{"id":"24ee24f3-b882-4de2-8251-ae3efdb87b6c","arxiv_id":"2607.05919","paper_version":1,"verdict":"ACCEPT","confidence":"UNKNOWN","novelty_score":5.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":2,"one_line_summary":"Positive distribution solutions of $-Δu + a²Δ²u = u^q$ in ℝ³ exist only for q>5, super-solutions only for q>3, and all such solutions are radial, differentiable, and decay like the Coulomb-type kernel $K_a(x)=(1-e^{-|x|/a})/|x|$.","lead":"The paper proves Liouville theorems, regularity, and decay estimates for positive distribution solutions of the fourth-order equation $-Δu + a²Δ²u = u^q$ in ℝ³, establishing critical exponents q>3 (super-solutions) and q>5 (solutions) that match the classical Lane-Emden case. A smart generalist would read this to understand when certain embedding inequalities have non-attainable best constants and how mixed-dispersion Schrödinger static equations behave.","discovery_kind":"unclear","skeptic_critique":{"model":"glm-5.2","headline":"Pohozaev identity derivation (Theorem 2.2) assumes distribution solution is a critical point of E without independent verification.","rationale":"The reader identified the weakest assumption as the membership of K_a * ψ in D (Theorem 2.5), which is a technical but verifiable estimate. The estimates in Step 1 of Theorem 2.5 appear correct upon inspection: the key bound |ΔΦ| ≤ C/|x| near the origin (equation 2.17) gives L²-integrability since ∫_{B_δ} |x|^{-2} dx < ∞ in ℝ³, and the exponential decay at infinity handles the tail. So the reader's concern, while reasonable to flag, does not appear to be where the actual vulnerability lies. The more load-bearing issue is the Pohozaev identity derivation in Theorem 2.2, which is the linchpin of the main Liouville theorem (q>5). The paper's approach — asserting that the distribution solution is a critical point of E and then differentiating under scaling — skips the justification that would normally be provided by either a direct Pohozaev computation or a verification of differentiability of E along the scaling path. The integral-form Pohozaev identity (Theorem 6.1) is derived with more care, including explicit convergence of improper integrals, which suggests the authors are aware of the subtleties but did not apply the same rigor to the PDE-side identity. That said, the paper does have a fallback: Theorem 1.1(i) establishes equivalence between PDE and integral formulations, and Theorem 6.1 proves q>5 for the integral equation with the more careful Pohozaev derivation. So even if the PDE-side identity in Theorem 2.2 has a gap, the conclusion q>5 for distribution solutions may still hold via the integral equation route, provided the equivalence (Theorem 1.1) is solid. This is why I recommend CONDITIONAL rather than REJECT: the main conclusion likely survives, but the specific proof in Theorem 2.2 needs either a direct Pohozaev computation or explicit justification of the scaling argument. The paper's overall structure is sound, the estimates are mostly careful, and the results are interesting. The concern is with a specific step in a specific proof, not with the overall framework.","tokens_in":29193,"tokens_out":1175,"duration_ms":2782685,"concrete_test":"Independently re-derive the Pohozaev identity for the PDE (1.5) by the classical method: multiply the equation by (x·∇u), integrate over B_R, integrate by parts, and track all boundary terms explicitly. Verify that the boundary terms vanish as R→∞ along a sequence R_j (using Theorem 2.1 and the decay estimates from Theorem 1.5, which are available via the equivalence). If the boundary terms do not vanish, the identity in Theorem 2.2 is unjustified.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The Liouville theorem q>5 (Theorem 2.2) is the central claim. Its proof derives the Pohozaev identity by computing d/dμ E(u(x/μ))|_{μ=1} = 0, justified by the statement 'the distribution solution u is the critical point of E(u)' (page 8, between equations (2.10) and the identity). However, E(u) as written in (1.3) includes the term -1/(2(σ+1)) ∫|u|^{2(σ+1)} dx, which corresponds to the nonlinearity u^{2σ+1}. The PDE (1.5) has nonlinearity u^q. The paper sets 2σ = q-1 (page 2), so 2(σ+1) = q+1, making the exponents consistent. But the critical-point property requires that the first variation of E vanishes at u in a suitable sense. The paper establishes u ∈ L^{q+1} (Theorem 2.1) and that (2.4) holds, but (2.4) is the weak formulation with test functions in C_0^∞, not a statement that the functional derivative of E vanishes. The step from 'u satisfies the weak equation' to 'd/dμ E(u(x/μ))|_{μ=1} = 0' requires either: (a) justification that the scaling variation μ ↦ u(x/μ) is an admissible variation in the function space where E is differentiable, or (b) a direct Pohozaev-type computation from the PDE. The paper asserts (a) without verification. The integral-form Pohozaev identity (Theorem 6.1) is derived more carefully with explicit convergence arguments, but the PDE-side identity in Theorem 2.2 is the one used for the main Liouville theorem for distribution solutions. If the scaling argument has a gap — for instance, if the boundary terms from the cut-off procedure used to establish (2.4) do not vanish in the right way under the scaling u(x/μ) — the identity 0 = G1/2 - G2/2 - 3G3/(q+1) may not hold, and the conclusion q>5 would be unsupported for PDE solutions (though it might still hold via the integral equation route for solutions with sufficient integrability).","agreement_with_reader":"partial"},"referee_report":{"model":"glm-5.2","summary":"This paper studies distribution solutions of the fourth-order semilinear equation $-Δu + a^2 Δ^2 u = u^q$ in $R^3$, which arises as the static equation of a mixed-dispersion Schrödinger equation and as the Euler–Lagrange equation for an embedding inequality in the space $D = {u ∈ D^{1,2} : Δu ∈ L^2}$. The authors establish: (i) equivalence between the PDE and an integral equation with the modified Coulomb kernel $K_a(x) = (1-e^{-|x|/a})/|x|$; (ii) Liouville theorems giving critical exponents $q>3$ (super-solutions) and $q>5$ (solutions), implying non-attainability of the best constant in the embedding inequality at $q=5$; (iii) regularity (differentiability, $L^p$ integrability, radial symmetry via moving planes) and decay estimates for positive solutions; (iv) an integral-form Pohozaev identity yielding the $q>5$ threshold independently; and (v) a Liouville theorem for an Allen–Cahn-type integral equation. The proofs span 36 pages and combine variational arguments, HLS estimates, regularity lifting, and moving-plane methods.","tokens_in":29509,"tokens_out":2230,"duration_ms":802954,"significance":"The paper addresses a natural and well-motivated problem. The equivalence between the PDE and the integral formulation (Theorem 1.1) is a useful structural result, and the Liouville theorems (Theorems 1.2–1.3) provide clean critical exponents that parallel the classical Lane–Emden theory. The implication that the best constant of the embedding inequality (1.6) is non-attainable at $q=5$ is a concrete, falsifiable consequence. The integral-form Pohozaev identity (Theorem 6.1) is developed with explicit convergence arguments, which is a strength. The extension to the Allen–Cahn-type equation (§7) broadens the scope. The results are parameter-free in the sense that the critical exponents emerge from the analysis rather than being assumed.","major_comments":[{"comment":"Theorem 2.2 (Pohozaev identity, PDE side), which is the main route to the Liouville theorem $q>5$ for distribution solutions, contains a gap in justification. The proof states (page 8, between (2.10) and the identity): 'In view of (2.4), the distribution solution $u$ is the critical point of $E(u)$. Therefore, we have the following Pohozaev identity $0 = [d/dμ E(u(x/μ))]_{μ=1}$.' Equation (2.4) is the weak formulation with test functions in $C_0^∞(R^3)$, i.e., $∫(∇u·∇φ + a^2 Δu·Δφ) = ∫u^q φ$. The step from this weak formulation to $d/dμ E(u(x/μ))|_{μ=1} = 0$ requires justification that the scaling variation $μ ↦ u(·/μ)$ is an admissible variation in the energy space and that the first variation of $E$ vanishes along it. The paper asserts this without verification. The authors should either: (a) provide a direct Pohozaev-type computation from the PDE using cut-off functions and passage to","section":null},{"comment":"the limit (justifying that boundary terms vanish), or (b) verify that the scaling $u(·/μ)$ lies in the function space where $E$ is Fréchet-differentiable and that the weak formulation (2.4) suffices to conclude $δE(u)[v] = 0$ for the specific variation $v = x·∇u$. Note that the integral-form Pohozaev identity (Theorem 6.1) is derived more carefully with explicit convergence arguments, but Theorem 2.2 is the one used for the main Liouville theorem for distribution solutions (Theorem 1.2(ii)). The gap is load-bearing because Theorem 1.2(ii) and its corollary (non-attainability of the best constant at $q=5$, Remark 1.2) depend on it.","section":null},{"comment":"Theorem 2.5 (equivalence, Step 1): the key estimate $|ΔΦ| ≤ C/|x|$ for small $|x|$ (equation 2.17) is used to conclude $ΔΦ ∈ L^2(R^3)$ via $∫_{B_δ} |x|^{-2} dx < ∞$ in $R^3$. This integral indeed converges. However, the estimate (2.17) itself is derived by splitting $B_R$ into $B_{|x|/2}(x)$ and $B_R ∖ B_{|x|/2}(x)$; the bound on the first piece uses $∫_0^{|x|} r dr ≤ C|x|^2$ divided by $|x|$, giving $C|x|$, while the second uses $e^{-|x|/(2a)}/|x| ≤ C/|x|$. The $L^2$-integrability then follows. This appears correct, but the reader's concern about whether $ΔΦ ∈ L^2$ near the origin is the load-bearing premise for the equivalence (Theorem 1.1) and hence for all results derived via the integral formulation. The authors should verify and, if necessary, explicitly confirm that the case $ψ$ supported near the origin (where $|x|/2$ may not be small relative to the support) is handled. As far I","section":null},{"comment":"can verify, the estimates do go through because $ψ ∈ C_0^∞$ is bounded and the kernel estimates (2.2)–(2.3) are uniform, but a sentence clarifying this would strengthen the proof.","section":null}],"minor_comments":[{"comment":"Page 2: the notation $D^{1,2}(R^3)$ is used without definition; a brief reference to the homogeneous Sobolev space would help readers.","section":null},{"comment":"Equation (1.3): the functional $E(u)$ is written with $α$ and $β$, but the PDE (1.5) uses $α=1, β=-a^2$. The relationship between the general functional and the specific case studied should be stated more explicitly.","section":null},{"comment":"Theorem 1.5(iii), equation (1.13): the decay estimate $C^{-1}K_a(x) ≤ u(x) ≤ CK_a(x)$ is stated for 'sufficiently large $|x|$' but the constant $C>1$ is not quantified. This is acceptable but could note dependence on $u$.","section":null},{"comment":"Section 4.4, proof of Theorem 1.5(iii): the choice $m(x) = 1 + (s/q) log_8 log[c_*^{-1} K_a(x)|x|^{3q/s-2}]$ requires $c_*^{-1} K_a(x)|x|^{3q/s-2} > 1$ for the logarithm to be positive. Since $K_a(x) ~ |x|^{-1}$ for large $|x|$, this requires $|x|^{3q/s-3} > c_*$, which holds for large $|x|$ when $3q/s - 3 > 0$, i.e., $s < q$. The text states $s ∈ (3, q)$, so this is consistent, but the verification should be explicit.","section":null},{"comment":"Typo on page 5: 'Theoerm' should be 'Theorem' (Remark 1.5).","section":null},{"comment":"Page 14, proof of Theorem 3.2: the notation $c_1(x)$ for a 'double bounded function' is introduced but the term 'double bounded function' is defined on page 9. A forward reference or brief reminder would help.","section":null},{"comment":"Theorem 6.1, Step 2: the differentiation under the integral sign in $u(μx) = μ^2 ∫ [(1-e^{-μ|x-y|/a})/|x-y|] u^q(μy) dy$ should note that the convergence of the resulting integrals is guaranteed by Step 1 (equations 6.4–6.5). This is mentioned but could be more explicit.","section":null},{"comment":"Reference [53] (Xu, 2005) is cited for the result that $a^2 Δ^2 u = u^q$ has $C^4$-solutions iff $q=-7$. This seems surprising for $q>0$; the authors should verify the citation context.","section":null}],"recommendation":"major_revision","confidential_remarks":"The skeptic's concern about the Pohozaev identity derivation in Theorem 2.2 is the most substantive issue. The gap is real: the paper jumps from the weak formulation (2.4) to the vanishing of $d/dμ E(u(x/μ))|_{μ=1}$ without justifying that the scaling variation is admissible. This is fixable — either by a direct Pohozaev computation with cut-offs (standard for such equations) or by verifying the Fréchet differentiability of $E$ along the scaling path — but it requires revision. The integral-form Pohozaev identity (Theorem 6.1) is more carefully done and could potentially serve as an alternative route if the PDE-side argument cannot be completed, but Theorem 1.2(ii) as stated relies on Theorem 2.2. The reader's concern about Theorem 2.5 (equivalence) is less serious: the estimates appear to hold upon inspection, though a clarifying sentence would help. Overall, the paper is a solid contribution with a specific gap to address."},"author_rebuttal":{"model":"glm-5.2","summary":"We thank the referee for a careful reading and for identifying two specific points where the exposition can be strengthened. Both comments are substantive and we address them in detail below.","responses":[{"response":"The referee is correct that the justification in the current manuscript is insufficient. The step from the weak formulation (2.4) to the Pohozaev identity via scaling of the energy functional is stated too tersely, and the admissibility of the scaling variation is not verified. We will revise the proof of Theorem 2.2 along the following lines. First, we note that Theorem 2.1 already establishes u ∈ L^{q+1}(R^3), so all three terms in E(u) are finite. The weak formulation (2.4), combined with the density of C_0^∞(R^3) in D, shows that δE(u)[φ] = 0 for all φ ∈ D. To obtain the Pohozaev identity, one needs the specific test function v = x·∇u, which does not lie in C_0^∞(R^3). We will add a direct Pohozaev-type computation using standard cut-off functions: multiply the PDE by (x·∇u)ζ_R^2 for a suitable cut-off ζ_R, integrate by parts, and pass to the limit R → ∞. The key boundary terms are controlled by the estimates already established in Theorem 2.1 (u ∈ L^{q+1}) and the fact that u ∈ D (so ∇u ∈ L^2 and Δu ∈ L^2). Specifically, the boundary terms on ∂B_R involve R∫_{∂B_R} u^{q+1} dS, R∫_{∂B_R} |∇u|^2 dS, and similar terms, all of which vanish along a sequence R_j → ∞ by standard arguments using the integrability already established. We will write out this computation explicitly. We also note that the integral-form Pohozaev identity (Theorem 6.1), which is derived with full convergence arguments, provides an independent route to the same conclusion q > 5 for solutions of the integral equation; the PDE-side Pohozaev identity in Theorem 2.2 is needed specifically for distribution solutions before the equivalence (Theorem 1.1) is fully established. The revised proof will make the logical dependency clear.","revision_made":"yes","referee_comment":"Theorem 2.2 (Pohozaev identity, PDE side) contains a gap: the step from the weak formulation (2.4) to d/dμ E(u(x/μ))|_{μ=1} = 0 is not justified. The scaling variation μ ↦ u(·/μ) must be shown to be admissible in the energy space, and the first variation must be shown to vanish along it. The referee suggests either (a) a direct Pohozaev computation with cut-off functions, or (b) verification that the scaling lies in the Fréchet-differentiability domain and that the weak formulation suffices for the specific variation v = x·∇u."},{"response":"The referee's concern is well-taken, and we agree that a clarifying sentence would strengthen the proof. The estimates do go through as written, but the reason is somewhat implicit and should be made explicit. The key point is the following. When |x| is small and ψ ∈ C_0^∞(B_R), the split into B_{|x|/2}(x) and B_R ∖ B_{|x|/2}(x) is valid regardless of where ψ is supported within B_R. On B_{|x|/2}(x), we use only that |ψ| ≤ ‖ψ‖_∞ and that |x−y| ≤ |x|/2, so the integral is bounded by C∫_0^{|x|/2} r dr / (a^2 |x|) ≤ C|x|, using (2.2). On B_R ∖ B_{|x|/2}(x), we have |x−y| ≥ |x|/2, and the exponential decay of ΔK_a gives the bound C e^{−|x|/(2a)} / |x| ≤ C/|x|. The boundedness of ψ (not its support properties) is what enters. The case where ψ is supported near the origin is therefore handled identically: the estimates depend only on ‖ψ‖_∞ and the kernel bounds (2.2)–(2.3), which are uniform. We will add a sentence after (2.17) making this explicit, stating that the estimates rely on the uniform bound |ψ| ≤ ‖ψ‖_∞ and the kernel estimates (2.2)–(2.3), and are therefore independent of the specific support of ψ within B_R.","revision_made":"yes","referee_comment":"Theorem 2.5 (equivalence, Step 1): the estimate |ΔΦ| ≤ C/|x| for small |x| (equation 2.17) is used to conclude ΔΦ ∈ L^2(R^3) via ∫_{B_δ} |x|^{-2} dx < ∞. The referee asks the authors to verify and explicitly confirm that the case ψ supported near the origin (where |x|/2 may not be small relative to the support) is handled."}],"tokens_in":29401,"tokens_out":1196,"duration_ms":2331933,"standing_objections":[]},"desk_editor":{"model":"glm-5.2","letter":"The main result is a pair of Liouville theorems for $-Δu + a²Δ²u = u^q$ in $ℝ³$: positive distribution super-solutions require $q > 3$, and positive distribution solutions require $q > 5$. These exponents match the classical Lane-Emden case, which is a clean structural statement — the second-order operator dominates the fourth-order one. The paper also establishes equivalence between the PDE and an integral equation with the Coulomb-type kernel $K_a$, proves regularity (differentiability, $L^{q+1}$ membership, radial symmetry via moving planes, and sharp decay $u(x) ∼ K_a(x)$), and extends the Pohozaev identity to an Allen-Cahn-type integral equation. The integral-form Pohozaev identity in Section 6 is the strongest part: the convergence arguments are explicit and careful, and the conclusion $q > 5$ there is well-supported. The iteration argument for the super-solution non-existence ($q ≤ 3$) in Section 3 is also solid and self-contained. The equivalence result (Theorem 2.5) rests on case-by-case estimates showing $K_a * ψ ∈ D$; I checked the $|ΔΦ| ≤ C/|x|$ estimate near the origin (equation 2.17) and the $L²$ integrability follows since $|x|^{-2}$ is integrable near zero in $ℝ³$. That load-bearing step holds up. Now the soft spot. The stress-test concern about Theorem 2.2 is real. The PDE-side Pohozaev identity is derived by asserting that $u$ is a critical point of $E(u)$ and then computing $d/dμ E(u(x/μ))|_{μ=1} = 0$. The paper establishes that $u ∈ L^{q+1}$ (Theorem 2.1) and that the weak formulation (2.4) holds, but the step from 'u satisfies the weak equation with test functions in $C_0^∞$' to 'the scaling variation $μ ↦ u(x/μ)$ is admissible in the function space where $E$ is differentiable' is not justified. The scaling $u(x/μ)$ changes the support properties and the boundary terms from the cut-off procedure could fail to vanish in the right way. The paper says it wants to 'avoid the tedious calculations' of a direct Pohozaev computation, but that shortcut leaves a gap. This matters because Theorem 2.2 is the stated Liouville theorem for distribution solutions. However, the conclusion $q > 5$ is independently recovered through the integral equation route: Theorem 1.1(i) converts PDE solutions to integral solutions, and Theorem 6.1 proves $q > 5$ for integral solutions with a careful, convergent Pohozaev identity. So the main theorem is likely correct — it just needs either a direct Pohozaev computation from the PDE or an explicit verification that the scaling variation is admissible. The Allen-Cahn extension (Section 7) is a nice addition and the Pohozaev identity there is derived more carefully, with explicit convergence of all integrals. This is a paper for researchers in nonlinear elliptic PDEs and integral equations. The techniques are standard (Chen-Li-Ou moving planes, HLS, regularity lifting) but the specific operator and the sharp exponents are new. The paper deserves a serious referee who can check the variational gap in Theorem 2.2 and confirm that the integral-equation route fully covers the PDE case.","headline":"Sharp Liouville theorems for a fourth-order elliptic operator with a second-order term: critical exponents match the Lane-Emden case, but the PDE-side Pohozaev identity has a gap in its variational justification.","tokens_in":30435,"tokens_out":869,"would_cite":false,"duration_ms":342706,"reading_group":"no","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35Q55","35Q60","35J91","45E10"],"pacs":[],"model":"glm-5.2","headline":"No positive solutions below q=5 for a mixed-dispersion Schrödinger equation","keywords":["Liouville theorem","mixed dispersion Schrödinger equation","distribution solutions","Pohozaev identity","Coulomb potential","critical exponent","integral equation","radial symmetry"],"falsifier":"If one could construct a positive distribution solution of $-Δu + a²Δ²u = u^q$ in D for some $q ≤ 5$, or a positive super-solution for $q ≤ 3$, the Liouville theorem would fail. Alternatively, if the equivalence between the PDE and the integral equation (Theorem 1.1) were shown to be incomplete—e.g., if there exist distribution solutions in D that do not satisfy the integral equation—then the regularity, symmetry, and Pohozaev arguments built on the integral formulation would not apply to all distribution solutions.","tokens_in":29254,"feed_emoji":"⚖️","tokens_out":1312,"duration_ms":194014,"temperature":0.7,"pith_summary":"The paper studies the equation $-Δu + a²Δ²u = u^q$ in $ℝ³$, which arises as the static form of a mixed-dispersion nonlinear Schrödinger equation and as the Euler–Lagrange equation for a Sobolev-type embedding inequality. The central claim is a Liouville theorem: if a positive distribution solution exists, then necessarily $q > 5$; if only a positive super-solution exists, then $q > 3$. These thresholds match the classical Lane–Emden exponents for $-Δu = u^q$, showing that the second-order operator $-Δ$ dominates the fourth-order term $a²Δ²$ in governing existence. The method hinges on proving equivalence between the PDE and an integral equation involving the kernel $K_a(x) = (1 - e^{-|x|/a})/|x|$, a modified Coulomb potential. Once equivalence is established, the authors deploy integral estimates, a regularity lifting lemma, the method of moving planes in integral form, and a Pohozaev identity derived from the integral formulation to obtain regularity, radial symmetry, decay rates, and the nonexistence results. A consequence is that the best constant in the embedding inequality $||u||_{L^{q+1}} ≤ C(||∇u||_{L²} + a||Δu||_{L²})$ is not attained at $q = 5$, and similarly the best constant in a Hardy–Littlewood–Sobolev-type inequality with the Coulomb kernel is not attained. The paper extends the same Pohozaev identity technique to an Allen–Cahn-type integral equation, obtaining Liouville theorems there as well.","feed_headline":"Mixed-dispersion Schrödinger equation has no solutions below q=5","feed_subtitle":"Liouville theorem shows the critical exponent matches classical Lane–Emden, proving embedding constants are unattainable.","key_machinery":"The modified Coulomb potential $K_a(x) = (1 - e^{-|x|/a})/|x|$, which serves as the Green's function for the operator $-Δ + a²Δ²$ and bridges the PDE to an equivalent integral equation. The Pohozaev identity in integral form, derived from the variational structure of the energy functional, yields the exponent bound $q > 5$.","core_discovery":"The critical exponent for positive distribution solutions of $-Δu + a²Δ²u = u^q$ in $ℝ³$ is $q = 5$: solutions exist only for $q > 5$, and super-solutions only for $q > 3$. This is established by proving equivalence between the PDE and the integral equation $u(x) = ∫ K_a(x-y) u^q(y) dy$ with $K_a(x) = (1-e^{-|x|/a})/|x|$, then applying a Pohozaev identity in integral form. The thresholds coincide with the Serrin exponent ($q=3$) and Sobolev exponent ($q=5$) of the Lane–Emden equation $-Δu = u^q$, demonstrating that the $-Δ$ operator governs the critical exponents despite the presence of the fourth-order term.","pith_inferences":[],"forward_implications":["The non-attainability of the best constant in the embedding inequality at $q=5$ means that extremal functions for this Sobolev-type embedding do not exist, which constrains the variational approach to finding ground states of the mixed-dispersion NLS.","The equivalence between the PDE and the integral equation with kernel $K_a$ provides a toolbox—regularity lifting, moving planes, decay estimates—that can be applied to other fourth-order elliptic problems whose Green's functions have similar Coulomb-type structure, such as the Bopp–Podolsky electrostatic theory.","The Allen–Cahn-type Liouville theorem (trivial solutions only for $1 < q ≤ 6$) suggests a phase-transition threshold: for the mixed-dispersion Allen–Cahn equation, nontrivial phase-coexistence solutions require sufficiently strong nonlinearity ($q > 6$).","The radial symmetry and precise decay rate $u(x) ∼ K_a(x) ∼ 1/|x|$ at infinity for positive solutions provide the asymptotic profile needed to study stability and scattering of standing waves in the time-dependent mixed-dispersion Schrödinger equation."],"fun_headline_variants":["Mixed-dispersion Schrödinger solutions require q > 5","Fourth-order Schrödinger shares critical exponent with Lane-Emden","Coulomb potential establishes q=5 threshold for static Schrödinger","Static dispersion Schrödinger lacks positive solutions below q=5","Pohozaev identity yields q=5 limit for mixed-dispersion Schrödinger"],"cache_read_input_tokens":0,"weakest_assumption_plain":"The load-bearing premise is that the PDE and the integral equation are equivalent—that for any test function ψ, the convolution $Φ = K_a * ψ$ belongs to the energy space D (meaning its gradient and Laplacian are both square-integrable). This is verified by case-by-case estimates of $|∇Φ|$ and $|ΔΦ|$ near the origin, at infinity, and in bounded annuli. If any of these estimates fail—particularly the square-integrability of $ΔΦ$ near the origin, where $|ΔΦ| ≤ C/|x|$—the bridge,","fun_headline_variants_meta":{"raw":{"variants":["Mixed-dispersion Schrödinger solutions require q > 5","Fourth-order Schrödinger shares critical exponent with Lane-Emden","Coulomb potential establishes q=5 threshold for static Schrödinger","Static dispersion Schrödinger lacks positive solutions below q=5","Pohozaev identity yields q=5 limit for mixed-dispersion Schrödinger"]},"model":"glm-5.2","effort":"high","cost_usd":0.0,"raw_usage":{"total_tokens":1773,"prompt_tokens":594,"completion_tokens":1179,"prompt_tokens_details":null},"tokens_in":594,"tokens_out":1179,"duration_ms":49756,"temperature":1.0,"reasoning_tokens":1025,"cache_read_input_tokens":0,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-08T20:27:20.034025+00:00","model_set":{"reader":"glm-5.2"},"falsifier":"If one could construct a positive distribution solution of $-Δu + a²Δ²u = u^q$ in D for some $q ≤ 5$, or a positive super-solution for $q ≤ 3$, the Liouville theorem would fail. Alternatively, if the equivalence between the PDE and the integral equation (Theorem 1.1) were shown to be incomplete—e.g., if there exist distribution solutions in D that do not satisfy the integral equation—then the regularity, symmetry, and Pohozaev arguments built on the integral formulation would not apply to all distribution solutions.","supporting_citations":[],"review_version":1}