{"id":"f10beaea-773c-438b-8b37-d2a1f779df9c","arxiv_id":"2502.04036","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"For holographic confining theories on spheres, a curvature-driven quantum phase transition occurs between a low-curvature branch with flat-space-like IR and a high-curvature regular branch; the transition is first-order above the Efimov bound and at least second-order below it.","lead":"Holographic models of confining quantum field theories on positively curved space-times admit two competing gravity solutions, and the free energy switches between them at a critical curvature. The paper derives when this switch is a sharp first-order transition and when it is continuous, which matters for whether curvature can deconfine strongly coupled sectors in cosmology.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The order-of-transition claim rests on the uncomputed linear map M in Eq. (4.19); without independent derivation or numerical construction of M, the predicted first- vs higher-order behavior is not established.","rationale":"The paper's central claim has two parts: existence of a curvature-driven transition, and its order. The existence part is supported by the IR classification, the dimensional uplift, and the numerical shooting in Section 5, and I do not see a fatal flaw there. The order part, however, is controlled by the free-energy expansion (1.9), which is obtained from the linearized analysis around the type-I solution. The linear map M in (4.19) is not computed; the paper explicitly states that determining it would require solving the linearized equation for all φ, which the authors are not able to do. Section 5.2 fits the functional forms (4.21)-(4.23) to numerical data, which checks consistency but cannot rule out a singular M or a breakdown of linearized propagation in the intermediate region. This is exactly the reader's weakest assumption, and it is a genuine soft spot rather than a manufactured concern. I therefore see no reason to move the CONDITIONAL verdict; an independent linearized-flow computation would settle whether the order prediction holds.","tokens_in":65427,"tokens_out":10146,"duration_ms":111532,"concrete_test":"Numerically construct the type-I background for the d=4 potential (5.1) at b=0.47 and b=0.65. Linearize Eq. (2.21) about SI(φ); at a large φ=φmatch where the IR expansion (3.36) is reliable, launch the two independent solutions δS=e^{β±φ} (with β± from (3.37)) and integrate the linear ODE backward to φ→0. Extract the coefficients of the UV modes φ^{-1+d/Δ-} and φ^{1+2/Δ-}; this gives M, its determinant, and the ratios m^-_C/m^-_R entering (4.35)-(4.36). Compare the predicted (δR(p), δC(p)) from (4.20) with the full nonlinear solutions used in Section 5.2, and check whether the numerical second-derivative jump F1 in (5.24) is reproduced. If the linearized evolution disagrees with the full solutions, or if M is near-singular, the order-of-transition claim is not supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing assumption is in Section 4.2: the relation (4.19) between the IR perturbation amplitudes (S_-, S_+) and the UV data (δR, δC) is asserted through an unknown 2×2 matrix M, on the basis that the type-I solution is an attractor. The linearized modes in (3.36) have Re β± > b for b > bc, so δS/SI grows toward φ→∞; linearization is therefore valid only in an intermediate region, and the matching that fixes S± from the type-II/III boundary data uses the higher-dimensional scaling (3.43)-(3.44), not a solution of the linearized equation across the full bulk. If the linearized flow from that intermediate region to the UV is not captured by the two IR modes, or if M has vanishing determinant or a vanishing m^-_R entry, the expansion (4.32), the exponent δ in (1.10)/(4.33), and the order predictions (1.11)/(4.34)-(4.38) do not follow. Section 5.2 fits the amplitudes in (4.21)-(4.23) to numerics; this verifies consistency of the assumed functional forms but does not independently determine M or test the attractor assumption. The paper is candid that M is not computed analytically, but the central claim about first- vs higher-order transitions (and the specific order in the monotonic case) is load-bearing on this matrix.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies holographic Einstein-scalar theories dual to confining QFTs on a sphere S^d. It classifies bulk solutions into three IR types, argues that the Euclidean path integral switches from the singular-but-acceptable type II branch to the regular type III branch at a critical curvature, and derives that, depending on the exponential potential exponent b relative to the Efimov bound b_E, the transition is first-order (b > b_E) or at least second-order (b < b_E), with a possible finite order 1+ceil(delta). The claims are supported by analytic IR expansions and by numerical solutions in d=4 for b=0.65 and b=0.47.","tokens_in":65704,"tokens_out":8094,"duration_ms":84487,"significance":"If correct, this is a generic curvature-driven quantum phase transition in holographic confining theories, with a sharp qualitative prediction for the order of the transition as a function of b and an explicit exponent delta. The paper's strengths are its combination of a higher-dimensional uplift, a clean IR classification, explicit free-energy expansions, and numerical verification of both regimes. The manuscript is also unusually candid about the points where its analytic control weakens, which is a genuine virtue.","major_comments":[{"comment":"The central derivation of the order of the transition in the monotonic case (ii) relies on the linear map M between the IR perturbation amplitudes (S_-, S_+) and the UV data (δR, δC). The paper explicitly states that M is not computed analytically, and it assumes invertibility and a non-vanishing m^-_R entry. These assumptions enter directly in Eq. (4.32) and in the discontinuity formula Eq. (4.36); if det M = 0 or m^-_R = 0, the predicted order and even the existence of the δ-term do not follow. The numerical fits in Section 5.2 determine only ratios such as m^-_C/m^-_R and the coefficients n_R^±, n_C^±; they do not determine det M or independently test the invertibility assumption. I request a numerical construction of M (for example, by integrating the linearized perturbation equation around the type I solution) or a clear restatement of the order predictions as conditional on these assumptions.","section":"§4.2, Eq. (4.19)"},{"comment":"The linearized solution (3.36) is obtained only in the IR region φ → ∞, where it diverges relative to S_I; the text therefore states that it is valid only in an intermediate range. The subsequent matching to the UV uses the assumption that the type I solution is an attractor, justified only by the numerical observation in footnote 15 that type II and III solutions remain close to type I in the UV. This observation concerns the nonlinear solutions, not the linearized flow that determines M. A quantitative check, such as integrating the linearized ODE for δS on top of the numerical type I solution and comparing the transfer matrix with the IR asymptotics, would directly test the attractor assumption and provide M. Without it, Eqs. (4.20)–(4.23) remain an ansatz with fitted coefficients.","section":"§3.3, Eq. (3.36), and §4.2 attractor assumption"},{"comment":"The finite-order formula 1+⌈δ⌉ is derived under the extra assumptions F1 = 0 and Fδ ≠ 0, with δ non-integer. The paper acknowledges this with 'assuming the transition is at least third-order' and 'generic situation', but Section 1.1 presents the formula as a definite prediction. Because F1 is built from the same fitted matrix elements, there is no argument that F1 vanishes; the numerical example b = 0.47 finds a non-zero F'' discontinuity, i.e. a second-order transition. I recommend separating the robust statement (at least second-order, under the M assumptions) from the conditional higher-order formula, and making explicit that the higher-order possibility is not realized in the numerical example.","section":"§4.3, Eqs. (4.36)–(4.38)"}],"minor_comments":[{"comment":"The phrase 'depending on the leading asymptotic exponent' could be read as saying the order is fixed by b alone; Section 4.3 makes clear that below b_E the order also depends on the subleading shape of the potential and on F1. Please align the abstract with that qualification.","section":"Abstract and §1.1"},{"comment":"'all they way to the UV' should be 'all the way to the UV'.","section":"§4.2, first paragraph"},{"comment":"'one of they main interests' should be 'one of the main interests'.","section":"Introduction, §1"},{"comment":"The coefficient B is used before it is introduced; please define it immediately before Eq. (4.10) or add a sentence after it.","section":"Eq. (4.10)"},{"comment":"The asymptotic values R_I^II and R_I^III differ by about 4×10^-4; although the quoted errors overlap, the overlap is marginal. A sentence explaining the fitting procedure and why the overlap is taken as evidence of equality would improve reproducibility.","section":"§5.2, Eq. (5.12)"}],"recommendation":"major_revision","confidential_remarks":"I see no grounds for rejecting on novelty or competence. The main risk is that the order-of-transition claim is presented more strongly than the analytic control justifies. A revision that either supplies a numerical construction of M or carefully conditionalizes the central claims should make this acceptable for JHEP."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Good paper, worth engaging. The genuinely new thing is the classification of IR branches for confining Einstein-scalar theories on S^d: type I, II, III, the curvature-driven competition between them, and the claim that the order of the transition is controlled by the exponent b of the exponential potential. Earlier work covered special compactifications or negative curvature; this paper handles generic exponential potentials with a single scalar and gives a concrete formula for the higher-order exponent below the Efimov bound. The numerical examples for b=0.65 and b=0.47 support the two regimes, and the dimensional-reduction uplift gives a useful independent perspective on where the branches come from. The authors are candid about exclusions: IHQCD/V-QCD sit exactly at b=bc where the IR expansion breaks down, and the N=1 toroidal case is left open.\n\nWhere I agree with the conditional verdict: the order-of-transition result in the monotonic regime rests on the matrix M in (4.19), which is not computed. The paper says so. The coefficients n± are fitted to numerics, and consistency of a fit is not an independent derivation. If M were singular or the linearized flow from the intermediate IR region to the UV were not captured by the two IR modes, the expansion (4.32) and the exponent δ would not follow. I see no positive evidence that M is singular, and the numerics are consistent with the assumed forms, so this is a gap, not a demonstrated failure.\n\nWhere I push back on the stress-test note: the first-order claim above the Efimov bound does not really depend on M. The sine oscillations in (4.21)-(4.22) produce infinitely many turning points whenever the coefficients are nonzero, independent of the entries of M. So the first-order branch is robust; the conditional part is the specific higher-order behavior below b_E.\n\nMinor points: no code is released, and the near-transition numerics have small overlap artifacts, though the authors identify them. The free-energy framework leans on the authors' earlier papers; that is legitimate because the machinery is established, and the differential relation (4.16) is a sensible check.\n\nThis paper deserves a serious referee. It is clear, honest, and advances the program. I would send it to review, with a request that the referee examine the matching assumption and ideally ask for a numerical construction of M or at least a third b value in the monotonic regime.","headline":"Solid, honest holography paper with a genuinely new branch classification and a plausible order-of-transition claim that still rests on an uncomputed matching matrix; worth refereeing, not fully settled.","tokens_in":66332,"tokens_out":3134,"would_cite":true,"duration_ms":36303,"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":"A holographic confining theory on a sphere has a curvature-driven phase transition: first-order above the Efimov bound, at least second-order below it.","keywords":["holographic QFT","quantum phase transitions","confinement","Einstein-scalar theory","Efimov oscillations","curvature-driven transition","sphere compactification","renormalized free energy"],"falsifier":"Integrate the linearized perturbation equation around the numerical type I solution for a monotonic-regime potential (for example the $d=4$ potential with $b=0.47$) from $\\phi\\to\\infty$ down to $\\phi=0$ and extract the matrix $M$ in (4.19). If $M$ is singular, or if $\\delta S(\\phi)/S_I(\\phi)$ fails to remain small in the intermediate region, then the free-energy expansion (1.9) and the finite-order bound (1.11) are wrong; a high-precision direct evaluation of $F''_{III}(R_c)-F''_{II}(R_c)$ would then decide whether the transition is second-order or higher.","tokens_in":65110,"feed_emoji":"🔄","tokens_out":11173,"duration_ms":100281,"temperature":0.7,"pith_summary":"The paper argues that a holographic gauge theory which confines on flat space undergoes a genuine quantum phase transition when placed on a sphere of sufficiently large curvature. On the gravity side, two competing bulk geometries exist: at low curvature the scalar runs to infinity much as in flat space, while above a critical dimensionless curvature $R_c$ a regular interior endpoint dominates. The order of the transition is controlled by the exponent $b$ in the leading exponential falloff of the bulk potential. Above the Efimov bound $b_E$ the transition is first-order; below it the transition is at least second-order, and when the second derivative of the free energy is continuous the order is exactly $1+\\lceil\\delta\\rceil$ with a computable exponent $\\delta>1$. This gives a concrete, model-independent signature of how curvature competes with the scale that drives confinement.","feed_headline":"A critical curvature flips a confining QFT between two phases","feed_subtitle":"Above a critical radius the bulk geometry changes character; below the Efimov bound the transition is second-order or higher.","key_machinery":"The load-bearing object is the first-order formulation of Einstein-scalar gravity in terms of $S(\\phi)=\\dot\\phi$, $W(\\phi)=-2(d-1)\\dot A$ and $T(\\phi)=d\\kappa e^{-2A}$, which reduces the equations of motion to a single second-order equation for $S(\\phi)$. IR boundary conditions divide solutions into three types: type III with a regular endpoint $\\phi\\to\\phi_0$, type II with $\\phi\\to\\infty$ that uplifts to a regular geometry in which the internal sphere shrinks, and type I, the critical interface solution where both spheres shrink. The Efimov bound $b_E$ appears in the exponents $\\beta_\\pm$ of linearized perturbations around the type I solution: for $b>b_E$ the exponents are complex and produce multi-valued Efimov spirals in the vev $C(R)$; for $b<b_E$ they are real and $C(R)$ can be single-valued. The renormalized free energy satisfies $F'(R)=N(2C(R)/R^3-1/(96R))$, which turns continuity of $C(R)$ into continuity of $F'(R)$ and yields the expansion (1.9) that fixes the order of the continuous transition.","core_discovery":"The central claim is that for any confining Einstein-scalar holographic theory on a sphere there exists a critical dimensionless curvature $R_c$ separating two classes of bulk saddle points. For $R>R_c$ the Euclidean path integral is dominated by type III solutions with a regular endpoint at a finite scalar value $\\phi_0$; for $R<R_c$ the dominant type II solutions have the same IR endpoint $\\phi\\to\\infty$ as the flat-space confining solution, so the low-curvature phase inherits the discrete gapped spectrum of flat-space confinement. The order of the transition is set by $b$ in $V\\sim V_\\infty e^{2b\\phi}$. For $b>b_E$, where $b_E=2/\\sqrt{(d-1)(9-d)}$ is the Efimov bound, the branches of solutions oscillate around the critical type I solution and the transition is first-order. For $b_c<b<b_E$, the transition may be first-order or continuous; in the continuous case the paper proves that the free-energy difference obeys $(F'_{III}(R)-F'_{II}(R))/(2N)=F_1(R-R_c)/R^3+F_\\delta(R-R_c)^\\delta/R^3+\\ldots$ with $\\delta=((d-1)b+2\\sqrt{1-(b/b_E)^2})/((d-1)b-2\\sqrt{1-(b/b_E)^2})>1$, so the transition is at least second-order and, if $F_1=0$, has order $1+\\lceil\\delta\\rceil$ rather than infinite order. Numerical solutions in $d=4$ with the potential $V=-d(d-1)/\\ell^2+(\\Delta_-(\\Delta_--d)/(2\\ell^2)-4V_\\infty b^2)\\phi^2+4V_\\infty\\sinh^2(b\\phi)$ show a first-order transition for $b=0.65$ and a second-order transition for $b=0.47$.","pith_inferences":["A natural extension, not pursued in the paper, is to test whether the topological susceptibility (from a bulk axion profile) distinguishes the two phases as it does in thermal deconfinement; the paper raises this possibility but leaves it open.","Since the paper states that its IR expansion breaks down at $b=b_c$ and for power-law prefactors such as $\\phi^{1/2}e^{2b_c\\phi}$, adapting the analysis to potentials of that form could show whether the curvature transition survives in models closest to QCD and whether its order changes.","The numerical conclusion that the $b=0.47$ transition is second-order relies on fitting the ratios $m^-_C/m^-_R$; a direct high-precision computation of the linearized map $M$ in (4.19) would reveal whether all monotonic-regime potentials give second-order transitions or whether some realize the third-or-higher-order case.","For uplift dimensions with $d+N>9$, the Efimov regime is absent and the paper does not settle whether first-order transitions can occur there; a dedicated higher-dimensional scan would close this gap."],"forward_implications":["Above the critical curvature the dominant geometry has a regular endpoint and the dual spectrum is continuous but gapped by the curvature, while below it the spectrum is discrete; the transition therefore changes the spectral character even when it is only second-order.","The existence and order of the transition are insensitive to subleading terms of the form $e^{2\\gamma\\phi}$ with $\\gamma<b$, as long as the leading exponent lies in the confining range $b_c<b<b_G$.","In the Efimov regime $b>b_E$, the multi-valued spiral in $C(R)$ implies a range of curvatures with coexisting bulk geometries and a swallow-tail free energy, so the dominant phase jumps discontinuously at $R_c$.","In the monotonic regime with no peaks, the type I solution is the unique saddle at $R_c$ and is thermodynamically stable there, and the transition order is at most $1+\\lceil\\delta\\rceil$, never infinite.","In the uplifted picture the transition is a change in which sphere of the warped product $S^d\\times S^N$ shrinks to zero size, so it can be read as a conifold-type transition in the higher-dimensional geometry."],"supporting_citations":[{"why":"It establishes the confinement criterion for exponential potentials and the Wilson-loop area law that defines the class of theories under study.","marker":"[12]"},{"why":"It classifies acceptable singular IR endpoints and the exponential asymptotics used to identify the physically relevant solutions.","marker":"[17]"},{"why":"It supplies the generalized dimensional reduction from pure Einstein gravity on an internal sphere to Einstein-scalar theory with an exponential potential.","marker":"[18]"},{"why":"It provides the companion reduction framework used to map the IR solutions to higher-dimensional geometries.","marker":"[19]"},{"why":"It develops the first-order formulation and curved-domain-wall solution method on which the classification of type I, II and III solutions rests.","marker":"[26]"},{"why":"It gives the warped $S^d\\times S^N$ solutions and the critical type I geometry, including the exponents producing Efimov oscillations.","marker":"[32]"},{"why":"It analyzes the Efimov spiral of holographic saddle points on $S^2\\times S^2$ and connects it to phase transitions.","marker":"[33]"},{"why":"It derives the renormalized free energy and the $F'(R)=N(2C/R^3-1/(96R))$ relation used to compute the order of the transition.","marker":"[44]"}],"fun_headline_variants":["Curvature flips confining QFT between two holographic phases","Critical curvature drives first- or higher-order holographic transition","On a sphere, curvature changes the phase of holographic confinement","Phase transition in confining QFT on a positively curved space"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The paper assumes that small perturbations of the critical (type I) solution stay small all the way from the deep interior to the boundary, so that the unknown linear map connecting interior and boundary data is well defined and invertible; if that map is singular or the perturbations grow, the predicted exponent and the order of the transition do not follow.","fun_headline_variants_meta":{"raw":{"variants":["Curvature flips confining QFT between two holographic phases","Critical curvature drives first- or higher-order holographic transition","On a sphere, curvature changes the phase of holographic confinement","Phase transition in confining QFT on a positively curved space"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00056,"raw_usage":{"total_tokens":2710,"prompt_tokens":1043,"completion_tokens":1667,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":659,"completion_tokens_details":{"reasoning_tokens":1595}},"tokens_in":659,"tokens_out":1667,"duration_ms":13125,"temperature":1.0,"reasoning_tokens":1595,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-08T23:46:37.235000+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Integrate the linearized perturbation equation around the numerical type I solution for a monotonic-regime potential (for example the $d=4$ potential with $b=0.47$) from $\\phi\\to\\infty$ down to $\\phi=0$ and extract the matrix $M$ in (4.19). If $M$ is singular, or if $\\delta S(\\phi)/S_I(\\phi)$ fails to remain small in the intermediate region, then the free-energy expansion (1.9) and the finite-order bound (1.11) are wrong; a high-precision direct evaluation of $F''_{III}(R_c)-F''_{II}(R_c)$ would then decide whether the transition is second-order or higher.","supporting_citations":[{"cited_title":"Holography for Einstein-Maxwell-dilaton theories from generalized dimensional reduction","cited_arxiv_id":"1110.2320","evidence_quote":"It provides the companion reduction framework used to map the IR solutions to higher-dimensional geometries."}],"review_version":1}