{"id":"af34a3b1-da8b-4f1b-9439-38c4f85db4fa","arxiv_id":"1908.06937","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"For K-ary trees with a visual metric and weighted Newtonian spaces, the trace space is exactly a dyadic Besov-type space B^{θ,λ}_p, with borderline cases mapping to L^p or a distinct Besov-type space B^{0,λ}_α.","lead":"This paper characterizes the boundary traces of weighted Sobolev spaces on regular trees in terms of new dyadic Besov-type spaces, including borderline cases. It gives analysts a precise way to decide which boundary functions extend to finite-energy functions on the tree.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified","rationale":"The reader's weakest-assumption identification points to Ahlfors regularity, which is indeed the load-bearing input for the dyadic estimates. However, this property is a standard theorem for regular trees with the visual metric and is cited from [3], so it is not a live objection. The two clarifications the reader requested (Proposition 2.13 and the domain of T∘E) are legitimate but do not touch the central mathematical validity of Theorem 1.1: Proposition 2.13 is unused in the main proofs, and the embedding N^{1,p}(X) ↪ N^{1,p}(X, μλ) for λ≥0 makes the composition T∘E well-defined. I therefore find no reason to change the reader's conditional verdict; the conditions are minor and reasonable, but the central claim appears correct as stated.","tokens_in":27708,"tokens_out":22750,"duration_ms":227264,"concrete_test":"Independently verify the pivotal measure normalization: for a K-ary tree with the visual metric (2.1), compute ν(I_x) for the level-n cylinders I_x and compare with e^{-εn logK/ε}=K^{-n}; also compute ν(B(ξ,r)) for r=e^{-εn}. If the Ahlfors Q-regularity of Prop. 2.10 (Q=logK/ε) is reproduced, the concern that the dyadic normalization in (3.10) is unsupported does not land. As a secondary check, supply the omitted proof of Prop. 2.13 or replace it with a precise citation containing the proof, which would remove the one explicitly stated gap in the exposition.","verdict_should_be":"UNCHANGED","load_bearing_attack":"After a careful pass through the proof of Theorem 1.1, I do not find a load-bearing flaw. The trace and extension chains are internally consistent: (3.3)-(3.4) express the dX-length element in terms of dμλ, the cancellation in the dyadic-energy estimate rests on ν(E(x))≈r_j^Q, and the equivalence (3.10) follows from ν(I)≈e^{-εnQ} together with θ=1-(β-logK)/(εp). The most sensitive input is the Ahlfors Q-regularity of ν (Prop. 2.10, quoted from [3]); if ν(I) were not comparable to e^{-n logK}, the dyadic normalization in (3.10) would break. This is the same point the Reader flagged, but it is a standard property of the visual metric on regular trees, and I see no internal inconsistency or unsupported leap in its use here. Proposition 2.13 is indeed stated without proof, but it only connects the dyadic Besov spaces to the classical Besov spaces and is not used in the proofs of Theorems 1.1-1.4. The composition T∘E in Theorem 1.2/1.4 is legitimate because N^{1,p}(X) embeds boundedly into N^{1,p}(X, μλ) for λ≥0. Thus the honest summary is that no significant objection has been identified.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces Besov-type spaces B^{θ,λ}_p(∂X) on the boundary of a K-ary regular tree, defined through dyadic energies with weights e^{εnθp}n^λν(I), and proves that for β>logK, ε>0, λ∈R, p≥1 with p>(β−logK)/ε and θ=1−(β−logK)/(εp), B^{θ,λ}_p(∂X) is the trace space of the weighted Newtonian space N^{1,p}(X,μλ), where dμλ=e^{-β|x|}(|x|+C)^λd|x|. The trace and extension operators are explicit and bounded linear (Theorem 1.1). The paper also treats the critical case p=(β−logK)/ε (Theorems 1.2–1.4), including a sharp condition λ>p−1 for traces in L^p(∂X), a nonlinear extension operator from L^p(∂X) to N^{1,p}(X), the identification of the trace of N^{1,1}(X,μλ) with the space B^{0,λ}_α, and the optimality of B^{0,λ}_p(∂X) for a Whitney-type extension operator. The proofs use the Ahlfors Q-regularity of the boundary, dyadic energy estimates, Fubini-type counting on the tree, and a Gagliardo-type gluing construction.","tokens_in":27895,"tokens_out":18921,"duration_ms":177557,"significance":"If Theorem 1.1 holds, it gives an exact dyadic characterization of boundary traces for a family of weighted Sobolev spaces on regular trees, extending the unweighted trace theorem of Björn–Björn–Gill–Shanmugalingam and complementing hyperbolic-filling trace results. The paper's main estimates are explicitly verified: the measure comparisons (3.3), the dyadic equivalence (3.10), the counting identity (3.16), and the norm equivalence (3.25) all check out, and the trace and extension operators are constructive. The main external input is the Ahlfors Q-regularity of the visual boundary measure (Proposition 2.10), which is a standard property of regular trees rather than a fragile assumption. I find no load-bearing error in the proof of Theorem 1.1; the issues below concern statement precision and presentation and are local to the statements of Theorems 1.2–1.3.","major_comments":[],"minor_comments":[{"comment":"As stated, the identity T∘E=Id is not well-defined because T is introduced as a bounded linear operator from N^{1,p}(X,μλ) to L^p(∂X), while E is constructed into the unweighted space N^{1,p}(X). For λ>0 the weighted space is strictly smaller than N^{1,p}(X), and for λ=0, p>1 the unweighted trace is not bounded on all of N^{1,p}(X); indeed Example 3.2 specialized to λ=0 gives an N^{1,p}(X) function whose geodesic limits are infinite. The proof of Proposition 3.5 actually establishes that the limit of E f along rays equals f for ν-a.e. ξ. Please restate the theorem accordingly, e.g., by declaring the trace operator on the union of the relevant spaces or by phrasing the right-inverse property as a pointwise trace identity rather than as T∘E on the full unweighted space. Also, the notation 'Lp(X)' at the end of the statement should be 'Lp(∂X)'.","section":"Section 1, Theorem 1.2 (second paragraph)"},{"comment":"This proposition is stated without proof. It is used in the introduction to justify the claim that Theorem 1.1 recovers the classical Besov trace results from [3] when λ=0, while not being used in the proofs of Theorems 1.1–1.4. Please either supply the proof, give a precise reference for the equivalence, or explicitly downgrade the recovery claim to a remark so that the omission does not leave a labeled proposition unsupported.","section":"Section 2, Proposition 2.13"},{"comment":"In the displayed chain estimating the B^{0,λ}_α energy, the factors α(n) and α(n+1) are missing the exponent λ in two of the sums; the subsequent line uses α(n+1)^λ and shows the intended expression. This should be corrected to α(n)^λ and α(n+1)^λ for consistency.","section":"Section 3.3, proof of Theorem 1.3"},{"comment":"The phrase 'for any λ > p−1 if p = 1 or for λ ≥ 0 if p = 1' is garbled; it should read 'for λ > p−1 if p > 1, or λ ≥ 0 if p = 1'.","section":"Section 3.2, proof of Proposition 3.1"},{"comment":"There are numerous small typos and OCR artifacts (e.g., 'A /greaterorsimilarB' in Section 1, 'classfication' in the MSC line, 'Haj/suppress lasz' in the references). A careful proofreading pass is needed, but these do not affect the mathematics.","section":"Throughout"}],"recommendation":"minor_revision","confidential_remarks":"The manuscript is within scope and in good mathematical shape. The only substantive issue I see is the ill-posed composition in Theorem 1.2; it is local and fixable by a restatement. I would also ask the authors to address the unsupported Proposition 2.13. No concerns about citation practice or novelty disclosure."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"I read the Koskela–Wang paper on dyadic Besov-type trace spaces on regular trees. The headline: the main trace characterizations look right and are genuinely new, but there is a hole in the second part of Theorem 1.2 that the stress-test note hand-waved past.\n\nWhat is new: the spaces B^{θ,λ}_p and B^{0,λ}_α, the weighted trace characterizations for Newtonian spaces (Theorems 1.1, 1.3, 1.4), the sharp λ > p−1 condition in Theorem 1.2, and the strict inclusion B_1^{0,λ} ⊊ B_α^{0,λ} with a clean probabilistic example. The dyadic norm approach is elegant, and the proofs of the main trace and extension estimates, especially the dyadic energy comparisons (3.10) and the borderline p = (β−log K)/ε case, are detailed and check out. The paper does a good job of honestly stating what it does not prove, e.g., Remark 3.6 on surjectivity for λ > p−1.\n\nSoft spots. First, the composition T∘E in Theorem 1.2 is not legitimate as written. The trace operator T is defined and bounded on the weighted space N^{1,p}(X, μ_λ). The extension E is constructed in Proposition 3.5 only into the unweighted space N^{1,p}(X). For λ > 0, the weighted space is a proper subset of the unweighted one, not a superset. The stress-test note claims an embedding in the wrong direction. The proof of Proposition 3.5 uses only unweighted estimates (Lemmas 3.3 and 3.4), so it does not show that E(f) lies in the domain of T. For p = 1, λ > 0, Theorem 1.3 tells us T is not surjective onto L^1, so a right inverse from L^1 into the weighted space cannot exist. The statement needs repair: either prove E maps into the weighted space under the stated λ conditions, or restrict the right-inverse claim to B_p^{0,λ}, as Theorem 1.4 already does. This is not fatal to the main theorems, but it is a real gap, not just a clarification. Second, Proposition 2.13 is stated without proof; it only connects the dyadic Besov spaces to the classical ones and is not used elsewhere, so this is minor.\n\nThe Ahlfors Q-regularity of the boundary measure is a standard input and I do not see a problem there.\n\nWho this is for: researchers working on trace theorems on metric measure spaces, especially on trees and Cantor-type boundaries. The main results deserve a serious referee. The issue in Theorem 1.2 is localized and fixable, so I would send it to review with a request for major revision.","headline":"Main trace results are new and mostly solid, but Theorem 1.2's right-inverse claim has a domain mismatch that the stress-test note got backwards.","tokens_in":28541,"tokens_out":7489,"would_cite":true,"duration_ms":67556,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["46E35","30L05"],"pacs":[],"model":"deepseek-v4-flash","headline":"On any regular tree, the boundary traces of weighted Sobolev functions form exactly a Besov-type space, with bounded linear trace and extension operators.","keywords":["Besov-type space","regular tree","trace space","dyadic norm","Newtonian space","weighted Sobolev space","Ahlfors regular measure","Cantor-type boundary"],"falsifier":"Take a fixed parameter triple in the theorem's range and a boundary function constant on one level-$n$ dyadic ball $I$ and zero elsewhere. Use the paper's affine extension (3.6)–(3.8) and compute the ratio $$R_n=\\frac{\\int_X |g_{\\tilde u}|^p\\,d\\mu_\\$\\lambda$}{$e^{{\\varepsilon n\\theta p}}$n^\\$\\lambda$\\nu(I)|u_I-u_{\\hat I}|^p}.$$ Theorem 1.1 predicts $\\sup_n R_n<\\infty$ for all choices of $I$; if any allowed parameter choice gives $R_n\\to\\infty$ along a sequence of dyadic balls, the claimed bounded linear extension fails.","tokens_in":27430,"feed_emoji":"🌳","tokens_out":12043,"duration_ms":110166,"temperature":0.7,"pith_summary":"The paper proves that on any regular (K-ary) tree, the boundary traces of weighted first-order Sobolev functions are exactly the functions in a Besov-type space defined by a dyadic energy on the tree's Cantor-type boundary. The trace space is $B^{\\theta,\\lambda}_p(\\partial X)$ with $\\theta = 1 - (\\beta-\\log K)/(\\varepsilon p)$, where $K$ is the branching number, $\\beta$ controls the weight $e^{-\\beta|x|}$ on the tree, and $\\varepsilon$ determines the visual metric. The identification is quantitative: there is a bounded linear trace operator and a bounded linear extension operator that is its right inverse. The paper also settles the borderline case $p=(\\beta-\\log K)/\\varepsilon$, where the trace picture splits, including a sharp range for $\\lambda$ and a different Besov-type space when $p=1$. This provides an exact, computable description of boundary values for weighted Sobolev spaces on trees, analogous to classical Euclidean trace theorems.","feed_headline":"Sobolev traces on regular trees equal a dyadic Besov-type space","feed_subtitle":"Weighted Sobolev functions on K-ary trees restrict exactly to functions with finite dyadic boundary energy.","key_machinery":"The central object is the dyadic Besov-type energy on the boundary,\n$$\\|f\\|^p_{\\dot $B^{{\\theta,\\lambda}}$_p(\\partial X)}= \\sum_{n=1}^\\infty $e^{{\\varepsilon n \\theta p}}$ n^\\$\\lambda$ \\sum_{I\\in Q_n} \\nu(I) |f_I - f_{\\hat I}|^p,$$\nwhere $Q_n$ lists the boundary sets $I_x$ corresponding to vertices $x$ at distance $n$ from the root, $\\hat I$ is the parent of $I$, and $f_I$ is the $\\nu$-average of $f$ on $I$. The workhorse identity is the comparability $\\int_X |g_{\\tilde u}|^p\\,d\\mu_\\lambda \\approx \\|u\\|^p_{\\dot B^{\\theta,\\lambda}_p(\\partial X)}$ for the piecewise-affine extension $\\tilde u$; it follows from the measure estimates $\\mu_\\lambda([x,y])\\approx e^{-\\beta n}n^\\lambda$ and $\\nu(I)\\approx e^{-\\varepsilon n Q}=e^{-n\\log K}$. This comparability, together with the Ahlfors $Q$-regularity of $\\nu$ with $Q=\\log K/\\varepsilon$, is what makes the trace and extension estimates close.","core_discovery":"The central claim is Theorem 1.1. For a $K$-ary tree $X$ with $K\\ge 2$, fixed $\\beta>\\log K$, $\\varepsilon>0$, $\\lambda\\in\\mathbb{R}$, and $p\\ge 1$ with $p>(\\beta-\\log K)/\\varepsilon$, the Besov-type space $B^{\\theta,\\lambda}_p(\\partial X)$ is the trace space of the weighted Newtonian space $N^{1,p}(X,\\mu_\\lambda)$ with $\\theta = 1-(\\beta-\\log K)/(\\varepsilon p)$. The trace operator sends a Sobolev function to its limits along geodesic rays, and the extension operator interpolates the dyadic averages of a boundary function affinely along edges; both are bounded linear and compose to the identity. In the borderline case $p=(\\beta-\\log K)/\\varepsilon\\ge 1$, a bounded linear trace into $L^p(\\partial X)$ exists for $\\lambda>p-1$ (or $\\lambda\\ge 0$ when $p=1$), and this range is sharp, while a bounded nonlinear extension from $L^p(\\partial X)$ to $N^{1,p}(X)$ exists. For $p=1$ and $\\lambda>0$, the trace space of $N^{1,1}(X,\\mu_\\lambda)$ is the space $B^{0,\\lambda}_\\alpha(\\partial X)$, which strictly contains $B^{0,\\lambda}_1(\\partial X)$. The fixed Whitney-type extension operator is bounded and linear exactly on $B^{0,\\lambda}_p(\\partial X)$, and this space is optimal for that operator.","pith_inferences":["The same dyadic trace criterion could be tested on non-regular trees by replacing $K^n$ with the growth function $v(n)$ of the tree; the expected critical exponent would involve $Q=\\lim_{n\\to\\infty}\\frac{\\log v(n)}{\\varepsilon n}$.","The logarithmic counterexample at the critical integrability suggests that similar sharp thresholds for traces should appear in other hyperbolic or negatively-curved settings where radial functions diverge logarithmically.","The splitting between $B^{0,\\lambda}_1$ and $B^{0,\\lambda}_\\alpha$ at $p=1$ indicates that when the trace operator is not surjective, the choice of extension operator genuinely selects the trace space; users needing a linear right inverse should use $B^{0,\\lambda}_p$, while the full trace space is larger.","One could verify the dyadic energy characterization computationally on small finite truncated trees, checking whether the ratio between the Newtonian extension energy and the boundary dyadic energy stays bounded as the truncation level grows."],"forward_implications":["At $\\lambda=0$ and $0<\\theta<1$, the dyadic norm is equivalent to the double-integral Besov norm $B^\\theta_{p,p}(\\partial X)$, so Theorem 1.1 recovers the known trace description for $N^{1,p}(X)$ without weights.","The explicit linear extension operator gives a constructive way to extend boundary data with finite dyadic energy to the whole tree, with the Newtonian energy controlled by the boundary norm.","In the borderline case $p=(\\beta-\\log K)/\\varepsilon$, the trace operator is bounded into $L^p$ for $\\lambda>p-1$ (or $\\lambda\\ge 0$ if $p=1$), and the range is sharp: for $\\lambda=p-1-\\delta$, traces can be infinite almost everywhere.","For $p=1$ and $\\lambda>0$, the true trace space of $N^{1,1}(X,\\mu_\\lambda)$ is $B^{0,\\lambda}_\\alpha$, which is strictly larger than $B^{0,\\lambda}_1$; setting $\\theta=0$ in the main theorem does not give the correct borderline trace space.","For the fixed Whitney-type extension operator, $B^{0,\\lambda}_p$ is the optimal domain: any Banach space of boundary functions on which the operator is bounded and linear embeds into $B^{0,\\lambda}_p$."],"supporting_citations":[{"why":"Supplies the Ahlfors Q-regularity of the boundary measure, the measure estimates for balls, and the trace result for N^{1,p}(X) that Theorem 1.1 generalizes.","marker":"[3]"},{"why":"Gives the uniformizing metric dX on the tree used to define distance, boundary, and geodesic-ray limits.","marker":"[5]"},{"why":"Provides the dyadic-norm and Whitney-extension method for traces of weighted function spaces, including the equivalence used in Proposition 2.13.","marker":"[23]"},{"why":"Supplies the metric-space trace/extension framework with p-Poincaré inequality and co-dimensional Ahlfors boundary that parallels Theorem 1.2.","marker":"[21]"},{"why":"Provides the Whitney-type extension operator whose linearity and boundedness are analyzed in Theorem 1.4.","marker":"[22]"},{"why":"Gives the layer-cake Lipschitz-approximation construction adapted in Proposition 3.5 for the nonlinear L^p extension.","marker":"[10]"}],"fun_headline_variants":["Sobolev traces on trees are exactly dyadic Besov-type spaces","Dyadic Besov-type spaces are exactly Sobolev traces on trees","Sobolev traces on regular trees equal dyadic Besov-type spaces","Dyadic norm Besov-type spaces are traces on regular trees","Dyadic Besov-type spaces capture Sobolev traces on K-ary trees"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the boundary measure $\\nu$ is Ahlfors $Q$-regular with $Q=\\log K/\\varepsilon$, meaning every boundary ball $B(\\xi,r)$ has measure comparable to $r^Q$; every dyadic estimate in the proof, in particular the comparability between the Sobolev gradient energy and the dyadic Besov energy, uses $\\nu(I)\\approx e^{-\\varepsilon n Q}$.","fun_headline_variants_meta":{"raw":{"variants":["Sobolev traces on trees are exactly dyadic Besov-type spaces","Dyadic Besov-type spaces are exactly Sobolev traces on trees","Sobolev traces on regular trees equal dyadic Besov-type spaces","Dyadic norm Besov-type spaces are traces on regular trees","Dyadic Besov-type spaces capture Sobolev traces on K-ary trees"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001394,"raw_usage":{"total_tokens":5625,"prompt_tokens":917,"completion_tokens":4708,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":533,"completion_tokens_details":{"reasoning_tokens":4610}},"tokens_in":533,"tokens_out":4708,"duration_ms":33140,"temperature":1.0,"reasoning_tokens":4610,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:31:56.836328+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a fixed parameter triple in the theorem's range and a boundary function constant on one level-$n$ dyadic ball $I$ and zero elsewhere. Use the paper's affine extension (3.6)–(3.8) and compute the ratio $$R_n=\\frac{\\int_X |g_{\\tilde u}|^p\\,d\\mu_\\$\\lambda$}{$e^{{\\varepsilon n\\theta p}}$n^\\$\\lambda$\\nu(I)|u_I-u_{\\hat I}|^p}.$$ Theorem 1.1 predicts $\\sup_n R_n<\\infty$ for all choices of $I$; if any allowed parameter choice gives $R_n\\to\\infty$ along a sequence of dyadic balls, the claimed bounded linear extension fails.","supporting_citations":[{"cited_title":"Bj ¨orn, J","cited_arxiv_id":null,"evidence_quote":"Supplies the Ahlfors Q-regularity of the boundary measure, the measure estimates for balls, and the trace result for N^{1,p}(X) that Theorem 1.1 generalizes."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the uniformizing metric dX on the tree used to define distance, boundary, and geodesic-ray limits."},{"cited_title":"Koskela, T","cited_arxiv_id":null,"evidence_quote":"Provides the dyadic-norm and Whitney-extension method for traces of weighted function spaces, including the equivalence used in Proposition 2.13."},{"cited_title":"Mal´ y, N","cited_arxiv_id":null,"evidence_quote":"Provides the Whitney-type extension operator whose linearity and boundedness are analyzed in Theorem 1.4."},{"cited_title":"Gagliardo: Caratterizzazioni delle tracce sulla frontiera relative a d alcune classi di funzioni in n variabili, Rend","cited_arxiv_id":null,"evidence_quote":"Gives the layer-cake Lipschitz-approximation construction adapted in Proposition 3.5 for the nonlinear L^p extension."}],"review_version":1}