{"id":"95cf6cfc-c44b-4f3d-b634-83f3f2b847c4","arxiv_id":"1908.01639","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":2.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"This is a review of Yau's conjecture covering the real-analytic solution, the new proof of the lower bound in the smooth case, and the polynomial upper bound for nodal sets.","lead":"Logunov and Malinnikova survey the proof and history of Yau's conjecture on the size of nodal sets of Laplace eigenfunctions. The review explains how the lower bound is now proved in the smooth case and how a polynomial upper bound replaces an exponential one.","discovery_kind":"review","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Section 11 states an optimal Dirichlet upper bound H^{n-1}(Z_{\\phi_\\lambda}) \\le C_\\Omega \\sqrt\\lambda without proof or reference; as a review of the state of the art, this is the weakest evidential link.","rationale":"The main survey of Yau's conjecture in the smooth case rests on the published results [55] and [56]. Those papers provide independent support for the key lemmas sketched in Sections 9 and 10, so I do not see a sound basis for doubting the central mathematical claims on that ground. The reader's weakest_assumption identified the asserted lemmas in the proof sketches, but their rationale also flagged Section 11; my concern is concentrated on Section 11, where an optimal Dirichlet bound is stated without any proof or citation. This is a concrete missing-support issue: a review article should either give a reference for a theorem or clearly mark it as an announcement. The anecdotal attribution to an IAS workshop is not a substitute for a citable source. The verdict should remain CONDITIONAL because the main body of the review is reliable and the Section 11 issue is localized; it does not invalidate the review's central exposition, but it does require correction or a citation before the review can be considered fully verified. If a published proof of the Dirichlet bound is found, the condition would be satisfied and the review could be accepted as is.","tokens_in":25459,"tokens_out":7147,"duration_ms":73946,"concrete_test":"Search the peer-reviewed literature, including later papers by Logunov, Malinnikova, Nadirashvili, and Nazarov, for the theorem that Dirichlet Laplace eigenfunctions in a smooth bounded domain satisfy H^{n-1}(Z_{\\phi_\\lambda}) \\le C_\\Omega \\sqrt\\lambda, and check whether the exact statement and proof appear in a citable publication. If the theorem exists, Section 11 should cite it and the concern is resolved; if no such proof is published, Section 11 should be revised to label the result as announced or unpublished, removing the unqualified theorem statement.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The review's main theorems about closed smooth manifolds are the polynomial upper bound H^{n-1}(Z_{\\phi_\\lambda}) \\le C\\lambda^{Cn} from [55] and the lower bound H^{n-1}(Z_{\\phi_\\lambda}) \\ge c\\sqrt\\lambda from [56]. These are published in the Annals and are surveyed via proof sketches; the review is not required to reproduce those proofs, and the reader's concern about asserted lemmas in Sections 9 and 10 is mitigated by the existence of the published papers. The genuinely load-bearing weakness is in Section 11, which states, as a theorem, that for any smooth bounded domain \\Omega the Dirichlet eigenfunctions satisfy H^{n-1}(Z_{\\phi_\\lambda}) \\le C_\\Omega \\sqrt\\lambda, with a matching lower bound, and attributes the improvement to an oral anecdote: 'During the talk Fedor Nazarov removed a half of the proof... and improved the bound to the optimal one.' No citation, proof, or reference to a published source is given. This is a strong, optimal claim that is central to the Section 11 narrative and is presented without any way for the reader to verify it. If the result is not in fact proved in the cited literature, or if the precise stated form differs from the published result, the review overstates the current knowledge. The main closed-manifold claims are not affected, but the review's factual reliability is undermined by this uncited assertion.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This survey reviews Yau's conjecture on the Hausdorff measure of zero sets of Laplace eigenfunctions. It covers the classical real-analytic results of Donnelly and Fefferman, the recent polynomial upper bound and the lower bound for smooth closed manifolds obtained by the first author, and two-dimensional methods. The paper includes proof sketches of the new results and discusses related open problems, including a section on Dirichlet eigenfunctions in bounded domains.","tokens_in":25704,"tokens_out":17080,"duration_ms":153279,"significance":"The main theorems surveyed, the lower bound H^{n-1}(Z_{\\phi_\\lambda}) \\ge c\\sqrt{\\lambda} and the polynomial upper bound H^{n-1}(Z_{\\phi_\\lambda}) \\le C\\lambda^{Cn} for smooth closed manifolds, are published in Annals of Mathematics [55],[56], and the survey accurately conveys their statements and the structure of the proofs. The proof sketches in Sections 9 and 10 rely on asserted lemmas, but the authors label these as sketches and the underlying results are independently checkable in the cited publications, so this is not a serious weakness. The principal caveat is Section 11, which states an optimal upper bound for Dirichlet domains without proof or citation; this is the main reason the manuscript needs revision.","major_comments":[{"comment":"The Theorem in Section 11 and the paragraph that follows state that for any bounded smooth domain \\Omega, Dirichlet eigenfunctions satisfy H^{n-1}(Z_{\\phi_\\lambda}) \\le C_\\Omega \\sqrt{\\lambda}, with the improvement over the displayed log bound attributed to an oral remark by Nazarov. No proof or reference is given for this optimal bound, for the preceding log bound, or for the accompanying lower bound. Since this section is part of a survey of the state of the art, an unverifiable claim of an optimal result is load-bearing for the paper's factual reliability. Please supply a citation to a published or preprint source, or explicitly label these statements as unpublished announcements and state the precise hypotheses.","section":"11"},{"comment":"The lower-bound assertion 'H^{n-1}(Z_{\\phi_\\lambda}) \\ge c_\\Omega \\sqrt{\\lambda} (if we include the boundary of \\Omega)' is ambiguous because H^{n-1}(\\partial\\Omega) is a fixed constant; as stated, the lower bound cannot be read as a single inequality valid for all \\lambda unless the intended meaning is that the interior nodal set plus the boundary has measure growing like \\sqrt{\\lambda}. Please clarify the statement and provide references for the two proofs attributed to Donnelly and Fefferman and to Nadirashvili's conjecture.","section":"11"}],"minor_comments":[{"comment":"The text states that the bound H^{n-2}(S(u) \\cap B) \\le C N^2 was obtained by Naber and Valtorta [68] and then immediately states that it is not known whether this estimate holds even for harmonic functions in R^3. These statements cannot both be correct; please reconcile the wording, indicating which bound is known and which is conjectural.","section":"8.3"},{"comment":"The composition 'h = g \\circ u' is not well-defined as written because u is real-valued and g is a self-map of the disk; presumably the intended statement is that a solution of the transformed divergence-form equation is a harmonic function composed with a quasiconformal map (for instance u = h \\circ g or u \\circ g is harmonic). Please correct the direction of composition.","section":"3.7"},{"comment":"In the statement of the key lemma, the factor 'c[\\sqrt{N}]n-1' should be typeset as c(\\sqrt{N})^{n-1}. Since the section is restricted to harmonic functions in R^3, the dimension n should either be fixed to 3 or the statement should explicitly allow general dimension.","section":"10.2"},{"comment":"The manuscript contains numerous typographical errors, including 'Remannian' and 'Reimannian' for 'Riemannian', 'functons', 'extesnion', 'propogation', and inconsistent spacing around citations. A careful proofreading pass is needed.","section":"Global"}],"recommendation":"major_revision","confidential_remarks":"The Section 11 claim appears to announce an unpublished optimal bound in a review article without a reference; the editor may wish to ask the authors to clarify whether this result is published elsewhere, and if not, to mark it explicitly as an announcement. The rest of the survey is reliable, though the presentation would benefit from a proofreading pass."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The short version: this is a useful review of the Yau conjecture by the two people who proved the main smooth-case theorems. It contains no new results, but it does something valuable: it lays out the history from Brüning to Donnelly–Fefferman to the recent polynomial upper bound and matching lower bound, with enough detail that a nonspecialist can see the shape of the arguments. The proof sketches in Sections 9 and 10 are honest about what they omit, and the key lemmas they assert (distribution of doubling index, simplex and hyperplane lemmas, stable growth) are backed by the authors' Annals papers [55] and [56]. For a survey, that is enough.\n\nThe one thing I would press on is Section 11. It states, as a theorem, that for a bounded domain with smooth boundary the Dirichlet eigenfunction nodal set satisfies H^{n-1}(Z_{φλ}) ≤ C_Ω √λ, with the matching lower bound, and attributes the sharp upper bound to an oral comment by Nazarov during a 2017 IAS talk. No citation, no proof, no reference to a written source. That is an optimal, load-bearing claim in that section. If the result is in the literature, the review needs to say where; if it is not, the review is presenting folklore as established fact. The main closed-manifold survey is unaffected, but this undermines the review's reliability as a reference.\n\nMinor point: the proof sketches rely on a chain of lemmas that are asserted rather than demonstrated. That is acceptable in a review because the full proofs are cited, but a reader who wants to verify the survey's account has to go to the original papers. There are also a handful of typos (\"Remannian\", \"wtih\") that a copyeditor would catch.\n\nI would send this to peer review. The survey is written by the right people, covers the area responsibly, and will be a standard reference. The referee should ask for the Section 11 statement to be either cited or toned down.","headline":"Useful survey of Yau's conjecture by the authors who proved the smooth case; main body is reliable, but Section 11's uncited optimal Dirichlet bound needs a citation or a caveat.","tokens_in":26265,"tokens_out":2941,"would_cite":true,"duration_ms":27518,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35B60","35J05","58J50"],"pacs":[],"model":"deepseek-v4-flash","headline":"The review reports that in the smooth case the zero set of a Laplace eigenfunction has size at least $c\\sqrt{\\lambda}$ and at most $C\\lambda^{Cn}$, with the sharp upper bound still open in dimension 2.","keywords":["Laplace eigenfunctions","nodal sets","Yau's conjecture","Hausdorff measure","doubling index","Nadirashvili's conjecture","unique continuation","Riemannian manifolds"],"falsifier":"One concrete check is to take the simplified key lemma from Section 10.2 (Proposition 6.1 of [56]) at face value: a harmonic function with stable growth of order $N$ in a ball $B_r(x)$ is claimed to contain at least $c[\\sqrt{N}]^{2c\\log N/\\log\\log N}$ disjoint balls of radius $r/\\sqrt{N}$ with zeros at their centres. A numerical search among high-degree harmonic polynomials (for instance $\\Re(x_1+ix_2)^N$ or zonal harmonics) at large $N$ could confirm or contradict that multiplicity. If a single harmonic function in the unit ball with $u(0)=0$ had zero-set area tending to zero along a sequence, the harmonic-function conjecture and the lower bound in Yau's conjecture would fail.","tokens_in":25226,"feed_emoji":"📐","tokens_out":19035,"duration_ms":163223,"temperature":0.7,"pith_summary":"This review lays out the proof of Yau's conjecture, which predicts that the zero set of a Laplace eigenfunction $\\phi_\\lambda$ on a smooth closed $n$-manifold has $(n-1)$-dimensional Hausdorff measure comparable to $\\sqrt{\\lambda}$. The central news is that in the smooth case the lower bound is now a theorem, $H^{n-1}(\\{\\phi_\\lambda=0\\})\\ge c\\sqrt{\\lambda}$, obtained by solving Nadirashvili's conjecture about harmonic functions in $\\mathbb{R}^3$. The upper bound is proved in the weaker polynomial form $H^{n-1}(\\{\\phi_\\lambda=0\\})\\le C\\lambda^{Cn}$, while the original sharp upper bound $C\\sqrt{\\lambda}$ remains open even for surfaces. For real-analytic metrics the conjecture is fully proved, and the review explains the machinery (doubling index, frequency function, harmonic extension, multiscale iteration) that carries the arguments. The result ties nodal geometry quantitatively to eigenfunction growth: the size of the nodal set is controlled by how fast the eigenfunction doubles across concentric balls.","feed_headline":"Yau's zero-set conjecture: lower bound solved, upper bound polynomial","feed_subtitle":"The review lays out the doubling-index and harmonic-extension machinery behind the smooth-case proof.","key_machinery":"The central object is the doubling index $N_f(B)=\\log_2(\\sup_{2B}|f|/\\sup_B|f|)$, a single number that measures how fast a solution grows from a ball to a concentric ball of twice the radius; locally it controls the vanishing order, and it is nearly interchangeable with Almgren's frequency function, whose monotonicity in the radius is the main analytic engine. The harmonic extension $u(x,t)=\\phi_\\lambda(x)e^{\\sqrt{\\lambda}t}$ converts an eigenfunction on $M$ into a harmonic function on $M\\times\\mathbb{R}$ whose zero set is the cylinder over the nodal set, so every estimate can be made for solutions of a fixed divergence-form elliptic equation with smooth or Lipschitz coefficients. The proofs then run on two multiscale devices: the lemma on the distribution of doubling index, which says that a cube of high index can be subdivided so that almost all small subcubes have index at most half the original; and the key lemma on stable growth, which says that a harmonic function with high stable growth contains many disjoint balls of radius $r/\\sqrt{N}$ where it vanishes at the centre. These feed a recursion $F(N)\\le 2KF(N/2)$ for the upper bound and a contradiction $F(N)>2F(N)$ for the lower bound, the latter needing the simplex lemma, the hyperplane lemma, and quantitative Cauchy uniqueness to control how high-index points can be arranged.","core_discovery":"The review's central claim is that Yau's conjecture is now largely resolved in the smooth case. Papers [55] and [56] are credited with the polynomial upper bound $H^{n-1}(\\{\\phi_\\lambda=0\\})\\le C\\lambda^{Cn}$ for Laplace eigenfunctions on closed $C^\\infty$ Riemannian $n$-manifolds, and with the sharp lower bound $H^{n-1}(\\{\\phi_\\lambda=0\\})\\ge c\\sqrt{\\lambda}$, the latter obtained by proving a conjecture about harmonic functions: any solution of a uniformly elliptic equation with Lipschitz coefficients in the unit ball, vanishing at the centre, has zero-set $(n-1)$-measure at least a constant times $r^{n-1}$ on every ball $B_r$. The review presents the lower-bound proof as a multiscale contradiction argument built on a key lemma about stable growth, and the upper-bound proof as a recursion from a lemma on the distribution of doubling indices. It credits [29] with the full sharp bound in the real-analytic case and reports that in dimension two the best known upper bound is $C\\lambda^{3/4-\\varepsilon}$, leaving the original upper bound open. It also asserts, as a theorem stated without proof or citation, that for a bounded domain with smooth boundary the Dirichlet eigenfunction's nodal set satisfies $H^{n-1}(Z_{\\phi_\\lambda})\\le C_\\Omega\\sqrt{\\lambda}$ once the boundary is included, with a matching lower bound.","pith_inferences":["Extension: if the optimal Dirichlet-domain bound asserted in Section 11 were supplied with a complete proof, it would immediately improve the state of the art for manifolds with boundary and would test whether the sharp upper bound transfers to all smooth settings.","Extension: the multiscale recursion in the upper-bound proof suggests a general template for bounding other geometric measures attached to solutions of elliptic equations, such as critical sets; the known singular-set bound $H^{n-2}(S(u)\\cap B)\\le CN^2$ sits in that same family.","Extension: the lower-bound proof's dependence on stable growth suggests that the constant $c$ in Yau's lower bound should be effectively computable from the metric and dimension once the distribution of doubling indices is understood.","Extension: one could test the key lemma numerically on high-degree harmonic polynomials, checking whether the predicted number of disjoint zero-centred balls at scale $r/\\sqrt{N}$ actually appears; this would give independent evidence for the lower-bound theorem."],"forward_implications":["On every smooth closed manifold, the $(n-1)$-dimensional measure of a nodal hypersurface is at least $c\\sqrt{\\lambda}$, so nodal sets cannot become sparse as the eigenvalue grows.","The same measure is at most $C\\lambda^{Cn}$, so nodal sets cannot explode faster than a fixed power of the eigenvalue; no exponentially large nodal sets occur.","For real-analytic metrics the full Yau bound $c\\sqrt{\\lambda}\\le H^{n-1}(\\{\\phi_\\lambda=0\\})\\le C\\sqrt{\\lambda}$ holds, covering the spherical harmonics and other analytic examples.","For Dirichlet eigenfunctions in smooth bounded domains, with the boundary included in the nodal set, the review asserts the optimal bound $H^{n-1}(Z_{\\phi_\\lambda})\\le C_\\Omega\\sqrt{\\lambda}$ along with a matching lower bound.","In dimension two the sharp upper bound remains open; the current record for general surfaces is $C\\lambda^{3/4-\\varepsilon}$, leaving a gap to the conjectured $C\\sqrt{\\lambda}$."],"supporting_citations":[{"why":"It supplies the polynomial upper bound $H^{n-1}(\\{\\phi_\\lambda=0\\})\\le C\\lambda^{Cn}$ that is the subject of Section 9.","marker":"[55]"},{"why":"It supplies the proof of the relevant harmonic-function conjecture and the sharp lower bound $H^{n-1}(\\{\\phi_\\lambda=0\\})\\ge c\\sqrt{\\lambda}$ reviewed in Section 10.","marker":"[56]"},{"why":"It proves the full Yau bound in the real-analytic case and provides the holomorphic-extension and doubling-index framework.","marker":"[29]"},{"why":"It formulates the harmonic-function conjectures that motivate the smooth-case strategy, one of which is solved in [56].","marker":"[74]"},{"why":"It provides the earlier exponential bound $H^{n-1}(Z\\cap B)\\le CN^{CN}$ that the polynomial upper bound improves.","marker":"[42]"},{"why":"It establishes the two-dimensional upper bound $C\\lambda^{3/4}$ and develops the Carleman-inequality method.","marker":"[31]"},{"why":"It proves monotonicity of the frequency function for elliptic equations in divergence form, used throughout the review.","marker":"[36]"}],"fun_headline_variants":["Yau's nodal set conjecture: sharp lower bound, polynomial upper","Review: Yau's zero-set conjecture largely resolved in smooth case","Lower bound solved, upper polynomial: Yau conjecture review","Yau's conjecture: sharp lower bound, polynomial upper in smooth case","Zero-set bounds: sharp lower, polynomial upper; 2D still open"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The review's account of the smooth-case theorems rests on a chain of technical lemmas about the distribution of doubling indices and about stable growth, whose complete proofs are not included here; the optimal Dirichlet-domain upper bound in Section 11 is additionally presented without proof or citation.","fun_headline_variants_meta":{"raw":{"variants":["Yau's nodal set conjecture: sharp lower bound, polynomial upper","Review: Yau's zero-set conjecture largely resolved in smooth case","Lower bound solved, upper polynomial: Yau conjecture review","Yau's conjecture: sharp lower bound, polynomial upper in smooth case","Zero-set bounds: sharp lower, polynomial upper; 2D still open"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000727,"raw_usage":{"total_tokens":3259,"prompt_tokens":946,"completion_tokens":2313,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":562,"completion_tokens_details":{"reasoning_tokens":2221}},"tokens_in":562,"tokens_out":2313,"duration_ms":17337,"temperature":1.0,"reasoning_tokens":2221,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T15:06:54.551262+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"One concrete check is to take the simplified key lemma from Section 10.2 (Proposition 6.1 of [56]) at face value: a harmonic function with stable growth of order $N$ in a ball $B_r(x)$ is claimed to contain at least $c[\\sqrt{N}]^{2c\\log N/\\log\\log N}$ disjoint balls of radius $r/\\sqrt{N}$ with zeros at their centres. A numerical search among high-degree harmonic polynomials (for instance $\\Re(x_1+ix_2)^N$ or zonal harmonics) at large $N$ could confirm or contradict that multiplicity. If a single harmonic function in the unit ball with $u(0)=0$ had zero-set area tending to zero along a sequence, the harmonic-function conjecture and the lower bound in Yau's conjecture would fail.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It supplies the polynomial upper bound $H^{n-1}(\\{\\phi_\\lambda=0\\})\\le C\\lambda^{Cn}$ that is the subject of Section 9."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It supplies the proof of the relevant harmonic-function conjecture and the sharp lower bound $H^{n-1}(\\{\\phi_\\lambda=0\\})\\ge c\\sqrt{\\lambda}$ reviewed in Section 10."},{"cited_title":"Donnelly and C","cited_arxiv_id":null,"evidence_quote":"It proves the full Yau bound in the real-analytic case and provides the holomorphic-extension and doubling-index framework."},{"cited_title":"Nadirashvili","cited_arxiv_id":null,"evidence_quote":"It formulates the harmonic-function conjectures that motivate the smooth-case strategy, one of which is solved in [56]."},{"cited_title":"Hardt and L","cited_arxiv_id":null,"evidence_quote":"It provides the earlier exponential bound $H^{n-1}(Z\\cap B)\\le CN^{CN}$ that the polynomial upper bound improves."},{"cited_title":"Donnelly and Ch","cited_arxiv_id":null,"evidence_quote":"It establishes the two-dimensional upper bound $C\\lambda^{3/4}$ and develops the Carleman-inequality method."},{"cited_title":"Garofalo and F","cited_arxiv_id":null,"evidence_quote":"It proves monotonicity of the frequency function for elliptic equations in divergence form, used throughout the review."}],"review_version":1}