SEMISMALL MORPHISMS 1. Motivation Let's Recall Some Results

SEMISMALL MORPHISMS 1. Motivation Let's Recall Some Results

SEMISMALL MORPHISMS YEHAO ZHOU 1. Motivation Let's recall some results concerning t-exactness of pushforward from last time: Proposition 1.1. Let f : X ! Y be a morphism between varieties, then (1) If f is quasi-finite, then f∗ is left t-exact and f! is right t-exact; (2) If f is affine, then f∗ is right t-exact and f! is left t-exact; (3) As a result, if f is quasi-finite affine, then f∗ and f! are t-exact. Unfortunately, proper morphisms, which commonly show up in applications (e.g. reso- lutions of singularities), are not in this list (except for finite morphisms). Heuristically, this comes from the fact that for a proper variety Z of dimension d, the top degree co- homology H2d(Z; C) 6= 0 (Lemma 2.3), so if there is a proper morphism f : X ! Y such that there is a subvariety W ⊂ Y with dimensions of fibers over W being ≥ d, 2d−dim X p ≤0 then the H (fjf −1W )∗C[dim X] 6= 0, i.e. (fjf −1W )∗C[dim X] 2 Dc (Y ) only if dim W ≥ 2d − dim X. So we restrict to proper morphisms with a nice property on the dimension of fibers: Definition 1.2. Let X be an irreducible variety, f : X ! Y is called semismall if there exists a stratification fYtgt2T of Y , such that for each stratum Yt and each point y 2 Yt\f(X) −1 2 dim f (y) + dim Yt ≤ dim X f is called small with respect to a dense open subset W ⊂ Y if 8y 2 W , f −1(y) is a finite set, and there exists a stratification fYtgt2T , such that W is a union of strata and such that 8y 2 Yt \ f(X) ⊂ Y − W −1 2 dim f (y) + dim Yt < dim X Proposition 1.3 (Equivalent definition of semismallness). Let X be an irreducible variety. Then f : X ! Y is semismall if and only if dim X ×Y X ≤ dim X Proof. If f is semismall, then there is a stratification fYtgt2T , such that for each stratum Yt and each point y 2 Yt \ f(X) −1 −1 dim(f × f) (y) + dim Yt = 2 dim f (y) + dim Yt ≤ dim X −1 −1 which implies that dim(f × f) Yt ≤ dim X. Since X ×Y X = qt(f × f) Yt, we conclude that dim X ×Y X ≤ dim X. Let's prove the "if" part by constructing a stratification. Let's prove a lemma first. 1 2 YEHAO ZHOU Lemma 1.4. If f : X ! Y is a morphism, then there exists a stratification fYtgt2T such −1 that 8t 2 T , the fibers of fjf Yt are either empty or of dimensions d(t) which only depend on t, and −1 −1 dim f Yt = d(t) + dim Yt ; dim(f × f) Yt = 2d(t) + dim Yt Proof. We prove it by Noetherian induction. First of all, we replace Y with the closure of f(X), endowed with reduced structure sheaf. Suppose that the lemma is true for any proper closed subvariety of Y . According to the generic flatness theorem [GD65] (EGA IV2, Th´eor`eme6.9.1) there is an irreducible open subvariety U ⊂ Y such that fjf −1U is flat. Shrink U if necessary, we assume that each irreducible component of f −1U is mapped surjectively to U. Then −1 we apply Theorem 15.1 of [Mat89] to fjf −1U and we see that 8y 2 U and 8x 2 f (y), −1 −1 dim OX;x ≤ dim f (y) + dim OY;y, and there is at least one x 2 f (y) such that the equality is obtained. In other words −1 −1 sup dim OX;x = dim f (y) + dim OY;y = dim f (y) + dim U x2f −1(y) Since each irreducible components of f −1U is mapped surjectively to U, there exists x 2 f −1(y) such that x lies in the component with maximal dimension, i.e. dim f −1U, so −1 sup dim OX;x = dim f U x2f −1(y) As a consequence, dim f −1(y) = dim f −1U − dim U is constant on U. −1 −1 Apply Theorem 15.1 of [Mat89] to the flat morphism (f × f)jf U×Y f U , we see that −1 −1 −1 sup dim OX×Y X;x = dim f (y) × f (y) + dim OY;y = 2 dim f (y) + dim U x2(f×f)−1(y) The RHS is constant on U, so −1 −1 dim(f × f) U = sup sup dim OX×Y X;x = 2 dim f (y) + dim U y2U x2(f×f)−1(y) According to the lemma, there is a stratification fYtgt2T such that 8t 2 T , fibers of −1 fjf Yt are either empty or of dimensions d(t) which only depend on t and −1 dim(f × f) Yt = 2d(t) + dim Yt −1 If dim X ×Y X ≤ dim X, then (f × f) Yt ≤ dim X ×Y X ≤ dim X, which implies that 2d(t) + dim Yt ≤ dim X, so f is semismall. Exercise 1.5. For any f : X ! Y , inequality dim X ×Y X ≥ dim X always hold. Exercise 1.6. Consider morphisms f : X ! Y and g : Y ! Z, if g ◦ f is semismall, then f is semismall. Exercise 1.7 (Composition of semismall maps can fail to be semismall). Blowing up subvariety fx = w = 0g inside the affine cone fxy = zwg ⊂ C4, followed by blowing up the preimage of f0g, each blow up is semismall, but the composition is not. Show that the composition is directly blowing up f0g. SEMISMALL MORPHISMS 3 2. Semismall and small pushforward Theorem 2.1. Assume that X is smooth, f : X ! Y is a semismall morphism, then p ≤0 p ≤0 (1) If F 2 Dloc(X), then f!F 2 Dc (Y ); p ≥0 p ≥0 (2) If F 2 Dloc(X), then f∗F 2 Dc (Y ). Note that (2) follows from (1) by Verdier duality. p ≤0 Proof. It suffices to prove that if F 2 Loc(X)[dim X] then f!F 2 Dc (Y ), i.e. i dim Supp(H f!F ) ≤ −i From the definition of semismallness, we know that there is a stratification fYtgt2T , such that −1 for each stratum it : Yt ,! Y and each point y 2 Yt \ f(X), 2 dim f (y) + dim Yt ≤ dim X, −1 −1 or equivalently − dim X + 2 dim f (y) ≤ − dim Yt. We see that F jf Yt is of cohomological i degree − dim X, hence the homological bound of f! [Bai18], i.e. H f!K = 0 for i > 2d and K 2 Shvc(X), implies that ∗ ≤− dim Yt ∼ −1 −1 it f!F = (fjf Yt )!F jf Yt 2 Dc (Yt) ∗ i i ∗ Since it H f!F = H it f!F , we have following inequality on strata ∗ i dim Supp(it H f!F ) ≤ −i i The LHS is dim(Supp(H f!F ) \ Yt). Let t runs through T , then we have the desired inequality i dim Supp(H f!F ) ≤ −i Corollary 2.2. Assume that X is smooth, f : X ! Y is a proper morphism, TFAE: (a) f is semismall; (b) f∗ : Loc(X)[dim X] ! Perv(Y ); (c) f∗C[dim X] 2 Perv(Y ). Proof. By the previous theorem, (a) implies (b). (b) implies (c) tautologically. It remains to prove that (c) implies (a). Assume that (a) is false. −1 By Lemma 1.4, there is a stratification fYtgt2T such that 8t 2 T , fibers of fjf Yt are either empty or of dimensions d(t) which only depends on t and −1 dim(f × f) Yt = 2d(t) + dim Yt Since f is not semismall, there must be at least one stratum Yt such that 2d(t) + dim Yt > −1 dim X, or equivalently − dim X + 2 dim f (y) > − dim Yt. We need the following lemma Lemma 2.3. If X is a proper variety of dimension d, then H2d(X; C) 6= 0. 2d −2d Proof. By Verdier duality, the statement H (X; C) 6= 0 is equivalent to H (X; !X ) 6= 0, where !X is the dualizing complex on X. Take a smooth open subvariety j : U,! X with complement i : Z,! X such that dim Z < d, consider the distinguished triangle i∗!Z ! !X ! j∗!U ! 4 YEHAO ZHOU Apply the functor RΓ(X; −) and we get exact sequence −2d −2d −2d −2d+1 H (Z; !Z ) ! H (X; !X ) ! H (U; !U ) ! H (Z; !Z ) _ _ Note that RΓ(Z; !Z ) = RΓc(Z; C) RΓ(Z; C) because Z is proper, and since dim Z < i −2d+1 −2d d,H (Z; C) = 0 for i > 2d − 2, hence H (Z; !Z ) = H (Z; !Z ) = 0. We see that −2d ∼ −2d H (X; !X ) = H (U; !U ). Since there is at least one connected component of U with −2d 0 dimension d, denote it by U0, then !U0 = C[2d], so H (U0;!U0 ) = H (U0; C) = C 6= 0, −2d hence we conclude that H (X; !X ) 6= 0. − dim X+2 dim f −1(y) ∗ −1 Apply the lemma and we get H (fjf Yt )∗C 6= 0, so that it f∗C[dim X] 2= ≤− dim Yt Dc (Yt), hence f∗C[dim X] 2= Perv(Y ), i.e. (c) is false. It turns out that we can say more for proper small morphisms. Theorem 2.4.

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    11 Page
  • File Size
    -

Download

Channel Download Status
Express Download Enable

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

  • Not to be reproduced or distributed without explicit permission.
  • Not used for commercial purposes outside of approved use cases.
  • Not used to infringe on the rights of the original creators.
  • If you believe any content infringes your copyright, please contact us immediately.

Support

For help with questions, suggestions, or problems, please contact us