{"id":"cfb5ea5e-0cae-4918-8734-a5d010575bd6","arxiv_id":"1908.10691","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For stationary ergodic conductances on Z^d with a (p,q)-moment condition and reflection invariance, the space of harmonic functions growing slower than |x|^(1+alpha) has dimension d+1.","lead":"This paper proves a discrete Liouville theorem: in a random lattice with arbitrarily large or small edge weights, every slowly growing harmonic function is a linear combination of constants and coordinates. It is the first such result for degenerate random conductance models and adapts continuum proofs plus finite-element smoothing ideas to the lattice.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Most load-bearing gap: Corollary 2.3 relies on Setting 6.1 assumptions (91)-(92) and (94), deferred to companion [37]; if those fail, Theorem 1.4 is not established.","rationale":"The reader and I identify the same point: the proof of Theorem 1.4 is conditional on deferred boundary regularity. This is not an internal contradiction; the argument is explicit about the dependency. It is nevertheless load-bearing because the excess decay is the only deterministic route to the Liouville theorem, and Step 2 of that proof cannot be completed without comparing normal and tangential gradients of the harmonic extensions. The corrector section is mostly self-contained (with a sketched Meyer-type estimate), and the smoothing operators are constructed in detail, so the missing piece is localized. The suggested check is concrete: compare [37] statement-by-statement with Setting 6.1. If the companion paper is correct, acceptance is reasonable; if not, the theorem is unverified. Thus the appropriate action is to keep the reader's CONDITIONAL verdict unchanged.","tokens_in":47110,"tokens_out":7079,"duration_ms":70187,"concrete_test":"Fetch arXiv:1905.08151 ([37]) and verify that it states and proves precisely the inequalities (91) and (92) from Setting 6.1: for the constant-coefficient Dirichlet and Neumann extensions defined by (89)-(90), for every u on D_R, for all s in {2p/(p+1),2q/(q+1)} and the high exponent max{4p/(p-1),4q/(q-1)}, with a constant independent of R and u. Also confirm (94) is either proved in this paper or follows from a standard discrete Sobolev inequality on the boundary cube. If any of these is absent, the gap remains and Theorem 1.4 should stay conditional on [37]. As a minimal independent check, test (91) numerically for p=q=2, d=3, R=16,32,64 on random harmonic Dirichlet data and inspect whether the ratio bound degrades with R.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim Theorem 1.4 is derived from Corollaries 2.2 and 2.3, and Corollary 2.3 (excess decay) is not proved in this manuscript. Section 2.2 explicitly says the boundary regularity estimates comparing normal and tangential gradients are \"written in another paper (see [37]). Here, we consider it as assumptions (see (91) and (92) in Setting 6.1)\". Setting 6.1 also lists (94), a Sobolev inequality on the boundary, as an assumption. Lemmas 6.7 and 6.8 use (91) directly to bound B(u,Du) and B(u,Nu); these bounds propagate to Corollary 6.10, then to estimate (18), the boundary term in (15), and ultimately to the excess decay (13) and Corollary 2.3. The required inequalities are quantitative: for every u, the L^{2p/(p+1)} and L^{2q/(q+1)} norms of the normal and tangential discrete gradients of the harmonic extensions Du and Nu must be comparable by a constant c independent of R and u, and (92) must hold in the high exponent max{4p/(p-1),4q/(q-1)} over all edges in E_R. If [37] does not prove exactly these norms with uniform constants, or if (94) is not a valid discrete Sobolev inequality in this form, the excess decay and hence Theorem 1.4 are unsupported. The corrector construction in Section 3 is largely self-contained, so the gap is localized to this deferred boundary estimate, but it is load-bearing.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims a first-order Liouville theorem for the random conductance model on Z^d with positive but possibly unbounded conductances. Under stationarity, ergodicity, reflection invariance, and moment bounds ||mu_e||_{L^p} + ||mu_e^{-1}||_{L^q} < infinity with 1/p + 1/q <= 2/d, the space of omega-harmonic functions satisfying a sublinear-growth condition of order o(R^{1+alpha}) in an averaged L^{2p/(p-1)} norm is asserted to have dimension d+1. The proof follows the continuum strategy of Bella, Fehrman, and Otto: construct first- and second-order correctors with sublinearity, then combine them with a deterministic excess-decay estimate. The corrector construction in Section 3 is largely self-contained, while the excess decay is based on boundary regularity estimates stated as assumptions in Setting 6.1 and deferred to the companion paper [37].","tokens_in":47584,"tokens_out":5895,"duration_ms":66720,"significance":"If the deferred boundary estimates hold, the theorem is a meaningful extension of known Liouville results: it covers degenerate random conductances that are not bounded away from zero or infinity, requires only stationarity and ergodicity together with reflection invariance, and reaches the endpoint moment condition 1/p + 1/q <= 2/d. The paper also gives a careful discrete treatment of the corrector construction, including a density argument for sublinearity, and a detailed finite-element-style smoothing construction on the discrete boundary. The presentation is honest in flagging the assumptions borrowed from [37], and the central derivation is deterministic and structurally sound. However, the manuscript as submitted is not fully self-contained: the proof of the main theorem depends on quantitative boundary estimates that are not proved here, and the Meyer-type estimate is only sketched.","major_comments":[{"comment":"The proof of Corollary 2.3, and hence of Theorem 1.4, relies on boundary regularity estimates that are assumed but not proved in this manuscript. Corollary 6.10, which controls the boundary term (18), uses Lemmas 6.7 and 6.8, both of which invoke the normal/tangential comparability (91). Corollary 6.13, which controls the annulus term (19), uses Lemma 6.12, which invokes the high-exponent bound (92). Corollary 6.11, used for the interior control of the harmonic extensions, uses the discrete Sobolev inequality (94) through Lemma 6.6. These estimates must hold with constants independent of R and u in the specific norms 2p/(p+1), 2q/(q+1), and max{4p/(p-1),4q/(q-1)}. Section 2.2 explicitly states that they are written in another paper and are considered as assumptions here. If [37] does not prove exactly these inequalities with uniform constants, or if (94) is not valid in the stated form, the excess decay and the main theorem are not established. This is a load-bearing gap that should be resolved by including full proofs or by verifying the exact statements against [37] in an appendix.","section":"Section 2.2 and Setting 6.1, Eqs. (91), (92), (94)"},{"comment":"The Meyer-type estimate for the projection Hf onto L^2_{\\nabla^*} is an important step in proving that the second-order corrector belongs to L^{2p/(p+1)} and is sublinear. Its proof is only sketched: it refers to a 'standard argument with good and bad boxes' in [10], uses Green-function derivative bounds from [35], and then says that a duality argument completes the proof. Since this estimate is used to ensure the correctors have the integrability needed for Corollary 2.2, the proof should be written out in full or replaced by a precise statement with a complete derivation. As written, this is a second place where a central ingredient is deferred rather than proved.","section":"Section 3.2, Lemma 3.5"},{"comment":"Corollary 2.3 is stated as a main deterministic ingredient, but its proof is not fully contained in the paper. Section 2.2 sketches the energy estimates and then says that Steps 3, 4, and 5 'can be easily adapted' from the continuum proof of Bella, Fehrman, and Otto. Since the excess decay is the bridge between the sublinear correctors and the Liouville conclusion, the manuscript should provide a complete deterministic argument for Corollary 2.3, including the induction/iteration over scales, rather than relying on an external paper for the final steps. At minimum, the exact discrete analogues of the omitted steps should be stated with their hypotheses and proof.","section":"Section 2.2, Corollary 2.3"}],"minor_comments":[{"comment":"In the proof of Corollary 6.11(i) the text reads '2c^6 R Lambda(u)' where the statement has '2c^6 Lambda(u)'; the spurious factor R should be removed or the inequality should be restated consistently.","section":"Corollary 6.11"},{"comment":"There is a typo 'haft-space' in the description of the Fischer--Raithel result; it should read 'half-space'.","section":"Introduction, Section 1.1"},{"comment":"The notation in (13) is dense and the free parameter gamma appears both in a favorable and an unfavorable power; a short sentence explaining the final choice of gamma would improve readability.","section":"Section 2.2, Eq. (13)"},{"comment":"Reference [37] is described as 'submitted to Potential Analysis' with an arXiv identifier; since the current paper's main theorem is conditional on results in [37], the author should provide a precise pointer to the theorem or equation numbers in [37] that imply (91), (92), and (94).","section":"References, [37]"}],"recommendation":"major_revision","confidential_remarks":"The paper is transparent about its reliance on the companion paper [37], and the corrector part is solid, but the central theorem is not self-contained. I would ask the editor to require that the deferred boundary estimates be either fully proved in this paper or matched exactly against results in [37], and that the Meyer-type estimate be completed. If the companion estimates remain unavailable in published or verifiable form, the acceptance of this manuscript would be premature."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is the first Liouville principle for the random conductance model on Z^d with degenerate unbounded conductances, under the usual (p,q)-moment condition with 1/p+1/q <= 2/d. The corrector construction is solid, and the Scott-Zhang-style smoothing for discrete boundary estimates is a real piece of technical work. The problem is that the main theorem rests on Corollary 2.3, the excess decay, and that corollary is not proved in this manuscript. The proof explicitly imports the boundary regularity estimates (91), (92), and the Sobolev inequality (94) from the author's companion paper [37]. Section 2.2 says exactly that these estimates are \"written in another paper\" and are treated as assumptions. If they fail, Theorem 1.4 is unsupported. This is a load-bearing dependency, not a cosmetic one.\n\nWhat is genuinely new: the statement itself, and the route — constructing first and second correctors with a density argument to hit the endpoint 1/p+1/q <= 2/d, then adapting the Bella-Fehrman-Otto machinery with discrete duality and a mollifier built from Scott-Zhang interpolation. The corrector part in Section 3 is mostly self-contained, and the discrete smoothing construction in Section 5 is a nice contribution. The author is honest about what is deferred, which is to his credit.\n\nWhere the gaps are: (i) Corollary 2.3 is the heart of the Liouville theorem and depends on (91), (92), and (94) from [37]; the present text does not contain those proofs. (ii) The Meyer-type estimate (Lemma 3.5) is only sketched via Green's functions and a \"standard argument\", so a referee would need to check the discrete details. (iii) Lemma 5.2 (smoothing on Lipschitz boundaries) is stated without proof, though that is less serious since it is standard and not the main novelty. There is no circular reasoning and no invented entities; the dependency on [37] is explicit and plausible.\n\nWho this is for: people working on homogenization, Liouville theorems for random elliptic equations, or quantitative discrete regularity. A serious referee should engage with it, but the recommendation should be conditional: the manuscript needs to be revised once [37] is available and the deferred estimates are shown to hold in exactly the form used here. If they do, this is a substantial result.","headline":"A genuinely new discrete Liouville theorem for degenerate random conductances, but the excess-decay core is explicitly imported from a companion paper and not proved here.","tokens_in":47965,"tokens_out":2000,"would_cite":true,"duration_ms":21446,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60K37","35B53","35J15"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that in a degenerate random conductance model on $\\mathbb{Z}^d$, the space of harmonic functions with sub-$(1+\\alpha)$ growth is exactly the $(d+1)$-dimensional space of affine functions.","keywords":["random conductance model","Liouville theorem","degenerate conductances","homogenization","correctors","discrete harmonic functions","excess decay","finite-element interpolation"],"falsifier":"Compute, for a reflection-invariant stationary ergodic conductance law satisfying the moment bound, the discrete Dirichlet and Neumann extensions on the box $D_R$ and test whether inequalities (91) and (92) hold with a constant independent of $R$. A sequence of boxes where the normal-to-tangential gradient ratio blows up would falsify the proof's key input; more decisively, exhibiting an $\\omega$-harmonic function with $o(R^{1+\\alpha})$ growth that is not affine would falsify the theorem itself.","tokens_in":46859,"feed_emoji":"🎲","tokens_out":6070,"duration_ms":67175,"temperature":0.7,"pith_summary":"This paper proves a first-order Liouville theorem for the random conductance model on $\\mathbb{Z}^d$ when conductances may be arbitrarily large or close to zero. Under stationarity, ergodicity, reflection invariance, and a two-sided moment bound with $1/p+1/q \\le 2/d$, the only harmonic functions that grow more slowly than $|x|^{1+\\alpha}$ are the affine functions: constants plus the $d$ coordinate functions. The theorem matters because it extends a known continuum result to discrete random media and covers degenerate environments where uniform ellipticity fails, including conductances given by the exponential of a massless Gaussian free field. The proof constructs sublinear correctors and adapts an excess-decay argument to the lattice, with the key boundary regularity estimates assumed from a companion paper.","feed_headline":"Random conductance media have no slow harmonic modes beyond d+1","feed_subtitle":"Under only moment conditions, unbounded random conductances still force all slowly growing harmonic functions to be affine.","key_machinery":"The argument runs on three objects: the first-order correctors $\\varphi_i$ (harmonic coordinates) and second-order correctors $\\sigma_{ijk}$ solving discrete Poisson equations; the excess decay estimate of Corollary 2.3, which controls the homogenization error in a box of radius $R$ by $(r/R)^\\alpha$; and a discrete smoothing toolkit built from a finite-element interpolation projection and a dual smoothing operator, which transfers continuum boundary estimates to the lattice. Reflection invariance of the environment makes the homogenized matrix $a^h$ diagonal, so the only affine directions are the $d$ coordinate axes.","core_discovery":"The central claim is Theorem 1.4: for $\\mathbb{P}$-almost every environment $\\omega$, the set $S(\\omega)$ of $\\omega$-harmonic functions $u$ satisfying the growth condition $\\lim_{R\\to\\infty} R^{-(1+\\alpha)} (\\sum_{|y|_\\infty<R} |u(y)|^{2p/(p-1)})^{(p-1)/(2p)} = 0$ is a linear space of dimension exactly $d+1$. Equivalently, every such function is of the form $u(x) = a + b\\cdot x$ with $a\\in\\mathbb{R}$ and $b\\in\\mathbb{R}^d$. The paper establishes this by constructing first-order and second-order correctors with sublinear growth, proving a deterministic excess decay estimate on boxes, and using reflection invariance to reduce the homogenized matrix to a diagonal matrix. The excess decay proof adapts the continuum argument to the discrete setting, with boundary regularity estimates treated as assumptions supplied by the author's companion paper.","pith_inferences":["The paper relaxes the strict inequality $1/p+1/q<2/d$ used in the continuum to $1/p+1/q\\le 2/d$, which suggests that boundary regularity assumptions, rather than interior estimates, are the true bottleneck; if those assumptions hold under weaker moments, the Liouville dimension may persist beyond the stated range.","The method likely extends to non-diagonal homogenized matrices and to random-graph settings once discrete $L^p$ boundary trace theory is developed; the author explicitly notes the non-diagonal case as an open difficulty.","A testable consequence is that numerical computation of the dimension of slowly growing harmonic functions on large boxes should return $d+1$ for any reflection-invariant stationary ergodic conductance law satisfying the moment bound; deviations would indicate a missing assumption.","The finite-element interpolation and dual smoothing tools developed here could be reused for higher-order Liouville theorems or quantitative homogenization on boxes in degenerate random environments."],"forward_implications":["If Theorem 1.4 is correct, every $\\omega$-harmonic function with sub-$(1+\\alpha)$ growth is affine, so the medium cannot host non-trivial slowly growing harmonic modes.","The result covers conductances with unbounded upper and lower ratios, showing Liouville behavior is stable under degeneracies that satisfy the moment condition.","The discrete excess decay and corrector construction provide a route to quantitative homogenization and higher-order Liouville statements in the same degenerate setting.","The dimension $d+1$ matches the classical lattice Liouville theorem, so the random medium does not alter the space of harmonic coordinates even when ellipticity constants are absent."],"supporting_citations":[{"why":"Supplies the continuum excess-decay and corrector framework that the paper adapts to the lattice.","marker":"[5]"},{"why":"Contains the boundary regularity estimates (91), (92) and Sobolev inequality (94) assumed without proof in Setting 6.1.","marker":"[37]"},{"why":"Provides the finite-element interpolation projection used to construct discrete smoothing operators on box boundaries.","marker":"[40]"},{"why":"Establishes the quenched invariance principle and the $(p,q)$-moment condition that motivates the assumptions.","marker":"[2]"},{"why":"Shows reflection invariance makes the homogenized matrix diagonal, which selects the $d$ coordinate directions.","marker":"[14]"}],"fun_headline_variants":["Random conductances: degenerate bounds still force affine harmonics","Liouville theorem holds for random conductances without boundedness","Unbounded random conductances: harmonic growth must be affine","Degenerate random conductance model: only d+1 harmonic directions"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof assumes, without proving it here, that on box boundaries the discrete gradients of harmonic extensions in tangential and normal directions are quantitatively comparable, together with a discrete Sobolev inequality; if these boundary regularity estimates fail, the excess-decay argument and hence the theorem are not established.","fun_headline_variants_meta":{"raw":{"variants":["Random conductances: degenerate bounds still force affine harmonics","Liouville theorem holds for random conductances without boundedness","Unbounded random conductances: harmonic growth must be affine","Degenerate random conductance model: only d+1 harmonic directions"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000556,"raw_usage":{"total_tokens":2620,"prompt_tokens":894,"completion_tokens":1726,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":510,"completion_tokens_details":{"reasoning_tokens":1658}},"tokens_in":510,"tokens_out":1726,"duration_ms":13824,"temperature":1.0,"reasoning_tokens":1658,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T10:36:18.128725+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute, for a reflection-invariant stationary ergodic conductance law satisfying the moment bound, the discrete Dirichlet and Neumann extensions on the box $D_R$ and test whether inequalities (91) and (92) hold with a constant independent of $R$. A sequence of boxes where the normal-to-tangential gradient ratio blows up would falsify the proof's key input; more decisively, exhibiting an $\\omega$-harmonic function with $o(R^{1+\\alpha})$ growth that is not affine would falsify the theorem itself.","supporting_citations":[{"cited_title":"Bella, B","cited_arxiv_id":null,"evidence_quote":"Supplies the continuum excess-decay and corrector framework that the paper adapts to the lattice."},{"cited_title":"An $L^p$-comparison, $p\\in (1,\\infty)$, on the finite differences of a discrete harmonic function at the boundary of a discrete box","cited_arxiv_id":"1905.08151","evidence_quote":"Contains the boundary regularity estimates (91), (92) and Sobolev inequality (94) assumed without proof in Setting 6.1."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the finite-element interpolation projection used to construct discrete smoothing operators on box boundaries."},{"cited_title":"Andres, J.-D","cited_arxiv_id":null,"evidence_quote":"Establishes the quenched invariance principle and the $(p,q)$-moment condition that motivates the assumptions."},{"cited_title":"De Masi, P","cited_arxiv_id":null,"evidence_quote":"Shows reflection invariance makes the homogenized matrix diagonal, which selects the $d$ coordinate directions."}],"review_version":1}