{"id":"a6a16384-1ca4-4394-8d5a-3e34c6a35571","arxiv_id":"1908.05988","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"A d-dimensional tropical variety with an l-dimensional lineality space remains connected through codimension one after fewer than d-l facets are removed.","lead":"The paper proves that tropicalizations of irreducible varieties stay connected even after removing many facets, and that generic hyperplane sections of tropical varieties are again tropicalizations of irreducible varieties. This gives new tools for deciding whether a polyhedral complex is tropically realizable.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Claim 12's key intersection assertion is unproved and the stated reason is false for general balanced complexes; Theorem 1's induction has a load-bearing gap unless a missing lemma is supplied.","rationale":"The reader's weakest_assumption identifies exactly the same step, and I agree. The theorem may well be true, and the surrounding structure—tropical Bertini, the induction, and the reduction to the pointed case—is coherent. But the proof as written relies on a statement that is not derived and is false if pointedness or other side conditions are ignored. This gap sits in the central induction rather than in a peripheral lemma, so a conditional verdict is appropriate. I do not see grounds for outright rejection: there is no indication the theorem is false, and the missing statement may be supplied. There are no data, reproducibility, or parameter-count concerns for this theory paper.","tokens_in":11115,"tokens_out":45835,"duration_ms":517128,"concrete_test":"Test the exact implication on a pointed balanced fan. One concrete check: take Σ to be the standard tropical plane in R^3, set Hp={z=1}, and enumerate all rational affine hyperplanes Hq that meet a chosen facet Q, avoid a third facet F, and for which A=Σ∩Hp is not contained in a hyperplane parallel to Hq; compute whether Hp∩Hq∩Σ is always nonempty. The unpointed two-parallel-planes example shows the naked implication is false outside the pointed hypothesis, so the test should focus on pointed fans. If a pointed counterexample exists, Claim 12 and the proof of Theorem 1 need revision; if none exists, the missing lemma should be extracted and proved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In the proof of Claim 12 (§3), the assertion 'Since the balanced positive-dimensional polyhedral complex Σ ∩ Hp is not contained in any hyperplane parallel to Hq, some point in Σ must lie in Hp ∩ Hq' is the only bridge from the perturbed hyperplanes to Hp∩Hq∩Σ≠∅. This implication is not a consequence of the premises as written: balancedness plus 'not contained in any hyperplane parallel to Hq' does not force an intersection. For example, in R^3 the pure 2-dimensional complex A consisting of the two parallel planes x=1 and x=2, each with weight 1, is balanced; for Hq={x=0}, A∩Hq=∅, yet A is not contained in any single plane x=c. This particular example has lineality, so it does not refute the pointed version used in the proof, but it shows that the asserted implication is not a direct consequence of balance and non-parallel containment, and no pointed replacement or citation is supplied. If the implication fails in the pointed setting, the equivalence relation ∼F is not shown to be one class, Claim 11 cannot be applied, and the induction for Theorem 1 collapses. At minimum the proof must state and prove a lemma saying that a pointed balanced positive-dimensional complex avoiding a hyperplane must be contained in a parallel hyperplane; no such lemma appears.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves that the tropicalization of a d-dimensional irreducible subvariety of (K*)^n over a characteristic-zero field is (d-ℓ)-connected through codimension one, where ℓ is the dimension of the lineality space of the tropicalization. The proof reduces to the pointed case by quotienting by the lineality subtorus, then proceeds by induction on dimension. The inductive step is organized around two claims: a tropical Bertini theorem (Proposition 4) supplies hyperplane sections that are themselves tropicalizations of irreducible varieties, and an equivalence relation ∼F on facets is used to route paths around any prescribed set of removed facets. The paper also derives a higher connectivity statement for skeleta of rational polytopes and for fine subdivisions of Bergman fans, and gives an example of a balanced fan that is not realizable as the tropicalization of an irreducible variety.","tokens_in":11389,"tokens_out":23311,"duration_ms":256416,"significance":"If the proof can be completed, Theorem 1 is a substantial strengthening of the classical result that tropicalizations are connected through codimension one, and it yields a tropical analogue of Balinski's theorem. Proposition 4 answers a question of Cartwright and Payne and is of independent interest. The argument is coherent and makes careful use of established tools (FMZ18, AS17, Pay12, CP12, OP13, JY16, Rin13); it is not circular, and the paper is explicit about the characteristic-zero hypothesis and about open problems. The main weakness is a missing and nontrivial balancing lemma in the proof of Claim 12, together with a false assertion about pointedness in the reduction step; both are local and fixable.","major_comments":[{"comment":"The sentence \"Since the balanced positive-dimensional polyhedral complex Σ ∩ Hp is not contained in any hyperplane parallel to Hq, some point in Σ must lie in Hp ∩ Hq\" is the sole justification for the crucial intersection statement (3), and it is not proved. The implication is false for arbitrary balanced complexes: in R^3, the union of the two planes x=1 and x=2 with weight one on each is balanced, avoids Hq={x=0}, and is not contained in any single plane parallel to Hq. The additional hypotheses present in the proof, such as pointedness and connectedness, may make the statement true, but that requires a lemma, for instance that a pointed balanced polyhedral complex contained in an open halfspace must lie in the boundary hyperplane. No such lemma or citation is supplied. Without it, Claim 12 is not established and the induction for Theorem 1 collapses. The gap is local and likely fixable, but it is load-bearing.","section":"§3, proof of Claim 12, equation (3)"},{"comment":"The assertion \"Since trop(X) is connected, the triviality of the lineality space implies that every face of trop(X) is pointed\" is false as stated. The connected one-dimensional complex consisting of the two coordinate axes in R^2 has trivial lineality space but has faces containing affine lines; it is in fact the tropicalization of the irreducible curve V(1+2x1+3x2+4x1x2) in (C*)^2. The proof should insert an explicit refinement step replacing Σ by a pointed subdivision and explain why k-connectedness of the subdivision implies k-connectedness of the original complex. This matters because the perturbation argument in Claim 12 uses the full-dimensionality of inner normal cones, which requires pointedness.","section":"§3, paragraph after the quotient by the lineality space"}],"minor_comments":[{"comment":"There is a typo: \"restriction to X is also free. is Let ~X\" contains a stray \"is\" that should be deleted.","section":"§3, first paragraph of the main proof"},{"comment":"The symbol G is used both for the chosen set of removed facets and for the facet-ridge hypergraph of Σ; please rename one of them to avoid confusion.","section":"§3, proof of Claim 11"},{"comment":"The clause \"meets both P and Q in their relative interior\" should read \"relative interiors\".","section":"§3, definition of ∼F"},{"comment":"The phrase \"dense in the Euclidean topology on P^n_Q\" is informal; since P^n_Q is not a Euclidean space, clarify that the density is in the real points of P^n_Q. Also, the use of H both for a hyperplane in R^d and for its preimage in R^n is confusing and should be disambiguated.","section":"§2, Proposition 4"}],"recommendation":"major_revision","confidential_remarks":"The central theorem is novel and the overall strategy is credible. The two issues I raise are both fixable within the scope of the manuscript: a missing balancing lemma in Claim 12 and an explicit pointed refinement in the reduction step. I do not see grounds for rejection; the paper is a good fit for the journal and the result, once the proof is completed, would be of broad interest."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear [Colleague],\n\nThis is a strong paper, and the main theorem is genuinely new. Maclagan and Yu show that the tropicalization of an irreducible d-dimensional variety is (d-l)-connected through codimension one, upgrading the classical connectivity theorem and yielding a tropical Bertini theorem that answers Question 9 of Cartwright-Payne. The Balinski theorem consequence is a nice bonus, and the Bergman fan result (Proposition 14) is a solid addition. The proof strategy—induce on dimension using the tropical Bertini theorem—is clever and new. The citations are standard tools (MS15, JY16 for stable intersections, Pay12, FMZ18), not recycled results; there is no circularity.\n\nThe honest weakness is in Claim 12. The proof asserts that because the balanced positive-dimensional complex Sigma ∩ Hp is not contained in any hyperplane parallel to Hq, some point of Sigma must lie in Hp ∩ Hq. That implication is not a consequence of balancedness alone. The stress-test note gives a clean counterexample: two parallel planes in R3, each weighted 1, form a balanced complex but can avoid a parallel hyperplane. The example has lineality, so it doesn't refute the pointed case used in the proof, but it does show the stated reason is incomplete. The proof needs a lemma specific to pointed complexes—something like: a pointed balanced complex of positive dimension that is not contained in a hyperplane parallel to H must meet H. I suspect such a lemma is true, and the proof can be patched, but as written this is a load-bearing gap. The perturbation step that ensures Sigma ∩ Hp is not contained in a hyperplane parallel to Hq is also terse; it relies on the inner normal cone being full-dimensional, which is fine, but the consequence needs spelling out.\n\nEverything else checks out. The base case and induction are coherent, and the reduction to pointed complexes via the lineality action is standard. The characteristic-zero caveat is flagged honestly (Remark 10). There are a few typos (e.g., 'is Let' on page 7), but nothing substantive.\n\nI agree with the reader's conditional verdict. This deserves a serious referee. The gap is fixable and the result is significant. I'd send it for review; the referee should ask for a proof of the missing lemma in Claim 12 and a slightly fuller justification of the perturbation.\n\nBest,\n[You]","headline":"Strong new higher-connectivity theorem with a load-bearing gap in Claim 12 that is likely patchable.","tokens_in":11872,"tokens_out":10588,"would_cite":true,"duration_ms":107658,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14T05"],"pacs":[],"model":"deepseek-v4-flash","headline":"Tropicalizations of irreducible varieties stay connected after removing many facets.","keywords":["tropical geometry","tropicalization","connectivity through codimension one","facet-ridge hypergraph","Tropical Bertini theorem","Balinski's theorem","Bergman fans","polyhedral complexes"],"falsifier":"Try to realize the two-plane fan of Example 3 (two standard tropical planes meeting along the ray e1) as the tropicalization of an irreducible surface over C, for example by computing Gröbner bases of candidate parametrizations; since the fan disconnects after removing a facet containing e1, any such realization would disprove Theorem 1.","tokens_in":10889,"feed_emoji":"","tokens_out":12820,"duration_ms":136821,"temperature":0.7,"pith_summary":"This paper proves that the tropicalization of an irreducible d-dimensional variety is highly connected: no matter how one chooses a polyhedral complex presenting it, the complex stays connected through codimension one after any fewer than d−ℓ closed facets are removed, where ℓ is the dimension of the lineality space. The proof introduces a tropical version of Bertini's theorem, showing that a generic rational affine hyperplane cuts the tropicalization of an irreducible variety into the tropicalization of another irreducible variety, and then inducts on dimension. A sympathetic reader should care because this sharpens the classical connectedness of tropicalizations into a quantitative statement with concrete consequences: it recovers Balinski's theorem on polytope graphs, gives a connectivity obstruction to tropical realizability, and yields higher connectivity for Bergman fans and for skeleta of normal fans of rational polytopes.","feed_headline":"Tropicalizations stay connected after many facets are removed","feed_subtitle":"The result generalizes Balinski's graph theorem and gives a new obstruction to realizing fans tropically.","key_machinery":"The load-bearing object is the facet-ridge incidence hypergraph of a pure polyhedral complex: its vertices are the facets and its hyperedges are the ridges, and being k-connected through codimension one means the hypergraph remains connected after deleting any k−1 vertices together with their incident hyperedges. The proof's engine is the new Tropical Bertini Theorem, which guarantees that a dense set of rational affine hyperplanes intersect the tropicalization in the tropicalization of an irreducible variety. In the induction, hyperplane sections create lower-dimensional irreducible tropical varieties; paths inside those sections are then lifted to facet-ridge paths in Σ. The final step relies on an equivalence relation ∼F, where two facets are related if a hyperplane avoiding F meets both in their relative interiors, together with stable intersection results that make Σ∩Hp a balanced positive-dimensional complex.","core_discovery":"The central claim is Theorem 1: if K has characteristic zero and is algebraically closed, complete, or real closed with convex valuation ring, and X ⊂ (K*)n is irreducible of dimension d, then any polyhedral complex Σ with support trop(X) and ℓ-dimensional lineality space is (d−ℓ)-connected through codimension one. Equivalently, the facet-ridge hypergraph of Σ remains connected after deleting any d−ℓ−1 closed facets, and the bound is sharp because a simplicial facet is isolated by removing its d−ℓ neighbouring facets. The proof reduces to the pointed case by quotienting by the lineality space, proves the Tropical Bertini Theorem (Proposition 4) for generic rational affine hyperplanes, and then runs an induction in which hyperplane slices supply lower-dimensional irreducible tropical varieties whose connectivity is transferred back to paths in the original facet-ridge hypergraph. Along the way the paper obtains Corollary 2 on skeleta of normal fans of rational polytopes and Proposition 14 on the fine subdivision of Bergman fans.","pith_inferences":["Because the characteristic-zero assumption enters only through the Tropical Bertini Theorem, a positive-characteristic version of that theorem would immediately extend Theorem 1; conversely, if Theorem 1 holds in positive characteristic despite Bertini failing there, the missing ingredient must lie elsewhere in the induction.","The connectivity bound supplies a cheap necessary condition for realizability that could be checked combinatorially before attempting any algebraic construction, and it may be useful for pruning searches for tropical bases or parametrizations.","The unproved geometric assertion inside Claim 12—that a balanced positive-dimensional slice not parallel to a plane must meet it—is a natural target for a counterexample or a separation theorem; until it is settled, the induction has a delicate point."],"forward_implications":["Balinski's theorem is recovered: applying the result to the complete normal fan of a full-dimensional polytope in R^d shows its edge graph is d-connected.","The k-skeleton of the normal fan of a rational full-dimensional polytope is k-connected through codimension one (Corollary 2).","A pure d-dimensional fan with ℓ-dimensional lineality space that fails to be (d−ℓ)-connected through codimension one cannot be the tropicalization of an irreducible variety over characteristic 0; Example 3 gives a concrete non-realizable fan.","The fine subdivision of the Bergman fan of a rank d+1 matroid is d-connected, whether or not the matroid is representable over a field (Proposition 14).","The Tropical Bertini Theorem answers affirmatively the question of whether a general hyperplane section of a tropicalization is again the tropicalization of an irreducible variety."],"supporting_citations":[{"why":"Supplies the base case that tropicalizations are connected through codimension one and poses the question answered by the Tropical Bertini Theorem.","marker":"[CP12]"},{"why":"Provides the toric Bertini theorem for finite dominant maps with property PB, the main input to Proposition 4.","marker":"[FMZ18]"},{"why":"Supplies the property PB criterion and the isogeny modification used to arrange that a monomial projection satisfies PB.","marker":"[AS17]"},{"why":"Gives the finiteness criterion for monomial maps whose tropicalization is injective on maximal faces, used in Lemma 8.","marker":"[Pay12]"},{"why":"Provides the Transverse Intersection Lemma that lifts a hyperplane intersection to the tropicalization of a fiber.","marker":"[OP13]"},{"why":"Shows that stable intersections with hyperplanes are balanced, used to keep Claim 12's induction running.","marker":"[JY16]"},{"why":"States Balinski's theorem, which Theorem 1 generalizes by way of complete normal fans.","marker":"[Bal61]"},{"why":"Supplies standard facts about tropicalizations, connectedness, and transverse intersections used throughout the proof.","marker":"[MS15]"},{"why":"Describes local tropical linear spaces as homeomorphic to R^d, the key ingredient in the Bergman fan connectivity proof.","marker":"[Rin13]"},{"why":"Defines the coarse and fine subdivisions of Bergman fans used in Proposition 14.","marker":"[AK06]"}],"fun_headline_variants":["Tropical varieties stay connected after many cuts","Higher connectivity for tropicalizations proven","Tropical Bertini yields sharp connectivity bound","Balinski's graph theorem extends to tropical fans","New obstruction to tropically realizing polytopal fans"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof of Claim 12 assumes, without proof, that a positive-dimensional balanced slice of the tropical variety that is not parallel to a chosen plane must actually meet that plane; if this geometric assertion fails, the argument that any two facets are equivalent can collapse.","fun_headline_variants_meta":{"raw":{"variants":["Tropical varieties stay connected after many cuts","Higher connectivity for tropicalizations proven","Tropical Bertini yields sharp connectivity bound","Balinski's graph theorem extends to tropical fans","New obstruction to tropically realizing polytopal fans"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000286,"raw_usage":{"total_tokens":1617,"prompt_tokens":818,"completion_tokens":799,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":434,"completion_tokens_details":{"reasoning_tokens":732}},"tokens_in":434,"tokens_out":799,"duration_ms":7776,"temperature":1.0,"reasoning_tokens":732,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:59:29.830703+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Try to realize the two-plane fan of Example 3 (two standard tropical planes meeting along the ray e1) as the tropicalization of an irreducible surface over C, for example by computing Gröbner bases of candidate parametrizations; since the fan disconnects after removing a facet containing e1, any such realization would disprove Theorem 1.","supporting_citations":[],"review_version":1}