{"id":"2c0409e7-8db9-4b26-8694-bcd616720c70","arxiv_id":"2412.11759","paper_version":2,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The paper proves Speyer's 2005 tropical f-vector conjecture by showing that the matroid invariant ω(M) is non-negative for every matroid, via a Cohen-Macaulay external activity complex of a pair of matroids.","lead":"Two mathematicians prove a 2005 conjecture in tropical geometry by building a new 'external activity complex' from pairs of matroids. The result bounds the number of faces in subdivisions of matroid base polytopes, settling a long-open question connecting tropical geometry, commutative algebra, and combinatorics.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Final step to Speyer's conjecture is outsourced to the concurrent companion [FSS24]; acceptance should be conditional on that reduction theorem being independently verified.","rationale":"I read the paper as proving two things: a nonnegative homology formula for ω(M), and, via the external Fink–Shaw–Speyer theorem, the tropical f-vector conjecture. The internal route to formula (1.1) is impressive and, as far as I can tell, sound: the Cohen-Macaulay property of Δ_w(M,M), the bivaluative K-polynomial, and the Hochster/Betti-number argument in Section 7 fit together, and the sign bookkeeping in the proof of Theorem E checks out when D(M,M) has expected rank and M is connected. I found no internal contradiction or obvious gap in the main construction. The single most load-bearing weakness is the quoted reduction theorem: it is the only step that converts ω(M) ≥ 0 for all matroids into nonnegativity of every coefficient of g_M(t), hence into Speyer's conjecture. Because that theorem is not proved in the present paper, is concurrent, and shares an author, an unconditional acceptance is not fully warranted; the right verdict is conditional on independent verification of [FSS24] and absence of circular dependence on this paper. This matches the reader's weakest-assumption identification, so I agree with the reader's diagnosis but would adjust the verdict from ACCEPT to CONDITIONAL rather than leaving it unconditional.","tokens_in":58844,"tokens_out":19845,"duration_ms":195604,"concrete_test":"Perform an independent, line-by-line proof check of the Fink–Shaw–Speyer reduction theorem (stated in §1.4 of this paper, proved in [FSS24, §1.4]), deriving it only from the definition of g_M(t) in [FS12] and the standard valuative properties of matroid polytopes, and recording every use of this paper's results. If the derivation succeeds without invoking Theorem E or Lemma 7.6, the concern is resolved; if it fails, or if it depends essentially on results of the present paper, then the combined argument is circular and Theorem E should be downgraded to 'ω(M) ≥ 0', with Speyer's conjecture remaining conditional on [FSS24].","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Theorem E has two components: (i) a new, self-contained nonnegative formula for ω(M) via homology of links in the external activity complex Δ_w(M,M), and (ii) the inference from ω(M) ≥ 0 to nonnegativity of all coefficients of g_M(t), via the quoted theorem of Fink, Shaw, and Speyer. Component (ii) is not proved in this paper: it is stated as 'Theorem (Fink–Shaw–Speyer [FSS24])' in Section 1.4 and the proof is relegated to a concurrent preprint (arXiv:2411.19521) with overlapping authorship. This reduction is load-bearing because it is the unique bridge from a single nonnegative coefficient of g_M(t) to the full tropical f-vector conjecture. If the reduction were false, the paper would still establish a striking nonnegative formula for ω(M), but not Speyer's 2005 conjecture. The internal proof of formula (1.1) appears coherent: the signs in Theorem D and the connected case of Lemma 7.6 are consistent, and the bivaluativity/Cohen-Macaulay arguments are developed in detail. The unresolved point is purely the external reduction step, which the present text neither proves nor verifies.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces two related objects: the Schubert variety of a pair of linear subspaces of C^n and the external activity complex Δ_w(M1,M2) of a pair of matroids. The authors prove that this complex is Cohen-Macaulay, that its finely graded K-polynomial is bivaluative, and that its Z2-graded K-polynomial matches an Euler-characteristic formula involving exterior powers of tautological quotient classes. From these results they derive a nonnegative formula for the Fink-Shaw-Speyer invariant ω(M) as a sum of dimensions of reduced homology groups of links in Δ_w(M,M). Combining this with a theorem of Fink-Shaw-Speyer, they conclude that Speyer's 2005 tropical f-vector conjecture holds.","tokens_in":59011,"tokens_out":5713,"duration_ms":57669,"significance":"If correct, this resolves a long-standing conjecture and provides a new, explicitly homological mechanism for the nonnegativity of ω(M). The main technical engine—the external activity complex, the diagonal Dilworth truncation, and the tropical cell-complex description of Chern class products—is substantial and likely to be influential. The paper also contains several structurally interesting intermediate results, such as bivaluativity of the K-polynomial and the Cohen-Macaulay property, which are established by a novel combination of Gröbner degeneration, Kempf collapsing, and tropical intersection theory. The chief caveat is that the final step to the full f-vector conjecture is outsourced to a concurrent, overlapping-authorship preprint.","major_comments":[{"comment":"The inference from ω(M) ≥ 0 to nonnegativity of all coefficients of g_M(t) is stated as \"Theorem (Fink–Shaw–Speyer [FSS24])\" but is not proved in this manuscript; it is imported from the concurrent preprint arXiv:2411.19521, which shares an author with the present paper. This inference is load-bearing for the resolution of Conjecture 1.1, because the paper's internal results establish only nonnegativity of ω(M) (Equation (1.1)), not nonnegativity of the other coefficients of g_M(t). The authors should either include a self-contained proof of the reduction theorem or explicitly mark Theorem E and the resolution of Speyer's conjecture as conditional on independent verification of [FSS24].","section":"Section 1.4, proof of Theorem E"}],"minor_comments":[{"comment":"There is a typo: \"linear susbspaces\" should be \"linear subspaces\".","section":"Section 1, first paragraph"},{"comment":"The phrase \"Gröbner\" is typeset as \"Gr¨obner\" in the running text; please use the correct umlaut rendering consistently.","section":"Section 1, item (3) of Theorem A"},{"comment":"The text \"when proving proving Theorem A\" contains a duplicated word; it should read \"when proving Theorem A\".","section":"Section 5, paragraph before Definition 5.1"},{"comment":"The heading \"tatutological bundles\" should be \"tautological bundles\".","section":"Section 3.2 heading"},{"comment":"The word \"uneffected\" should be \"unaffected\".","section":"Theorem 7.2 proof"},{"comment":"The table of circuit decompositions is dense; adding one sentence pointing out that the row for circuit 123567 illustrates the difference between Δ_w(F,F) and Δ_w(F-,F-) would improve readability.","section":"Example 4.20"}],"recommendation":"major_revision","confidential_remarks":"The technical core of the paper appears sound and impressive, and the nonnegative formula for ω(M) is a genuine result regardless of the final reduction. The only substantive concern is the logical dependence on [FSS24], a concurrent preprint with overlapping authorship. If the editor is willing to accept the headline theorem as contingent on that companion result, the changes needed are minor; otherwise the paper should be revised to prove the reduction or to state the dependence explicitly."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Berget and Fink prove Speyer's 2005 tropical f-vector conjecture. The genuinely new machinery is the external activity complex Δ_w(M1,M2) of a pair of matroids, defined from the diagonal Dilworth truncation D(M1,M2). They prove it is Cohen–Macaulay, its finely graded K-polynomial is bivaluative, and its Z2-graded K-polynomial is computed by tautological quotient classes. The payoff is a homology formula for ω(M), the top coefficient of Speyer's g-invariant, as a sum of dimensions of reduced homology groups of links in Δ_w(M,M). Because the summands are nonnegative, ω(M)≥0 for all matroids. The internal logic is coherent: the realizable case uses Kempf collapsing and Gröbner bases, then Section 6 computes products of Chern classes via compactly supported Euler characteristics of tropical cells and extends by bivaluativity. The introduction is a model of clarity, and the paper is honest about what is not proved: shellability, integer coefficients, and positive characteristic analogues.\n\nNow the soft spot. Theorem E splits into two claims: (i) the nonnegative formula for ω(M), proved here; (ii) the reduction from ω(N)≥0 for all minors N to nonnegativity of all coefficients of g_M(t). Claim (ii) is quoted verbatim from [FSS24], a concurrent preprint with overlapping authorship. It is not proved or even sketched in this paper. It is load-bearing: without it, the paper gives a strong nonnegative formula for ω(M), but does not establish Speyer's conjecture. I found no circularity in the main argument, and the sign checks in Theorem D and Lemma 7.6 are consistent. Still, a referee should verify the companion result, and acceptance should be conditional on it.\n\nMinor concerns: Section 6 is intricate; I followed the structure but did not verify every line. The authors' admission that shellability is unproved is a real limitation: it costs integer coefficients and a cleaner resolution, but it does not affect the main theorem. The citation pattern is fine; citations to their own earlier work are to results actually used.\n\nWho should read this: matroid theorists, commutative algebraists, tropical geometers. It deserves a serious referee; desk rejection would be wrong. I would take it to reading group and would cite it once the FSS companion is confirmed. Recommendation: send to peer review, and make sure [FSS24] is part of the reviewed package or independently checked.","headline":"Berget and Fink resolve Speyer's 2005 tropical f-vector conjecture with a new pair-of-matroids external activity complex, but the last step to the conjecture is outsourced to a concurrent companion paper that needs independent checking.","tokens_in":59691,"tokens_out":3422,"would_cite":true,"duration_ms":33255,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["05B35","14T05","13D02","05E45"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves the tropical f-vector conjecture by expressing the matroid invariant $\\omega(M)$ as a sum of reduced homology dimensions of links in a new external activity complex for pairs of matroids.","keywords":["matroids","external activity complex","tropical f-vector conjecture","g-invariant","Cohen-Macaulay simplicial complexes","tautological classes of matroids","diagonal Dilworth truncation","K-polynomials"],"falsifier":"Compute both sides of Equation (1.1) for a connected non-realizable matroid such as the non-Fano matroid: if the sum of reduced homology dimensions of the links in $\\Delta_w(M,M)$ differs from $\\omega(M)$ computed from the exterior-power definition of the $g$-invariant, the central formula is false.","tokens_in":58484,"feed_emoji":"🌴","tokens_out":7605,"duration_ms":69627,"temperature":0.7,"pith_summary":"This paper proves the tropical f-vector conjecture, a 2005 bound on how many faces of each dimension can appear when a hypersimplex is subdivided into matroid base polytopes; such subdivisions describe tropical linear spaces, so the bound controls their combinatorial complexity. The proof works for all matroids, realizable or not, by introducing a simplicial complex attached to a pair of matroids: the external activity complex. The central result is a formula expressing the matroid invariant $\\omega(M)$, the top coefficient of the $g$-invariant, as a sum of dimensions of reduced homology groups of links in this complex, making nonnegativity manifest. Because nonnegativity of $\\omega$ on all minors is known to force nonnegativity of all coefficients of the $g$-invariant, the face-count conjecture follows.","feed_headline":"Matroid homology proves tropical f-vector conjecture","feed_subtitle":"A new Cohen-Macaulay complex gives every matroid a nonnegative omega invariant, settling the 2005 face-count bound.","key_machinery":"The external activity complex of a pair of matroids $(M_1,M_2)$ is the simplicial complex whose facets are the monomials $x^{B\\cup E_1(B)}y^{B\\cup E_2(B)}$ for each basis $B$ of the diagonal Dilworth truncation $D(M_1,M_2)$, the matroid whose circuits are the minimal nonempty subsets $C$ with $\\operatorname{rank}_{M_1}(C)+\\operatorname{rank}_{M_2}(C)=|C|$; the sets $E_1(B)$ and $E_2(B)$ record external activity with respect to a weight vector $w$. This complex generalizes the external activity complex of a single matroid and arises as a Gr\\\"obner degeneration of the Schubert variety of a pair of linear spaces. The paper proves that $\\Delta_w(M_1,M_2)$ is Cohen-Macaulay, that its finely graded $K$-polynomial is bivaluative in the two matroids, and that this $K$-polynomial equals $\\sum_{i,j}\\chi(\\wedge^i[Q^\\vee_{M_1}]\\cdot\\wedge^j[Q^\\vee_{M_2}])(-U_1)^i(-U_2)^j$. Hochster's formula then identifies the coefficients of total degree $n$ with sums of reduced homology dimensions of links inside the complex.","core_discovery":"The paper establishes that for every matroid $M$ on $[n]$ of rank $r$, the invariant $\\omega(M)$ is a sum of nonnegative integers: $\\omega(M)=\\sum_{B\\in\\binom{[n]}{r}}\\dim \\widetilde{H}_{2r-2}(\\operatorname{link}_{\\Delta_w(M,M)}(x_{[n]\\setminus B}\\,y_B))$. Since nonnegativity of $\\omega$ on all minors implies every coefficient of the $g$-invariant is nonnegative, this proves the tropical f-vector conjecture. The central construction is the external activity complex $\\Delta_w(M_1,M_2)$ of a pair of matroids, defined combinatorially by facets $x^{B\\cup E_1(B)}y^{B\\cup E_2(B)}$ for each basis $B$ of the diagonal Dilworth truncation $D(M_1,M_2)$, together with the proof that this complex is Cohen-Macaulay for every pair. Its finely graded $K$-polynomial is shown to be bivaluative and equal to an Euler-characteristic expression in exterior powers of dual tautological quotient classes, and Hochster's formula converts the top-degree coefficients of that polynomial into the homology sums above.","pith_inferences":["The proof suggests a general recipe: matroid invariants expressible as Euler characteristics of tautological classes might be proved nonnegative by finding Cohen-Macaulay initial degenerations whose links compute them; the same bivaluativity bridge could be used for other coefficients of the $g$-invariant, not just the top one.","The authors conjecture that $\\Delta_w(M_1,M_2)$ is shellable; if true, the homology groups in the formula for $\\omega(M)$ would have integer coefficients and likely explicit bases, connecting to the external-activity bases used for single matroids.","The log-concavity of external activity counts in the expected-rank case has combinatorial content independent of the tropical f-vector conjecture, and may extend to the full bivariate activity distribution rather than just one marginal.","The tropical cell complex dual to $\\Delta_w(M_1,M_2)$, built from intersections of translated Chern class fans, appears to carry enough structure to encode the full minimal free resolution of the Stanley-Reisner ring, which would give finer homological information than the top-coefficient formula."],"forward_implications":["The number of $(n-i)$-dimensional interior faces in any subdivision of $\\Sigma(r,n)$ into matroid base polytopes is at most $(n-i-1)!/((r-i)!(n-r-i)!(i-1)!)$.","Every matroid $M$ satisfies $\\omega(M)\\geq 0$, with $\\omega(M)$ realized as a sum of Betti numbers of links in $\\Delta_w(M,M)$.","All coefficients of the $g$-invariant are nonnegative for every matroid, not only for matroids realizable over $\\mathbb{C}$.","For pairs with $D(M_1,M_2)$ of expected rank, the sequence counting bases of $D(M_1,M_2)$ by external 1-activity is log-concave.","The $K$-polynomial identity yields the positivity statement $(-1)^{\\operatorname{rank}D}\\chi(\\wedge^p[Q^\\vee_{M_1}]\\cdot\\wedge^q[Q^\\vee_{M_2}])\\geq 0$ whenever $p+q=n$.","The Cohen-Macaulay property and bivaluativity of the external activity complex give a matroidal formula for the higher cohomology of exterior powers of dual tautological bundles when the pair is realizable."],"supporting_citations":[{"why":"States the tropical f-vector conjecture and defines the matroid base polytope subdivisions whose face counts are bounded.","marker":"[Spe05]"},{"why":"Introduces the $g$-invariant and proves the conjecture for realizable matroids via Kawamata-Viehweg vanishing, the case the present paper extends.","marker":"[Spe09]"},{"why":"Supplies the external reduction used in the final step: nonnegativity of $\\omega$ on all minors implies nonnegativity of every coefficient of the $g$-invariant.","marker":"[FSS24]"},{"why":"Provides the tautological sub- and quotient classes $[S_M]$, $[Q_M]$ and the equivariant localization tools used in the $K$-polynomial formulas.","marker":"[BEST23]"},{"why":"Introduced the Schubert variety of one linear space and the external activity complex that the pair construction generalizes.","marker":"[AB16]"},{"why":"Supplies the geometric method converting the minimal free resolution of a Kempf collapsing into Euler characteristics of exterior powers.","marker":"[Wey03]"},{"why":"Provides Hochster's formula and Reisner's criterion, connecting reduced homology of links to Betti numbers and Cohen-Macaulayness.","marker":"[MS05]"},{"why":"Supplies Cartwright-Sturmfels ideal machinery used to identify initial ideals and transfer Betti numbers from realizable pairs to arbitrary pairs.","marker":"[CDNG20]"},{"why":"Gives the valuative expansion of matroid base polytopes into Schubert matroids, used to prove bivaluativity and extend realizable results to all matroids.","marker":"[DF10]"}],"fun_headline_variants":["Cohomology settles Speyer's tropical f-vector conjecture","New complex proves nonnegative matroid invariants","External activity complex confirms tropical bound","Matroid pair complex settles 2005 face-count conjecture"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is the theorem, proved in [FSS24], that nonnegativity of $\\omega$ on all minors forces every coefficient of the $g$-invariant to be nonnegative; if that theorem failed, the paper would still prove $\\omega(M)\\geq 0$ but not the full tropical f-vector conjecture.","fun_headline_variants_meta":{"raw":{"variants":["Cohomology settles Speyer's tropical f-vector conjecture","New complex proves nonnegative matroid invariants","External activity complex confirms tropical bound","Matroid pair complex settles 2005 face-count conjecture"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000629,"raw_usage":{"total_tokens":2896,"prompt_tokens":926,"completion_tokens":1970,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":542,"completion_tokens_details":{"reasoning_tokens":1908}},"tokens_in":542,"tokens_out":1970,"duration_ms":11804,"temperature":1.0,"reasoning_tokens":1908,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T14:38:27.977222+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute both sides of Equation (1.1) for a connected non-realizable matroid such as the non-Fano matroid: if the sum of reduced homology dimensions of the links in $\\Delta_w(M,M)$ differs from $\\omega(M)$ computed from the exterior-power definition of the $g$-invariant, the central formula is false.","supporting_citations":[],"review_version":1}