{"id":"2213d060-e223-4fda-bdde-bc92a492de29","arxiv_id":"2608.10083","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The cusp anomalous dimension at strong coupling is obtained for all rapidity separations from two competing string saddles, and the transversely-separated two-cusp Wilson loop scales as (b_perp^2 Lambda^2)^{-Re Gamma_cusp}.","lead":"This paper computes the strong-coupling cusp anomalous dimension in N=4 super Yang-Mills theory from string theory saddle points, covering all rapidity separations and discovering an exchange of dominance between two saddle families. It then uses the same machinery to derive the transverse-momentum dependence of a heavy-quark fragmentation Wilson loop, connecting it to the TMD soft function.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Branch-cut contributions to the Lagrange-multiplier integral (app. A.3) are an admitted gap: if non-negligible, they add terms to Gamma_cusp beyond the saddle action (3.49); current numerical checks cover only two Delta-eta values, Delta-theta=0, and only the u_max=1 saddle.","rationale":"The reader's weakest_assumption identifies exactly the branch-cut contribution issue in appendix A.3, and this is the most load-bearing concern because the one-cusp anomalous dimension in eq. (3.49) is the input to the two-cusp result in eq. (4.34). Even if the two-cusp construction were made fully rigorous, it would inherit any error in Re Gamma_cusp. The paper deserves credit for reproducing known small- and large-Delta_eta limits, for the internally consistent Lagrange-multiplier formalism, and for explicitly flagging this gap rather than hiding it. However, the numerical checks in appendix B are restricted to Delta_theta = 0, to Delta_eta = 1.5 and 2, and to the u_max = 1 term; the u_max = u_+ branch is not checked, and the claim is for all Delta_eta and Delta_theta. The proposed direct numerical evaluation of the deformed-contour integral at previously unchecked parameter points would settle whether the branch-cut segments actually contribute. Until either that check is performed or an analytic bounding contour is found, CONDITIONAL remains the correct verdict; if the test reveals a finite branch-cut contribution, the all-Delta_eta form of eq. (3.49) would need to be revisited.","tokens_in":46464,"tokens_out":4382,"duration_ms":46289,"concrete_test":"Evaluate numerically the full integral in eq. (3.29) along a contour deformed to include the magenta dot-dashed branch-cut segments, for example at Delta_eta = 3 with Delta_theta = 0 and at Delta_eta = 2 with Delta_theta = pi, extracting the coefficient of ln(Lambda L) from the logarithm of the integral; compare that coefficient with the saddle-point prediction in eq. (3.49). A discrepancy would show that branch-cut contributions modify Gamma_cusp in a regime not covered by appendix B; agreement would support the saddle result where no numerical check currently exists.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing concern is the unresolved treatment of branch-cut contributions in the steepest-descent evaluation of the Lagrange-multiplier integral eq. (3.29). The actions ~S_eff^(+) and ~S_eff^(1) have a branch cut along the imaginary C_eta (and C_theta) axis, so a valid contour deformation from the real axis to the contributing saddle must include the magenta dot-dashed segments in figure 16. The authors state in appendix A.3 and footnote 10 that they have not found a systematic argument that these segments do not contribute to the large-sqrt(lambda) behavior; they instead check numerically only for Delta_eta = 1.5 and 2, with Delta_theta = 0, and only for the u_max = 1 term (eq. B.2). The u_max = u_+ integral is explicitly left unexamined. Since Gamma_cusp in eq. (3.49) is read off from the saddle-point action alone, any additional contribution from these segments would change the coefficient of ln(Lambda L) in eq. (3.28), and through eqs. (4.33)-(4.34) would alter the two-cusp prediction chi_1 ~ (b_perp^2 Lambda^2)^{-Re Gamma_cusp}. The authors themselves flag the centrality of this gap: 'the situation would be more satisfactory if one could find an argument that systematically shows that these segments do not modify the asymptotic behavior of the integral.' This is not an internal inconsistency, but it is the weakest link in the chain from eq. (3.29) to eq. (3.49) and hence to the paper's main quantitative claims.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies cusped Wilson loops in N=4 super Yang-Mills theory at strong coupling through AdS/CFT, motivated by the heavy-quark TMD fragmentation matrix element chi_1(b_perp). The one-cusp calculation is reduced, after a Lagrange-multiplier reformulation, to a finite-dimensional integral over the multipliers (eq. 3.29). The authors identify two families of saddle points, u_max = u_+ and u_max = 1, derive the saddle-point conditions (eqs. 3.32-3.33 and 3.39-3.40), and extract the cusp anomalous dimension as the coefficient of ln(Lambda L) in the regularized effective action (eq. 3.49). They find that the two families exchange dominance as a function of Delta eta, reproduce the known large-rapidity slope sqrt(lambda)/(4 pi), recover the small-angle bremsstrahlung behavior, and satisfy the zig-zag symmetry at Delta theta = pi. For the two-cusp configuration, the authors introduce a conserved transverse momentum flux conjugate to b_perp, compute the transverse profile of the extremal surface in the large-rapidity limit, and obtain chi_1(b_perp; Delta eta, Delta theta) proportional to (b_perp^2 Lambda^2)^{-Re Gamma_cusp} (eqs. 4.33-4.34). They also extract the Collins-Soper kernel at large rapidity, eq. (4.36).","tokens_in":46830,"tokens_out":23037,"duration_ms":224084,"significance":"If the result holds, the paper provides a nonperturbative strong-coupling prediction for a TMD-type soft function and the first full-Delta-eta formula for the Minkowski-signature cusp anomalous dimension at strong coupling. The manuscript has real strengths: the Lagrange-multiplier formulation is explicit and the saddle-point equations are given in closed form; the predictions are parameter-free in the sense that the multipliers are fixed by solving the saddle-point conditions rather than fitted; known limits (large Delta eta, small angle, zig-zag symmetry) are reproduced; and the two-cusp computation is organized around a conserved momentum flux, giving a physical interpretation of the transverse separation. The main risk is the unresolved treatment of branch-cut contributions in the steepest-descent evaluation of the Lagrange-multiplier integral, which the authors themselves flag in appendix A.3 and footnote 10. A secondary but concrete issue concerns scheme dependence in the Fourier transform that leads to eqs. (4.34) and (4.36).","major_comments":[{"comment":"The steepest-descent evaluation of the integral in eq. (3.29) requires deforming the integration contours in the complex C_eta, C_theta planes. Because the effective actions have branch cuts along the imaginary axes, a valid deformation must include the magenta dot-dashed segments in figure 16. The authors state in appendix A.3 and footnote 10 that they have not found a systematic argument showing that these segments do not contribute to the large-sqrt(lambda) behavior. The numerical checks in appendix B cover only Delta eta = 1.5 and 2, only Delta theta = 0, and only the u_max = 1 term (eq. B.2); the u_max = u_+ integral is explicitly left unexamined. Since Gamma_cusp in eq. (3.49) is read off from the saddle-point action alone, any extra contribution from these segments would change the coefficient of ln(Lambda L) and hence propagate into the two-cusp predictions (4.33)-(4.34). This is the weakest link in the chain from eq. (3.29) to the paper's main quantitative claims, and it needs either a rigorous argument or substantially expanded numerical verification, including the u_+ family and nonzero Delta theta.","section":"Appendix A.3, footnote 10, eq. (3.29)"},{"comment":"The step from eq. (4.32) to eq. (4.33) fixes the b_perp dependence up to a Lambda-independent prefactor that may depend on Delta eta. The Fourier transform from eq. (4.33) to eq. (4.34) is then evaluated in the saddle-point approximation; for large Re Gamma_cusp the saddle point produces an additional prefactor of order exp[2 Re Gamma_cusp ln(2 Re Gamma_cusp) - 2 Re Gamma_cusp], whose logarithm is O(sqrt(lambda) ln lambda), not O(lambda^0). The text says that O(lambda^0) terms in the exponent are neglected, but this prefactor is parametrically larger. If eq. (4.34) is intended as a scheme choice for the b-space soft function, that should be stated explicitly; if eq. (4.36) is meant to be the full rapidity derivative d ln chi_1 / d Delta eta, the b-independent term 2 Re Gamma_cusp' ln(2 Re Gamma_cusp) should be included or shown to be removable by a stated scheme. As written, eq. (4.36) is therefore either incomplete or implicitly scheme-dependent in a way that the manuscript does not explain.","section":"Section 4.3, eqs. (4.33)-(4.36)"}],"minor_comments":[{"comment":"The direct numerical verification is presented only for the u_max = 1 term for Delta eta = 1.5 and 2 with Delta theta = 0. The text should state explicitly that the u_max = u_+ integral and the Delta theta = pi case remain unchecked numerically, so the reader does not overestimate the empirical support for the branch-cut assumption.","section":"Appendix B, eq. (B.2)"},{"comment":"Several figures, in particular figures 12, 17, and 18, lack axis labels and legends in the displayed material; adding them would make the numerical comparisons easier to verify.","section":"Figures 12, 17, 18"},{"comment":"There is a typo 'comapring' that should read 'comparing'. Also, the notation for the turning point alternates between u_max and u_max; the paper should use a single symbol consistently.","section":"Section 3.4, text near eq. (3.28)"},{"comment":"The bullet list contains the fragment 'in int is a convex function' which appears to be a typo for 'it is a convex function'. This should be corrected.","section":"Section 3.6, bullet list"},{"comment":"The derivation of the saddle-point value of q treats q as a scalar in the exponent; since q is a two-dimensional transverse vector, the stationary condition should be written for the vector q and the direction of b_perp. The result is unaffected, but the notation should be clarified.","section":"Section 4.3, eq. (4.35)"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is ambitious and generally careful, and the central idea is worth publishing if the branch-cut issue is resolved. The authors themselves flag the gap in appendix A.3, which is commendable, but the current numerical checks are too narrow to close it. I would be willing to look at a revised version that either provides a systematic argument for the branch-cut segments or substantially extends the numerical verification, and that clarifies the scheme dependence in eqs. (4.34) and (4.36)."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, this is a substantial holographic calculation: the all-rapidity cusp anomalous dimension in N=4 SYM at strong coupling, including the dominance exchange between the two saddle families, is genuinely new and checks out against known limits (large-rapidity slope sqrt(lambda)/(4 pi), small-angle bremsstrahlung behavior, zig-zag symmetry at Delta theta = pi). The two-cusp result chi_1 ~ (b_perp^2 Lambda^2)^{-Re Gamma_cusp} and the Collins-Soper kernel coefficient -2 ln(Lambda b_perp) sqrt(lambda)/(4 pi) are also new. Second, the branch-cut contribution to the Lagrange-multiplier integral (App. A.3, footnote 10) is admittedly unresolved. The authors state they have not found a systematic argument that those segments do not modify the asymptotic behavior; the numerical checks in App. B cover only Delta eta = 1.5 and 2, Delta theta = 0, and only the u_max = 1 saddle. The u_max = u_+ integral is left unexamined. This is not an internal contradiction, but it is the weakest link between eq. (3.29) and eq. (3.49), and therefore the main quantitative claims.\n\nWhat the paper does well deserves emphasis. The saddle-point machinery with Lagrange multipliers is careful, explicit, and reproducible: the saddle conditions (3.32)-(3.46) are written out, the subtraction and regularization steps are clear, and the paper consistently flags its own technical gaps rather than hiding them. It also engages the existing Euclidean and lightlike literature rather than ignoring it, and reproduces the established limits. That is real evidence of solid thinking.\n\nThe soft spots are in proportion. The branch-cut issue is load-bearing but not demonstrated false; it is a missing proof plus limited numerics. A referee should ask for either a systematic argument that those segments do not affect the large sqrt(lambda) asymptotics, or a much broader numerical scan covering more Delta eta values, nonzero Delta theta, and both saddle families. The route to the two-cusp result also starts from the 'natural guess' in eq. (4.1) and proceeds via a linearized analysis and dimensional arguments rather than a full two-cusp worldsheet solution. That is heuristic, though reasonable, and it is a further gap. Neither concern strikes me as fatal at this stage, but both need to be addressed before this becomes an established result.\n\nWho is this for: holographers and TMD practitioners who want a strong-coupling prediction for soft functions. It deserves a serious referee, not a desk reject. Send it to review, and push the authors to close the contour issue before it is cited as settled.","headline":"First all-rapidity strong-coupling cusp anomalous dimension from holography, with an unresolved branch-cut contribution as the load-bearing soft spot.","tokens_in":727,"tokens_out":1081,"would_cite":true,"duration_ms":29501,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["11.25.Tq","12.38.-t"],"model":"deepseek-v4-flash","headline":"At every rapidity gap, the strongly coupled cusp anomalous dimension is the smaller of two saddle-point actions; the two families swap dominance at a finite gap, and the two-cusp soft function follows as a power law.","keywords":["cusp anomalous dimension","AdS/CFT correspondence","Wilson loops","Nambu-Goto action","TMD soft function","Collins-Soper kernel","heavy-quark TMD fragmentation","N=4 super Yang-Mills"],"falsifier":"Numerically evaluate the integral in eq. (3.29) directly, following appendix B, at values of $\\Delta\\eta$ other than 1.5 and 2 — for instance 3, 4, and 6 — and at an intermediate internal angle such as $\\Delta\\theta = 8\\pi/9$, then compare both the growth rate and the phase of the result with the least-action prediction of eq. (3.49); a mismatch would show that the branch-cut segments of the deformed contour contribute and that the saddle-point formula is incomplete. A separate check: compute $-\\mathrm{Re}\\,\\Gamma_{\\mathrm{cusp}}$ from the power-law exponent of $\\chi_1(b_\\perp)$ by an independent strong-coupling method, such as a lattice construction or an integrability-based proposal, and see whether it equals the minimum of the two saddle actions across the crossover region.","tokens_in":46202,"feed_emoji":"🧵","tokens_out":24913,"duration_ms":197449,"temperature":0.7,"pith_summary":"Using the AdS/CFT correspondence, this paper establishes how the cusp anomalous dimension — the universal growth rate controlling ultraviolet divergences where two Wilson lines meet at an angle — behaves at strong coupling in $\\mathcal{N}=4$ super Yang-Mills for every rapidity separation $\\Delta\\eta$. The central claim is that for a timelike cusp this quantity is everywhere the minimum of the actions of two families of extremal string surfaces, which exchange dominance at a finite $\\Delta\\eta$; the same machinery is then applied to the vacuum expectation value of two cusped Wilson lines separated in the transverse direction, the configuration underlying heavy-quark transverse-momentum-dependent fragmentation and the TMD soft function. If the calculation is right, that matrix element is a pure power law in the transverse separation with exponent set by the real part of the cusp anomalous dimension, and its lightlike limit supplies the strong-coupling value of the Collins-Soper kernel.","feed_headline":"Two string shapes trade control of the cusp growth rate","feed_subtitle":"At strong coupling the cusp growth rate is the smaller of two saddle actions; a two-cusp soft function follows as a power law.","key_machinery":"The carrier of the argument is the Nambu-Goto action for a string in AdS$_5\\times S^5$, written in scale-invariant coordinates ($a$, $u=z/a$) so that each worldsheet profile reduces to two ordinary functions $\\eta(u)$, $\\theta(u)$. The rapidity gap $\\Delta\\eta$ and the internal angle $\\Delta\\theta$ are not imposed by boundary conditions directly but by Lagrange multipliers $C_\\eta, C_\\theta$ inserted into the action; after the string fields are integrated out, the Wilson loop expectation value becomes a finite two-dimensional integral over these multipliers, and the multipliers themselves become conserved momentum fluxes along the worldsheet, a quantity tied to momentum transfer in the field theory. The saddle points of the multiplier integral are complex for $\\Delta\\eta$ above a critical value and fall into two families distinguished by the turning point of the surface, $u_{\\max}=u_+(C_\\eta,C_\\theta)$ in the bulk versus $u_{\\max}=1$ on the null line; the steepest-descent analysis of which family dominates, and where the dominance switches, is the technical core of the one-cusp calculation. For the two-cusp configuration a third multiplier $C_\\perp$ enforces the transverse separation and equals the string momentum flux $\\Pi_a$; its conjugate variable $q$ fixes the transverse profile mode $x_\\perp = (q/\\sqrt{\\lambda})a^2\\tilde{c}_2 f_2(u)$, with $f_2$ the unique flux-carrying mode and $\\sqrt{\\lambda}/q$ serving as the IR regulator that turns the one-cusp logarithm into $\\ln(\\Lambda b_\\perp)$.","core_discovery":"For a single timelike cusp in strongly coupled $\\mathcal{N}=4$ super Yang-Mills, the paper claims that the cusp anomalous dimension is, for every rapidity gap $\\Delta\\eta$ and internal angle $\\Delta\\theta$, given by the saddle-point evaluation of a two-dimensional integral over Lagrange multipliers: $$\\Gamma_{\\mathrm{cusp}}^{(s)}[\\$\\Delta$\\eta,\\$\\Delta$\\$\\theta$] = \\frac{i\\sqrt{\\$\\lambda$}}{\\pi}\\left(\\int_{L(u_{\\max}^{(s)})} \\frac{du}{$u^{2}$}\\left[\\sqrt{\\frac{1-$u^{2}$}{1-(1+C_\\$theta^{2}$)$u^{2}$ - C_\\$eta^{2}$ $u^{4}$}} - 1\\right] - \\frac{1}{u_{\\max}^{(s)}}\\right),$$ with $s = +$ for surfaces whose turning point sits in the bulk and $s = 1$ for surfaces that reach the null line $u = z/a = 1$, the latter having no Euclidean counterpart. The value realized in the Wilson loop is the one with the smallest real part, and both families exchange dominance at a finite rapidity gap; at small $\\Delta\\eta$ the $u_+$ family reproduces the known small-angle behavior tied to the heavy quark-antiquark potential, while at large $\\Delta\\eta$ both families recover the established result $\\Gamma_{\\mathrm{cusp}} \\sim \\frac{\\sqrt{\\lambda}}{4\\pi}\\Delta\\eta$ with imaginary part tending to $-\\frac{\\sqrt{\\lambda}}{4}$, matching the lightlike computations of refs. [34, 40]. For the two-cusp vacuum matrix element, the paper obtains $\\chi_1(b_\\perp;\\Delta\\eta,\\Delta\\theta) = (b_\\perp^2\\Lambda^2)^{-\\mathrm{Re}\\,\\Gamma_{\\mathrm{cusp}}[\\Delta\\eta,\\Delta\\theta]}$ together with the large-rapidity transverse profile $x_\\perp(a,u) = \\frac{q}{\\sqrt{\\lambda}}a^2\\tilde{c}_2 f_2(u)$, and in the lightlike limit a Collins-Soper kernel $-2\\ln(\\Lambda b_\\perp)\\frac{\\sqrt{\\lambda}}{4\\pi}$ that matches the expected factorization equations (1.5)-(1.7).","pith_inferences":["The exchange of dominance between the two saddle families resembles a first-order transition in $\\Delta\\eta$; if that structure is physical, the derivative of $\\mathrm{Re}\\,\\Gamma_{\\mathrm{cusp}}$ with respect to $\\Delta\\eta$ should change sharply at the exchange point, a feature that an independent computation (for example a lattice or integrability-based one) of the same Wilson loop could test.","The paper offers the linearized profile $x_\\perp(a,u) = \\frac{q}{\\sqrt{\\lambda}}a^2\\tilde{c}_2 f_2(u)$ as an initial condition for integrating the full nonlinear string equations outward from the cusp; actually performing that integration would verify the midpoint matching between the two cusps and show where the one-cusp approximation of eq. (4.31) first fails.","The Lagrange-multiplier and conserved-flux machinery should transfer to non-conformal holographic backgrounds; there the power law would acquire an extra $b_\\perp$ dependence set by the new scale, plausibly turning $\\chi_1(b_\\perp)$ into a holographic diagnostic of flux-tube breaking that could be compared with heavy-hadron energy-energy correlator measurements."],"forward_implications":["The timelike cusp anomalous dimension is now fixed at strong coupling across the entire range of rapidity separation: the $u_+$ saddle gives the small-angle $\\sim -V/\\phi'$ behavior tied to the heavy quark-antiquark potential, both families reproduce the linear lightlike growth $\\frac{\\sqrt{\\lambda}}{4\\pi}\\Delta\\eta$, and the crossover between them is controlled by the $u=1$ family.","The heavy-quark fragmentation matrix element $\\chi_1(b_\\perp)$ is a pure power law, $\\chi_1(b_\\perp;\\Delta\\eta,\\Delta\\theta) = (b_\\perp^2\\Lambda^2)^{-\\mathrm{Re}\\,\\Gamma_{\\mathrm{cusp}}[\\Delta\\eta,\\Delta\\theta]}$, at leading order in the strong-coupling expansion, with the complex part cancelling between the amplitude and its conjugate as expected.","In the lightlike limit the Collins-Soper kernel takes the value $-2\\ln(\\Lambda b_\\perp)\\frac{\\sqrt{\\lambda}}{4\\pi}$, and because the limit is insensitive to whether one or both lines are lightlike, the same value applies to the lightlike-lightlike TMD soft function.","The $u=1$ family — complex saddle surfaces with no Euclidean counterpart — is the one that dominates at large rapidity separation, so Euclidean constructions of cusped Wilson loops necessarily miss the dominant contribution in the lightlike regime."],"supporting_citations":[{"why":"Define the heavy-quark TMD fragmentation matrix element $\\chi_1(b_\\perp)$ that the paper computes at strong coupling, and supply its weak-coupling evolution equations as the consistency check.","marker":"[1, 2]"},{"why":"The Euclidean minimal-surface computation of cusped Wilson loops in AdS/CFT that anchors the method and the small-angle saddle family.","marker":"[33]"},{"why":"The Minkowski-signature lightlike Wilson loop calculation that established the large-rapidity linear growth $(\\sqrt{\\lambda}/4\\pi)\\Delta\\eta$ which both saddle families must reproduce.","marker":"[34]"},{"why":"The lightcone Wilson loop benchmark used to check the $\\Delta\\eta\\to\\infty$ asymptotics of the cusp anomalous dimension.","marker":"[40]"},{"why":"Introduced Lagrange multipliers to enforce string boundary conditions and interpreted them as momentum flow along the worldsheet, the technique the whole calculation is built on.","marker":"[51]"},{"why":"The generalized heavy quark-antiquark potential that fixes the small-angle $\\sim -V/\\phi'$ behavior the $\\Delta\\theta=0$ limit must match.","marker":"[35]"},{"why":"The real-time closed-time-path holography prescription used to give the non-time-ordered two-cusp correlator a string-theory formulation.","marker":"[66, 67]"},{"why":"The Wilson-line derivation of the velocity-dependent cusp anomalous dimension that defines the object being computed at strong coupling.","marker":"[22]"}],"fun_headline_variants":["Two string saddles swap control of cusp growth","Cusp anomaly from min of two saddle actions","Strong coupling cusp: saddle dominance switch","Competing saddles dictate cusp growth rate","Saddle switch sets cusp anomalous dimension"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the cusp anomalous dimension is set entirely by the saddle points of the integral over the constraint-enforcing multipliers: the extra segments of the deformed integration contour that run along the branch cut in the complex plane must contribute nothing at any rapidity gap or angle, a fact the paper verifies numerically at only two values of the rapidity gap and states it cannot yet prove in general.","fun_headline_variants_meta":{"raw":{"variants":["Two string saddles swap control of cusp growth","Cusp anomaly from min of two saddle actions","Strong coupling cusp: saddle dominance switch","Competing saddles dictate cusp growth rate","Saddle switch sets cusp anomalous dimension"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000428,"raw_usage":{"total_tokens":2379,"prompt_tokens":1322,"completion_tokens":1057,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":938,"completion_tokens_details":{"reasoning_tokens":987}},"tokens_in":938,"tokens_out":1057,"duration_ms":10531,"temperature":1.0,"reasoning_tokens":987,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T04:14:24.519427+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Numerically evaluate the integral in eq. (3.29) directly, following appendix B, at values of $\\Delta\\eta$ other than 1.5 and 2 — for instance 3, 4, and 6 — and at an intermediate internal angle such as $\\Delta\\theta = 8\\pi/9$, then compare both the growth rate and the phase of the result with the least-action prediction of eq. (3.49); a mismatch would show that the branch-cut segments of the deformed contour contribute and that the saddle-point formula is incomplete. A separate check: compute $-\\mathrm{Re}\\,\\Gamma_{\\mathrm{cusp}}$ from the power-law exponent of $\\chi_1(b_\\perp)$ by an independent strong-coupling method, such as a lattice construction or an integrability-based proposal, and see whether it equals the minimum of the two saddle actions across the crossover region.","supporting_citations":[{"cited_title":"Light-Cone Wilson Loops and the String/Gauge Correspondence","cited_arxiv_id":"hep-th/0210256","evidence_quote":"The lightcone Wilson loop benchmark used to check the $\\Delta\\eta\\to\\infty$ asymptotics of the cusp anomalous dimension."}],"review_version":1}