{"id":"0f0ea7b5-ce64-4c14-bb6f-16288709241d","arxiv_id":"2506.13603","paper_version":5,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Under a constant anti-canonical volume condition on a Zariski-dense set of Fano type fibers, the geometric generic fiber is Fano type; the dimension-2 case is unconditional.","lead":"This paper proves that a family of Fano type varieties has a Fano type generic fiber when all the good fibers have the same anti-canonical volume, and it proves the surface case with no volume condition. It also proves a reduction mod p version of the Schwede and Smith conjecture under the same constant volume assumption.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The (b)⇒(a) direction of Theorem 1.2 rests on the unproved assertion that the restricted sequence (5.4) is a D-MMP on every closed fiber; Remark 5.5 concedes this and cites support only over C.","rationale":"The reader's weakest assumption matches the point I would stress: the converse direction of Theorem 1.2 depends on the restricted sequence (5.4) being a D-MMP on every closed fiber, and Remark 5.5 admits this is only anticipated while the cited support is over C. This is genuinely load-bearing because Lemma 5.3 and Lemma 5.4 are proved only for actual D-MMP steps, and an arbitrary spread of a generic MMP step need not be an MMP step on a closed fiber; without it, the constancy of anti-canonical volumes on all fibers, and hence the full equivalence stated in Theorem 1.2, does not follow. The gap is addressable rather than a contradiction, so it supports the same conditional verdict rather than rejection. I also note the external dependence on Jiao's volume theorem [30] in the (a)⇒(b) direction, but that direction is less problematic: the volume statement is plausible, the paper proves a mod p analog, and the explicit self-acknowledged MMP restriction is the sharper internal gap. The dimension-2 theorem and the positive-characteristic theorem follow the cited arguments and appear substantially sound, which further supports keeping the paper under conditional status pending the missing MMP-spreading lemma.","tokens_in":23829,"tokens_out":23592,"duration_ms":258923,"concrete_test":"Write out the missing step in (b)⇒(a): after shrinking S and taking a finite cover, prove from [9, Lemma 5.5] or by an algebraic argument that the spread of any −(K_{X_η}+Δ_η)-MMP restricts to a −(K_{X_s}+Δ_s)-MMP on every closed fiber, over every algebraically closed field of characteristic zero and not only over C. A decisive concrete check is to take a family over a curve defined over Q̄, run a generic threefold flip in the MMP, and verify on a closed fiber whether the induced rational map is a flip of that fiber; if some fiber fails, the proof has a counterexample, and if all pass, the missing hypothesis reduces to a field-independent relative MMP statement.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The weakest load-bearing point is in the proof of (b)⇒(a) of Theorem 1.2, around display (5.4). After running a −(K_{X_η}+Δ_η)-MMP on the generic fiber and spreading the maps over S, the proof asserts that for every closed fiber the restricted sequence X_{0s}⇢X_{1s}⇢...⇢X_{ns} is a sequence of −(K_{X_{(i-1)s}}+Δ_{(i-1)s})-negative maps, and then concludes that each X_{is}, and hence Y_s, is Fano type by Lemma 5.3 and Lemma 5.4. Those lemmas are only proved for actual D-MMP steps; an arbitrary D-negative birational map need not preserve Fano type, and a generic fiber MMP step need not restrict to an MMP step on every closed fiber because the contracted locus, the extremal ray, and the induced flip can behave differently on special fibers. The proof gives no argument for this fiberwise MMP statement, and Remark 5.5 explicitly says the assertion is only 'anticipated', referring to [9, Lemma 5.5] with S a variety over C. Theorem 1.2 is stated for arbitrary algebraically closed fields of characteristic zero; no reduction from that generality to C is supplied, and the cited lemma is not reproduced. Without (5.4) being an MMP on fibers, the constancy of vol(−(K_{X_s}+Δ_s)) asserted in the final paragraph of the proof does not follow from the displayed arguments. This is not a contradiction, but it is a load-bearing unproved premise of the stated equivalence.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies generic invariance of the Fano type property in families of projective pairs. Theorem 1.2 asserts equivalence between (a) existence of a Zariski-dense set of closed fibers that are Fano type with constant anti-canonical volume v>0, and (b) the geometric generic fiber is Fano type, over any algebraically closed field of characteristic 0. Theorem 1.3 proves Conjecture 1.1 for Fano type surfaces. Corollary 1.4 derives the DCC property for anti-canonical volumes of bounded families of Fano type surfaces. Theorem 1.6 gives a reduction-mod-p analogue for globally F-regular type. The proofs combine a log canonical threshold criterion for Fano type (Lemma 5.2), quasi-monomial valuation computations after Xu, and volume semicontinuity results (Theorem 4.2 from Jiao and Theorem 4.3). The reverse direction of Theorem 1.2 spreads a generic-fiber MMP to all fibers and uses Lemmas 5.3 and 5.4.","tokens_in":24098,"tokens_out":7362,"duration_ms":66427,"significance":"If the proofs can be completed, Theorem 1.2 would be a strong and useful criterion: pointwise Fano type plus constancy of anti-canonical volume would force the generic fiber to be Fano type, and would imply Conjecture 1.1 in this case. The lct-based strategy and the positive-characteristic parallel are natural and elegant. Theorem 1.3 and the explicit counterexample in Example 5.6 are valuable contributions. The paper is not circular: the target statements are not used as inputs, and the external theorems cited are independent of the main claims. The main caveat is that the (b)⇒(a) direction relies on an unproved fiberwise MMP assertion (Remark 5.5), and the volume-semicontinuity foundation includes an unreviewed preprint and a terse proof; the stated generality is therefore not yet established.","major_comments":[{"comment":"The proof asserts that after spreading a −(K_{X_η}+Δ_η)-MMP over S, for every closed fiber the restricted sequence (5.4) is a sequence of −(K_{X_{(i-1)s}}+Δ_{(i-1)s})-negative maps, and then applies Lemma 5.3 and Lemma 5.4 to conclude that each X_{is} and Y_s is Fano type. This inference is not justified: Lemma 5.3 requires actual MMP steps, and an arbitrary D-negative birational map need not preserve Fano type or anti-canonical volumes. A generic-fiber MMP step need not restrict to an MMP step on every closed fiber, because the contracted locus, the extremal ray, and the induced flip can behave differently on special fibers. Remark 5.5 explicitly concedes that the statement is only 'anticipated' and points to [9, Lemma 5.5] for S a variety over C, whereas Theorem 1.2 is stated for arbitrary algebraically closed fields of characteristic 0. The same gap affects the final volume-constancy conclusion and therefore Corollary 1.4, and it also affects the analogous display (5.9) in the proof of Theorem 1.6. This load-bearing premise needs a proof, or the theorems need to be restricted to the setting where it is known.","section":"§5, proof of Theorem 1.2, (b)⇒(a), display (5.4); see also display (5.9) and Remark 5.5"},{"comment":"The paper relies on [30, Theorem 1.1] for the volume semicontinuity that underpins Theorem 1.2, but [30] is an unreviewed arXiv preprint and its statement is not reproduced. The proof of the analog Theorem 4.3 contains a terse and partly illegible inequality chain: after combining (4.1)–(4.3) the text writes 'A^{d-1}_s·(A_s+E_s)≤(A_s+E_s)≤A^d_s+C′√ε', where the middle term is not a number and is dimensionally inconsistent; the intended intersection-theoretic inequality must be supplied. The passage to the limit as ε and ε′ tend to 0, and in Case 1 the interchange of inf over s∈S′ with the limit t→0+, are also only sketched. Since Theorem 4.3 is the basis for Theorem 1.6 and Theorem 4.2 for Theorem 1.2, these gaps need to be closed, or the external theorem should be quoted with a complete proof.","section":"§4, Theorem 4.3 and the use of Theorem 4.2"}],"minor_comments":[{"comment":"The text contains a stray '=skip' marker between the heading 'Nakayama’s asymptotic order' and the following paragraph; this appears to be an editing artifact and should be removed.","section":"§2.4"},{"comment":"The phrase 'reduction modp' should be typeset as 'reduction mod p'; the spacing error appears in the abstract and in the introduction.","section":"Abstract and §1"},{"comment":"The 'if' direction of Lemma 5.2 is dismissed as being the same as [47, Lemma 3.1], but it is used in both main theorems; a self-contained proof or a precise statement of the referenced lemma would improve the paper.","section":"§5, Lemma 5.2"},{"comment":"In the chain of inequalities following display (5.2), the line 'vol(D_s)=vol(D_η)' uses the hypothesis that the volume is a constant v on S′ together with Theorem 4.2; this step should be spelled out, since Theorem 4.2 only gives vol(D_η)=inf_{s∈S′} vol(D_s) in general.","section":"§5, proof of Theorem 1.2, (a)⇒(b)"}],"recommendation":"major_revision","confidential_remarks":"The main theorem's reverse direction depends on an assertion the author labels 'anticipated' and refers to a preprint [9] that is not reproduced in the manuscript. The manuscript also contains a stray '=skip' marker, suggesting the arXiv version was not fully cleaned. Given the importance of Theorem 1.2, I would ask the editor to require the author either to prove the fiberwise MMP statement or to state the theorem only in the range where it is known, and to make the volume-semicontinuity input self-contained."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The genuinely new content is the constant-volume equivalence in Theorem 1.2 and the mod-p analog in Theorem 1.6. The high-level strategy is coherent: use volume semicontinuity to get bigness, then compare log canonical thresholds via quasi-monomial valuations and weighted blowups. The dimension-2 case follows Gongyo–Takagi as the author says, and Corollary 1.4 is a nice consequence. Example 5.6 is a useful counterexample to a natural DCC question. The paper is honestly written, and the reliance on Jiao's volume theorem, though an unreviewed preprint, is explicit and appears mathematically reasonable.\n\nThe main soft spot is exactly where the stress-test points: the (b)⇒(a) direction of Theorem 1.2. After spreading the generic fiber MMP to the family, the proof asserts that (5.4) is a sequence of negative maps on every closed fiber, and then uses Lemmas 5.3 and 5.4 to conclude Fano type. But a D-negative birational map on the generic fiber need not restrict to an MMP step on special fibers: contracted loci, extremal rays, and flips can behave differently. Remark 5.5 concedes the assertion is only anticipated, and the cited support is over C while the theorem is stated over any algebraically closed field of characteristic 0. No reduction is supplied. This is a load-bearing gap, not a cosmetic one, and the constancy of volumes at the end of the proof depends on it.\n\nThat said, the (a)⇒(b) direction and the dimension-2 theorem appear sound. Theorem 4.3's limit interchange is terse but standard in spirit, and the dependence on Jiao's theorem is secondary.\n\nWho is this for? Birational geometers and moduli theorists who work with Fano type properties and boundedness. The paper deserves a serious referee: the main statements are significant, the gap is addressable, and the author is upfront about it. If the fiberwise MMP claim can be proved, or the theorem restricted to C, the result would be solid. I would send it to peer review.\n\nYes, I would take it to a reading group; the gap is worth discussing, and Example 5.6 is instructive.","headline":"A plausible and useful constant-volume criterion for Fano type fibers, with a real but openly acknowledged gap in the (b)⇒(a) direction of Theorem 1.2.","tokens_in":24702,"tokens_out":1675,"would_cite":true,"duration_ms":17041,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14E05","14E30","14J45","14G17"],"pacs":[],"model":"deepseek-v4-flash","headline":"Sparse Fano type fibers whose anti-canonical volumes are constant force the geometric generic fiber to be Fano type, and the same principle works in dimension two and in reduction mod p.","keywords":["Fano type varieties","anti-canonical volume","generic invariance","log canonical threshold","quasi-monomial valuations","globally F-regular type","reduction mod p","DCC property"],"falsifier":"Take the family constructed in the proof of (b)$\\Rightarrow$(a) over an algebraically closed field of characteristic zero that is not C, and check on one closed fiber whether the restricted contractions form an MMP and whether the anti-canonical volume equals the constant v; a single fiber where the volume differs or the contractions are not an MMP would refute Theorem 1.2 in the stated generality.","tokens_in":23538,"feed_emoji":"🧮","tokens_out":9714,"duration_ms":84505,"temperature":0.7,"pith_summary":"The paper establishes that the Fano type property, which is classically known to propagate from an open set of fibers, also propagates from a Zariski-dense but possibly sparse set of fibers, provided the anti-canonical volumes on those fibers are a fixed positive constant. It proves the analogous statement without any volume condition when the Fano fibers have dimension 2. In the reduction mod p direction, it proves the same propagation when the dense fibers are globally F-regular with constant anti-canonical volume. A reader should care because this converts pointwise boundedness data into a generic geometric conclusion, and it yields a descending chain condition for anti-canonical volumes in bounded families of Fano type surfaces.","feed_headline":"Fano type spreads from sparse fibers when volume is fixed","feed_subtitle":"Constant anti-canonical volume, or dimension 2 alone, makes the Fano type property generic; a mod p version follows.","key_machinery":"The central test is the log canonical threshold criterion: a projective pair with big anticanonical class is Fano type exactly when $\\operatorname{lct}_\\sigma(X,\\Delta,-(K_X+\\Delta))>1$, where $\\operatorname{lct}_\\sigma$ is the infimum of log discrepancies divided by Nakayama's asymptotic order along valuations. To make this test work across a family, the paper uses the theorem that a minimizing valuation for the log canonical threshold is quasi-monomial, so only a single log-smooth model is needed, and Jiao's theorem that the volume of a divisor on the geometric generic fiber is the infimum of its volumes over any Zariski-dense set of closed fibers. Weighted blow-ups of the log-smooth model extract the quasi-monomial valuation as a divisor simultaneously on the generic and every closed fiber, and the volume-drop criterion for asymptotic orders then lets the author compare log canonical thresholds on closed fibers with those on the generic fiber. For the surface case, the machinery is Zariski decomposition of the anticanonical divisor, specialized to fibers and used to identify a klt boundary on the generic fiber.","core_discovery":"The paper's main equivalence is Theorem 1.2: for a projective surjective morphism $(X,\\Delta)\\to S$ between normal varieties over an algebraically closed field of characteristic zero, condition (a)---there is a Zariski-dense set $S'\\subseteq S$ and a positive constant $v$ such that every fiber $(X_s,\\Delta_s)$ for $s\\in S'$ is Fano type and $\\operatorname{vol}(-(K_{X_s}+\\Delta_s))=v$---is equivalent to condition (b), namely that the geometric generic fiber $(X_\\eta,\\Delta_\\eta)$ is Fano type. The direction (a)$\\Rightarrow$(b) is proved by showing that a log canonical threshold criterion for Fano type pairs, stated as $\\operatorname{lct}_\\sigma>1$, passes from the dense fibers to the generic fiber. The reverse direction shows that after shrinking the base, all fibers are Fano type with anti-canonical volume $v$, by running an MMP on the generic fiber and restricting its steps to fibers. The paper also proves the dimension-2 case of the conjectured generic invariance without a volume hypothesis, and a reduction-mod-p analogue in which the role of Fano type is played by global F-regularity.","pith_inferences":["The volume-constancy hypothesis may be relaxable to a DCC condition on the set of volumes: Corollary 1.4 already extracts constancy along a subsequence from boundedness, and the same mechanism may work in higher dimensions under boundedness assumptions.","The gap between the theorem and its stated generality is a relative MMP statement; proving that restricted D-MMP sequences specialize to D-MMPs over arbitrary algebraically closed fields of characteristic zero would complete the (b)$\\Rightarrow$(a) direction.","The same $\\operatorname{lct}_\\sigma$ comparison could test the mod p conjecture directly: a globally F-regular type variety with a model whose mod p fibers have constant anti-canonical volumes should be Fano type, using Theorem 1.6's method without further hypotheses.","For surfaces, the proof treats the negative part of the Zariski decomposition of $-(K_{X_\\eta}+\\Delta_\\eta)$ as a boundary after specialization; a family version of the Zariski decomposition in higher dimensions would be the natural route to extend the dimension-2 theorem."],"forward_implications":["If Theorem 1.2 stands, then a family whose Fano type fibers form any Zariski-dense subset, with all anti-canonical volumes equal to the same positive number, has a Fano type geometric generic fiber; by the reverse direction, the same volume then propagates to a neighborhood of the base.","Fano type surfaces are generically invariant with no volume hypothesis: Conjecture 1.1 holds in relative dimension two.","Anti-canonical volumes over a bounded family of Fano type surfaces satisfy the descending chain condition, so strictly decreasing infinite sequences of such volumes are impossible.","In the reduction mod p setting, constant anti-canonical volume on a Zariski-dense set of globally F-regular fibers forces the characteristic-zero limit pair to be Fano type.","The log canonical threshold criterion used here provides a practical fiberwise test for Fano type that needs only one log resolution, not all of them."],"supporting_citations":[{"why":"Supplies the theorem that the volume on the geometric generic fiber equals the infimum of volumes over any dense set of closed fibers, used to prove bigness and compare volumes in both directions of Theorem 1.2.","marker":"[30]"},{"why":"Proves that a minimizing valuation for the log canonical threshold of a graded sequence is quasi-monomial, letting the lct comparison run on a single log-smooth model.","marker":"[46]"},{"why":"Gives the log canonical threshold characterization of Fano type pairs, restated as Lemma 5.2 and applied fiberwise.","marker":"[47]"},{"why":"Provides the lemma that, over C, the restricted sequence (5.4) is a D-MMP; Remark 5.5 cites it for the (b)$\\Rightarrow$(a) direction.","marker":"[9]"},{"why":"Supplies the weighted blow-up compatibility result, Lemma 2.16, used to realize a quasi-monomial valuation as a divisor on both generic and closed fibers.","marker":"[4]"},{"why":"Supplies Lemma 2.11, the volume-drop criterion for Nakayama's asymptotic order, used to compare asymptotic orders on closed and generic fibers.","marker":"[16]"},{"why":"Gives the surface case of the conjectures and the Zariski decomposition lemmas that the proof of Theorem 1.3 follows.","marker":"[19]"},{"why":"Establishes the structure of globally F-regular pairs used to translate global F-regularity into the lct estimate in Theorem 1.6.","marker":"[40]"}],"fun_headline_variants":["Fano type generic if anti-canonical volume is constant","Constant anti-canonical volume makes Fano type generic","Dimension 2 fibers alone give generic Fano type","Fano type generic from fixed volume or dim 2","Volume constancy or dimension 2 forces Fano type generic"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that a fixed sequence of birational contraction steps (an MMP) run on the generic fiber, when restricted to each closed fiber, is still a valid MMP there and preserves both the Fano type property and the anti-canonical volume; the paper states this is anticipated and cites a proof only over C, not over arbitrary algebraically closed fields of characteristic zero.","fun_headline_variants_meta":{"raw":{"variants":["Fano type generic if anti-canonical volume is constant","Constant anti-canonical volume makes Fano type generic","Dimension 2 fibers alone give generic Fano type","Fano type generic from fixed volume or dim 2","Volume constancy or dimension 2 forces Fano type generic"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00054,"raw_usage":{"total_tokens":2548,"prompt_tokens":865,"completion_tokens":1683,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":481,"completion_tokens_details":{"reasoning_tokens":1601}},"tokens_in":481,"tokens_out":1683,"duration_ms":11659,"temperature":1.0,"reasoning_tokens":1601,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T20:04:33.936702+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the family constructed in the proof of (b)$\\Rightarrow$(a) over an algebraically closed field of characteristic zero that is not C, and check on one closed fiber whether the restricted contractions form an MMP and whether the anti-canonical volume equals the constant v; a single fiber where the volume differs or the contractions are not an MMP would refute Theorem 1.2 in the stated generality.","supporting_citations":[{"cited_title":"The volume function is upper semicontinuous on families of divisors","cited_arxiv_id":"2504.16676","evidence_quote":"Supplies the theorem that the volume on the geometric generic fiber equals the infimum of volumes over any dense set of closed fibers, used to prove bigness and compare volumes in both directions of Theorem 1.2."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Proves that a minimizing valuation for the log canonical threshold of a graded sequence is quasi-monomial, letting the lct comparison run on a single log-smooth model."},{"cited_title":"Alg´ ebrique (2023), Art","cited_arxiv_id":null,"evidence_quote":"Gives the log canonical threshold characterization of Fano type pairs, restated as Lemma 5.2 and applied fiberwise."},{"cited_title":"Variation of cones of divisors in a family of varieties -- Fano type case","cited_arxiv_id":"2504.04109","evidence_quote":"Provides the lemma that, over C, the restricted sequence (5.4) is a D-MMP; Remark 5.5 cites it for the (b)$\\Rightarrow$(a) direction."},{"cited_title":"J.171(2022), no","cited_arxiv_id":null,"evidence_quote":"Supplies the weighted blow-up compatibility result, Lemma 2.16, used to realize a quasi-monomial valuation as a divisor on both generic and closed fibers."},{"cited_title":"J.65(2016), no","cited_arxiv_id":null,"evidence_quote":"Supplies Lemma 2.11, the volume-drop criterion for Nakayama's asymptotic order, used to compare asymptotic orders on closed and generic fibers."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the surface case of the conjectures and the Zariski decomposition lemmas that the proof of Theorem 1.3 follows."},{"cited_title":"Smith,Globally F -regular and log Fano varieties, Adv","cited_arxiv_id":null,"evidence_quote":"Establishes the structure of globally F-regular pairs used to translate global F-regularity into the lct estimate in Theorem 1.6."}],"review_version":2}