{"id":"d621a57c-57bb-48dd-963d-1d8aa685a97b","arxiv_id":"2411.15335","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Viscosity solutions of fully nonlinear free transmission problems are shown to be C^{1,log-Lipschitz at the boundary and W^{2,p} in a boundary neighborhood, under a closeness condition between the phase operators.","lead":"This paper proves boundary regularity estimates for a class of two-phase diffusion problems where the governing equation switches depending on the sign of the solution. The results extend known single-operator boundary regularity theory to free transmission problems, without assuming convexity of the operators.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 2 claims W^{2,d} for f∈L^d, but the proof's maximal-function estimate only works for p>d; the endpoint case is not proved.","rationale":"The paper's main advertised results are Theorems 1 and 2. Theorem 1's C^{1,Log-Lip} estimate follows a standard approximation-and-bootstrapping scheme, and its main caveat is the unquantified closeness constant in Assumption A4, which is likely repairable by stating the smallness condition explicitly. Theorem 2, however, contains a concrete gap: the proof's only mechanism for summability of the dyadic distribution function is the strong-type maximal inequality, which requires the exponent p on f to exceed d. The theorem's hypothesis f∈L^d forces p=d, so the final estimate ∑ M^{pk}α_k is not controlled. This is not a matter of consensus or a missing technical refinement; it is an internal mismatch between the statement and the proof. The proposed check—repeating the argument at p=d via weak-type (1,1)—would either reveal a missing endpoint argument or force the theorem to be corrected to p>d. Either way the current version overclaims. The reader's CONDITIONAL verdict is appropriate; the paper should be accepted only with the Sobolev statement corrected or an endpoint proof supplied.","tokens_in":16138,"tokens_out":5597,"duration_ms":49533,"concrete_test":"Re-run the final summation in the proof of Theorem 2 with p=d. Replace the strong-type bound by the weak-type (1,1) estimate for M(f^d) and attempt to bound ∑_{k≥0} M^{dk} β_k. In particular, test the inequality ‖M(g)‖_{L^1} ≤ C‖g‖_{L^1} for g=f^d∈L^1; since this is false (e.g., g(x)=1_{B(0,1)}(x)/(|x|^d log^2(2/|x|))), the series is not controlled, confirming that Theorem 2 as stated is not proved.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"In the proof of Theorem 2 (Section 5), the authors control the dyadic sums via the strong-type maximal inequality: ‖M(f^d)‖_{L^{p/d}(Ω)} ≤ C‖f^d‖_{L^{p/d}(Ω)} = C‖f‖_{L^p(Ω)}^d. This bound is valid only for p/d > 1, i.e., p > d. It is then used to assert equation (20), ∑_{k≥0} M^{pk} β_k ≤ C. The theorem, however, states the conclusion u∈W^{2,d} for f∈L^d, which corresponds to p=d. In that case M(f^d) is only of weak type (1,1), and the series ∑ t |{M(f^d) ≥ t}| need not be finite for f^d∈L^1. Thus the displayed argument does not justify the endpoint p=d; the proof establishes W^{2,p} only for some p>d. This is an internal inconsistency between the statement of Theorem 2 and its proof.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies boundary regularity for the fully nonlinear free transmission problem (1), in which the operator switches between F1 and F2 according to the sign of the solution. Working with Lp-viscosity solutions of the viscosity inequality formulation (2)-(3), the authors prove two main results under different closeness assumptions to a common uniformly elliptic operator F: Theorem 1 establishes C^{1,Log-Lip} boundary estimates when F1 and F2 are uniformly close to F in the sense of Assumption A4, and Theorem 2 claims W^{2,d} boundary regularity when the operators satisfy the stronger closeness condition A3 and the limiting operator is differentiable (A5). The proof strategy combines an approximation lemma (Proposition 4 and Corollary 1) with boundary regularity results for the limiting equation F=0 and Calderón-Zygmund-type arguments.","tokens_in":16341,"tokens_out":6600,"duration_ms":59485,"significance":"If correct, the results would extend boundary regularity theory from single fully nonlinear equations to free transmission problems without convexity assumptions, which is a genuine contribution to an active area. The paper is clearly structured, builds on established tools such as the ABP maximum principle and the boundary regularity results of Silvestre-Sirakov, and Theorem 1's proof is plausible. The main reservation concerns Theorem 2: the proof as written does not justify the claimed endpoint W^{2,d} estimate. Since Theorem 2 is one of the two stated headline results, the paper's overall significance is conditional on repairing that gap or adjusting the theorem.","major_comments":[{"comment":"The proof of Theorem 2 invokes the strong-type maximal inequality in the form ||M(f^d)||_{L^{p/d}(Ω)} ≤ C ||f^d||_{L^{p/d}(Ω)}, which is valid only when p/d > 1, i.e. p > d. The theorem statement assumes only f ∈ L^d, so the displayed estimate leading to (20), ∑ M^{pk} β_k ≤ C, is not justified at the endpoint p=d. Consequently, Lemma 1 yields the conclusion u ∈ W^{2,p} only for some p>d, not the stated W^{2,d}. Please either provide a genuine endpoint argument (for example, using the weak-type (1,1) bound together with an additional summability argument) or restate Theorem 2 with the weaker integrability assumption that the proof actually supports.","section":"Section 5, Proof of Theorem 2"},{"comment":"In the proof of Proposition 4, the step 'Proposition 3 applied to h − h(x0) in B_{θ/2}(x0) yields θ² ||D²h(x)|| ≤ C θ^α (1 + ||f||_{L^d})' is not justified as written. Proposition 3 gives a C^{2,α} bound on h, which implies ||D²h||_{L∞} ≤ C and hence θ² ||D²h|| ≤ C θ², not C θ^α. The subsequent estimate (10), controlling |F_i(D²h, Dh, x)| through Assumption A3, depends on this θ^{α-2} bound; without a correct derivation, the approximation-error estimate in Proposition 4 lacks support. This estimate is load-bearing for Lemma 2 and therefore for Theorem 2, so the argument needs to be clarified or corrected.","section":"Section 3, Proposition 4"}],"minor_comments":[{"comment":"In the induction step of the proof of Theorem 1, the sentence 'Since Assumption 3 holds uniformly for F_i, it also holds for F^k_i' appears to refer to Assumption A4 rather than Assumption A3; please correct the reference.","section":"Section 4, Step 3"},{"comment":"The sentence 'Let M > 0 and C0 > 0 be the same as in Lemma 7' should refer to Proposition 7, not Lemma 7, since the relevant statement is Proposition 7.","section":"Section 5, Proof of Theorem 2"},{"comment":"Proposition 5 assumes f ∈ L^p(B^+_{14√d}) but the asserted estimate uses the L^d norm of f; please state the intended integrability assumption consistently.","section":"Section 5, Proposition 5"},{"comment":"There are minor typographical issues (e.g., 'onde completes the proof' at the end of Proposition 4 and inconsistent use of B^+_1 versus B^+_{14√d} after rescaling) that should be corrected in a revision.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within the scope of math.AP and the first theorem appears plausible, but the proof of Theorem 2 contains a genuine endpoint gap: the strong-type maximal function argument requires p>d, while the theorem states W^{2,d} for f∈L^d. This is not a cosmetic issue; it affects a central advertised result. The issue with Proposition 4's θ^α estimate is also concerning and should be addressed in the revision. I do not recommend outright rejection because the errors are localized and the authors may either fix the endpoint argument or revise the statement to a weaker but still meaningful W^{2,p} result. The self-citations are contextually appropriate and there is no indication of circularity."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThis paper does a real job: it carries boundary regularity theory for fully nonlinear equations over to a genuinely free transmission problem, where the interface is solution-dependent and sits on the boundary. The main new tool is a pair of approximation lemmas that compare the two phase operators to a single limiting operator F, and the authors avoid convexity assumptions, which is a plus. The C^{1,Log-Lip} estimate (Theorem 1) is plausible and follows the established approximation-bootstrap program; I don't see any hidden circularity or fitted parameters. The example in Remark 1, showing that C^{2,α} boundary data cannot propagate to a continuous Hessian, is a nice sanity check.\n\nThe soft spot is Theorem 2. The statement claims W^{2,d} for f ∈ L^d, but the proof only runs when f ∈ L^p for some p>d. In the dyadic-sum argument, the authors invoke the strong-type maximal inequality on M(f^d) in L^{p/d}, which requires p/d>1. For p=d, M(f^d) is only weak type (1,1), and the series in (20) need not be finite. So the endpoint case is not proved. This is not a minor typo: the argument as written establishes W^{2,p} in a neighborhood of the boundary for any p>d (under the same assumptions), but it does not reach p=d. The authors should either correct the statement to p>d or supply a genuine endpoint argument (e.g., via potential estimates or a different summation). This is the single load-bearing problem.\n\nMinor points: Proposition 4's proof has a few sketchy parameter choices (the θ^{-2}κ regime is handled hastily) but nothing that looks fundamentally broken. The closeness conditions A3/A4 are unquantified, but that's normal in this approximation style; the results are conditional on those constants.\n\nOverall, the paper is a solid extension of Silvestre-Sirakov and Winter to a new class of problems, and the citation pattern looks honest. It deserves a serious referee, but I would not accept it until the Theorem 2 gap is resolved. The fix may be simple — state the Sobolev result for p>d — and then it would be a useful contribution for the PDE regularity community.","headline":"Solid boundary regularity toolkit for free transmission problems; Theorem 2 overclaims the endpoint p=d.","tokens_in":16846,"tokens_out":3655,"would_cite":false,"duration_ms":32307,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35B65","35R35","35B30"],"pacs":[],"model":"deepseek-v4-flash","headline":"A fully nonlinear free transmission problem driven by two sign-dependent operators has $C^{1,{\\rm Log-Lip}}$ boundary regularity when the operators are close to a common limit, and $W^{2,d}$ boundary estimates when the limit is…","keywords":["free transmission problem","fully nonlinear elliptic equations","boundary regularity","viscosity solutions","C^{1,Log-Lip} regularity","W^{2,d} estimates","approximation methods","free boundary"],"falsifier":"Take a domain satisfying A1, choose two operators satisfying A2 and A4 with arbitrarily small $\\tau$, put $f\\in L^p$, $p>d$, and prescribe boundary data that change sign on the flat boundary so that both phases meet at a boundary point $x_0$; if a normalized viscosity solution of (2)-(3) at such a point has boundary Taylor error whose ratio to $r^2\\ln(1/r)$ is unbounded as $r\\to 0$, Theorem 1 is false. A concrete place to look is a small derivative perturbation of the two-phase model in Remark 1, checking the observed decay against the claimed log-Lipschitz modulus.","tokens_in":15900,"feed_emoji":"📐","tokens_out":14193,"duration_ms":116634,"temperature":0.7,"pith_summary":"This paper aims to prove that solutions of a fully nonlinear free transmission problem—where the diffusion operator switches between two operators depending on the sign of the solution—are as regular at the boundary as solutions of a single uniformly elliptic equation, provided the two operators are uniformly close to one limiting operator. If the theorems are right, the free boundary itself does not destroy boundary regularity: near boundary points where both phases meet, a normalized solution is $C^{1,{\\rm Log-Lip}}$, meaning it has a tangent gradient and the Taylor error is at most $C|x-x_0|^2\\ln(1/|x-x_0|)$ whenever $f\\in L^p$, $p>d$. When the limiting profile is differentiable in the Hessian variable, the same solutions belong to $W^{2,d}$ in a boundary strip with $f$ merely in $L^d$. The argument is an approximation method: compare $u$ with a smooth solution of the limiting homogeneous equation and import known boundary estimates back to the transmission problem, with no convexity assumptions on the operators.","feed_headline":"Free transmission solutions stay C^{1,Log-Lip} at the boundary","feed_subtitle":"When the phase operators nearly match one limit, solutions are differentiable at the boundary; extra smoothness gives W^{2,d}.","key_machinery":"The load-bearing mechanism is the approximation lemma (Proposition 4 and Corollary 1): near the boundary, one solves the limiting homogeneous problem $F(D^2h,Dh)=0$ with $h=u$ on the boundary of a smaller half-ball, obtaining a $C^{2,\\alpha}$ comparison function; the difference $w=u-h$ then satisfies the extremal-Pucci-type inequalities of the class $\\mathcal S^*(\\varphi)$ with data controlled by $\\kappa^\\gamma+\\|f\\|_{L^d}$, where the extremal Pucci operators are the minimal and maximal envelopes of uniformly elliptic operators. In Theorem 1, the comparison drives an induction over dyadic scales that constructs affine and quadratic polynomials tracking $u$ at the boundary, producing the log-Lipschitz modulus of the gradient. In Theorem 2, the same comparison feeds a Calderón–Zygmund decomposition argument that turns the measure decay of the bad sets $A_{M^k}$ into $L^d$-integrability of $D^2u$.","core_discovery":"The central claim is that the boundary regularity of the two-phase system is controlled by the limiting homogeneous problem. Theorem 1 states that a normalized viscosity solution of the two-inequality system, under the domain-regularity assumption A1, the structural ellipticity assumption A2, the uniform closeness assumption A4, and $f\\in L^p$, $p>d$, is $C^{1,{\\rm Log-Lip}}$ at the boundary: for every boundary point $x_0$, $|u(x)-u(x_0)-Du(x_0)(x-x_0)|\\leq C|x-x_0|^2\\ln(1/|x-x_0|)$. Theorem 2 states that if the limiting operator is also differentiable in the Hessian entry (A5) and the closeness allows Hessian growth (A3), then $f\\in L^d$ suffices to place $u$ in $W^{2,d}$ on a boundary strip, with a universal estimate. The paper also shows that this is near the optimal scale: even with convex, smooth operators that agree on the boundary, the Hessian can be discontinuous, so one cannot expect $C^{2,\\alpha}$ propagation.","pith_inferences":["The paper does not pursue it, but the same two-inequality comparison should force boundary regularity for any finite number of sign-selected operators, as long as all of them are uniformly close to a common limiting operator.","The approximation lemma suggests quantitative stability of the free-boundary set: small changes in the phase operators should move the interface $\\partial\\{u>0\\}\\cap\\partial\\{u<0\\}$ by an amount controlled by the closeness parameters, because the comparison solution $h$ is pinned down by the limiting operator.","A testable extension in the subcritical range $d/2<p<d$ is that the same approximation method should yield $W^{1,q}$ boundary estimates for every $q<dp/(d-p)$; the paper's Remark 4 sketches the interior mechanism but does not prove the boundary version."],"forward_implications":["If Theorem 1 is correct, every $L^p$-viscosity solution with $p>d$ is differentiable at every boundary point, and the distance from the solution to its tangent plane is bounded by $C|x-x_0|^2\\ln(1/|x-x_0|)$.","If Theorem 2 is correct, adding $C^1$ differentiability of the limiting profile upgrades the boundary regularity to $W^{2,d}$ without any convexity assumption on the phase operators.","The results apply to strong solutions of the original transmission problem (1), since an $L^d$-strong solution is automatically an $L^d$-viscosity solution of the two-inequality system (2)-(3).","The optimality example in Remark 1 shows that the Hessian cannot be expected to be Hölder continuous even with smooth boundary data and convex, small-perturbation operators; the $C^{1,{\\rm Log-Lip}}$ and $W^{2,d}$ scales are therefore natural stopping points."],"supporting_citations":[{"why":"Supplies the boundary regularity theorem for the limiting homogeneous problem $F(D^2h,Dh)=0$, used in Propositions 2 and 3.","marker":"[27]"},{"why":"Provides the viscosity-solution toolbox: comparison principles, the class $\\mathcal S^*$, and the Calderón–Zygmund decomposition arguments.","marker":"[3]"},{"why":"Provides the boundary $W^{2,p}$ and second-derivative $L^\\mu$ estimates whose boundary version is adapted in Proposition 5 and the proof of Theorem 2.","marker":"[31]"},{"why":"Initiates the modern transmission regularity program for $C^{1,\\alpha}$ interfaces that this paper extends to the free-boundary setting.","marker":"[4]"},{"why":"Establishes regularity up to the interface for fully nonlinear transmission problems with a fixed interface, the baseline being generalized here.","marker":"[28]"},{"why":"Contains the original second-derivative $L^p$ estimate for nondivergence equations that underlies the measure-decay argument.","marker":"[18]"},{"why":"Supplies the small-perturbation principle motivating the transfer of regularity from the limiting operator $F$.","marker":"[25]"},{"why":"Provides the interior $W^{1,p}$ estimates invoked in Remark 4 for the subcritical integrability regime.","marker":"[30]"}],"fun_headline_variants":["Free transmission solutions achieve C^{1,Log-Lip} at the boundary","C^{1,Log-Lip} boundary regularity for free transmission problems","Free transmission: C^{1,Log-Lip} boundary estimates","Log-Lipschitz boundary regularity for free transmission","Free transmission: C^{1,Log-Lip} and W^{2,d} boundary results"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole argument depends on the two phase operators being uniformly close to a single limiting elliptic operator $F$, with the smallness constants $\\tau$ and $\\kappa$ fixed only after the approximation parameter is chosen; if that closeness fails, the comparison with $F=0$ carries no information and both boundary theorems collapse, and the proof of Theorem 2 also secretly uses integrability of $f$ strictly above $d$ although the theorem is stated for $f\\in L^d$.","fun_headline_variants_meta":{"raw":{"variants":["Free transmission solutions achieve C^{1,Log-Lip} at the boundary","C^{1,Log-Lip} boundary regularity for free transmission problems","Free transmission: C^{1,Log-Lip} boundary estimates","Log-Lipschitz boundary regularity for free transmission","Free transmission: C^{1,Log-Lip} and W^{2,d} boundary results"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.002376,"raw_usage":{"total_tokens":9093,"prompt_tokens":840,"completion_tokens":8253,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":456,"completion_tokens_details":{"reasoning_tokens":8156}},"tokens_in":456,"tokens_out":8253,"duration_ms":52339,"temperature":1.0,"reasoning_tokens":8156,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T14:26:52.094907+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a domain satisfying A1, choose two operators satisfying A2 and A4 with arbitrarily small $\\tau$, put $f\\in L^p$, $p>d$, and prescribe boundary data that change sign on the flat boundary so that both phases meet at a boundary point $x_0$; if a normalized viscosity solution of (2)-(3) at such a point has boundary Taylor error whose ratio to $r^2\\ln(1/r)$ is unbounded as $r\\to 0$, Theorem 1 is false. A concrete place to look is a small derivative perturbation of the two-phase model in Remark 1, checking the observed decay against the claimed log-Lipschitz modulus.","supporting_citations":[{"cited_title":"Boundary regularity for viscosity solu- tions of fully nonlinear elliptic equations","cited_arxiv_id":null,"evidence_quote":"Supplies the boundary regularity theorem for the limiting homogeneous problem $F(D^2h,Dh)=0$, used in Propositions 2 and 3."},{"cited_title":"Caffarelli and Xavier Cabré","cited_arxiv_id":null,"evidence_quote":"Provides the viscosity-solution toolbox: comparison principles, the class $\\mathcal S^*$, and the Calderón–Zygmund decomposition arguments."},{"cited_title":"W 2,p and W 1,p -estimates at the boundary for solutions of fully nonlinear, uniformly elliptic equations","cited_arxiv_id":null,"evidence_quote":"Provides the boundary $W^{2,p}$ and second-derivative $L^\\mu$ estimates whose boundary version is adapted in Proposition 5 and the proof of Theorem 2."},{"cited_title":"Caffarelli, María Soria-Carro, and Pablo Raúl Sti nga","cited_arxiv_id":null,"evidence_quote":"Initiates the modern transmission regularity program for $C^{1,\\alpha}$ interfaces that this paper extends to the free-boundary setting."},{"cited_title":"Regularity of viscosity solutions to fully nonlinear elliptic transmission problems","cited_arxiv_id":null,"evidence_quote":"Establishes regularity up to the interface for fully nonlinear transmission problems with a fixed interface, the baseline being generalized here."},{"cited_title":"Second derivative Lp-estimates for elliptic equations of non- divergent type","cited_arxiv_id":null,"evidence_quote":"Contains the original second-derivative $L^p$ estimate for nondivergence equations that underlies the measure-decay argument."},{"cited_title":"Small perturbation solutions for ellipt ic equations","cited_arxiv_id":null,"evidence_quote":"Supplies the small-perturbation principle motivating the transfer of regularity from the limiting operator $F$."},{"cited_title":"W 1,p -interior estimates for solutions of fully nonlinear, uniformly elliptic equations","cited_arxiv_id":null,"evidence_quote":"Provides the interior $W^{1,p}$ estimates invoked in Remark 4 for the subcritical integrability regime."}],"review_version":1}