{"id":"7be14006-8b4a-402d-aafd-f126889bec28","arxiv_id":"2501.15936","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For even d > 2, the paper derives the spectral dimension of Liouville Brownian motion, shows it depends on the point thickness, and constructs higher-dimensional quantum cones.","lead":"Researchers computed the spectral dimension of Liouville Brownian motion in dimensions greater than two, showing it depends on both the coupling and the local thickness of the random field, unlike in two dimensions. They also built higher-dimensional analogues of quantum cones and proved that these cones are stationary along the diffusion.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem B proves a Laplace-transform threshold (Definition 2.8), not the short-time heat-kernel diagonal exponent claimed in the abstract; the missing Tauberian step leaves the headline 'depends on thickness' unproven.","rationale":"The paper's core mathematical content appears sound: Theorem A's identification of the spherical average via a Langevin equation is plausible and the power-spectrum computation checks out (modulo a small proof typo where 'b = c(d)(d-2)!pi' should be 'b^2/2 = c(d)(d-2)!pi'); the LBM construction follows the 2D roadmap; Proposition 5.17's lower and upper bounds are coherent, with the key exponent arising from the Riesz-potential estimate Lemma B.3. The reader's weakest_assumption targeted the imported GMC small-ball estimates (Lemmas A.1-A.4). I agree these are the main external dependencies, but the paper does provide proofs for A.3 and A.4, and the exponents are the standard multifractal ones. The single most load-bearing issue I see is instead the gap between Definition 2.8 (a Laplace-transform threshold) and the abstract's 'short-time asymptotics of the heat kernel along the diagonal'. The paper does not prove that the threshold chi yields p_t(0,0) ≈ t^{-(chi+1)}; it only computes the abscissa of convergence of the Laplace transform. This is a real interpretive gap, and it is the reason the reader's CONDITIONAL verdict is appropriate. A Tauberian theorem or direct heat-kernel bounds would settle whether the abstract's claim holds. Because the precise theorem is internally consistent, the verdict should remain CONDITIONAL rather than move to ACCEPT or REJECT.","tokens_in":54219,"tokens_out":43849,"duration_ms":393897,"concrete_test":"Prove a Tauberian theorem for the measure t p_{h,alpha,t}(0,0) dt: show that if the Laplace transform ∫_0^∞ t^chi e^{-t} p_t(0,0) dt is finite for chi > chi and infinite for chi < chi, then p_t(0,0) = t^{-chi-1+o(1)} as t→0, using the semigroup's sub-Markovianity and the Brownian-bridge representation (2.11). Alternatively, directly bound p_t(0,0) from below and above using (2.12) and the estimates in Section 5.3; if p_t(0,0) is not regularly varying, the abstract's 'short-time asymptotics' claim must be weakened to 'Laplace-transform spectral dimension'.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central advertised claim is that the spectral dimension of the LBM, i.e. 'the short-time asymptotics of the heat kernel along the diagonal', depends on both gamma and the thickness of the starting point. The actual Theorem B, via Proposition 5.17, proves a threshold in the Laplace transform of the heat-kernel measure: it shows that m_{h,alpha,chi}(0) = ∫_0^∞ E_{0,0,t}[(F_{h,alpha}(t))^chi e^{-F_{h,alpha}(t)}] p_t(0,0) dt is finite for chi > chi and infinite for chi < chi, with chi = (d-2)/(2+alpha^2/2 - alpha beta). Definition 2.8 then sets the spectral dimension to 2(chi+1). To identify this with the usual short-time decay p_{h,alpha,t}(0,0) ≈ t^{-(chi+1)}, one needs a Tauberian argument or direct two-sided heat-kernel bounds; the paper provides neither. Without that step, the abstract's phrasing overstates what is proved: the computed quantity is the abscissa of convergence of a Laplace transform, not necessarily the exponent of p_t(0,0). The paper is transparent about this (Sections 2.2 and 5.3), and Remark 2.10 explicitly leaves the thick-point version as a conjecture, so the 'depends on thickness' claim for the original field h is not fully established. This concern does not invalidate Theorem B as a precise statement, but it is load-bearing for the paper's headline interpretation.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper studies analogues of two-dimensional Liouville quantum gravity in dimensions d>2. The authors construct the Liouville Brownian motion (LBM) associated with the whole-space log-correlated Gaussian field, using a clock process defined by Gaussian multiplicative chaos (GMC). Their main result (Theorem B) computes the spectral dimension of this LBM at the origin for the field h - beta log|.|, in the sense of Definition 2.8, obtaining d_{h-beta log|.|,alpha}(0) = 2 + 2(d-2)/(2 + alpha^2/2 - alpha beta). They also prove Theorem A, which identifies the spherical average process of the field in even dimensions as the integral of a stationary Gaussian Markov process satisfying a Langevin equation; Theorems C and D, which construct the beta-quantum cone in even dimensions as a total-variation limiting object; and Theorem E, which proves that the law of the alpha-quantum cone is invariant under shifts along its LBM trajectories. The proofs use GMC estimates, stochastic calculus, and Markov process theory, and the paper is transparent about the distinction between the proved statements and the conjectural thick-point extension.","tokens_in":54491,"tokens_out":10854,"duration_ms":88535,"significance":"If the results are correct, this is a significant step in extending LQG-type results beyond two dimensions. The spectral dimension formula is explicit and parameter-free, and the identification of the spherical average process in even dimensions opens the door to constructions (quantum cones, invariance properties) previously available only for d=2. The paper also provides a precise statement of what is proved, leaving the thick-point version of the spectral dimension as a conjecture. The proofs are detailed and follow established techniques, and the paper credits its reliance on standard GMC multifractal results. The main weakness is that the advertised interpretation of the spectral dimension as the short-time heat-kernel asymptotics is not directly proved.","major_comments":[{"comment":"The spectral dimension is defined in Definition 2.8 via the finiteness threshold of m_{h,alpha,chi}(x) in (2.13), not via the short-time heat-kernel asymptotic (2.9). The abstract states that the paper computes 'the spectral dimension, i.e., the short-time asymptotics of the heat kernel along the diagonal,' but no Tauberian theorem or two-sided heat-kernel bounds are provided to show that the threshold in (2.13) determines the exponent of p_{h,alpha,t}(0,0) as t tends to 0. This is a load-bearing gap: without such a step, the headline claim that the short-time heat-kernel decay depends on both gamma and beta is not established; what is computed is the abscissa of a Laplace-type integral. Please add a Tauberian argument or revise the abstract and introduction to describe exactly the quantity that is proved.","section":"2.2 and Abstract"},{"comment":"The abstract's claim that the spectral dimension depends on 'the thickness of the starting point' is not proved for the original field h. Theorem B is stated for the field h - beta log|.| at the origin, and the transfer to beta-thick points of h is explicitly left as a conjecture in Remark 2.10. The abstract and Section 1 should be rephrased to distinguish between the proved statement and the conjectural extension, e.g., by saying that the spectral dimension for h - beta log|.| is computed and that the thick-point version is expected to hold.","section":"2.2, Remark 2.10"}],"minor_comments":[{"comment":"In the sentence 'By choosing b >0 sufficiently large (depending on u but not on b)', the symbol b appears to be used for two different parameters; the intended meaning is likely 'not on \\tilde{b}'. Please fix the notation.","section":"6.2, proof of Theorem C"},{"comment":"The notation \\mathbf{h} = h - beta log|.| is introduced and used in Proposition 5.17 and the lemmas, but Theorem B and the abstract use the notation h - beta log|.|; please ensure the notation is defined at first use and used consistently.","section":"5.3"},{"comment":"In the proof of Lemma 5.18, the notation 'ot(1)' is used without definition; please define o_t(1) or use standard Landau notation.","section":"5.3, Lemma 5.18"},{"comment":"The abstract contains a missing space in 'dimensiond >2'; please correct the typo.","section":"Abstract"},{"comment":"The proof of Lemma A.4 relies on the multifractal spectrum of GMC measures at q = sqrt(2d)/gamma, citing [BP24, Theorem 3.26]. Since this estimate is load-bearing for the finiteness proof in Section 5.3, please state this specific external input explicitly in the main text or in the appendix.","section":"Appendix A, Lemma A.4"},{"comment":"The caption of Figure 1 would benefit from a description of the simulation parameters (e.g., time step and finite horizon) for reproducibility.","section":"Figure 1"}],"recommendation":"major_revision","confidential_remarks":"This is a solid paper with a clear and honest exposition of its main results. The central issue is that the abstract and introduction overstate what is proved: the spectral dimension is defined as a Laplace-transform threshold, not as the short-time heat-kernel exponent, and the thick-point dependence is conjectured, not proved. The missing Tauberian step is not merely cosmetic, since it is the bridge between Theorem B and the advertised physical interpretation. I would recommend major revision, requiring either a Tauberian argument or a careful revision of the claims. The remaining results (Theorems A, C, D, E) are interesting and appear correct, though their proofs rely on standard GMC estimates; the authors should make those dependencies explicit. The paper is within the scope of the journal and would be a valuable addition if the claims are aligned with the proofs."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Know this going in: Theorem B is not a short-time heat-kernel asymptotic theorem, despite the abstract. What is actually proved is a threshold in the Laplace transform of the heat-kernel measure (Definition 2.8). The step from that threshold to the claimed diagonal exponent p_t(0,0) ~ t^{-(chi+1)} is missing—no Tauberian theorem, no two-sided heat kernel bounds. The paper is transparent about this: Remark 2.10 explicitly leaves the thick-point version as conjecture. So the advertised 'spectral dimension depends on thickness' reading is an interpretation, not a theorem.\n\nThat said, the paper is a serious technical contribution. Theorem A, identifying the spherical average process in even d>2 as the integral of a stationary Gaussian Markov process / multivariate OU process, is new and clean. Theorem B is a precise statement about the abscissa of convergence of the Laplace transform of the LHK; if that is taken as the definition of spectral dimension, the formula is the result. The quantum cone construction (Theorems C–E) is substantial, and the proof strategy—tightness of the recentred process, total-variation convergence via mixing, then the local limit—looks sound. The paper relies on standard GMC small-ball estimates (Lemmas A.1–A.4) as black boxes; they are imported from the literature, not re-derived, but they are well-established.\n\nThe soft spot is the one above: the gap between the abstract and Theorem B is load-bearing for the paper's headline. This needs fixing in revision, either by supplying the Tauberian argument or by softening the claims to match what is proved. There's also a smaller caveat that the spectral dimension exponent inherits any hidden corrections in the GMC multifractal estimates. Neither issue undermines the precise statements of Theorem B or the cone theorems.\n\nWho should read it: people working on higher-dimensional LQG, Gaussian multiplicative chaos, or diffusion in random media. It deserves a serious referee. I'd send it to peer review with a request to address the Tauberian gap or revise the abstract.","headline":"A serious technical paper whose abstract overstates Theorem B: the proved spectral-dimension result is a Laplace-transform threshold, not short-time heat-kernel asymptotics, though the gap is clearly flagged in the text.","tokens_in":55097,"tokens_out":2831,"would_cite":true,"duration_ms":26862,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60J65","60G60","60J60","60G57"],"pacs":[],"model":"deepseek-v4-flash","headline":"For $d \\geq 2$, the spectral dimension of the Liouville Brownian motion is $2 + 2(d-2)/(2 + \\alpha^2/2 - \\alpha\\beta)$, and even-dimensional quantum cones are constructed and shown to be LBM-stationary.","keywords":["Liouville Brownian motion","spectral dimension","Gaussian multiplicative chaos","log-correlated Gaussian field","quantum cone","whole-space LGF","even-dimensional Langevin representation","Liouville heat kernel"],"falsifier":"Compute, by simulation of Brownian bridges in $d=4$ with a fixed $\\beta$ and $\\alpha$, the threshold $\\chi$ at which $\\mathbb{E}^{0,0,t}[(F_{h-\\beta \\log|\\cdot|,\\alpha}(t))^\\chi]$ switches from diverging to converging as $t\\to 0$; if the threshold is not $\\chi(\\beta) = (d-2)/(2+\\alpha^2/2-\\alpha\\beta)$, the formula fails.","tokens_in":53938,"feed_emoji":"📐","tokens_out":9214,"duration_ms":78411,"temperature":0.7,"pith_summary":"This paper works out the short-time behaviour of the heat kernel of Liouville Brownian motion (LBM) in dimension $d > 2$, the diffusion obtained by time-changing a Brownian motion by the Gaussian multiplicative chaos measure of a log-correlated Gaussian field. It proves that the spectral dimension\\u2014the exponent in the diagonal decay of the heat kernel\\u2014is no longer always 2, as in $d = 2$, but is given explicitly by $2 + 2(d-2)/(2 + \\alpha^2/2 - \\alpha\\beta)$, where $\\alpha = Q - \\sqrt{Q^2-4}$ is the coupling parameter of the Brownian clock and $\\beta$ is the local thickness of the starting point. If correct, this says that in high dimensions the local geometry of the random metric remembers both the strength of the field and how \\u201cthick\\u201d the starting point is. The paper also identifies the spherical average of the field in even dimensions with the integral of a stationary Gaussian Markov process, and uses this to construct higher-dimensional analogues of $\\beta$-quantum cones and to prove a stationarity property of the associated diffusion.","feed_headline":"Spectral dimension in d>2 depends on local thickness","feed_subtitle":"In d>2 the short-time heat decay is set by both the coupling γ and the local thickness β, unlike in 2D.","key_machinery":"The argument runs on two objects. First, the GMC measure $\\mu_{h,\\gamma}(dx) = \\lim_{\\varepsilon\\to0} \\varepsilon^{\\gamma^2/2} e^{\\gamma h_\\varepsilon(x)} dx$, used to define the clock process $F_{h,\\alpha}(t) = \\lim_{\\varepsilon\\to0}\\int_0^t \\varepsilon^{\\alpha^2/2}e^{\\alpha h_\\varepsilon(B_s)}ds$; the LBM is the Brownian motion time-changed by the inverse of this clock. The spectral-dimension computation converts the diagonal heat kernel into Laplace transforms of bridge moments $m_{h,\\alpha,\\chi}(0) = \\int_0^\\infty \\mathbb{E}^{0,0,t}[(F_{h,\\alpha}(t))^{\\chi} e^{-F_{h,\\alpha}(t)}] p_t(0,0)\\,dt$, so the problem reduces to locating the finiteness threshold in $\\chi$, which is controlled by multifractal small-ball estimates for GMC measures near thick points. Second, for even $d>2$, the spherical average process $(h_{e^{-t}}(0))_{t\\in\\mathbb{R}}$ is shown to be the integral of the unique stationary solution of a $(d-2)/2$-dimensional Langevin equation whose drift is a Frobenius companion matrix with characteristic roots $-(d-2k)$; this \\u201csmoothed Brownian motion\\u201d representation powers the construction of quantum cones, whose radial part is built by recentring the spherical average at a hitting time and letting the recentring level tend to infinity.","core_discovery":"The central claim is Theorem B: for $d \\ge 2$, $Q > \\sqrt{2d}$, and $\\beta \\in (-\\infty, Q)$, the spectral dimension at the origin of the Liouville Brownian motion driven by the field $h - \\beta \\log|\\cdot|$ is $$d_{h-\\$\\beta$\\log|\\cdot|,\\$\\alpha$}(0) = 2 + \\frac{2(d-2)}{2 + \\$alpha^{2}$/2 - \\$\\alpha$\\$\\beta$},$$ with $\\alpha = Q - \\sqrt{Q^2-4}$ and $Q = d/\\gamma + \\gamma/2$. The exponent comes from a sharp threshold in the moments of the clock process over Brownian bridges: quantities $\\mathbb{E}[F_{h,\\alpha}(t)^{\\chi}]$ are infinite below $\\chi = (d-2)/(2 + \\alpha^2/2 - \\alpha\\beta)$ and finite above it. In dimension 2 the formula collapses to 2, independent of $\\alpha$ and $\\beta$; in dimensions $d>2$ it varies with both, equalling $d$ when $\\beta = \\alpha/2$ and increasing in $\\beta$. The paper also constructs the $d$-dimensional quantum cone for even $d>2$ as a local limit of recentred fields, and proves that for $\\alpha = Q - \\sqrt{Q^2-4}$ the law of the $\\alpha$-quantum cone is invariant under shifts along the trajectories of the corresponding Liouville Brownian motion.","pith_inferences":["Editorial inference: because the spectral-dimension formula uses only multifractal small-ball exponents that are dimension-independent in form, the same explicit value probably holds in odd dimensions too, even though the quantum-cone construction here is restricted to even $d$.","Editorial inference: if the small-ball power laws acquire $\\log$ corrections at the boundary value $\\beta = Q$, the threshold in $\\chi$ could shift; checking the rate of divergence at the boundary would test the robustness of the formula.","Editorial inference: the stationarity modulo scaling proved for the $\\alpha$-quantum cone suggests that a higher-dimensional analogue of the mated-CRT scaling limit, if it exists, would converge to this LBM; the Voronoi random-walk problem listed in the paper is a concrete instance where the clock exponent $\\alpha$ should appear.","Editorial inference: the spectral dimension at a thick point should equal the formula for every $x\\in T_\\beta$, not just at the origin for $h-\\beta\\log|\\cdot|$; the paper states this as a plausible stronger version, and it would follow if a uniform-in-space version of the bridge-moment estimates could be proved."],"forward_implications":["In $d>2$ the spectral dimension of the Liouville heat kernel is a genuinely local quantity: starting the diffusion at points of different thickness $\\beta$ gives different short-time decay exponents, with thicker points having larger spectral dimension.","At $\\beta = \\alpha/2$ the spectral dimension equals the Euclidean dimension $d$, and in the limit $\\alpha\\to0$ the formula returns $d$ for every $\\beta$; the heat kernel therefore interpolates between Brownian and genuinely random-metric behaviour.","The even-dimensional quantum cone construction gives a concrete local-limit object for higher-dimensional Liouville quantum gravity, with a radial law built from the stationary Langevin solution rather than from a Brownian motion as in $d=2$.","For $\\alpha = Q-\\sqrt{Q^2-4}$, the law of the quantum cone seen from the Liouville particle is stationary up to spatial rescaling, which identifies a natural diffusion to study on these random geometries."],"supporting_citations":[{"why":"Supplies the d=2 spectral-dimension method via bridge moments and Laplace transforms that Theorem B extends to d>2.","marker":"[R V14b]"},{"why":"Establishes LBM as a Markov process through positive continuous additive functionals; the d>2 construction and heat-kernel continuity arguments are adapted from it.","marker":"[GR V16]"},{"why":"Introduces the time-changed Brownian construction of planar LBM used here as Definition 2.4.","marker":"[Ber15]"},{"why":"Provides the thick-point dimension theorem and GMC multifractal results used in the small-ball estimates behind Proposition 5.17.","marker":"[R V14a]"},{"why":"Supplies the GMC convergence theory, multifractal spectrum, and coordinate-change formula used throughout Sections 5 and A.","marker":"[BP24]"},{"why":"Gives the covariance kernel of the spherical average process, the starting point for Theorem A's Langevin identification.","marker":"[LSSW16]"},{"why":"Defines the 2D quantum cone as a recentring limit, which Theorem D generalises to even d>2.","marker":"[DMS21]"},{"why":"Offers the earlier higher-dimensional LBM construction whose differences are discussed and from which Lemma 5.7 is imported.","marker":"[SHKS24]"}],"fun_headline_variants":["In d>2, spectral dimension depends on gamma and beta","Spectral dimension in d>2: gamma and beta both matter","Alpha-quantum cone invariant under LBM shifts in d>2","For d>2, LBM spectral dimension varies with local thickness","In d>2, short-time heat decay depends on gamma and beta"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof assumes that the Gaussian multiplicative chaos measure of a small ball near a point of thickness $\\beta$ scales exactly as $r^{d + \\gamma^2/2 - \\beta\\gamma \\pm \\epsilon}$, with no hidden logarithmic corrections, on the very small scales that control the diagonal heat kernel.","fun_headline_variants_meta":{"raw":{"variants":["In d>2, spectral dimension depends on gamma and beta","Spectral dimension in d>2: gamma and beta both matter","Alpha-quantum cone invariant under LBM shifts in d>2","For d>2, LBM spectral dimension varies with local thickness","In d>2, short-time heat decay depends on gamma and beta"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001764,"raw_usage":{"total_tokens":7024,"prompt_tokens":1075,"completion_tokens":5949,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":691,"completion_tokens_details":{"reasoning_tokens":5857}},"tokens_in":691,"tokens_out":5949,"duration_ms":38355,"temperature":1.0,"reasoning_tokens":5857,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T13:51:17.598759+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute, by simulation of Brownian bridges in $d=4$ with a fixed $\\beta$ and $\\alpha$, the threshold $\\chi$ at which $\\mathbb{E}^{0,0,t}[(F_{h-\\beta \\log|\\cdot|,\\alpha}(t))^\\chi]$ switches from diverging to converging as $t\\to 0$; if the threshold is not $\\chi(\\beta) = (d-2)/(2+\\alpha^2/2-\\alpha\\beta)$, the formula fails.","supporting_citations":[],"review_version":1}