{"id":"92db9acd-2ab5-482c-a3f8-34be8d7c8dff","arxiv_id":"2412.02741","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The AdS3 × S3 magnons are identified with 'long' collective eigenstates of the perturbed symmetric orbifold, matching the BMN dispersion relation from the CFT side.","lead":"This paper identifies the AdS3 × S3 string excitations with certain 'long' collective eigenstates in the perturbed symmetric orbifold CFT. If correct, it completes the map between boundary CFT states and all BMN string modes, supporting the AdS3/CFT2 correspondence.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The AdS3xS3 dictionary rests on the large-w extrapolation of truncated mixing matrices, and the unexplained sum-of-anticommutators prescription for the Lambda^- sector is checked only at w=12.","rationale":"The reader's conditional verdict is appropriate. The paper's evidence is substantial: explicit numerical diagonalisations, two independent sectors (unphysical bilinears and physical three-fermion states), absence of long states in the positive sector, and direct comparison with the independent [29] full calculation at w=12, with code provided. Nevertheless the central dictionary is a large-w statement, and all large-w statements are obtained from truncated matrices with a prescription (Section 3.2) that the authors state they do not understand. The specific mapping (5.6) uses Lambda^-, which is exactly one of the sectors in which the two definitions of the anomalous-dimension matrix disagree at O(1/w) and produce a spurious long state. This is not a disagreement with consensus; it is an internal gap between the evidence supplied and the strength of the claim ('precisely', 'one-to-one correspondence'). The proposed w>=14 full-calculation test would settle whether the prescription is correct or merely a small-w coincidence. A successful test would raise confidence; a failure would require weakening the identification to a conjecture. Verdict should remain conditional.","tokens_in":22908,"tokens_out":9084,"duration_ms":103058,"concrete_test":"Using the full higher-magnon technology of [29] (including five-magnon intermediate states), compute the exact low-lying eigenstates of the neutral L-sector mixing problem for w=14 or 16 and n=1, and compare with the long eigenvector predicted by diagonalising the two-magnon matrix with the sum-of-anticommutators prescription of Section 3.2. The test passes if (a) the predicted vector has squared overlap >0.9 with the full eigenstate's three-magnon head, (b) the eigenvalue tracks eps_2(1/w) within the expected O(1/w), and (c) no second long state appears when using only one anticommutator. If any condition fails, the a^AdS <-> Lambda^- identification in eq. (5.6) is not established; if all pass, rerun at one larger w to confirm the trend.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing step is the claim in Section 5 that the long modes, especially Lambda^- in eq. (3.9), are genuine large-w eigenstates whose dispersion reproduces the BMN result. This depends on the truncated two-/three-magnon mixing matrices of Sections 2-4 and, in the Lambda^-/G^-/G'^- sectors, on the ad hoc sum-of-anticommutators prescription introduced in Section 3.2. The paper states there that the two anticommutators differ elementwise at O(1/w) and that the extra state obtained with a single anticommutator is an artefact of cumulative O(1/w) errors, but that the sum prescription is not understood. Since each long state is a superposition of O(w) basis states, elementwise O(1/w) errors can accumulate to O(1) in overlaps and eigenvalues; therefore the w=12 checks with higher-magnon states in Sections 2.3 and 4.3, while encouraging, do not control the w->infinity limit in which the BMN dictionary is formulated. If the sum prescription fails at larger w, the Lambda^- <-> a^AdS identification in eq. (5.6), and with it the one-to-one correspondence, is unsupported.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper studies the spectrum of the symmetric orbifold CFT perturbed away from the tensionless point, working in a large-w twisted sector where the perturbation problem becomes tractable. The authors diagonalize numerically, at finite w, the anomalous-dimension mixing matrices describing two-magnon and three-magnon sectors, and identify a class of 'long' eigenstates that are coherent superpositions of O(w) fractional torus modes. They show that these long states have dispersion relations matching ϵ2(p) = sin²(πp)/(1−p), that they appear only in the expected sectors, and that they are absent from the positive sectors where no AdS3×S3 states are expected. On this basis they propose in eq. (5.6) a one-to-one identification of the long modes with the massive BMN oscillators of AdS3×S3, and claim that the symmetric orbifold at k=1 contains all AdS3 and S3 excitations, not just torus modes.","tokens_in":23256,"tokens_out":2571,"duration_ms":30831,"significance":"If the central identification is correct, the paper resolves a long-standing question about the spacetime interpretation of symmetric-orbifold states at k=1: it shows that the AdS3×S3 excitations are present as collective long modes and are independent from the torus modes once the theory is deformed. The claim is supported by several independent numerical checks, including finite-w diagonalization, inclusion of higher-magnon intermediate states up to w=12, a physical three-fermion sector, L1-norm scaling diagnostics, and comparison with the exact higher-magnon calculation of the companion paper [29]. The ancillary Jupyter and Mathematica notebooks make the numerical evidence reproducible, and the paper is careful to distinguish checked statements from conjectured extrapolations. The main weakness is that the large-w limit is not controlled analytically: the long states are constructed from finite-w numerics, the truncation to low-magnon sectors is justified only by small-w comparisons, and one-sector prescription (the sum of anticommutators in Section 3.2) is explicitly admitted to be not understood.","major_comments":[{"comment":"The sum-of-anticommutators prescription is load-bearing for the Λ−, G−, and G′− long states, yet the paper states that the two anticommutators differ elementwise at O(1/w) and that 'we do not really understand why this prescription works'. Since a long state is a superposition of O(w) basis states, elementwise O(1/w) errors can in principle accumulate to O(1) corrections to overlaps and eigenvalues. The w=12 comparisons against the full higher-magnon calculation of [29] are encouraging but only probe a single small value of w and a restricted set of states. The claim in Section 5 that Λ−1+p corresponds to the AdS oscillator therefore needs either a quantitative bound on the accumulated error as w→∞, an analytic understanding of why the sum prescription selects the true eigenstate, or a more extensive finite-w scaling study in these sectors.","section":"§3.2, eq. (3.9) and surrounding text"},{"comment":"The higher-magnon checks are restricted to w=12 (and n=1 in the two-magnon case), where the number of states is only 892 and 1778, respectively. The paper's ultimate claim is formulated in the w→∞ limit, where the BMN dictionary applies. The observed agreement at w=12 does not by itself control the extrapolation: the coefficients ξ(n/w; a/w) show singular behavior near the boundaries, and the paper notes that these boundary values are ambiguous at infinite w. To make the central identification robust, the authors should show that the overlap between the truncated-state prediction and the exact eigenstate improves systematically with w, or give some argument that the O(1/w) corrections to the mixing matrix cannot reorganize the long-state structure at arbitrarily large w.","section":"§2.3 and §4.3"},{"comment":"The identification with the BMN oscillators is based on a dispersion relation matched only to O(p²) in the small-momentum, small-g expansion, and the dictionary (5.5) is stated without derivation. At this order the Ξ− and Λ− modes have the same leading dispersion, so the assignment aS ↔ Ξ− versus aAdS ↔ Λ− is not distinguished by the energy alone; it relies on the symmetry charges and the reduction to integer modes. Since the one-to-one correspondence is the central claim, the paper should either extend the dispersion comparison to higher order in p (or to finite w/n) or present an independent quantum-number argument that fixes the assignment uniquely. As written, the O(p²) match is suggestive but does not uniquely pin down the proposed dictionary.","section":"§5, eqs. (5.5)–(5.6)"}],"minor_comments":[{"comment":"The 'inert boson' weighting factor δ(w)(m;n) is introduced as an ad hoc device for unphysical states. The paper notes that different choices (e.g. m=n versus m=n−1) give essentially the same results; it would be helpful to state explicitly which prescription was used for the figures and to quantify the sensitivity.","section":"§2.1"},{"comment":"The normalization prefactor 1/√(w+1−n) makes the ξ coefficients O(1), but the boundary behavior of ξ is stated to be ambiguous because of degeneracy with short states. The discussion would be clearer if the figure captions specified which boundary convention was used, and whether the higher-magnon corrections of §2.3 resolve the ambiguity.","section":"§2.2, eq. (2.20)"},{"comment":"The statement that 'the number of long states in the negative sector agrees with the expected number' is checked only for a few values of w up to 140 and for one energy cutoff. It would be useful to present a table of observed versus expected counts for several w and cutoffs.","section":"§4.2"},{"comment":"The paper honestly lists open questions such as an analytic determination of ξ and the action of right-moving supercharges. These are appropriate for a conclusion, but some of them directly bear on the validity of the dictionary and could be previewed in the introduction.","section":"§6"}],"recommendation":"major_revision","confidential_remarks":"This is a serious and careful numerical study from a leading group, and the companion papers [22,29] provide much of the technical machinery. The main risk is not that the numerical work is sloppy, but that the central claim rests on an uncontrolled large-w extrapolation and on one unexplained prescription. The paper would be strengthened substantially if the authors could give even a partial analytic argument for the sum-of-anticommutators prescription or a systematic scaling analysis of the overlap between truncated and exact eigenstates. Given the strength of the existing checks, I do not see this as a reason to reject, but the load-bearing points need to be addressed before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Let me cut to the chase: this paper gives the most direct evidence so far that the massive AdS3 x S3 BMN modes have concrete counterparts in the symmetric orbifold. The central object is the 'long' eigenstate Ξ− of eq. (2.20), a superposition of O(w) fractional-momentum bilinears with a non-constant coefficient profile ξ(n/w; a/w). The authors show numerically that these states are eigenstates of the anomalous dimension matrix at finite w, and that in the large-w limit their dispersion matches the BMN formula to O(p^2), with the dictionary in eq. (5.6). This goes beyond the old proposals of Lunin-Mathur, Gomis-Motl-Strominger, and Gava-Narain, which identified the modes with fractional N=4 generators but couldn't produce the dispersion; it also directly contradicts Frolov-Sfondrini's claim that such modes are absent.\n\nThe paper is careful. There are multiple independent checks: finite-w diagonalization in the bilinear sectors, higher-magnon corrections up to w=12 using the technology of the companion paper [29], a physical three-fermion sector with exact momentum conservation, L1-norm scaling to distinguish short and long states, and counting of the number of long states against the expected BMN oscillators. The authors are transparent about the weak points: they state openly that they do not understand why the sum-of-anticommutators prescription in Sec. 3.2 works, and that the large-w limit is not analytically controlled. They provide the notebooks, so the numerical claims can be checked.\n\nThe soft spots are exactly where the stress-test note points. The long states are superpositions of O(w) basis states, so the elementwise O(1/w) errors from the truncated mixing matrix could in principle accumulate to O(1) in the infinite-w limit. The w=12 check with all higher-magnon states, while encouraging, does not control w→∞. The sum-of-anticommutators prescription is genuinely ad hoc: the two anticommutators differ at O(1/w), and the sum is justified only by the fact that it reproduces the full calculation at small w. And the BMN dispersion is matched only to O(p^2), so the dictionary is tested at low momenta.\n\nI don't think these caveats warrant rejection. The identification is concrete, falsifiable, and supported by several independent strands of evidence. But the paper should be honest in the final version that the dictionary is a well-motivated conjecture pinned to a numerical extrapolation, not a proven equality. A good referee should push for either a higher-w check with the full [29] machinery or an analytic argument for the coefficient profile, and for some understanding of the anticommutator-sum rule.\n\nWho is this for? People working on AdS3/CFT2, symmetric orbifolds, and BMN limits. It is not a broad-audience paper, but it answers a specific open question that has been debated in the recent literature. I would send it to review, and I would cite it. I'd also bring it to the group meeting, because it's a nice example of how far numerical diagonalization plus honest caveats can go in a subject where exact methods are scarce.","headline":"A careful numerical identification of AdS3 x S3 BMN modes as 'long' collective eigenstates in the symmetric orbifold, with the main caveat the uncontrolled large-w extrapolation and an unexplained ad hoc prescription.","tokens_in":23737,"tokens_out":3293,"would_cite":true,"duration_ms":30821,"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":"The paper argues that long collective eigenmodes of the deformed symmetric orbifold are in one-to-one correspondence with the AdS3×S3 BMN string excitations, and that their dispersion relation matches the BMN result to second order.","keywords":["AdS3/CFT2 duality","symmetric orbifold","BMN limit","anomalous dimensions","long collective modes","fractional N=4 generators","twisted sector","magnons"],"falsifier":"Diagonalise the full anomalous-dimension matrix with all higher-magnon channels included for a moderately large w, for example w=20, in the negative bilinear sector: if the long eigenstate's eigenvalue deviates from $\\epsilon_2(n/w)$ at leading order, or if a spurious long state survives in the $G^-$/L sectors when one anticommutator is used, the identification would fail.","tokens_in":22720,"feed_emoji":"🌀","tokens_out":8457,"duration_ms":80560,"temperature":0.7,"pith_summary":"This paper claims that the massive string excitations along AdS3 and S3, which appear in the BMN limit of the dual spacetime, are already present in the symmetric orbifold CFT, and that perturbation lifts the degeneracy enough to identify them unambiguously. The authors study the eigenvalues of the anomalous-dimension matrix in the large-w twisted sector of the deformed orbifold and find two classes of eigenstates: 'short' states built from a few fractional torus modes, and one 'long' collective state per momentum built from O(w) products of fractional torus modes. They show that the long states are deformations of the fractional N=4 generators, reduce to those generators for integer momentum, and carry the same second-order anomalous dimension as a single torus magnon. Matching the momentum map and the dispersion relation to O($p^{2}$), they argue that the long modes are in one-to-one correspondence with the BMN oscillators of AdS3 and S3, and that no such mode exists in the positive sectors where none is expected. If correct, this settles whether the AdS3×S3 modes are independent of the torus modes in the dual CFT: they become independent once the theory is deformed.","feed_headline":"Long orbifold modes match the AdS3×S3 BMN oscillators","feed_subtitle":"Deforming the symmetric orbifold lifts the degeneracy and reveals collective states with exactly the BMN dispersion, to second order.","key_machinery":"The load-bearing object is the long collective eigenmode $\\Xi^{-}_{n/w}|w\\rangle$, defined as a sum over $O(w)$ products of two fractional torus fermions with coefficients $\\xi(n/w;a/w)$, together with its supercharge-generated partners $\\Gamma^-$, $\\Gamma'^-$, and $\\Lambda^-$. The calculation is carried by the large-w anomalous-dimension mixing matrix obtained from the supercharge commutation relations of the companion paper, keeping only two-magnon (or three-magnon) intermediate states; the paper distinguishes short from long eigenstates by the scaling of the $L^1$-norm of the normalised state, $O(1)$ versus $O(w^{1/2})$. The key identity is that the long state's second-order anomalous dimension equals $\\epsilon_2(p)=\\frac{\\sin^2(\\pi p)}{1-p}$, the same function as a single torus magnon, and that under the identification in eq. (5.6) this reproduces the BMN dispersion relation to $O(p^2)$.","core_discovery":"The central discovery is that the fractional N=4 modes, which were proposed years ago as the AdS3×S3 excitations, are not themselves the eigenstates for non-integer momentum. Instead, the true large-w eigenstates are long collective modes such as $\\Xi^{-}_{n/w}|w\\rangle = -\\frac{1}{\\sqrt{w+1-n}}\\sum_{a=n}^{w}\\xi(n/w;a/w)\\, \\bar\\psi^{-}_{1/2+n/w-a/w}\\psi^{-}_{-1/2+a/w}|w\\rangle$, with coefficients $\\xi$ converging to a continuous function of $a/w$ and reducing to a constant when $n/w$ is an integer. Acting with supercharges generates the partner long states $\\Gamma^-$, $\\Gamma'^-$, and $\\Lambda^-$, all of which share the second-order dispersion $\\epsilon_2(p)=\\frac{\\sin^2(\\pi p)}{1-p}$. The paper then identifies these modes with the BMN oscillators $a^{S}(n)$, $b^{S}(n)$, $a^{\\mathrm{AdS}}(n)$, and $b^{\\mathrm{AdS}}(n)$ through the dictionary in eq. (5.6), using the momentum map $p(n)=\\frac{n}{\\alpha' p_+(1+4\\pi^2 g^2)}$ and reproducing the mixed-flux BMN dispersion $\\sqrt{\\sin^2\\alpha + (\\cos\\alpha \\pm n/(\\alpha' p_+))^2}$ to $O(p^2)$. Because long modes appear only in the negative sectors and in the physical three-fermion sector, the correspondence is claimed to be exact: these are the AdS3×S3 excitations of the dual string.","pith_inferences":["If the identification is right, the coefficient function $\\xi(p;q)$ should be determinable analytically, and a closed form would replace the numerical diagonalisation with a proof of the large-w eigenstate structure.","The unexplained success of the sum-of-anticommutators prescription hints at an exact symmetry in the truncated supercharge algebra; identifying it could turn the numerical long-state construction into an analytic argument.","A natural test is to repeat the analysis starting from the bottom BPS state instead of the top one; symmetry reversal might produce long modes in the positive sectors, which would sharpen the chiral/anti-chiral interpretation.","If the correspondence extends beyond $O(p^2)$, the exact dispersion of the long states would be a direct CFT prediction for the massive AdS3×S3 worldsheet S-matrix at finite coupling."],"forward_implications":["The k=1 symmetric orbifold contains one-to-one counterparts of all massive AdS3 and S3 BMN oscillators, not merely the torus modes.","The AdS3×S3 modes become unambiguous once the orbifold is deformed; at the free point they are hidden among degenerate states.","The BMN dispersion relation with the correct sign choice is reproduced from the CFT side to second order in the momentum.","The absence of long states in the positive sectors explains why the BMN analysis contains no extra oscillators to match.","The long modes are present already at zero coupling, but can only be singled out after perturbation."],"supporting_citations":[{"why":"Supplies the large-w supercharge commutation relations, the anomalous dimension formula $\\epsilon_2(p)$, and the dispersion relation on which the paper's mixing-matrix analysis is built.","marker":"[22]"},{"why":"Provides the BMN pp-wave spectrum and dispersion relation that the paper reproduces from the orbifold side to $O(p^2)$.","marker":"[23]"},{"why":"Gives the full higher-magnon calculation used to check that the truncated two- and three-magnon analysis yields the correct long eigenstates.","marker":"[29]"},{"why":"Proposed the identification of AdS3×S3 modes with fractional N=4 generators, which the paper refines into its long-mode proposal.","marker":"[5–7]"},{"why":"Argued that no such modes are present in the symmetric orbifold, the position this paper directly addresses and overrides.","marker":"[28]"},{"why":"Corrects the BMN dispersion relation by the sign choice for half the oscillators, which the identification in eq. (5.6) relies on.","marker":"[30]"}],"fun_headline_variants":["Long orbifold modes are the AdS3×S3 magnons","Orbifold's long collective states match BMN oscillators","Symmetric orbifold yields long-magnon dictionary","Fractional modes spawn long AdS3×S3 excitations"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is that the truncated anomalous-dimension matrix, computed from the large-w supercharge commutation relations with only two- or three-magnon intermediate states and the unexplained sum-of-anticommutators prescription, gives the correct large-w eigenstates; the authors checked this against the full higher-magnon calculation only at small w.","fun_headline_variants_meta":{"raw":{"variants":["Long orbifold modes are the AdS3×S3 magnons","Orbifold's long collective states match BMN oscillators","Symmetric orbifold yields long-magnon dictionary","Fractional modes spawn long AdS3×S3 excitations"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000494,"raw_usage":{"total_tokens":2485,"prompt_tokens":1062,"completion_tokens":1423,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":678,"completion_tokens_details":{"reasoning_tokens":1348}},"tokens_in":678,"tokens_out":1423,"duration_ms":14627,"temperature":1.0,"reasoning_tokens":1348,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T23:09:17.648526+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Diagonalise the full anomalous-dimension matrix with all higher-magnon channels included for a moderately large w, for example w=20, in the negative bilinear sector: if the long eigenstate's eigenvalue deviates from $\\epsilon_2(n/w)$ at leading order, or if a spurious long state survives in the $G^-$/L sectors when one anticommutator is used, the identification would fail.","supporting_citations":[],"review_version":1}