{"id":"43dc8475-0ac3-4546-a764-45bee10f7b43","arxiv_id":"1908.07138","paper_version":5,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The fractional porous medium equation has a unique global strong solution on compact conical manifolds for all m>0 and sigma in (0,1), with comparison, contraction, and mass conservation.","lead":"This paper proves that a nonlinear nonlocal diffusion equation, the fractional porous medium equation, has a unique global strong solution on spaces with cone-shaped singularities, for every positive exponent and bounded initial data. The proof builds a new functional-analytic toolkit for such singular spaces, which is the main reason a generalist should care.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Strong regularity rests on an unverified import: the Bénilan–Crandall estimate (5.20) is applied to the regularized operators A_ω without checking its hypotheses, so the claimed strong-solution property in Theorem 1.1 may not follow.","rationale":"The reader's weakest_assumption identifies the same load-bearing concern: the use of [9, Theorem 1] at equation (5.20) and in Section 7 without verifying its hypotheses in the conical setting. My reading confirms that this is the most fragile step: the strong-solution claim in Theorem 1.1, as well as the ω-uniform bound (5.23), depends on it. The domain-decomposition import from [48] is also noted by the reader, but it is a published theorem and even if it needed adjustment, the compactness for D((−Δ_F)^{σ/2}) follows from the same argument. I do not see an internal contradiction in the rest of the proof, and the overall strategy is coherent. Therefore the appropriate verdict is CONDITIONAL: accept after the authors verify or supply a proof of the regularizing estimate for the conical operator, uniformly in ω, or replace the argument with a direct derivation from the discretized resolvent estimates (5.4) and (5.9).","tokens_in":31473,"tokens_out":13967,"duration_ms":137268,"concrete_test":"Verify the hypotheses of [9, Theorem 1] for the family A_ω = (ω − Δ_F,1)^σ Φ on L1(M), ω > 0: check that each A_ω is m-completely accretive (i.e., for every convex φ, ∫ φ((I+λ A_ω)^{-1}u − (I+λ A_ω)^{-1}v) ≤ ∫ φ(u−v)) and positively homogeneous of degree m, and confirm that the regularizing estimate (5.20) holds with a constant independent of ω as ω → 0; if the constant degenerates or complete accretivity fails, Theorem 7.2 does not follow from the given argument.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim includes existence of a unique global strong solution (Theorem 1.1). The proof that the weak solution is strong depends on the regularizing estimate (5.20), imported from [9, Theorem 1], and on the same theorem used again in Section 7 to assert that ∂_t u is a Radon measure with the L1 time-derivative bound. For (5.20) to hold, the operator A_ω = (ω − Δ_F,1)^σ Φ must satisfy the hypotheses of [9, Theorem 1]: m-accretivity, positive homogeneity of degree m, and complete accretivity on L1(M). The paper proves m-accretivity only for the regularized family, ω > 0, and never verifies complete accretivity or the homogeneity/tangent conditions for this family, nor constructs an m-accretive realization of (−Δ)^σ Φ on L1 to which [9] would directly apply to the limit solution u. Moreover, the bound in (5.20) is used uniformly in ω; even if [9] applies for each fixed ω, the constants could in principle degenerate as ω → 0. Without (5.20), the uniform bound (5.23), the conclusion that Φ(u_ω) ∈ L∞_loc((0,T), D((−Δ_F)^{σ/2})), and the strong-solution regularity in Theorem 7.2 are unsupported. A secondary, easily repaired gap is that the compactness used at (5.26) is stated in (4.8) for D((−Δ_F)^σ), while the available bound is in D((−Δ_F)^{σ/2}); the same interpolation argument gives the needed compactness, but it is not stated. The imported domain decomposition (4.9) from [48] is a credible published result and is not the primary concern.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the fractional porous medium equation ∂_t u + (−Δ_g)^σ(|u|^{m−1}u)=0 on compact (n+1)-dimensional Riemannian manifolds with isolated conical singularities, for σ∈(0,1) and m>0. It first develops Mellin–Sobolev space techniques, including a Rellich–Kondrachov theorem and Sobolev–Poincaré, Nash, and super-Poincaré inequalities. It then studies closed extensions of the conical Laplacian, proves that the Friedrichs extension is the unique Markovian extension, constructs fractional powers of the conical Laplacian, and establishes compactness properties for their domains. The main result, Theorem 1.1, asserts that for any u0∈L∞(M) there is a unique global strong solution, with continuous dependence in C([0,T),L1(M)), comparison, Lp-contraction, and conservation of mass. The proof proceeds by regularizing with ω>0, applying nonlinear semigroup theory (Brézis–Strauss and Crandall–Liggett) to obtain ω-mild solutions, passing to the limit ω→0 to obtain a weak solution, and finally using Bénilan–Crandall regularizing estimates and Steklov averages to upgrade to strong solutions. Section 8 states analogous results for the classical porous medium equation σ=1.","tokens_in":31838,"tokens_out":14139,"duration_ms":134481,"significance":"If the main theorem is correct, this is a substantial contribution: it gives the first complete well-posedness and qualitative theory for the fractional porous medium equation on conical singularities, and it combines several nontrivial tools—Mellin–Sobolev compact embeddings, Dirichlet-form/Markovian-extension theory, fractional powers of the conical Laplacian, and nonlinear semigroup methods—in a reusable way. The paper is explicit about the role of the Markovian extension and provides concrete functional inequalities that are likely to be useful beyond this problem. However, the strong-solution claim depends on imported regularizing estimates whose hypotheses are not verified in the conical setting, and the ω→0 compactness step uses a domain mismatch that is repairable but not stated. These issues are load-bearing for the central claim and require correction before the result can be regarded as established.","major_comments":[{"comment":"Equation (5.20) is a crucial regularizing estimate: it is used in (5.23) to obtain a uniform-in-ω bound on ||(ω−Δ_g)^{σ/2}Φ(u_ω)(t)||_2 and thereby to justify the ω→0 passage. The text says it holds by following the proof of [9, Theorem 1] and using (5.19). No verification is given that the operator A_ω(u)=(ω−Δ_{F,1})^σ Φ(u) satisfies the hypotheses of [9, Theorem 1] for each fixed ω—in particular complete accretivity and positive homogeneity of degree m—and the estimate is needed uniformly in ω. The paper proves only m-accretivity of A_ω via Proposition 4.4, which is not sufficient for the cited theorem. This is a load-bearing gap in the proof of weak-solution regularity.","section":"§5, Eq. (5.20)"},{"comment":"The proof of strong regularity asserts, by [9, Theorems 1 and 2], that ∂_t u is a Radon measure and that the limsup bound (ii) holds for the limiting solution u of (1.1). This requires an m-accretive realization of A_0(u)=(−Δ_g)^σ Φ(u) on L^1(M), or an equivalent argument that the regularizing estimates for A_ω pass to the limit with constants independent of ω; neither is supplied. The subsequent application of [11, Theorem 1.1] to conclude ∂_t u∈L^∞_loc((0,T),L^1(M)) from u∈BV((τ,T),L^1(M)) also assumes hypotheses that are not checked for the conical setting. Consequently the 'strong solution' clause of Theorem 1.1 is not established by the text as written.","section":"§7, after Definition 7.1"},{"comment":"The compactness passage Φ(u_{ω_k})(t)→v(t) in L^2(M) is justified by Proposition 4.6, whose compact embedding statement (4.8) is for D((−Δ_F)^σ). The uniform bound available from (5.24) is in D((−Δ_F)^{σ/2}). The needed compactness for the σ/2-domain follows from the same interpolation and Rellich–Kondrachov argument used in Proposition 4.5, but it is not stated. This is a repairable but real gap in the ω→0 limiting argument.","section":"§5, around (5.26)"}],"minor_comments":[{"comment":"In the displayed inequality after (5.21), the time integral on the left is written with 'dt' but the integrand contains 's'; it should be 'ds' (or the outer variable should be renamed).","section":"§5, Eq. (5.22)"},{"comment":"The estimate corresponding to (4.4) is used with the half-power σ/2, but the analogous inequality for (ω−Δ)^{σ/2}−(−Δ)^{σ/2} is not displayed. It is true by the same argument and should be stated.","section":"§5, Eqs. (5.15)–(5.16)"},{"comment":"There is a typo 'in the the norm' in the statements of both theorems; it should be 'in the norm'.","section":"Theorem 7.3 and Theorem 8.2"},{"comment":"The remark refers to an 'anti-cone' and to conservation of mass on components without defining the term or giving a proof; it should either be expanded or removed.","section":"Remark 6.3"},{"comment":"The statement 'D((−Δ_F)^σ) ↪ H^{2σ−ε,σ+δσ−ε}_2(M)+Cω for any ε>0' should specify that ε is taken small enough so that the displayed Sobolev exponents remain meaningful; otherwise the notation is misleading when σ is small.","section":"Proposition 4.5"}],"recommendation":"major_revision","confidential_remarks":"The paper is a serious contribution and the main architecture of the proof is credible, but the strong-solution result relies on two imported abstract regularizing results ([9] and [11]) whose hypotheses are not verified for the conical operators, and the ω→0 compactness step has a domain mismatch. These issues are load-bearing but appear repairable within the manuscript's scope: the authors could verify complete accretivity and homogeneity for the regularized operators, prove the regularizing estimate directly, and state the correct compact embedding for D((−Δ_F)^{σ/2}). The dependence on the companion paper [45] and prior work [43,44,48] is heavy but not circular."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a real advance. It treats the fractional porous medium equation on compact manifolds with conical singularities for all m > 0 and L^infinity initial data, and gets uniqueness, Lp-contraction, comparison, and conservation of mass. Section 8's parallel treatment of sigma = 1 is a useful extension. The Mellin-Sobolev inequalities and the Markovian extension analysis are independent contributions of their own. The overall strategy—Dirichlet forms, Crandall-Liggett generation, the Brezis-Strauss lemma, and the Schrohe-Seiler domain decomposition—is coherent and credible. The citation pattern is normal: the paper leans on its own companion papers and on standard nonlinear semigroup references, and that is appropriate here.\n\nThe main soft spot is exactly where the stress-test note points. The step from weak to strong solution in Section 7 uses the regularizing estimate (5.20), imported from Bénilan-Crandall [9, Theorem 1] and [38, Proposition 8.1]. The hypotheses of that theorem are not verified for the regularized operators A_omega = (omega - Delta_F,1)^sigma Phi. The paper proves m-accretivity and density of the domain, but Bénilan-Crandall needs more, and I do not see a check that the family is completely accretive or that the estimate is uniform as omega -> 0. If that uniformity fails, (5.23) and the conclusion partial_t u in L^infty_loc((0,T), L^1) do not follow. I think the gap is closable: the semigroup is Markovian, which should give the needed complete accretivity, and homogeneity is built into Phi. But the authors need to spell it out rather than citing the theorem as though the conical setting were the Euclidean one.\n\nThe secondary gap at (5.26) is real but minor: the compact embedding stated in (4.8) is for D((-Delta_F)^sigma), while the available bound is in D((-Delta_F)^(sigma/2)). Interpolation and compactness of D(Delta_F) in L^2 fix it, but that argument should be written down. There are also small typos, for example in the norm display in Lemma 3.8, and the application of [11, Theorem 1.1] in Section 7 is asserted rather than checked.\n\nNone of this made me doubt the main theorem. It made me want a careful referee. This is a paper for people working on nonlinear nonlocal diffusion on singular spaces, and also for anyone needing Mellin-Sobolev inequalities. It deserves serious refereeing. I would send it out and ask the authors to verify the imported regularity estimates or, failing that, to state the strong-solution result in the weaker form that their verified estimates actually support.","headline":"First real well-posedness result for the fractional porous medium equation on conical manifolds; the proof is mostly solid, but the weak-to-strong step rests on an unverified imported regularity estimate that a good referee should make the authors address.","tokens_in":32396,"tokens_out":2988,"would_cite":true,"duration_ms":34020,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["26A33","35K65","35K67","35R01","35R11","76S05"],"pacs":[],"model":"deepseek-v4-flash","headline":"The fractional porous medium equation has a unique global strong solution on manifolds with conical singularities for every bounded initial datum, every $m>0$, and every $\\sigma\\in(0,1)$.","keywords":["fractional porous medium equation","conical singularities","Mellin-Sobolev spaces","Markovian extensions","fractional Laplacian powers","nonlinear semigroups","L^p contraction","conservation of mass"],"falsifier":"Run the implicit time-discretization scheme (5.4) on a flat cone with $n=1$, $B=S^1$, $m=1/2$, $\\sigma=1/2$, and an $L^\\infty$ initial datum, and check whether $\\frac{1}{h}\\|u_\\omega(t+h)-u_\\omega(t)\\|_1$ stays below $\\frac{2}{|m-1|t}\\|u_0\\|_1+o(1)$; a violation would falsify the strong-solution theorem. Alternatively, verify whether $D((-\\Delta_F)^\\sigma)$ compacts into $L^p(M)$ at an indicial root on the critical line; failure there breaks the $\\omega\\to0$ identification of the weak limit.","tokens_in":31241,"feed_emoji":"📐","tokens_out":10130,"duration_ms":90658,"temperature":0.7,"pith_summary":"This paper proves that the fractional porous medium equation $\\partial_t u+(-\\Delta_g)^\\sigma(|u|^{m-1}u)=0$ is globally well-posed on compact Riemannian manifolds with isolated conical singularities. For every $u_0\\in L^\\infty(M)$, every $\\sigma\\in(0,1)$, and every $m>0$, there is a unique global strong solution, and the solution has the qualitative properties one expects from a porous-medium flow: continuous dependence on the initial data in $C([0,T),L^1(M))$, a comparison principle, $L^p$-contraction for $1\\le p\\le\\infty$, and conservation of mass. The paper's contribution is to carry the theory of this nonlocal, degenerate diffusion equation from Euclidean and smooth-manifold settings to spaces with a singular geometry, where the fractional Laplacian has no canonical closed extension. A sympathetic reader would care because conical singularities are the simplest model of a nonsmooth space, and the proof isolates exactly which analytic ingredients are needed to make sense of the equation there.","feed_headline":"Unique global solutions for fractional diffusion on cones","feed_subtitle":"Bounded initial data give comparison, L^p contraction, and mass conservation for every m>0 and sigma in (0,1).","key_machinery":"The load-bearing machinery is a chain of four constructions. First, the Mellin-Sobolev spaces $H^{s,\\gamma}_p(M)$--weighted Sobolev spaces whose weights are powers of the distance to the singularity--provide compact embeddings and Sobolev-Poincar\\'e, Nash, and Super-Poincar\\'e inequalities with weights. Second, the conical Laplacian $\\Delta_g$ is shown to have a unique Markovian extension $\\Delta_F$, the minimal self-adjoint extension generating a positive $L^\\infty$-contractive semigroup, and this extension transfers to $L^p$ as $\\Delta_{F,p}$. Third, the fractional powers $(\\omega-\\Delta_{F,p})^\\sigma$ are built by resolvent-integral formulas and verified to be Markovian, positive, and resolvent-regular, with the domain inclusion $D((-\\Delta_F)^\\sigma)\\hookrightarrow L^p(M)$ compact. Fourth, the nonlinear operator $u\\mapsto(\\omega-\\Delta_{F,1})^\\sigma\\Phi(u)$ is shown to be $m$-accretive and densely defined in $L^1(M)$, so the nonlinear semigroup theorem yields global mild solutions; a careful $\\omega\\to0$ limit, using the compact embedding and a regularizing estimate for $L^1$ time increments, turns these into weak and then strong solutions.","core_discovery":"The central discovery is Theorem 1.1: on an $(n+1)$-dimensional conical manifold $(M,g)$, for any $u_0\\in L^\\infty(M)$, $\\sigma\\in(0,1)$, and $m>0$, equation (1.1) has a unique global strong solution, with continuous dependence on $u_0$ in $C([0,T),L^1(M))$, comparison principle, $L^p$-contraction for all $1\\le p\\le\\infty$, and conservation of mass. The construction works by writing the equation in $L^1(M)$ with the accretive operator $A(u)=(-\\Delta_{F,1})^\\sigma\\Phi(u)$, where $\\Phi(u)=|u|^{m-1}u$; proving $A$ is $m$-accretive and densely defined; generating a unique mild solution by implicit time discretization; then showing these solutions converge as the regularizing parameter $\\omega\\to0$ in $(\\omega-\\Delta_{F,1})^\\sigma$ to a weak solution of the original problem, and finally upgrading the weak solution to a strong one using time-regularity estimates. The same result is obtained for $\\sigma=1$ in Section 8, recovering the classical porous medium equation on conical manifolds.","pith_inferences":["If the Markovian-extension route is as robust as the paper suggests, the same $\\omega\\to0$ approximation should prove well-posedness for other fractional nonlinear diffusion equations, such as a fractional $p$-Laplacian flow; a first test would be to rerun the Section 4 construction with a uniformly elliptic cone operator that has a self-adjoint Markovian extension.","The compact embedding $D((-\\Delta_F)^\\sigma)\\hookrightarrow L^p(M)$ is likely the quantitative engine behind all smoothing effects; tracking the constants in the Nash and Super-Poincar\\'e inequalities could yield explicit decay rates such as $\\|u(t)\\|_\\infty\\le Ct^{-\\alpha}$ for $m>1$, which the paper defers to a sequel.","Conservation of mass combined with the comparison principle suggests that for $m<1$ the conical geometry suppresses finite-time extinction and instead produces a nonlocal smoothing profile; this could be tested numerically on a flat cone $(0,1)\\times S^n$."],"forward_implications":["For every bounded initial datum, every $m>0$, and every $\\sigma\\in(0,1)$, the flow is globally defined and unique, so no finite-time blow-up or non-uniqueness can occur for bounded data on conical manifolds.","The solution map is continuous in $C([0,T),L^1(M))$ and contractive, so approximating and numerical schemes built on the implicit discretization are stable.","Order is preserved and all $L^p$ norms decay or stay constant, while total mass is conserved; for $m<1$, this mass conservation distinguishes the conical setting from Euclidean or bounded-domain fractional porous media, where finite-time extinction can occur.","The same construction with $\\sigma=1$ gives the analogous well-posedness theorem for the classical porous medium equation on conical manifolds, as stated in Theorem 8.1.","The method does not rely on Euclidean integral representations of the fractional Laplacian, so the framework extends to manifolds with more general singularities, including cuspidal, edge, and corner singularities."],"supporting_citations":[{"why":"Supplies the exact domain decomposition $D(\\Delta_F)=D(\\Delta_{F,\\min})\\oplus\\bigoplus E_{q_j}$ that yields the compact embedding (4.8).","marker":"[48]"},{"why":"Gives the nonlinear semigroup generation theorem that turns the $m$-accretive operator into a global mild solution.","marker":"[23]"},{"why":"Proves solvability and $L^1$-contraction for the resolvent equation $\\lambda(\\omega-\\Delta_g)^\\sigma u+\\beta(u)=f$, the core of the $m$-accretivity argument.","marker":"[16]"},{"why":"Supplies the regularizing estimate used in (5.20) to control time increments and promote weak solutions to strong solutions.","marker":"[9]"},{"why":"Provides the functional calculus and resolvent formulas defining $(\\omega-\\Delta_{F,p})^\\sigma$ and its domain.","marker":"[51]"},{"why":"Gives the Dirichlet-form criterion used to prove Markovianity and uniqueness of the Markovian extension.","marker":"[29]"},{"why":"Provides the test-function argument used to prove uniqueness of $L^1$-weak solutions.","marker":"[52]"},{"why":"Supplies the Euclidean fractional porous medium theory whose Steklov-average technique is adapted in Section 7 to prove strong regularity.","marker":"[38]"}],"fun_headline_variants":["Fractional diffusion on cones: unique strong solutions","Unique global solutions for porous media on singular cones","Global well-posedness for fractional PME on cones","Fractional porous media on cones: existence and uniqueness","On conical manifolds, fractional PME is well-posed"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof leans on two imported facts: the conical Laplacian's Markovian extension has a domain made of the minimal domain plus finitely many special singular functions, and the fractional-power domains embed compactly into $L^p$; if either fails on a particular cone, the passage from approximating solutions to a true strong solution can break.","fun_headline_variants_meta":{"raw":{"variants":["Fractional diffusion on cones: unique strong solutions","Unique global solutions for porous media on singular cones","Global well-posedness for fractional PME on cones","Fractional porous media on cones: existence and uniqueness","On conical manifolds, fractional PME is well-posed"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000993,"raw_usage":{"total_tokens":4242,"prompt_tokens":1012,"completion_tokens":3230,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":628,"completion_tokens_details":{"reasoning_tokens":3153}},"tokens_in":628,"tokens_out":3230,"duration_ms":24357,"temperature":1.0,"reasoning_tokens":3153,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:25:48.767550+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the implicit time-discretization scheme (5.4) on a flat cone with $n=1$, $B=S^1$, $m=1/2$, $\\sigma=1/2$, and an $L^\\infty$ initial datum, and check whether $\\frac{1}{h}\\|u_\\omega(t+h)-u_\\omega(t)\\|_1$ stays below $\\frac{2}{|m-1|t}\\|u_0\\|_1+o(1)$; a violation would falsify the strong-solution theorem. Alternatively, verify whether $D((-\\Delta_F)^\\sigma)$ compacts into $L^p(M)$ at an indicial root on the critical line; failure there breaks the $\\omega\\to0$ identification of the weak limit.","supporting_citations":[{"cited_title":"Schrohe, J","cited_arxiv_id":null,"evidence_quote":"Supplies the exact domain decomposition $D(\\Delta_F)=D(\\Delta_{F,\\min})\\oplus\\bigoplus E_{q_j}$ that yields the compact embedding (4.8)."},{"cited_title":"Crandall, T.M","cited_arxiv_id":null,"evidence_quote":"Gives the nonlinear semigroup generation theorem that turns the $m$-accretive operator into a global mild solution."},{"cited_title":"Br´ ezis, W.A","cited_arxiv_id":null,"evidence_quote":"Proves solvability and $L^1$-contraction for the resolvent equation $\\lambda(\\omega-\\Delta_g)^\\sigma u+\\beta(u)=f$, the core of the $m$-accretivity argument."},{"cited_title":"B´ enilan, M.G","cited_arxiv_id":null,"evidence_quote":"Supplies the regularizing estimate used in (5.20) to control time increments and promote weak solutions to strong solutions."},{"cited_title":"Tanabe, Equations of evolution","cited_arxiv_id":null,"evidence_quote":"Provides the functional calculus and resolvent formulas defining $(\\omega-\\Delta_{F,p})^\\sigma$ and its domain."},{"cited_title":"Fukushima, Y","cited_arxiv_id":null,"evidence_quote":"Gives the Dirichlet-form criterion used to prove Markovianity and uniqueness of the Markovian extension."},{"cited_title":"V´ azquez, The Porous Medium Equation","cited_arxiv_id":null,"evidence_quote":"Provides the test-function argument used to prove uniqueness of $L^1$-weak solutions."},{"cited_title":"Pablo, F","cited_arxiv_id":null,"evidence_quote":"Supplies the Euclidean fractional porous medium theory whose Steklov-average technique is adapted in Section 7 to prove strong regularity."}],"review_version":1}