{"id":"efbc4a6c-2c7d-4930-a22d-29f6913066cd","arxiv_id":"2411.16536","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"The paper proves conditional a priori bounds for the dynamic fractional Phi^4 equation on the three-torus for every subcritical exponent s in (3/4,1), assuming a suitable renormalised model exists.","lead":"This paper proves a mathematical bound for solutions of a singular random partial differential equation, the fractional phi^4 model on a three-dimensional torus, across the full range of exponents where the equation is just barely solvable. The result matters because such bounds are the step needed to show the equation has eternal solutions and a stable probability distribution, a candidate for a quantum field theory measure.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The conditional bound rests on a weakly admissible fractional-heat model that is not constructed, so the advertised application to (1.1) is not yet established.","rationale":"The reader's weakest_assumption identifies exactly the point I would press. Theorem 1.1 is a conditional analytic statement; its proof is detailed and appears coherent once a smooth weakly admissible model is assumed. But the paper's stated goal of treating the dynamic fractional Phi^4 model requires such a model to be constructed from space-time white noise. The authors themselves flag after Theorem 1.1 that the construction is not fully contained in the cited literature and that the relevant random variables are only expected to exist. For the fractional heat operator this is not a formality: Hairer's original framework does not cover the kernel because of its lack of smoothness, and the new Schauder estimates supply the missing analytic ingredient for the small-scale argument but not the stochastic renormalisation. The omitted proof of Lemma 3.10 is a smaller internal gap; it is used in the blow-up proof of the Schauder estimate, but the referenced analogous result in [FR17] makes it likely repairable. Since the reader already judged the paper CONDITIONAL on this basis, my concern does not move the verdict. I would keep CONDITIONAL with the explicit requirement that either the model construction be supplied or the theorem be stated purely as a conditional a priori bound without the global-existence and invariant-measure conclusion.","tokens_in":77676,"tokens_out":8405,"duration_ms":90924,"concrete_test":"Check whether the existing BPHZ-type stochastic estimates in [LOTT24, HS24] can be run for the fractional heat kernel: replace the heat semigroup by the Fourier multiplier e^{-t|k|^{2s}} and verify the kernel estimates used there, in particular |∂^n K(t,x)| ≲ t^{-(3+|n|)/(2s)}(1 + |x| t^{-1/(2s)})^{-N} for the needed N>0, plus the 2s-scaling of the associated convolution. If these estimates close at the required homogeneities for s∈(3/4,1) and γ∈(3−2s,2s), the missing model construction is routine; if not, Theorem 1.1 has no verified application to equation (1.1).","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 1.1 is an implication for an arbitrary smooth weakly admissible model, but the advertised consequences—global existence and an invariant measure—require a model constructed from the space-time white noise, with the model seminorms on the RHS of (1.2) finite and possessing the same stretched-exponential stochastic integrability. The paper explicitly admits, immediately after Theorem 1.1, that this construction is 'not (fully) contained' in [CH16, LOTT24, HS24] because the fractional heat operator is not covered there, and it only 'expects' the relevant random variables to exist. This is a substantive missing ingredient, not a routine check: Hairer's admissible-model framework does not apply directly to the fractional heat kernel because of its lack of smoothness, as the paper itself notes via [CL22], and the new Schauder estimates in Section 3 do not by themselves renormalise the noise or establish the model bounds in Assumption 4.1. A secondary gap is Lemma 3.10, whose proof is omitted and which is used to pass to the limit in the blow-up proof of Lemma 3.4; this is probably repairable but should be supplied. The primary unresolved step is the model construction.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves a conditional a priori bound for the dynamic fractional Phi^4 model on T^3 in the subcritical regime s in (3/4,1), following the strategy of CMW23 and MW20 within Hairer's regularity structures. Theorem 1.1 states that for any smooth, 1-periodic, weakly admissible model for the fractional heat operator and any 1-periodic modelled distribution Phi solving the algebraic and PDE equations (2.9)-(2.10), the coefficient v = <1,Phi> satisfies a deterministic bound on (t^{2s},1] x T^3 by t^{-s} and by finitely many model seminorms. The proof develops localised multilevel Schauder estimates for the fractional heat operator, uses a maximum-principle argument for the large scales, and concludes with a bootstrap that converts the a priori bounds into global bounds. The abstract and introduction further claim that, assuming model bounds, these estimates imply global existence of solutions and existence of an invariant measure.","tokens_in":77903,"tokens_out":8247,"duration_ms":81301,"significance":"If all ingredients are completed, this would be a substantial contribution: it extends the a priori bound methodology of Chandra-Moinat-Weber from the standard heat operator to the fractional heat operator, covers the full subcritical regime, and supplies new multilevel Schauder estimates for a kernel not covered by Hairer's original framework. The analytic core is detailed, the constants are stated to depend only on s, d and gamma, and there is no parameter fitting or hidden reduction of the target bound to itself. The paper also streamlines the algebraic part by using decorated trees and the duality formula from BM23. However, the advertised application to equation (1.1) is conditional on a stochastic model construction that the paper explicitly does not provide, and one exchange-of-limits lemma used in the Schauder proof is stated without proof. These gaps are load-bearing for the central claims, although they do not appear to invalidate the conditional theorem itself.","major_comments":[{"comment":"The advertised application to equation (1.1) is not established: Theorem 1.1 is an implication for an arbitrary smooth weakly admissible model, but global existence and the existence of an invariant measure require a model constructed from space-time white noise whose seminorms in Assumption 4.1 are finite and have the required integrability. The manuscript itself states, immediately after Theorem 1.1, that this construction is 'not (fully) contained' in [CH16, LOTT24, HS24] because the fractional heat operator is not covered there, and that one only 'expects' the relevant random variables to exist. This is not a routine check: the new Schauder estimates in Section 3 do not by themselves renormalise the noise or verify Assumption 4.1, and Hairer's admissible-model framework does not apply directly to the fractional heat kernel, as the paper notes via [CL22]. A complete version should either construct the model (or reduce it to a finite, explicit list of stochastic estimates), or state the main theorem as purely conditional and remove the global existence/invariant measure consequences from the abstract and introduction.","section":"After Theorem 1.1 and Section 2.2"},{"comment":"Lemma 3.10 is a modified exchange-of-limits result for the fractional heat operator, but its proof is omitted with the sentence 'We omit the proof since it is analogous to the referenced one'. This lemma is used in Step 4 of the proof of Lemma 3.4 to pass to the limit in the blow-up argument and to conclude that the limit v satisfies Lv=0. Since Lemma 3.4 is the core of Theorem 3.1, Lemma 3.10 is load-bearing. The exchange is nontrivial because the operator is nonlocal and the hypotheses involve uniform convergence of the approximate solutions, weak convergence of the right-hand sides, and a growth condition. The proof should either be supplied in full or replaced by a precise reference whose hypotheses are checked against the objects constructed in Lemma 3.4.","section":"Section 3, Lemma 3.10"},{"comment":"Lemma 4.15 is used to handle the time intervals on which Assumption 4.1 fails, and it converts the model seminorms appearing on the right-hand side of Theorem 1.1 into the final bound. The proof, however, is only written explicitly for T = T_{1,c}; for T_{2,c} and T_{3,c} the text says the argument is 'completely analogous'. Given that the final theorem involves three different types of model seminorms, including the supremum over x of ||Xi_x I(tau)|| on a non-compact spatial set, the proof should at least sketch the T_{2,c} and T_{3,c} cases and verify that the same exponent kappa(tau) is used consistently with (1.2).","section":"Section 4, Lemma 4.15 and Definition 4.14"}],"minor_comments":[{"comment":"The exponents in display (1.2) are difficult to read: the notation involving l(tau), kappa(tau), and the powers of the model seminorms should be defined once, close to the display, and cross-referenced with Assumption 4.1 and Lemma 4.15 to avoid ambiguity about which exponent applies to which seminorm.","section":"Theorem 1.1, display (1.2)"},{"comment":"The abstract states that the estimates imply global existence and existence of an invariant measure, but Theorem 1.1 as stated is only an a priori bound for a given modelled distribution. The paper should explicitly say that these consequences are conditional not only on the model bounds but also on the existence of a weakly admissible model constructed from the noise, and should indicate where the standard arguments for global existence and invariant measure are proved or deferred.","section":"Abstract and Introduction"},{"comment":"The notation B_r(x) is used both for Euclidean balls in R^d and for half-parabolic balls in R^{1+d}, and the distinction is left to context. In Section 3, where both types appear in the same estimates, a notational distinction (for example B_r^par or an explicit subscript) would improve readability.","section":"Section 3, Lemma 3.15 and surrounding notation"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a serious analytic contribution and the conditional theorem appears defensible, but the advertised application to the stochastic equation is incomplete because the required weakly admissible model is not constructed. The paper is honest about this gap, which suggests it can be addressed by reframing or by an additional construction; it is not a case of a hidden circular argument. I would be willing to look at a revision that supplies the proof of Lemma 3.10, gives the missing model construction or clearly states the theorem as conditional, and adjusts the claims in the abstract accordingly."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Punchline: Theorem 1.1 is a clean conditional estimate, and Section 3 is the genuine contribution: localised multilevel Schauder estimates for the fractional heat operator for s in (0,1). Those are not covered by Hairer's framework, and the kernel-free scaling proof via the FR17 Liouville theorem looks sound. The advertised consequences for (1.1) remain conditional, because the weakly admissible model is not constructed.\n\nWhat the paper does well: it faithfully extends the CMW23 strategy. Small scales via regularity structures plus the new Schauder estimates; large scales via the maximum principle, which is correct for s in (0,1) and explains the torus setting. Constants depend only on s and gamma. No parameter fitting, no circularity. The paper is also honest: after Theorem 1.1 it states that the model bounds are not fully contained in CH16/LOTT24/HS24 and only expects the Wiener chaos variables to exist.\n\nSoft spots in proportion: the primary one is exactly that gap. Assumption 4.1 assumes a weakly admissible model; the theorem is an implication for arbitrary such models. To apply to space-time white noise, someone has to construct the model for the fractional heat operator with the right stochastic integrability. That is substantive, not routine: Hairer's admissibility notion doesn't cover this kernel, and the new Schauder estimates alone don't renormalise the noise or produce Assumption 4.1. So the route to global existence and an invariant measure for (1.1) is open, not closed. Secondary: Lemma 3.10's proof is omitted (\"analogous\") and it is used in the blow-up limit; likely repairable but a referee should ask for it. The algebraic sections are not machine-checked, but they lean on BHZ19/BCCH20 and BM23 and look fine.\n\nVerdict and recommendation: this paper is for the singular SPDE / regularity structures audience. The conditional bound and the Schauder estimates deserve refereeing; the model-construction gap is a missing ingredient, not a flaw in the main argument. I would send it to peer review and ask for the Lemma 3.10 proof plus a realistic assessment of the model construction. I would cite it if I worked on fractional Phi^4.","headline":"A real conditional a priori bound with genuinely new Schauder estimates for the fractional heat operator; the advertised application to (1.1) stays conditional because the weakly admissible model is not constructed.","tokens_in":78414,"tokens_out":2587,"would_cite":true,"duration_ms":26439,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35R60","60H15","35R11","35B45"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves a priori bounds for the dynamic fractional $\\Phi^4_3$ equation on $\\mathbb{T}^3$ across the whole subcritical range $s\\in(3/4,1)$, conditional on a stochastic model.","keywords":["fractional Laplacian","Phi^4 model","a priori bounds","regularity structures","Schauder estimates","stochastic quantisation","invariant measure","subcritical SPDE"],"falsifier":"A concrete way to test the claim: for some $s\\in(3/4,1)$ and $\\gamma\\in(3-2s,2s)$, construct a weakly admissible model and a modelled distribution solving (2.9)–(2.10) with all model seminorms finite, and check whether $\\|v\\|_{(t^{2s},1]\\times\\mathbb{T}^3}$ grows faster than $C\\max\\{t^{-s},\\text{model terms}\\}$; any such violation would disprove Theorem 1.1. A cheaper check on the auxiliary Schauder estimate is to exhibit a bounded germ satisfying the hypotheses of Theorem 3.1 but violating the estimate (3.10) on a sequence of shrinking balls.","tokens_in":77466,"feed_emoji":"📐","tokens_out":8161,"duration_ms":76110,"temperature":0.7,"pith_summary":"This paper proves uniform a priori bounds for solutions of the dynamic fractional $\\Phi^4_3$ equation $(\\partial_t+(-\\Delta)^s)\\varphi=-\\varphi^3+\\xi$ on $\\mathbb{T}^3$ for every $s\\in(3/4,1)$, the full subcritical regime. The central result, Theorem 1.1, bounds the leading coefficient $v=\\langle 1,\\Phi\\rangle$ of a modelled distribution $\\Phi$ on the time interval $(t^{2s},1]$ by a constant times $\\max\\{t^{-s},\\text{model seminorms}\\}$, with the constant depending only on $s$ and the regularity exponent. If a suitable random model exists with the right stochastic integrability, the bound implies global-in-time existence and the existence of an invariant measure, making it a stochastic-quantisation route to the fractional $\\Phi^4_3$ measure in this regime. To carry this out, the paper develops localised multilevel Schauder estimates for the fractional heat operator $L=\\partial_t+(-\\Delta)^s$, which are not available from standard kernel smoothness because the fractional symbol is non-smooth at the origin.","feed_headline":"Fractional Phi^4 gets full-regime a priori bounds","feed_subtitle":"New Schauder estimates for the fractional heat operator turn model bounds into global existence and an invariant measure.","key_machinery":"The load-bearing mechanism is a localised multilevel Schauder estimate for the fractional heat operator $L=\\partial_t+(-\\Delta)^s$ on germs $U(x,y)$, which are families of functions indexed by a base point $x$ with running variable $y$ in the past of $x$. Theorem 3.1 says that if a germ is $\\gamma$-H\\\"older in a diagonal sense, satisfies a three-point continuity condition, and has $LU$ controlled in a distributional $(\\gamma-2s)$-seminorm, then $[U]_\\gamma$ on a half-size ball is bounded by those ingredients plus a nonlocal $L^\\infty$ term. The proof uses scaling, a Liouville theorem for $L$-harmonic functions saying that controlled solutions are at most linear polynomials in space, and an abstract absorption lemma. This estimate replaces the smoothness of the heat kernel, which fails for the fractional symbol, and it is what makes small-scale control of the modelled solution possible. On the algebraic side, the paper uses a duality formula for the tree product to streamline the decorated-tree bookkeeping of the Da Prato–Debussche remainder.","core_discovery":"The paper's central claim is Theorem 1.1: fix $s\\in(3/4,1)$ and $\\gamma\\in(3-2s,2s)$. For every smooth, 1-periodic, weakly admissible model $(\\Xi,\\Pi)$ for $L$, and every 1-periodic modelled distribution $\\Phi$ solving the algebraic and analytic equations (2.9)–(2.10), the coefficient $v=\\langle1,\\Phi\\rangle$ satisfies $$\\|v\\|_{($t^{{2s}}$,1]\\times\\mathbb{T}^3}\\lesssim \\max\\Bigl\\{$t^{{-s}}$,\\; \\max_{\\tau}[\\Xi;\\tau]^{1/\\$\\beta$(\\tau)},\\; \\max_{\\tau,k}[\\Pi\\mathcal{I}(\\tau);X_k]^{1/\\$\\beta$(\\tau)},\\; \\max_{\\tau}\\sup_x\\|\\Xi_x\\mathcal{I}(\\tau)\\|^{1/\\$\\beta$(\\tau)}\\Bigr\\},$$ with implied constant depending only on $s$ and $\\gamma$, where $\\beta(\\tau)>0$ is an exponent built from the number of noises in the decorated tree $\\tau$. The paper argues that with such a model, this deterministic bound upgrades to global existence of the renormalised dynamics and to existence of an invariant measure, by a small-scale/large-scale argument: the small scales are controlled by the new Schauder estimates on germs, the large scales by the damping of the cubic nonlinearity through a maximum principle valid for $s\\le 1$.","pith_inferences":["The paper leaves the construction of the weakly admissible model open; if the expected random model is constructed with the right integrability, Theorem 1.1 is ready to deliver a full stochastic-quantisation proof of the fractional $\\Phi^4_3$ measure.","Because the Schauder estimate does not rely on kernel smoothness, it may also apply to other nonlocal or anisotropic parabolic operators whose Green's functions are non-smooth at the origin but share the same scaling and Liouville behaviour.","The streamlined tree algebra could lower the barrier to proving a priori bounds in other singular SPDEs where the number of decorated trees grows quickly as criticality is approached.","A numerical test of the claimed $t^{-s}$ prefactor could be made on highly resolved regularised equations: decay no slower than $t^{-s}$ would be consistent with the bound, while systematically slower decay would point to a missing model term."],"forward_implications":["For any weakly admissible model satisfying Assumption 4.1, Theorem 1.1 gives the deterministic bound $\\|v\\|_{(t^{2s},1]\\times\\mathbb{T}^3}\\lesssim t^{-s}$ up to model seminorms.","If the model can be realised as random variables in the Wiener chaos of the noise with the expected stretched exponential integrability, the right-hand side of the bound has the same integrability, so global existence and an invariant measure follow for the renormalised equation.","The localised multilevel Schauder estimates are stated for every $s\\in(0,1)$, so they form a standalone tool for other parabolic problems driven by $(-\\Delta)^s$.","The maximum-principle large-scale control applies because the fractional Laplacian with $s\\le1$ has the relevant positivity; this is why the three-dimensional torus case works and the four-dimensional analogue with $s>1$ would not.","Periodicity enters only through the large-scale $L^\\infty$ control, not through the local estimates, so the small-scale part transfers to nonperiodic settings."],"supporting_citations":[{"why":"Supplies the regularity-structures framework, the reconstruction theorem, and the notion of modelled distributions used throughout the paper.","marker":"[Hai14]"},{"why":"Provides the a priori-bound method for the heat case that this paper extends to the fractional heat operator.","marker":"[CMW23]"},{"why":"Provides the small-scale/large-scale strategy and the maximum-principle argument for the cubic term that is adapted here.","marker":"[MW20]"},{"why":"Supplies the Liouville theorem and the scaling-based proof technique underlying the new multilevel Schauder estimates.","marker":"[FR17]"},{"why":"Supplies the duality formula for decorated trees used to streamline the algebraic bookkeeping.","marker":"[BM23]"},{"why":"Provides the decorated-tree formalism and the subcriticality rules on which the regularity structure is built.","marker":"[BCCH20]"}],"fun_headline_variants":["Fractional Φ^4 on T^3: full subcritical bounds","A priori bounds for fractional Φ^4 in full subcritical regime","Fractional Φ^4: new Schauder estimates control full regime","Full subcritical Φ^4: bounds imply global existence and invariant measure","Fractional heat Schauder bounds yield Φ^4 a priori estimates"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire result is conditional on the existence of a weakly admissible model for the fractional heat operator whose seminorms are finite and satisfy Assumption 4.1; the paper does not construct this model, and if it does not exist the theorem does not apply to the stochastic equation.","fun_headline_variants_meta":{"raw":{"variants":["Fractional Φ^4 on T^3: full subcritical bounds","A priori bounds for fractional Φ^4 in full subcritical regime","Fractional Φ^4: new Schauder estimates control full regime","Full subcritical Φ^4: bounds imply global existence and invariant measure","Fractional heat Schauder bounds yield Φ^4 a priori estimates"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001381,"raw_usage":{"total_tokens":5611,"prompt_tokens":979,"completion_tokens":4632,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":595,"completion_tokens_details":{"reasoning_tokens":4537}},"tokens_in":595,"tokens_out":4632,"duration_ms":31049,"temperature":1.0,"reasoning_tokens":4537,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T12:59:39.674677+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A concrete way to test the claim: for some $s\\in(3/4,1)$ and $\\gamma\\in(3-2s,2s)$, construct a weakly admissible model and a modelled distribution solving (2.9)–(2.10) with all model seminorms finite, and check whether $\\|v\\|_{(t^{2s},1]\\times\\mathbb{T}^3}$ grows faster than $C\\max\\{t^{-s},\\text{model terms}\\}$; any such violation would disprove Theorem 1.1. A cheaper check on the auxiliary Schauder estimate is to exhibit a bounded germ satisfying the hypotheses of Theorem 3.1 but violating the estimate (3.10) on a sequence of shrinking balls.","supporting_citations":[],"review_version":1}