{"id":"99b2e094-b882-4750-a7af-d73138f19570","arxiv_id":"2608.02445","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"P_G^± is Minkowski decomposable if and only if G is K_n, K_{2,n−2}, or K_{1,1,n−2}.","lead":"This paper classifies exactly which graphs give symmetric edge polytopes that can be written as the Minkowski sum of two smaller polytopes. Only complete graphs, complete bipartite graphs with one part of size 2, and complete tripartite graphs with two singleton parts are decomposable; all others are indecomposable.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 'only if' direction of Theorem 1.1 rests on Lemma 4.2(b)–(e), whose displayed transition sequences in T(G) are asserted without verification; if any intermediate triple violates (A2)/(A3), Proposition 4.1 and the classification collapse.","rationale":"I read the paper in good faith and identified what the central claim requires: for every connected graph not isomorphic to K_n, K_{2,n−2}, or K_{1,1,n−2}, the symmetric edge polytope is Minkowski indecomposable. The proof strategy is sound in outline: use McMullen's criterion with a strongly connected family of triangular faces. The explicit decompositions for the three exceptional families in Section 5 check out, the cycle case in Proposition 3.4 is correct, and the final touching argument via facet subgraphs is standard. The one place where the proof genuinely depends on an unverified assertion is Lemma 4.2(b)–(e). These local switches are crucial: Proposition 4.1 uses them in every branch to rule out the existence of a good vertex unless the graph is exceptional. Without explicit verification of the displayed transitions, the proof has a gap. I considered whether the transitions are routine enough that 'can be verified in the same way' suffices, but the sequences are sufficiently complex—especially case (d) with its three subcases—that an independent check is warranted. I also looked for other potential issues, such as the use of [1, Theorem 3(2)] for the touching argument or the difference-constraints construction in Proposition 2.2, but these are either well-cited or internally consistent. Therefore the load-bearing concern is precisely the one the reader identified. Since the theorem is likely true and the gap is fillable, the appropriate verdict remains CONDITIONAL.","tokens_in":10488,"tokens_out":22330,"duration_ms":199752,"concrete_test":"Write a script that, for each of the five cases in Lemma 4.2 and every displayed transition, verifies conditions (A1)–(A3) for each intermediate triple under the stated local assumptions. Treat the depicted configuration as an induced subgraph of an arbitrary graph: all edges not explicitly forbidden in the case description are allowed, and for each triple check whether any completion of the unspecified edges creates a directed cycle of length 3 or 4 containing two of the triple's directed edges (violating (A2)) or a directed cycle of length 5 or 6 containing all three (violating (A3)). If all triples pass for all completions, Lemma 4.2 is certified; if any triple fails, Proposition 4.1 has a counterexample and the proof of Theorem 1.1 needs revision.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The proof of Proposition 4.1—the heart of the 'only if' direction—shows that if no vertex is 'good', the graph must be one of the three exceptional families. The argument forces this conclusion by repeatedly invoking Lemma 4.2 to rule out local configurations. Lemma 4.2(a) is verified in detail, but cases (b)–(e) are dismissed with 'can be verified in the same way as in (a)' and no verification is supplied. The displayed transition sequences are long and involve intermediate triples that mix edges incident to u with edges far away; for instance, in case (b) the triple {au, qc, qr} must be checked to contain no pair lying on a common directed 4-cycle, which requires using the assumptions q,r ∉ N_G[u] and qc, qr ∈ E(G). Case (d) has three subcases, each with a multi-step sequence. If any of these intermediate triples violates (A2) or (A3), the good-vertex conclusion fails at that step, and the proof of Proposition 4.1 loses its footing. Since Theorem 1.1's indecomposability claim for all non-exceptional graphs depends directly on the existence of a good vertex (to build a strongly connected family of triangular faces touching every facet), this unproved combinatorial assertion is genuinely load-bearing. The issue is not a stylistic preference for more details; it is that the correctness of the classification is contingent on transitions that the text does not actually establish.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper gives a complete classification of Minkowski decomposability for symmetric edge polytopes of connected graphs: for a connected graph G with n ≥ 3, P_G^± is Minkowski decomposable if and only if G is isomorphic to K_n, K_{2,n-2}, or K_{1,1,n-2}. The proof uses a graph-theoretic description of triangular faces (Conditions (A1)-(A3)), McMullen's indecomposability criterion via strongly connected families of triangular faces touching all facets, and a transition analysis showing that every 2-connected non-exceptional graph has a 'good' vertex. Explicit Minkowski decompositions are provided for the three exceptional families.","tokens_in":10832,"tokens_out":23535,"duration_ms":208169,"significance":"If correct, the result is a clean and satisfying classification. It connects Minkowski decomposability of symmetric edge polytopes to earlier work on edge counts, and the use of McMullen's criterion is well suited to the problem. The explicit decompositions in Section 5 and the facet-subgraph argument in Section 6 are valuable. The main combinatorial engine, Proposition 4.1, is structurally convincing, but its correctness is contingent on Lemma 4.2(b)-(e), whose verification is asserted rather than supplied. This is a genuine, load-bearing gap, although my own spot-checks of the displayed transitions did not reveal an actual error.","major_comments":[{"comment":"The five local-switch cases (b)-(e) are used repeatedly in Proposition 4.1 to force the three exceptional graphs when no good vertex exists. For each case the paper displays a transition sequence and states only that validity 'can be verified in the same way as in (a)'. No verification of Conditions (A2) or (A3) for the intermediate triples is provided. Since Proposition 4.1 is the heart of the 'only if' direction of Theorem 1.1, any invalid intermediate triple would break the good-vertex argument and the classification. This is not a stylistic matter. The authors should give a complete proof, or at least a table of pairwise (A2)/(A3) checks for every intermediate triple in cases (b)-(e), including the three subcases of (d).","section":"§4, Lemma 4.2(b)-(e)"},{"comment":"The application of Lemma 4.2(d) at the point 'If some a ∈ N_G(w)∩N_G(u) had a neighbour in N_G(u)' assumes the existence of the vertex b in the statement of Lemma 4.2(d). In the application this b exists because N_G(w)∩N_G(u) has at least two elements and |N_G(u)| ≥ 3, but the lemma as stated should justify the choice 'Choose b ∈ N_G(u)\\setminus{a,c} so that wb∈E(G) or wc∈E(G)'. Without this clarification, the three cases in (d) do not cover all possibilities as cleanly as claimed.","section":"§4, Proposition 4.1 (case V(G)\\N_G[u] nonempty)"}],"minor_comments":[{"comment":"In the displayed definition of c(e), the middle line should read '1, e ∈ T' (i.e., the reverse directed edge lies in T). As typeset, it is indistinguishable from the first line.","section":"§2, Proposition 2.2 proof"},{"comment":"The connectedness argument for the sign patterns is compressed, particularly for n = 5, 6. The statement that it is enough to consider sign patterns for a fixed I after 'ignoring signs' should say explicitly that adjacency between triples with different underlying index sets preserves two signs. This is true, but it is not immediate from the Johnson graph sentence alone.","section":"§3, Proposition 3.4"},{"comment":"The proof uses two nontrivial facts without elaboration: that facets of a free sum of symmetric edge polytopes are joins of facets of the summands, and that [10, Theorem 3] applies to make these facets indecomposable. A sentence or precise reference for each would help the reader.","section":"§3, Corollary 3.3"},{"comment":"The claim 'Every such directed edge belongs to some member of F^- ∪ F^+' relies on deg(u) ≥ 3. This is true in the application, but it could be stated explicitly.","section":"§6, proof of Theorem 1.1"}],"recommendation":"major_revision","confidential_remarks":"The sole substantive obstacle is the unproved Lemma 4.2(b)-(e). I did not find an actual error in the displayed transitions, and the overall structure of the proof is credible. If the authors supply the missing verifications and clarify the small points above, I would support acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Hi — quick take on Higashitani–Mori. The main theorem is a clean, complete classification: P_G^± is Minkowski decomposable iff G is K_n, K_{2,n−2}, or K_{1,1,n−2}. That answers a natural question and fits neatly alongside Codenotti–Riccardi–Venturello's edge-count result. The proof uses McMullen's criterion via strongly connected families of triangular faces, which is the right tool, and the triangular-face characterization in §2 (with the difference-constraints argument) is well done. The explicit decompositions in §5 are concrete and check out.\n\nNow the soft spot. The 'only if' direction leans on Proposition 4.1, which in turn uses the five local-switch lemmas. Lemma 4.2(a) is proved in detail. Cases (b)–(e) are dismissed with 'can be verified in the same way as in (a)' and the transition sequences are just displayed. That is the load-bearing part: if any intermediate triple violates (A2) or (A3), the good-vertex conclusion fails and the classification argument collapses. I spent some time on cases (b) and (d); the sequences do appear to satisfy the conditions under the stated assumptions, and the method from (a) generalizes. But the text doesn't demonstrate it, and a conscientious referee will have to redo a lot of tedious checking. This is not a fatal flaw — more a missing appendix. The authors should either work through the checks or provide a small computational verification (the conditions are finitary).\n\nAlso minor: Proposition 3.4's 'we can extend e to some T' for cycles is a bit quick but fine; and the facet-touching argument in the main theorem relies on the facet-subgraph description, which is cited correctly. The small cases n=3,4 are handled.\n\nOverall: the result is plausible and likely true; the writing is clear; the references and citations look appropriate. The gap is real but repairable. I'd send it to a serious referee, with the expectation that the authors fill in Lemma 4.2. It's worth a reading group slot for anyone in lattice polytopes.","headline":"The classification is almost certainly correct and the proof outline is sound, but the omitted verification in Lemma 4.2(b)–(e) is an expositional gap that should be fixed before publication.","tokens_in":11344,"tokens_out":4062,"would_cite":true,"duration_ms":41230,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["52B12","52B20","05C75"],"pacs":[],"model":"deepseek-v4-flash","headline":"Symmetric edge polytopes decompose exactly for three complete multipartite graphs.","keywords":["Minkowski decomposability","symmetric edge polytopes","complete multipartite graphs","triangular faces","lattice polytopes","2-connected graphs","polytope indecomposability"],"falsifier":"Take a small 2-connected graph not isomorphic to the three exceptional families, compute the set T(G) of directed-edge triples satisfying (A1)–(A3), and check whether the displayed sequences in Lemma 4.2(b)–(e) remain inside T(G) at every step; a single violation would invalidate the proof. For the theorem itself, a connected graph outside the three families whose P_G± admits a nontrivial Minkowski decomposition would refute the classification.","tokens_in":10351,"feed_emoji":"🧩","tokens_out":8803,"duration_ms":73244,"temperature":0.7,"pith_summary":"A polytope is Minkowski decomposable when it can be written as the set of all sums q+r with q in one polytope and r in another, neither a mere scaled copy of the original. The paper settles this question completely for symmetric edge polytopes, the convex hulls of differences e_i−e_j coming from the edges of a graph. The central claim is that a connected graph's symmetric edge polytope is Minkowski decomposable exactly for three complete multipartite graphs: K_n, K_{2,n−2}, and K_{1,1,n−2}. For every other connected graph the polytope is indecomposable. The proof characterizes triangular faces as triples of directed edges obeying three conditions, assembles these triangles into a strongly connected family touching every facet for every non-exceptional 2-connected graph, and gives explicit decompositions for the three exceptions.","feed_headline":"Only three graphs give decomposable symmetric edge polytopes","feed_subtitle":"For every other connected graph, the polytope cannot be written as a nontrivial sum of two polytopes.","key_machinery":"The central object is the symmetric edge polytope P_G± = conv{±(e_i−e_j) : {i,j} in E(G)}, a centrally symmetric lattice polytope in the hyperplane sum x_i = 0. The load-bearing mechanism is the set T(G) of triples of directed edges satisfying conditions (A1), (A2), and (A3); by the paper's Proposition 2.2, exactly these triples give triangular faces. The proof then uses transitions between such triples, swapping one directed edge at a time, to build a strongly connected family of triangular faces that touches every facet. Together with the paper's Theorem 3.2, a standard criterion saying that a polytope with such a family is indecomposable, this forces indecomposability. The only graphs for","core_discovery":"The discovery, on the paper's own terms, is a complete classification: for a connected graph G with at least three vertices, the symmetric edge polytope P_G± is Minkowski decomposable if and only if G is K_n, K_{2,n−2}, or K_{1,1,n−2}. The engine of the proof is a characterization of the two-dimensional faces: a triple of directed edges forms a triangular face exactly when it contains no opposite pair, no two of its edges lie on a directed cycle of length 3 or 4, and the three edges are not together on a directed cycle of length 5 or 6. Using this characterization, the paper shows that every 2-connected graph outside the three exceptional families has a strongly connected family of triangula","pith_inferences":["If the classification is right, Minkowski indecomposability is the generic behavior for symmetric edge polytopes: the decomposable cases sit at three high-symmetry complete multipartite families, so almost every connected graph yields an indecomposable polytope.","The appearance of the same three families in the edge-count lower-bound result suggests a possible principle: for symmetric edge polytopes, Minkowski decomposability may occur exactly when the one-dimensional face structure is as sparse as allowed; testing whether this principle extends to related polytope classes would be a natural next step.","The four local transitions in Lemma 4.2 stated without detailed verification are the natural place to stress-test the proof; a computer search over small 2-connected graphs checking that every displayed triple satisfies (A1)–(A3) would either confirm the classification or expose a gap."],"forward_implications":["For every connected graph outside the three families, P_G± is Minkowski indecomposable, including all non-2-connected graphs and all cycles of length at least 5.","The exceptional graphs are exactly those attaining equality in a sharp lower bound on the number of edges of P_G± proved in a separate result, a coincidence the paper highlights as evidence of a common mechanism.","The complete graph decomposition P_K_n± = Δ_{n−1} + (−Δ_{n−1}) expresses the polytope as a sum of a simplex and its negative, showing that one exceptional family decomposes in the simplest possible way.","For K_{2,n−2} and K_{1,1,n−2}, the explicit decompositions make the classification constructive: the summands are written down, not merely asserted to exist.","The triangular-face characterization gives a finite, checkable list of conditions for when three vertices of P_G± form a face, which can be reused in further studies of the polytope's face structure."],"fun_headline_variants":["Three and only three: decomposable symmetric edge polytopes","Only three graphs make symmetric edge polytopes decomposable","Symmetric edge polytopes decompose for exactly three graph types","Decomposable edge polytopes: just the complete multipartite trio"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that Lemma 4.2's five local transitions are all valid: the paper verifies case (a) and asserts that cases (b)–(e) 'can be verified in the same way' without supplying the verification, and the only-if direction of the main classification depends on every one of those transitions staying inside the set of valid triangular faces.","fun_headline_variants_meta":{"raw":{"variants":["Three and only three: decomposable symmetric edge polytopes","Only three graphs make symmetric edge polytopes decomposable","Symmetric edge polytopes decompose for exactly three graph types","Decomposable edge polytopes: just the complete multipartite trio"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00129,"raw_usage":{"total_tokens":5067,"prompt_tokens":671,"completion_tokens":4396,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":415,"completion_tokens_details":{"reasoning_tokens":4323}},"tokens_in":415,"tokens_out":4396,"duration_ms":32027,"temperature":1.0,"reasoning_tokens":4323,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-04T07:13:39.746357+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a small 2-connected graph not isomorphic to the three exceptional families, compute the set T(G) of directed-edge triples satisfying (A1)–(A3), and check whether the displayed sequences in Lemma 4.2(b)–(e) remain inside T(G) at every step; a single violation would invalidate the proof. For the theorem itself, a connected graph outside the three families whose P_G± admits a nontrivial Minkowski decomposition would refute the classification.","supporting_citations":[],"review_version":1}