29]"},{"why":"Supplies the 3d critical field $h_c\\approx 5.2$.","marker":"[30]"},{"why":"Used to estimate the convex-roof genuine multipartite concurrence for small subregions.","marker":"[47]"}],"fun_headline_variants":["Genuine multipartite entanglement only for adjacent spins, peaks at criticality","3- and 4-spin GME is local, dimension-sensitive, and critical","Multipartite entanglement dies with distance, thrives at critical point","Entanglement microscopy: Ising GME is local, peaks at critical field","In Ising models, genuine multipartite entanglement is local and critical"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The only load-bearing assumption is that the numerical biseparability certification used for a finite list of separated geometries is reliable and that no untested separated geometry in 2d or 3d has a weak form of the entanglement; if either fails, the claim that genuine multipartite entanglement vanishes for all non-adjacent spins collapses.","fun_headline_variants_meta":{"raw":{"variants":["Genuine multipartite entanglement only for adjacent spins, peaks at criticality","3- and 4-spin GME is local, dimension-sensitive, and critical","Multipartite entanglement dies with distance, thrives at critical point","Entanglement microscopy: Ising GME is local, peaks at critical field","In Ising models, genuine multipartite entanglement is local and critical"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001122,"raw_usage":{"total_tokens":4848,"prompt_tokens":1307,"completion_tokens":3541,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":923,"completion_tokens_details":{"reasoning_tokens":3441}},"tokens_in":923,"tokens_out":3541,"duration_ms":25395,"temperature":1.0,"reasoning_tokens":3441,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T13:59:20.009642+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same tomography and $I_2$ maximization on a 24x24 square-lattice transverse-field Ising model at the critical field $h\\approx 3.044$ for a disjoint three-site geometry not covered by Table II, for example sites 1, 5 and 9 forming a diagonal chain two lattice steps long. If $I_2$ evaluates to a strictly positive number at a precision where the reduced-density-matrix errors are below $10^{-5}$, the claim that all non-adjacent subregions in 2d are free of GME is refuted; conversely, certified biseparability for every such geometry would support it.","supporting_citations":[{"cited_title":"Hesselmann and S","cited_arxiv_id":null,"evidence_quote":"Supplies the 2d critical field $h_c\\approx 3.044$ used to position the QMC data."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the 3d critical field $h_c\\approx 5.2$."},{"cited_title":"Röthlisberger, J","cited_arxiv_id":null,"evidence_quote":"Used to estimate the convex-roof genuine multipartite concurrence for small subregions."}],"review_version":1}