{"id":"a1533009-3072-4576-9560-e7634266377c","arxiv_id":"2608.04793","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The n-birational motivic homotopy category is made into a presheaf on a correspondence category, and the birational version of a scheme with finitely many generic points splits as the product of the categories of its generic points.","lead":"This paper gives the n-birational motivic homotopy category of a scheme a systematic functorial structure, and shows that the birational version of a scheme with finitely many generic points splits into a product of categories attached to those points. The results give new homotopy-theoretic tools for detecting birational equivalence and for comparing categories attached to a field and to rational extensions of it.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 4.3.2 proves n>0 flat-affine continuity using a codimension computation that needs NUC hypotheses absent from the statement; the n=0 generic decomposition may survive, but the theorem as stated is unsupported.","rationale":"The reader correctly locates the soft spot in the second paragraph of the proof of Theorem 4.3.2, but the precise reason is not that flat maps fail to preserve codimension of arbitrary closed images; for the saturated preimage Z = phi^{-1}(C) arising in the proof, flat pullback usually preserves codimension under the dimension-formula hypotheses of Theorem 3.2.2. The real gap is that Theorem 4.3.2 is stated for arbitrary Qcqs schemes with flat affine transition maps, while the cited method relies on NUC/catenary hypotheses that are absent from the statement and not re-established. This does not appear to destroy the paper's central generic decomposition theorem, because the n=0 descent used in Corollary 4.3.4 is proved independently via dominance and constructibility, and the pro-system of dense open affines of a variety is NUC anyway. The overclaim is nevertheless substantial: the paper advertises flat-affine continuity for all n and uses the full theorem as a key structural result. I therefore retain the reader's CONDITIONAL verdict: the paper should either supply a proof of the codimension inequality under the stated hypotheses or restrict Theorem 4.3.2 to NUC pro-systems. The reliance on the unpublished preprints [Mai26a,Mai26b] for foundational identifications is secondary but reinforces the conditional recommendation.","tokens_in":39336,"tokens_out":55863,"duration_ms":463620,"concrete_test":"Check the hypotheses of the flat dimension formula used in the proof (EGA IV 13.1.1 / Stacks 02I6): if it requires locally Noetherian or universally catenary schemes, then Theorem 4.3.2's invocation of 'the methods of the second paragraph of Theorem 3.2.2' is invalid in the stated Qcqs generality. Specifically, attempt to construct a flat affine pro-system over a non-catenary Qcqs scheme in which a closed subset C has preimage Z with cod_X(Z) > cod_Y(closure(f(Z))). If such an example exists, the theorem must be restricted to NUC pro-systems (which still covers the generic-decomposition application); if no such example exists, the proof needs only an explicit reference for the dimension formula in full generality.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The proof of Theorem 4.3.2 (second paragraph) descends an n-dense open U⊂X with complement Z by setting Z1 = closure(phi_alpha1(Z)) and claiming cod_{X_alpha1}(Z1) >= cod_X(Z), citing 'the methods of the second paragraph of Theorem 3.2.2'. That citation is not automatic: Theorem 3.2.2 works under the NUC (Noetherian universally catenary) hypothesis, and its second paragraph uses the flat dimension formula for codimensions of preimages. Theorem 4.3.2, however, is stated for arbitrary Qcqs schemes with flat affine transition maps, and no argument is supplied that the dimension formula remains valid without NUC (or at least locally Noetherian) hypotheses. The equality phi^{-1}(Z1)=Z is fine because Z is a saturated preimage of a closed subset, but the codimension lower bound is the delicate point. In the non-Noetherian or non-catenary setting, flat maps can fail to preserve codimensions of preimages in the needed direction, so the descent construction for n>0 is not justified. The n=0 case is argued separately via dominance/constructibility and does not use this step, so the generic decomposition of Theorem 4.3.4 (the paper's headline) may still be correct; however, the full flat-affine continuity theorem for all n is overclaimed in its present form.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper promotes the assignment S ↦ H^n(S), the n-birational motivic homotopy category of a Qcqs scheme, to a presheaf on a category of correspondences whose vertical maps are smooth and whose horizontal maps are universally generalization-lifting (UGLT). It establishes Nisnevich descent, deformation invariance, flat-affine continuity, and, as the main structural result, a generic decomposition: for a Qcqs scheme X with finitely many generic points, H^0(X) is canonically equivalent to the product of H^0(k(η)) over the generic points η, so that in particular H^{bA1}(V) ≃ H^{bA1}(k(V)) for a variety V. The paper also derives fibrewise criteria for birational equivalences, rational and pure-transcendence invariance of H^0, and a partial unstable slice conjecture relating loop functors across birational heights.","tokens_in":39697,"tokens_out":7958,"duration_ms":75799,"significance":"If the main results hold, the generic decomposition is a substantial structural insight: it reduces the birational motivic homotopy category of a scheme to its values on residue fields of generic points, thereby connecting birational motivic homotopy theory with classical birational geometry. The paper also gives a clean framework for the functoriality of the H^n categories under UGLT maps and smooth maps, and it reproves the identification H^0 ≃ H^b rather than merely citing it. The treatment of Nisnevich descent, deformation invariance, and the partial slice conjecture are useful contributions. However, the paper depends in places on two unpublished preprints by the author, and one central continuity theorem is stated in greater generality than its proof supports; these issues need to be resolved before the results can be fully trusted.","major_comments":[{"comment":"The proof of flat-affine continuity for all n is unsupported as written. In the second paragraph, after defining Z1 = closure(φ_{α1}(Z)), the paper asserts cod_{X_{α1}}(Z1) ≥ cod_X(Z) 'by the methods of the second paragraph of Theorem 3.2.2'. That citation is not automatic: Theorem 3.2.2 is stated only for Noetherian universally catenary (NUC) schemes, and its argument uses the flat dimension formula [BE21, Theorem 2.1] and the catenary chain-length lemma [Sta26, Lemma 02I6]. Theorem 4.3.2, by contrast, is stated for arbitrary Qcqs schemes with flat affine transition maps, with no NUC or local-Noetherian hypothesis. Flat maps do not in general preserve the codimension of closed images in the needed direction outside such hypotheses, and no proof is supplied for this specific pro-system. This step is exactly what is needed to descend n-dense open immersions for n>0; the n=0 case, which only uses dominance and constructible-image arguments, may well survive, but the theorem as stated is overclaimed.","section":"§4.3, Theorem 4.3.2"},{"comment":"A similar codimension-preservation issue occurs in the proof of Proposition 4.5.1. To show that the open immersion (E×U)∖(0_X×(U∖V)) ↪ E×U is (n+d)-dense, the proof asserts that because X is smooth over S, the closed immersion 0_X×(U∖V) ↪ X×U has codimension at least d. This is a codimension formula for preimages under a smooth base-change morphism, and it again requires dimension-theoretic hypotheses (such as NUC and flatness of the relevant maps) that are not present in the statement. Without such hypotheses the asserted inequality does not follow from smoothness alone. Since this is the key geometric input for Corollary 4.5.2 and the unstable slice discussion, the result needs either a complete proof in the stated generality or an explicit additional hypothesis.","section":"§4.5, Proposition 4.5.1"}],"minor_comments":[{"comment":"The paragraph beginning 'Suppose X/S is smooth. Since PpSq is cartesian closed...' is repeated verbatim; one copy should be deleted.","section":"§2.5, Remark 2.5.4"},{"comment":"The corollary is followed by two consecutive 'Proof.' environments, the second of which is an orphaned fragment; this should be cleaned up.","section":"§3.1, Corollary 3.1.7"},{"comment":"There are numerous typographical errors, including 'continuuity' in Theorem 4.3.2, 'Frudenthal' in the Introduction, 'tow oer' on page 2, 'coresponding' on page 5, and 'schemes having finitely many irreducible schemes' in §1.3, which should read 'irreducible components'.","section":"Throughout"},{"comment":"H^8(S) is first defined as a colimit of the inclusions H^n(S) ⊆ H^{n+1}(S), but Remark 2.2.3 writes H^8(S) := lim← H^n(S); the notation should be reconciled to avoid confusion.","section":"§2.2, Remark 2.2.3"},{"comment":"The phrase 'pullback along the pro-smooth map ∐ Spec(k(η)) → X' is not literally a morphism of schemes; the proof correctly works with the pro-system of dense affine open subschemes, and the statement should be phrased in those terms.","section":"§4.3, Corollary 4.3.4"}],"recommendation":"major_revision","confidential_remarks":"The paper relies on the author's two unpublished preprints [Mai26a] and [Mai26b] for several identifications and closure properties. Since parts of this reliance are reproved in the present text, this is not disqualifying, but the editor should require that those preprints are publicly available in stable form before publication. The main technical gap concerns codimension preservation under flat and smooth maps outside the NUC setting; if the author can supply a proof or restrict the relevant theorems accordingly, the paper's central generic decomposition for H^0 appears defensible."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know two things about arXiv:2608.04793. First, the paper likely proves a real structural result: H^0(X) of the birational motivic homotopy category of a Qcqs scheme with finitely many generic points is equivalent to the product of H^0(k(eta)) over generic points. This is a nice and genuinely new fact, and it seems to survive scrutiny. Second, the paper states a stronger flat-affine continuity theorem (Theorem 4.3.2) for all heights n, and the proof of the n>0 case has a gap that I don't see how to close as written.\n\nThe paper's genuine contributions are the presheaf on the correspondence category Corr(Sch)_{sm,uglt} (Theorem 3.1.32) and the generic decomposition for H^0 (Theorem 4.3.4). The correspondence-level structure is new; [BE21] constructs n-birational categories over locally Noetherian schemes but without this schematic functoriality. The H^0 decomposition relies on the n=0 special case of continuity, where the argument only needs dominance, not codimension preservation. That part looks defensible, assuming the input from the author's preprints [Mai26a, Mai26b] is sound. The paper also contains an honest discussion of what does not work (non-cartesian localization, lack of A^1-invariance), which I appreciate.\n\nThe soft spot is Theorem 4.3.2. In the second paragraph of its proof, to descend an n-dense open immersion for n>0, the author needs that the closure of the image of a closed subset of codimension at least n+1 under a flat map has codimension at least n+1 in the target. He cites 'the methods of Theorem 3.2.2', but that theorem assumes Noetherian universally catenary schemes. Theorem 4.3.2 is stated for arbitrary Qcqs schemes with flat affine transition maps, and flat maps don't preserve codimension in that generality without extra hypotheses; non-Noetherian and non-catenary failures are standard. No proof is supplied for this specific pro-system. The n=0 case avoids the issue, so the headline decomposition stands; but the full continuity theorem is overclaimed, and the paper should either prove the codimension step under weaker assumptions or restrict the statement to the NUC case (or to n=0).\n\nWho is this for? Anyone working on motivic homotopy theory or birational geometry. It's a serious paper that deserves a referee, not a desk reject. The referee should focus on Theorem 4.3.2 and on the dependence on the author's preprints. With that fixed, I'd be happy to cite the generic decomposition.\n\nRecommendation: send to peer review, and tell the referee to push on the codimension-preservation assertion.","headline":"A structurally useful generic decomposition for H^0, paired with an overclaimed flat-affine continuity theorem whose n>0 case rests on an unproved codimension-preservation step.","tokens_in":40159,"tokens_out":5346,"would_cite":true,"duration_ms":42727,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14F42","14E05"],"pacs":[],"model":"deepseek-v4-flash","headline":"The birational motivic homotopy category of a scheme with finitely many generic points is the product of the categories of those generic points.","keywords":["birational motivic homotopy","n-dense open immersion","generic decomposition","presheaf on correspondences","flat-affine continuity","Nisnevich descent","function field invariant","unstable slice conjecture"],"falsifier":"Check whether a flat affine map in the spreading-out diagram can shrink codimension as follows: take the projection $\\mathbb{A}^2_k\\to\\mathbb{A}^1_k$ and a closed point $Z\\subset\\mathbb{A}^2_k$; $Z$ has codimension $2$, while the closure of its image is a closed point of codimension $1$ in $\\mathbb{A}^1_k$. If such a map can be realized as a transition map in the pro-system of Theorem 4.3.2, the descent of $n$-dense open immersions for $n>0$ fails, and the full flat-affine continuity theorem is false as stated; the $n=0$ generic decomposition would then require a separate proof.","tokens_in":39019,"feed_emoji":"🧩","tokens_out":14894,"duration_ms":127807,"temperature":0.7,"pith_summary":"This paper establishes a functoriality and continuity package for the $n$-birational motivic homotopy categories $\\mathcal{H}^n(S)$, the localizations of the $\\mathbb{A}^1$-motivic homotopy category at open immersions whose closed complement has codimension greater than $n$. Its main structural result, Theorem 4.3.4, states that for a quasi-compact quasi-separated scheme $X$ with finitely many generic points, the canonical map $\\mathcal{H}^0(X)\\to\\prod_{\\eta\\in X^{(0)}}\\mathcal{H}^0(k(\\eta))$ is an equivalence of $\\infty$-categories; since $\\mathcal{H}^0$ is the birational motivic homotopy category $\\mathcal{H}^{b\\mathbb{A}^1}$, a variety's birational motivic homotopy type is completely determined by that of its function field. A reader should care because this turns birational motivic questions over a scheme into questions over residue fields, yields pointwise criteria for birational equivalences, and gives functorial embeddings for stably birational and purely transcendental maps. The route passes through a new presheaf structure on a correspondence category, Nisnevich descent, deformation invariance, and a flat-affine continuity theorem.","feed_headline":"Birational motivic homotopy splits by generic points","feed_subtitle":"For a variety V, the birational motivic homotopy category of V equals that of its function field k(V).","key_machinery":"The mechanism is the graded family of localization procedures indexed by the height $n$. An open immersion is called $n$-dense when its closed complement has codimension $>n$; localizing the $\\mathbb{A}^1$-motivic homotopy category at the $n$-dense open immersions gives $\\mathcal{H}^n(S)$, with $n=0$ recovering the birational motivic homotopy category. The paper's main technical instrument is the promotion of this assignment to a presheaf on the correspondence category whose vertical maps are smooth morphisms and whose horizontal maps are universally generalization-lifting morphisms, together with the refined pushforward along closed immersions that shifts the height by the maximal codimension of the immersion. These functorialities feed a flat-affine continuity theorem for pro-systems with affine flat transition maps, from which the generic-point decomposition follows by taking the pro-system of dense affine open subschemes.","core_discovery":"The central claim is Theorem 4.3.4: for a quasi-compact quasi-separated scheme $X$ with finitely many generic points, the canonical map $\\mathcal{H}^0(X)\\to\\prod_{\\eta\\in X^{(0)}}\\mathcal{H}^0(k(\\eta))$ is an equivalence of $\\infty$-categories. In particular, for a variety $V$, $\\mathcal{H}^{b\\mathbb{A}^1}(V)\\simeq\\mathcal{H}^{b\\mathbb{A}^1}(k(V))$. The paper proves this by first promoting $S\\mapsto\\mathcal{H}^n(S)$ to a $\\mathrm{Pr}^L$-valued presheaf on the correspondence category $\\mathrm{Corr}(\\mathrm{Sch})_{\\mathrm{sm},\\mathrm{uglt}}$, then establishing that this presheaf satisfies Nisnevich descent, deformation invariance, and flat-affine continuity. The continuity step collapses the pro-system of dense affine open subschemes of $X$ to $\\mathrm{Spec}(k(X))$, while dense-locality and additivity reduce the general case to a single irreducible reduced variety. This is the statement the author is trying to establish: the birational motivic homotopy category is a function-field invariant.","pith_inferences":["A natural extension the paper leaves open is whether $\\mathcal{H}^n$ also splits over generic points for every $n>0$; the only barrier is the flat-affine codimension step, so one can test $\\mathcal{H}^n(X)\\simeq\\prod_\\eta \\mathcal{H}^n(k(\\eta))$ directly once that gap is resolved.","If the generic decomposition survives, every invariant computed from $\\mathcal{H}^{b\\mathbb{A}^1}$ becomes a function-field invariant, which would give a new arithmetic test for rationality: compare $\\mathcal{H}^{b\\mathbb{A}^1}$-enriched objects over $V$ and over $k(V)$ rather than over the whole scheme.","The paper leaves the inverse limit of the $\\Omega_{\\mathbb{P}^1}$ tower uncharacterized; a testable possibility is that this limit carries transfer maps, which would tie the birational tower to a different layer of birational invariants."],"forward_implications":["For a variety $V$, $\\mathcal{H}^{b\\mathbb{A}^1}(V)$ is canonically equivalent to $\\mathcal{H}^{b\\mathbb{A}^1}(k(V))$, so birational motivic invariants of $V$ are invariants of its function field.","A morphism $f:X\\to Y$ in $\\mathrm{Sm}_S$ whose generic fibers are birationally contractible is a birational equivalence, giving a pointwise detection criterion for birational motivic equivalences.","Stably birational morphisms and purely transcendental field extensions induce fully faithful embeddings $\\mathcal{H}^0(S)\\hookrightarrow\\mathcal{H}^0(X)$ and $\\mathcal{H}^0(k)\\hookrightarrow\\mathcal{H}^0(k(t_1,\\dots,t_n))$.","The presheaf $\\mathcal{H}^n$ on the correspondence category is a Nisnevich sheaf and is deformation-invariant, and $\\Omega_{\\mathbb{P}^1}$ gives a tower from higher to lower heights, providing a partial unstable analogue of the slice filtration."],"supporting_citations":[{"why":"It supplies the $\\infty$-categorical definition of the $n$-birational motivic homotopy categories and the localization machinery used throughout the paper.","marker":"[BE21]"},{"why":"It identifies $\\mathcal{H}^0$ with the birational homotopy category, so the generic decomposition is a statement about $\\mathcal{H}^{b\\mathbb{A}^1}$.","marker":"[Mai26a]"},{"why":"It introduced the unstable $n$-dense localization that the graded categories $\\mathcal{H}^n$ formalize.","marker":"[Pel14]"},{"why":"It provides Nisnevich descent for motivic homotopy categories and the base-change formalism that the presheaf structure extends.","marker":"[Hoy17]"},{"why":"It gives the fully faithful pushforward along closed immersions used in the refined pushforward and in monoidal embeddings.","marker":"[Kha16]"},{"why":"It supplies the codimension, spreading-out, and generalization-lifting lemmas that the continuity and descent arguments rely on.","marker":"[Sta26]"},{"why":"It provides the descent of open immersions and étale lifts used in deformation invariance and in spreading out $n$-dense opens.","marker":"[Gro66]"}],"fun_headline_variants":["Birational motivic homotopy: just the function field","Motivic homotopy of varieties reduces to k(V)","Birational homotopy categories: generic points enough","Function field is the whole birational motivic homotopy","Motivic homotopy: generic points split it"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof assumes that, along the flat affine maps appearing in the spreading-out pro-system, taking the closure of the image of the closed complement of an $n$-dense open cannot lower its codimension below $n+1$; flat maps do not generally preserve codimension in this way, and the paper supplies no proof for this specific pro-system.","fun_headline_variants_meta":{"raw":{"variants":["Birational motivic homotopy: just the function field","Motivic homotopy of varieties reduces to k(V)","Birational homotopy categories: generic points enough","Function field is the whole birational motivic homotopy","Motivic homotopy: generic points split it"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000704,"raw_usage":{"total_tokens":3191,"prompt_tokens":980,"completion_tokens":2211,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":596,"completion_tokens_details":{"reasoning_tokens":2128}},"tokens_in":596,"tokens_out":2211,"duration_ms":14269,"temperature":1.0,"reasoning_tokens":2128,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T14:39:00.143228+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Check whether a flat affine map in the spreading-out diagram can shrink codimension as follows: take the projection $\\mathbb{A}^2_k\\to\\mathbb{A}^1_k$ and a closed point $Z\\subset\\mathbb{A}^2_k$; $Z$ has codimension $2$, while the closure of its image is a closed point of codimension $1$ in $\\mathbb{A}^1_k$. If such a map can be realized as a transition map in the pro-system of Theorem 4.3.2, the descent of $n$-dense open immersions for $n>0$ fails, and the full flat-affine continuity theorem is false as stated; the $n=0$ generic decomposition would then require a separate proof.","supporting_citations":[],"review_version":3}