{"id":"a059a7bd-b6e5-4111-a3ac-02cd549e7297","arxiv_id":"2504.16235","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A weight-sum condition, called (T), exactly characterizes when the Kirwan blow-up and the toroidal compactification of each of the 85 Deligne-Mostow varieties are naturally isomorphic.","lead":"This paper classifies all 85 Deligne-Mostow spaces and gives a simple weight rule for when two natural ways of adding boundary to each space give the same result. The rule settles, case by case, a question that had previously been answered only for special families, and it also changes the known birational-geometry status of the nonmatching cases.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proposition 3.5(1) is false: the paper's own Table 4 contains ordered Deligne–Mostow pairs where (T) holds for the smaller pair and fails for the larger pair, invalidating the reduction method.","rationale":"The reader's conditional verdict focused on the omitted |T1|=3 and 5 local computations in Proposition 2.2(2). That is a legitimate gap, but the stress-test found a more decisive and checkable obstruction: Proposition 3.5(1), a key step in the paper's reduction method, is false. The two pairs used as a counterexample are both listed in the paper's own Table 4, with opposite (T) status, and they satisfy the partial order in Definition 3.1 after sorting weights as the definition requires. This invalidates Reduction Method 3.8, the extremal classification, and the statement that it suffices to prove the main theorem for minimal and maximal elements. I did not find a counterexample to Theorem 2.6 itself, whose direct proof in Section 2 does not rely on Proposition 3.5; the local computation gap flagged by the reader still needs to be checked. However, a false proposition supporting an advertised main contribution means the manuscript cannot be accepted as is. A major revision that corrects or removes the reduction claim could change this assessment.","tokens_in":21201,"tokens_out":26364,"duration_ms":259077,"concrete_test":"Recompute condition (1.3) for rows (5, 32111, N{3,5}) and (8, 11111111, N3) of Table 4 using sorted representatives, and verify Definition 3.1 gives (5, 32111, N{3,5}) ≺ (8, 11111111, N3). The first satisfies (T), the second violates it, falsifying Proposition 3.5(1).","verdict_should_be":"REJECT","load_bearing_attack":"Proposition 3.5(1) is contradicted by rows of Table 4. Let A be the pair (5, 32111, N{3,5}), i.e. weights (3/4, 1/2, 1/4, 1/4, 1/4) with S={3,4,5}; its sorted representative is (1/4, 1/4, 1/4, 1/2, 3/4) with S={1,2,3}. For A the only subset T1⊂S with |T1|≥3 is T1=S, whose weight sum is 3/4; the complement {1/2,3/4} has no subset of weight 1/4, so (T) holds, as Table 4 records. Let B be (8, 11111111, N3), i.e. eight 1/4-weights with S={1,2,3}. Taking T1=S and T2={4} gives total weight 1, so (T) fails, also as Table 4 records. Yet Definition 3.1 gives A≺B: 5≤8, each weight of B is ≤ the corresponding weight of A for i=1,...,5, |S|=|S'|=3, and w(S)=w'(S')=3/4. Thus Proposition 3.5(1) is false, and Reduction Method 3.8 together with the extremal counts in Table 2 are unsupported as stated. Theorem 2.6 is proved in Section 2 without Proposition 3.5, so the iff criterion may survive; but the advertised reduction to minimal/maximal elements is not valid.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies Deligne-Mostow pairs (w,S), where w is a weight vector for n points on P^1 and S is a subset of indices acted on by a symmetric group. The main result, Theorem 2.6, gives a numerical criterion (T) for the Deligne-Mostow period map to extend to an isomorphism between the Kirwan blow-up M^K_{w,S} and the toroidal compactification X^T_{w,S}. The paper also introduces a partial order on the 85 Deligne-Mostow pairs and claims, in Reduction Method 3.8, that it suffices to verify Theorem 2.6 on minimal and maximal elements. Finally, the paper derives consequences for the log minimal model program, showing in Corollary 1.9 that when (T) fails the Kirwan blow-up is not a semi-toroidal compactification and not log K-equivalent to the toroidal compactification.","tokens_in":21529,"tokens_out":5994,"duration_ms":54995,"significance":"If Theorem 2.6 is correct, it completely and uniformly settles the natural-isomorphism question for all 85 Deligne-Mostow pairs, extending the earlier results of Gallardo-Kerr-Schaffler, Hulek-Maeda, and Hulek-Kondo-Maeda. The criterion (T) is genuinely easy to check, and the tables in Section 5 are a useful census of the Deligne-Mostow universe. The LMMP consequences in Section 4 are natural applications of the main criterion. However, the current manuscript contains a false statement in the reduction method and an incomplete verification of a load-bearing local computation, so the advertised significance is not yet fully established.","major_comments":[{"comment":"Proposition 3.5(1) is false as stated. Let A be the pair (5, (3/4,1/2,1/4,1/4,1/4), N{3,5}), written with sorted weights (1/4,1/4,1/4,1/2,3/4) and S={1,2,3}, and let B be (8, (1/4)^8, N3). Then A≺B by Definition 3.1: 5≤8, each weight of B is at most the corresponding weight of A, |S|=|S'|=3, and w(S)=w'(S')=3/4. For A, the only subset T1⊂S with |T1|≥3 is T1=S, whose weight sum is 3/4, and no subset of the complement {1/2,3/4} has weight 1/4, so (T) holds; indeed Table 4 records this row as T. For B, taking T1=S and T2={4} gives total weight 1, so (T) fails; Table 4 also records this row as NT. This directly contradicts Proposition 3.5(1). Consequently, the proof of Reduction Method 3.8 and the extremal counts in Table 2 are not justified as they stand. The assertion in the proof that the claim follows from |S|=|S'| and w(S)=w'(S') is insufficient, because condition (T) also involves weights outside S, and these may decrease under ≺.","section":"§3.2, Proposition 3.5(1)"},{"comment":"The proof of Proposition 2.2(2) is incomplete in a load-bearing way. The non-transversality direction requires a local blow-up computation in a Luna slice, but the cases |T1|=3 and |T1|=5 are only asserted to follow from \"a similar computation,\" while the cases |T1|=4 and |T1|=6 are taken from earlier papers. Since Theorem 2.6's if-and-only-if statement depends on Proposition 2.2(2), the missing cases are not a cosmetic gap. The authors should either provide the full computation for |T1|=3 and |T1|=5, or supply a precise reference where these cases are proved. Since the computation is described as \"computer-based,\" an appendix with explicit coordinates or reproducible code would also be appropriate.","section":"§2, Proposition 2.2(2)"},{"comment":"Because Proposition 3.5(1) fails, the claimed reduction to minimal and maximal elements needs to be re-examined even if Theorem 2.6 is independently proved in Section 2. The reduction method relies on the monotonicity of (T) along ≺ to guarantee that a minimal element below a non-(T) pair is itself non-(T), and that a maximal element above a (T) pair is itself (T). The counterexample in Proposition 3.5(1) shows that such monotonicity does not hold in general. The authors should state precisely which weaker form of Reduction Method 3.8 remains valid, or prove the extremal reduction by a different argument that does not use Proposition 3.5(1).","section":"§3, Reduction Method 3.8 and Theorem 3.7"}],"minor_comments":[{"comment":"There are numerous typos and misspellings that should be corrected in a revision, including \"rove\" (Section 1.3), \"Propoitipon\" (Section 2), \"thar\" (Section 3.2), \"unorderend\" (Section 3.2), \"a prior\" (Section 1.1), and \"poystable\" (proof of Theorem 2.6).","section":"Throughout"},{"comment":"The definition of ≺ is given on sorted representatives of equivalence classes. The authors should state explicitly that ≺ is well-defined on the equivalence classes in ADM, since the choice of representative is not completely fixed by the condition \"we can do so without any restrictions.\"","section":"Definition 3.1"},{"comment":"The labels Max and Min in Tables 4 and 5 are not formally defined in the text. Clarify the intended definition: are these the maximal elements among pairs satisfying (T) and minimal elements among pairs not satisfying (T), or are they minimal and maximal with respect to ≺ in the whole universe ADM?","section":"Tables 4 and 5"},{"comment":"The sentence beginning \"Our approach to out to K-equivalene\" contains a typo and should read \"Our approach to K-equivalence\".","section":"Remark 1.10(2)"}],"recommendation":"major_revision","confidential_remarks":"The counterexample to Proposition 3.5(1) is not subtle: it is contained in the paper's own Table 4. This suggests the partial-order machinery was not checked against the tables. The main theorem may well be salvageable, but the reduction method as advertised is not. I would also press the authors to complete or explicitly reference the |T1|=3,5 computations in Proposition 2.2(2), since the central iff relies on them."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper has a real new result: Theorem 2.6 gives a clean, checkable numerical criterion for when the Deligne–Mostow isomorphism lifts to an isomorphism between the Kirwan blow-up and the toroidal compactification, covering all 85 partially unordered pairs. The tables are a useful census, and the LMMP byproduct is a natural consequence. The proof of Theorem 2.6 in Section 2 is independent of Section 3, and I think it may well survive.\n\nBut the reduction method does not survive. The stress-test note is correct. Take A = (5, 32111, N{3,5}) and B = (8, 11111111, N3). After sorting, A has weights (1/4,1/4,1/4,1/2,3/4) with S={1,2,3} and w(S)=1/4. B has eight 1/4's with S={1,2,3} and w(S)=1/4. Definition 3.1 gives A ≺ B: n increases, each weight of B is ≤ the corresponding weight of A, and |S| and w(S) match. Yet Table 4 correctly marks A as T and B as NT (take T1=S and T2={4} for B). So Proposition 3.5(1) is false. Consequently Reduction Method 3.8 and the extremal counts in Table 2, which rely on (T) being preserved under ≺, are unsupported. A fix might be to strengthen the partial order or to prove propagation of the isomorphism property directly instead of via (T).\n\nThe reader's other soft spot is also real: Proposition 2.2(2) delegates the cases |T1|=3 and 5 to a \"similar computation\" without showing it. That is load-bearing for the only-if direction of Theorem 2.6, and it should be spelled out.\n\nThe paper is for specialists in moduli of points, ball quotients, and Kirwan resolutions. It deserves peer review—not a desk reject—because the main criterion is likely true and the enumeration is valuable, but the reduction section needs major revision. I'd send it to a referee with the counterexample in hand.","headline":"The new (T)-criterion and the 85-pair census are genuinely useful, but the paper's reduction method is invalid: its own Table 4 gives a counterexample to Proposition 3.5(1).","tokens_in":22067,"tokens_out":4463,"would_cite":false,"duration_ms":38588,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14L24","14E30","14G35"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that for all 85 Deligne–Mostow pairs the period map extends to an isomorphism between the Kirwan blow-up and the toroidal compactification exactly when condition (T) holds.","keywords":["Deligne-Mostow varieties","ball quotients","period maps","Kirwan blow-up","toroidal compactification","configuration spaces","log minimal model program","geometric invariant theory"],"falsifier":"Perform the omitted blow-up check for a block $T_1$ of size 3 or 5 whose weights sum to 1: at the corresponding two-point degeneration, write the discriminant as a product of the binary discriminants of a degree-$|T_1|$ polynomial and the remaining factor, blow up the origin, and determine whether the strict transform and the exceptional divisor meet transversally. The theorem predicts generic non-transversality; any transversal intersection would invalidate the only-if direction.","tokens_in":20930,"feed_emoji":"🌌","tokens_out":13160,"duration_ms":122581,"temperature":0.7,"pith_summary":"This paper asks when Deligne–Mostow period maps—built from the monodromy of hypergeometric differential forms on configurations of $n$ ordered or partially unordered points on $\\mathbb{P}^1$—lift from an open isomorphism to an isomorphism between two natural compactifications: the Kirwan blow-up of the GIT quotient and the toroidal compactification of the ball quotient. For the full universe of 85 Deligne–Mostow pairs, the paper proves a numerical criterion: the lift is an isomorphism exactly when condition (T) holds, meaning there is no block $T_1\\subset S$ with at least three points and no set $T_2\\subset S^{\\mathrm{c}}$ such that the weights on $T_1\\sqcup T_2$ sum to $1$. If (T) fails, neither the period map nor its inverse lifts, so the two compactifications are genuinely different. The result settles the compactification question uniformly, and a partial order on the universe reduces the verification to minimal and maximal entries.","feed_headline":"A weight check settles all 85 compactification cases","feed_subtitle":"Kirwan blow-ups and toroidal compactifications match exactly when no 3+ points carry total weight 1.","key_machinery":"The load-bearing mechanism is the comparison of two divisors locally at polystable points. On the GIT side, the Kirwan blow-up $M^K_{w,S}$ resolves the polystable locus; on the ball-quotient side, the toroidal compactification $X^T_{w,S}$ adds the cusps. Around a polystable point whose support has two points, a local transverse slice to the group orbit (a Luna slice) gives local coordinates in which the discriminant divisor is a product of a monomial and binary discriminants $\\operatorname{disc}(X^r+b_1X^{r-2}+\\cdots+b_{r-1})$; blowing up the origin, the strict transform meets the exceptional divisor normally exactly when every such $r$ is at most $2$. Condition (T) is precisely the absence of a block with $r\\ge 3$ and total weight $1$. Transversality up to finite quotients triggers the Borel extension theorem, producing the morphism, and a birational lemma using $Q$-factoriality (every divisor has a multiple that is Cartier) upgrades it to an isomorphism. A separate combinatorial device, the partial order $\\prec$, shows that a check on minimal and maximal elements suffices.","core_discovery":"In Theorem 2.6 the paper establishes the exact dichotomy for every Deligne–Mostow pair $(w,S)$. On one hand, when (T) holds, the period map $M_{w,S}\\to X_{w,S}$ extends to an isomorphism $M^K_{w,S}\\xrightarrow{\\sim} X^T_{w,S}$; the proof uses transversality of the strict transform of the discriminant with the Kirwan exceptional divisor, the Borel extension theorem, $Q$-factoriality, and a birational criterion for isomorphisms. On the other hand, when (T) fails, a local computation shows that some component of the strict transform of the discriminant meets the Kirwan exceptional divisor generically non-transversally, while the toroidal boundary always meets the discriminant transversally; since both spaces are $Q$-factorial, no lift exists in either direction. The paper also records Corollary 1.6 on quotients by the symmetric group and Corollary 1.9: in the non-(T) case the Kirwan blow-up is not a semi-toroidal compactification, the pair $(M^K_{w,S}, \\Delta^K_{w,S})$ is not a log canonical log minimal model, and it is not log $K$-equivalent to $(X^T_{w,S}, \\Delta^T_{w,S})$.","pith_inferences":["The same numerical criterion could be applied to other ball-quotient period maps, such as the generalized configurations mentioned in the paper's closing remarks; if the local normal forms are unchanged, the isomorphism question would again reduce to a weight-sum check.","The reduction to minimal and maximal elements makes the 85-case classification mechanically checkable: regenerate the weight tables, evaluate (T), and propagate; any mismatch with the paper's tables would indicate a hidden assumption in the enumeration.","The distinction between $|T_1|\\le 2$ and $|T_1|\\ge 3$ suggests that the local singularity type—an $A_1$ degeneration versus a worse one—is the actual geometric invariant behind the isomorphism statement; this could be tested by comparing exceptional-divisor geometry across pairs sharing the same $S$."],"forward_implications":["The classification is complete: for all 85 Deligne–Mostow equivalence classes, one can decide the natural-isomorphism question by checking (T) on the weights, with no further geometric input.","Whenever $|S|\\le 2$, condition (T) holds automatically, so the Kirwan blow-up and the toroidal compactification are always naturally isomorphic and the symmetric-group quotient commutes with the Kirwan blow-up.","Whenever (T) fails, the Kirwan blow-up is not a semi-toroidal compactification and is not log $K$-equivalent to the toroidal compactification, so it cannot serve as a log canonical log minimal model of the ball quotient.","By the reduction method, verifying the extremal elements (34 in the tables) is sufficient: failure at a minimal element propagates upward, and success at a maximal element propagates downward along $\\prec$."],"supporting_citations":[{"why":"Constructs the ordered-case period maps and the GIT-to-Baily-Borel isomorphism that the paper seeks to extend.","marker":"[DM86]"},{"why":"Builds the partially unordered period maps and the unitary group $\\Gamma_{w,S}$ used throughout.","marker":"[Mos86]"},{"why":"Proves the ordered-case isomorphism between Kirwan blow-up and toroidal compactification and supplies the weight tables defining the 85 pairs.","marker":"[GKS21]"},{"why":"Introduces the partial desingularization whose exceptional divisors are the subject of the transversality analysis.","marker":"[Kir85]"},{"why":"Identifies the Kirwan blow-up with the moduli of weighted pointed stable curves, connecting the GIT and moduli viewpoints.","marker":"[KM11]"},{"why":"Supplies the extension theorem that lifts the period map once local normal-crossing transversality is known.","marker":"[Bor72]"},{"why":"Carries out the Gaussian side of the local computations and the eight-point case that the present theorem generalizes.","marker":"[HM25]"},{"why":"Carries out the Eisenstein side of the local computations and the twelve-point case, including the LMMP statements adapted in Section 4.","marker":"[HKM24]"},{"why":"Provides one of the local normal-form computations used in the proof of Proposition 2.2.","marker":"[CMGHL23]"}],"fun_headline_variants":["Weight check settles 85 Deligne-Mostow cases","Triple weight 1 breaks Kirwan-toroidal match","85 compactifications: isomorphism iff no weight-1 triple","No weight-1 triple: Kirwan and toroidal agree"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof that non-isomorphism really happens exactly when condition (T) fails relies on a local blow-up calculation at degenerations with three or five coincident points; for those two sizes the paper states the calculation is the same as the published four- and six-point cases but does not show it.","fun_headline_variants_meta":{"raw":{"variants":["Weight check settles 85 Deligne-Mostow cases","Triple weight 1 breaks Kirwan-toroidal match","85 compactifications: isomorphism iff no weight-1 triple","No weight-1 triple: Kirwan and toroidal agree"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000216,"raw_usage":{"total_tokens":1484,"prompt_tokens":1046,"completion_tokens":438,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":662,"completion_tokens_details":{"reasoning_tokens":369}},"tokens_in":662,"tokens_out":438,"duration_ms":4391,"temperature":1.0,"reasoning_tokens":369,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T11:09:00.819812+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Perform the omitted blow-up check for a block $T_1$ of size 3 or 5 whose weights sum to 1: at the corresponding two-point degeneration, write the discriminant as a product of the binary discriminants of a degree-$|T_1|$ polynomial and the remaining factor, blow up the origin, and determine whether the strict transform and the exceptional divisor meet transversally. The theorem predicts generic non-transversality; any transversal intersection would invalidate the only-if direction.","supporting_citations":[],"review_version":1}