{"id":"3ec5b085-b4ce-40b1-ade4-0a6fe0417eba","arxiv_id":"1908.03386","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"For a fractional Laplacian equation with coefficient K, stable critical points of K, including saddle points, generate a solution with many bubbles for small perturbation exponent.","lead":"The paper proves that a fractional Laplacian equation with a small perturbation in the critical exponent has solutions made of many peaked bubbles, concentrated near stable critical points of the coefficient K, including saddle points. The result extends classical Laplacian concentration theory to nonlocal operators and uses local Pohozaev identities to determine bubble locations.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1.1 is stated for both signs in the exponent, but the proof is written only for the plus sign; the minus branch is dismissed in one sentence, and the reduction estimates are sign-sensitive. The central claim as stated is therefore not fully established.","rationale":"The reader's verdict is CONDITIONAL, and I agree that the paper should not be accepted as fully proving the stated theorem without an additional check. The reader's weakest_assumption, however, is the technical parameter range max{...} < s < 1 entering Lemma 2.5. That is genuinely a stated assumption of Theorem 1.1, and Remark 1.2 explicitly flags it as technical; since the theorem assumes it, its failure would only narrow the parameter range and is not by itself a correctness gap. The stronger, more load-bearing issue is the omitted -epsilon branch: Theorem 1.1 and Theorem 1.5 claim both signs, while the proof says only the plus case is treated and the minus case is dismissed with 'slightly modifying the arguments.' The estimates in Lemma 2.4, Lemma 2.5, and Lemma B.7 use +epsilon in exponents, logarithmic bounds, and coefficients; their analogues for -epsilon are not written down. If any of those sign-dependent bounds fails, the theorem is false as stated or at least unproved for half its scope. The plus-sign chain looks coherent: the finite-dimensional reduction, the Pohozaev identities, the degree argument, and the local estimates are consistent, and the technical inequalities in Remark 1.2 are satisfied inside the stated range. I see no internal contradiction in the plus-sign argument that would force rejection outright. The appropriate verdict is CONDITIONAL: require a written verification of the minus-sign branch, or amend the theorem to the plus sign only.","tokens_in":1092,"tokens_out":976,"duration_ms":338985,"concrete_test":"Recompute the full reduction for the sign p = 2_s* - 1 - epsilon: (i) redo Lemma 2.4 with min(2_s* - 1 - epsilon, 2); (ii) redo the J1 estimate in Lemma 2.5 with 2_s* - 1 - epsilon, checking the exponent and the epsilon ln(1/epsilon) bound used in (2.23); (iii) redo Lemma B.7 with U^{2_s* - epsilon} and 1/(2_s* - epsilon), verifying the sign of the B1/lambda^3 term and the form of the reduced equation (3.68). If all estimates hold with the appropriate 1 +/- iota exponents and the degree argument in Theorem 1.5 is unchanged, the missing branch is benign; otherwise Theorem 1.1 should either be restricted to the plus sign or supplemented with a separate proof for the minus sign.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim covers both signs 2_s* - 1 +/- epsilon, but the proof is carried out only for +epsilon. The text states that the case 2_s* - 1 - epsilon 'can be obtained by slightly modifying the arguments' (Section 1). This is a theorem-proof mismatch for a sign that is not a mere notational variant. Every key estimate in the reduction changes with the sign of epsilon. Lemma 2.4 uses the contraction exponent min(2_s* - 1 + epsilon, 2); replacing +epsilon by -epsilon changes the exponent and the contraction bound. Lemma 2.5 estimates J1 using Z^{2_s* - 1 + epsilon} and the inequality epsilon ln(1/epsilon) <= C epsilon^{(1+iota)/(N-2s)}, whose direction is specific to the positive perturbation; for the minus sign the analogous logarithmic term must be rechecked in (2.23). Lemma B.7 expands the reduced functional with U^{2_s* + epsilon} and 1/(2_s* + epsilon); with -epsilon the terms F2, F3 and the sign of the B1/lambda^3 coefficient in (3.68) need independent verification. None of these minus-sign versions is supplied. Since Theorem 1.1 is not restricted to the plus branch, the theorem as written is not established; the plus-sign construction itself appears coherent.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs many-bubble solutions for the fractional critical equation (-Δ)^s u = K(|y'|,y'') u^{(N+2s)/(N-2s) ± ε} in R^N, N≥4, under conditions (K1)-(K2) on K. The proof combines a finite-dimensional Lyapunov-Schmidt reduction with local Pohozaev identities introduced by Peng-Wang-Yan. The main theorem asserts that for a range of s and for small ε, there exist solutions with m ~ ε^{-(N-2s-2)/(N-2s)^2} bubbles concentrating near a stable critical point of K, which may be a saddle. The detailed proof is given for the plus sign, with the minus sign dismissed in one sentence.","tokens_in":45224,"tokens_out":20280,"duration_ms":175444,"significance":"If fully established, the result is a meaningful extension of multi-bubble constructions to the fractional Laplacian, allowing saddle-type concentration points and avoiding direct differentiation of the reduced functional. The paper contains a substantial amount of technical work: the weighted norms, the contraction argument, and Appendices A-D with the Pohozaev identities and auxiliary estimates. The plus-sign construction appears internally coherent. The main weakness is that the theorem as stated covers both signs while the proof only treats the plus sign; this is a theorem-proof mismatch that must be resolved before the result can be accepted in its present form.","major_comments":[{"comment":"Theorem 1.1 is stated for both exponents 2_s^*-1+ε and 2_s^*-1-ε, but the proof is carried out only for the plus sign; the text says the minus case can be obtained by slightly modifying the arguments. This is not a notational variant. The sign of ε enters Lemma 2.4 through the exponent min(2_s^*-1+ε,2), Lemma 2.5 through the estimate (2.23) for J1, Lemma B.7 through the expansion with 1/(2_s^*+ε), and the reduced equations (3.66)-(3.68) through the balance between B1/λ^3 and B3 m^{N-2s}/λ^{N-2s+1}. None of these minus-sign versions is supplied, and the degree argument in Section 3 depends on the actual sign of the leading term in (3.68). The theorem as stated is therefore not established; the authors should either prove the minus-sign case or restrict the theorem to the plus sign.","section":"Section 1, after Remark 1.3"},{"comment":"In the estimate of M12 for the case 2_s^* ≤ 3, the text states 'Noting that τ > 2s', but the standing assumption of the paper (Remark 1.2 and the inequalities used in Lemma 2.5) gives 2s > τ. Please clarify whether τ > 2s is a typo. If the estimate actually requires τ > 2s, then the proof of (2.12) is invalid under the stated assumptions on s, and the estimate of the coefficients c_l in Lemma 2.1 would need reworking.","section":"Appendix C, proof of (2.12)"}],"minor_comments":[{"comment":"The statement 'lim_{ε→0} λ^{(N-2s)/2} ε = c' with c a positive constant is incorrect under the scaling λ ~ ε^{-1/(N-2s)}; the product tends to 0. The subsequent bound λ^{(N-2s)/2} ε ≤ C is what is needed and is true.","section":"Remark 1.4"},{"comment":"In the definition of the domain of the reduced functional F, the upper endpoint is written as L1 ε^{1/(N-2s)}; the exponent should be -1/(N-2s) to match (1.9) and the rest of the paper.","section":"Page 5, outline of Section 3"},{"comment":"In the proof, the text says 'In order to estimate J12, first we define...'; there is no quantity J12, and the intended reference is likely J2.","section":"Lemma 2.5"},{"comment":"In the proof, u_ε is written as Z_{\\bar r,\\bar y'',\\mu} + φ with an undefined parameter μ; it should be λ.","section":"Lemma 3.4"},{"comment":"There is a typo 'the re exists a unique ϕ' that should read 'there exists'.","section":"Proposition 2.3 proof"}],"recommendation":"major_revision","confidential_remarks":"The core reduction for the plus sign seems substantial and carefully executed, but the stated theorem covers both signs while the proof does not. This needs to be fixed before acceptance. The Appendix C inequality 'τ > 2s' also contradicts the assumptions and should be corrected or explained."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The plus-sign construction in this paper looks like a real result, but Theorem 1.1 overclaims: it states both signs 2_s*−1±ε, and the proof only treats the plus case. The minus branch is dismissed in one sentence in Section 1, and the estimates that follow are genuinely sign-sensitive. As written, the theorem is not established.\n\nWhat is actually new: this is the first construction I know of that produces many bubbles for a fractional critical equation with the concentration driven by a saddle point of K, and it does so via local Pohozaev identities for the harmonic extension rather than by differentiating the reduced functional directly. The analytic machinery — the norms, the contraction argument, and the appendices — is detailed and, as far as I can tell, coherent. The technical restriction on s is unpleasant but the authors flag it as a limitation, which is honest.\n\nThe soft spot is the sign gap. The minus sign changes the exponent in the nonlinearity; the estimates in Lemma 2.4, Lemma 2.5, and Lemma B.7 use inequalities whose direction is tied to +ε, and the reduced equation (3.68) has a coefficient whose positivity was checked for +ε. The sentence \"can be obtained by slightly modifying the arguments\" is not enough for a nonlocal problem where the interactions are not perturbative in an obvious way. If the authors restrict the statement to the plus sign, I think the paper deserves serious consideration. If they want both signs, the minus branch has to be written out.\n\nThe citation pattern looks appropriate: the fractional Pohozaev machinery from [16] and the prescribed-curvature construction in [27] are credited, and the s=1 cases are properly distinguished.\n\nBottom line: this is a paper for specialists in concentration phenomena for nonlocal PDEs. It should not be desk-rejected, but the referee should insist on either a proof of the minus case or a corrected theorem. My own verdict on the stated theorem would be negative until that gap is closed.","headline":"Plus-sign construction is solid and new, but the theorem as stated covers a sign that is never proved.","tokens_in":45665,"tokens_out":2386,"would_cite":false,"duration_ms":28351,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35B05","35B45"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that a perturbed fractional critical equation has positive solutions made of a large, epsilon-dependent number of bubbles, concentrating at a stable critical point of K that may be a saddle.","keywords":["bubble solutions","fractional Laplacian","critical exponent","finite-dimensional reduction","local Pohozaev identities","saddle critical point","harmonic extension","concentration points"],"falsifier":"In dimension $N=4$, take $s=0.38$, just below the boundary value $(3-\\sqrt{5})/2\\approx0.382$, so that $\\tau=(1-s)/(2-s)\\approx0.383$ and the inequality $N>4+2\\tau-2s$ fails. Evaluate the annulus contribution to $J_3$ in Lemma 2.5: on the annulus $\\sigma\\epsilon^{(1/2+\\iota)/(N-2s)}\\le |(r,y'')-(r_0,y_0'')|\\le 2\\delta$, the factor $1/(1+\\lambda|y-x_j|)$ is at most $C\\epsilon^{(1/2-\\iota)/(N-2s)}$, and the final exponent inequality becomes negative when $s<\\tau$. A worked calculation at this $s$ would show precisely which estimate breaks, settling whether the lower bound is intrinsic or an artifact of the proof.","tokens_in":44667,"feed_emoji":"🧮","tokens_out":11524,"duration_ms":114544,"temperature":0.7,"pith_summary":"The paper proves that a nonlocal equation at critical Sobolev growth, perturbed by a small exponent shift $\\epsilon$, has positive solutions assembled from many concentrated bubbles. The number of bubbles is of order $\\epsilon^{-(N-2s-2)/(N-2s)^2}$ as $\\epsilon \\to 0$, and the energy diverges at the same rate. The key point is that the bubbles can concentrate at a stable critical point of the coefficient $K$ that is a saddle, so no minimization or maximization can locate them. The proof reduces the full problem to a finite-dimensional one and solves the reduced equations by a degree argument, using local Pohozaev identities to find the concentration points without directly differentiating the reduced functional.","feed_headline":"Perturbed fractional equation admits solutions with many bubbles","feed_subtitle":"Number of bubbles grows like $\\epsilon^{-(N-2s-2)/(N-2s)^2}$; concentration points may be saddle critical points of $K$.","key_machinery":"The construction is a finite-dimensional reduction (Lyapunov-Schmidt) around a sum of standard bubbles $U_{x_j,\\lambda}(y)=C_{N,s}(\\lambda/(1+\\lambda^2|y-x_j|^2))^{(N-2s)/2}$, the explicit positive solutions of the unperturbed critical problem. Weighted norms $\\|\\cdot\\|_*$ and $\\|\\cdot\\|_{**}$ measure the correction and the error; with the estimates of Lemmas 2.4 and 2.5, a contraction mapping produces $\\phi$ with $\\|\\phi\\|_* \\le C \\epsilon^{(1+\\iota)/(N-2s)}$. To choose the parameters, the paper uses local Pohozaev identities for the harmonic extension into the upper half-space; these turn stationarity into the algebraic system $\\nabla K \\approx 0$ and the scale balance above, and a degree argument on that system finishes the proof. The identities also avoid the long direct expansions of derivatives of the reduced functional.","core_discovery":"Theorem 1.1 states that under conditions (K1) and (K2), if $N \\ge 4$ and $s$ lies above an explicitly displayed lower bound, then for every sufficiently small $\\epsilon$ the problem (1.1) has a positive solution $u_\\epsilon$ with $m$ bubbles, where $m \\sim \\epsilon^{-(N-2s-2)/(N-2s)^2}$. The bubble centers are placed on a regular $m$-gon in the first two coordinates with a common remaining coordinate, the scale is $\\lambda \\sim \\epsilon^{-1/(N-2s)}$, and as $\\epsilon \\to 0$ the centers converge to $(r_0,y_0'')$, with the correction $\\phi_\\epsilon$ small in a weighted norm. The concentration condition reduces to $\\nabla K(\\bar r,\\bar y'') = o(\\epsilon^{(1-\\iota)/(N-2s)})$ together with the scale balance $-B_1/\\lambda^3 + B_3 m^{N-2s}/\\lambda^{N-2s+1} = o(1)$, and a nonzero degree of $\\nabla K$ at the critical point yields a solution of this system. Consequently the stable critical point of $K$ may be a saddle.","pith_inferences":["The technical lower bound on $s$ likely marks where the current proof stops rather than where the phenomenon stops; computing $\\|l_\\epsilon\\|_{**}$ at the boundary value of $s$ would show whether the restriction is removable.","The Pohozaev-identity route should transfer to other nonlocal critical problems whose coefficient has saddle critical points, such as fractional Nirenberg-type problems on domains or spheres, where direct derivative expansions are heavier.","A natural numerical check is to solve the reduced finite-dimensional system for small $\\epsilon$ in dimension $N=4$ with a $K$ chosen to have a saddle critical point and compare the predicted bubble count and energy scaling with the theorem's formulas."],"forward_implications":["For every sufficiently small $\\epsilon$ there is a positive solution with roughly $\\epsilon^{-(N-2s-2)/(N-2s)^2}$ bubbles, so the number of peaks tends to infinity as the perturbation vanishes.","The corresponding energy is of order $\\epsilon^{-(N-2s-2)/(N-2s)^2}$, so these are high-energy solutions produced by a single small exponent shift.","A saddle critical point of $K$ with $\\Delta K(y_0)<0$ and nonzero local degree is enough; nondegeneracy of the critical point is not required.","The same argument with minor changes treats the negative perturbation exponent $2_s^*-1-\\epsilon$, and the paper conjectures that the borderline case $N=3$ needs a logarithmic bubble count.","The explicit restriction on $s$ is automatic when $s=1$, so the result recovers the classical Laplacian phenomenon as a limit."],"supporting_citations":[{"why":"Supplies the finite multi-bubble reduction and the local-Pohozaev method this paper adapts from the prescribed scalar curvature problem.","marker":"[27]"},{"why":"Introduces the local Pohozaev identities used here to locate concentration points without differentiating the reduced functional.","marker":"[28]"},{"why":"Provides the weighted norms and the key interaction estimate (Lemma B.1) used to control bubble overlap.","marker":"[34]"},{"why":"Gives the original idea of constructing infinitely many bubble solutions, which the reduction follows.","marker":"[33]"},{"why":"Shows how to let the number of concentration points depend on a small parameter, guiding the choice $m\\sim\\epsilon^{-(N-2s-2)/(N-2s)^2}$.","marker":"[20]"},{"why":"Proves the analogous large-number-of-bubbles result for the classical Laplacian, the baseline this paper extends to the fractional setting.","marker":"[21]"},{"why":"Supplies the boundary estimates for harmonic extensions needed for local Pohozaev identities in the fractional case.","marker":"[16]"},{"why":"Establishes the harmonic-extension representation of the fractional Laplacian, on which the local identities are built.","marker":"[7]"}],"fun_headline_variants":["Many-bubble solutions found for fractional Laplacian","Fractional Laplacian: saddle-point bubbles multiply","Bubbles proliferate in perturbed fractional equation","Saddle points yield many bubbles in fractional PDE"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is the displayed lower bound on $s$, which guarantees the two technical inequalities $2s>\\tau$ and $N>4+2\\tau-2s$ with $\\tau=(N-2s-2)/(N-2s)$; the contraction estimate for the error term $l_\\epsilon$ does not close without them, and the authors state they do not know how to remove them.","fun_headline_variants_meta":{"raw":{"variants":["Many-bubble solutions found for fractional Laplacian","Fractional Laplacian: saddle-point bubbles multiply","Bubbles proliferate in perturbed fractional equation","Saddle points yield many bubbles in fractional PDE"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00045,"raw_usage":{"total_tokens":2358,"prompt_tokens":1128,"completion_tokens":1230,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":744,"completion_tokens_details":{"reasoning_tokens":1168}},"tokens_in":744,"tokens_out":1230,"duration_ms":11586,"temperature":1.0,"reasoning_tokens":1168,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:15:26.474094+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"In dimension $N=4$, take $s=0.38$, just below the boundary value $(3-\\sqrt{5})/2\\approx0.382$, so that $\\tau=(1-s)/(2-s)\\approx0.383$ and the inequality $N>4+2\\tau-2s$ fails. Evaluate the annulus contribution to $J_3$ in Lemma 2.5: on the annulus $\\sigma\\epsilon^{(1/2+\\iota)/(N-2s)}\\le |(r,y'')-(r_0,y_0'')|\\le 2\\delta$, the factor $1/(1+\\lambda|y-x_j|)$ is at most $C\\epsilon^{(1/2-\\iota)/(N-2s)}$, and the final exponent inequality becomes negative when $s<\\tau$. A worked calculation at this $s$ would show precisely which estimate breaks, settling whether the lower bound is intrinsic or an artifact of the proof.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the finite multi-bubble reduction and the local-Pohozaev method this paper adapts from the prescribed scalar curvature problem."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the local Pohozaev identities used here to locate concentration points without differentiating the reduced functional."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the weighted norms and the key interaction estimate (Lemma B.1) used to control bubble overlap."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the original idea of constructing infinitely many bubble solutions, which the reduction follows."},{"cited_title":"Lin, W.-M","cited_arxiv_id":null,"evidence_quote":"Shows how to let the number of concentration points depend on a small parameter, guiding the choice $m\\sim\\epsilon^{-(N-2s-2)/(N-2s)^2}$."},{"cited_title":"Liu, Large number of bubble solutions for the equation ∆ u + K(y)u N +2 N − 2 ±ǫ = 0 on RN","cited_arxiv_id":null,"evidence_quote":"Proves the analogous large-number-of-bubbles result for the classical Laplacian, the baseline this paper extends to the fractional setting."},{"cited_title":"Solutions for fractional operator problem via local Pohozaev identities","cited_arxiv_id":"1904.08316","evidence_quote":"Supplies the boundary estimates for harmonic extensions needed for local Pohozaev identities in the fractional case."},{"cited_title":"Caffarelli, L","cited_arxiv_id":null,"evidence_quote":"Establishes the harmonic-extension representation of the fractional Laplacian, on which the local identities are built."}],"review_version":1}