This paper investigates Ulrich bundles on the Segre fourfold, providing characterizations, constructions, and conditions for their properties, advancing understanding of vector bundles in algebraic geometry.
Contribution
It offers new characterizations and constructions of Ulrich bundles on the Segre fourfold, including those obtained as pullbacks and satisfying specific cohomological conditions.
Findings
01
Characterization of Ulrich bundles with certain cohomological vanishing
02
Construction of complex examples of Ulrich bundles
03
Identification of pullback Ulrich bundles from b2
Abstract
We study the resolution of an Ulrich bundle of arbitrary rank on the Segre fourfold \PP2×\PP2. We characterize the Ulrich bundles \Vv of arbitrary rank on \PP2×\PP2 with h1(\Vv⊗Ω⊠Ω)=0 or with h2(\Vv⊗Ω(−1)⊠Ω(−1))=0 or obtained as pullback from \PP2 and we construct more complicated examples.
\operatorname{Ext}^{k}(^{\vee}E_{i},E_{j})=\operatorname{Ext}^{k}(E_{i},E_{j}^{\vee})=\left\{\begin{array}[]{cc}\mathbb{C}&\textrm{\quad if $i+j=n$ and $i=k$}\\
0&\textrm{\quad otherwise}\end{array}\right.
\operatorname{Ext}^{k}(^{\vee}E_{i},E_{j})=\operatorname{Ext}^{k}(E_{i},E_{j}^{\vee})=\left\{\begin{array}[]{cc}\mathbb{C}&\textrm{\quad if $i+j=n$ and $i=k$}\\
0&\textrm{\quad otherwise}\end{array}\right.
\mathrm{Ext}^{k}(E_{i},F_{j})=H^{k+k_{i}}(\mathcal{E}_{i}\otimes\mathcal{F}_{j})=\left\{\begin{array}[]{cc}\mathbb{C}&\textrm{\quad if $i=j=k$}\\
0&\textrm{\quad otherwise}\end{array}\right.
\mathrm{Ext}^{k}(E_{i},F_{j})=H^{k+k_{i}}(\mathcal{E}_{i}\otimes\mathcal{F}_{j})=\left\{\begin{array}[]{cc}\mathbb{C}&\textrm{\quad if $i=j=k$}\\
0&\textrm{\quad otherwise}\end{array}\right.
E1p,q=Extq(E−p,A)⊗F−p=Hq+k−p(E−p⊗A)⊗F−p
E1p,q=Extq(E−p,A)⊗F−p=Hq+k−p(E−p⊗A)⊗F−p
E_{\infty}^{p,q}=\left\{\begin{array}[]{cc}A&\textrm{\quad if $p+q=0$}\\
0&\textrm{\quad otherwise.}\end{array}\right.
E_{\infty}^{p,q}=\left\{\begin{array}[]{cc}A&\textrm{\quad if $p+q=0$}\\
0&\textrm{\quad otherwise.}\end{array}\right.
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Full text
Ulrich bundles on the degree six Segre fourfold
F. Malaspina
Politecnico di Torino, Corso Duca degli Abruzzi 24, 10129 Torino, Italy
We completely characterize the bigraded resolutions
of Ulrich bundles of arbitrary rank on the Segre fourfold P2×P2.
We characterize the Ulrich bundles V of arbitrary rank on P2×P2 with h1(V⊗Ω⊠Ω)=0 or with h2(V⊗Ω(−1)⊠Ω(−1))=0 or obtained as pullback from P2 and we construct more complicated examples.
A locally free sheaf (or “bundle”) E on a projective varity X is aCM if it has not intermediate cohomology or if the module
E of global sections of
E is a maximal Cohen-Macaulay module.
There has been increasing interest on the classification of aCM bundles on various projective varieties, which is important in a sense that the aCM bundles are considered to give a measurement of complexity of the underlying space. Moreover the understanding of aCM bundles is crucial for the study of any bundle on X as it is showed on [22]. A special type of aCM sheaves, called the Ulrich sheaves, are the ones achieving the maximum possible minimal number of generators. These bundles
are characterized by the linearity of the minimal graded free resolution
over the polynomial ring of their module of global section.
Ulrich bundles, originally studied for computing Chow forms,
conjecturally exist over any variety (see [13]). But the conjecture has been checked only for a few varieties, e.g. in case of surfaces, del Pezzo surfaces, rational normal scrolls, rational aCM surfaces in P4, ruled surfaces and so on; see [9, 23, 24, 1]. The case of Veronese varieties has been studied in [13], [7] and [20] and the case of Hirzebruch surfaces in [3]. Although there are some occasions where the classification problem of Ulrich bundles of special type is done as in [2, 8, 10, 11], the completion of classification problem is difficult in usual.
In [12] and [23] it shown that Segre varieties and rational normal scroll (except P1×P1 and cubic and quartic scroll, see [15] and [17]) are of Ulrich wild representation type namely they support families of Ulrich bundles of arbitrary large. The representation type is determined by considering a certain type of family of Ulrich bundles. In [2] in given a classification of Ulrich vector bundles of arbitrary rank on rational normal scroll. A consequence of the main result is that the moduli spaces of Ulrich bundles are zero-dimensional. The case P2×P1 has been studied in more details in [16], where all the aCM bundles are classified. The first biprojective space which is not a rational scroll is P2×P2. Here the families of Ulrich bundles are much more complicated as can be deduced from the study of the rank two case achieved in [6].
In this article we pay our attention to the case of arbitrary rank. In order to do that we show that every Ulrich bundle V on P2×P2 is regular according to both the two different notions of Castelnuovo-Mumford regularity given in [4] and [19]. Then we compute the cohomology of V tensored with OP2⊠Ω, Ω⊠OP2 and Ω⊠Ω with suitable twists. In particular we obtain that V is of natural cohomology (as in [13] on Veronese varities) in a suitable range. More precisely we get that the only nonzero cohomology for V(m,n) with ∣m−n∣≤1 may be given by
[TABLE]
At this point we choose suitable full exceptional collections in order to apply a Beilinson type spectral sequence and to obtain the following resolution
[TABLE]
[TABLE]
Moreover we characterize the Ulrich bundles V of arbitrary rank on P2×P2 with h1(V⊗Ω⊠Ω)=0 or with h1(V⊗Ω(−1)⊠Ω(−1))=0 or obtained as pullback from P2 and we construct more complicated examples. We believe that the study of this variety will be the key step for the understanding of the families of Ulrich bundles over all the biprojective spaces.
Here we summarize the structure of this article. In section 2 we introduce the definition of Ulrich bundles and several notions in derived category of coherent sheaves to understand the Beilinson spectral sequence. In section 3 we recall the definition of Castelnuovo-Mumford given in [4] and [19] and we made the cohomological computations. In section 4 we study the examples of families of Ulrich bundles and we prove the main results.
In section 5 we discuss the case of the hyperplane section of P2×P2: the flag variety F(0,1,2).
The author wants to thank M. Aprodu and P. Rao for helpful discussions on the subject.
2. Preliminaries
Throughout the article our base field is the field of complex numbers C.
Definition 2.1**.**
A coherent sheaf E on a projective variety X with a fixed ample line bundle OX(1) is called arithmetically Cohen-Macaulay (for short, aCM) if it is locally Cohen-Macaulay and Hi(E(t))=0 for all t∈Z and i=1,…,dim(X)−1.
Definition 2.2**.**
For an initialized coherent sheaf E on X, i.e. h0(E(−1))=0 but h0(E)=0, we say that E is an Ulrich sheaf if it is aCM and h0(E)=deg(X)rank(E).
Given a smooth projective variety X, let Db(X) be the the bounded derived category of coherent sheaves on X. An object E∈Db(X) is called exceptional if Ext∙(E,E)=C.
A set of exceptional objects ⟨E0,…,En⟩ is called an exceptional collection if Ext∙(Ei,Ej)=0 for i>j. An exceptional collection is said to be full when Ext∙(Ei,A)=0 for all i implies A=0, or equivalently when Ext∙(A,Ei)=0 does the same.
Definition 2.3**.**
Let E be an exceptional object in Db(X).
Then there are functors LE and RE fitting in distinguished triangles
[TABLE]
[TABLE]
The functors LE and RE are called respectively the left and right mutation functor.
The collections given by
[TABLE]
are again full and exceptional and are called the right and left dual collections. The dual collections are characterized by the following property; see [18, Section 2.6].
[TABLE]
Theorem 2.4** (Beilinson spectral sequence).**
Let X be a smooth projective variety and with a full exceptional collection ⟨E0,…,En⟩ of objects for Db(X). Then for any A in Db(X) there is a spectral sequence
with the E1-term
[TABLE]
which is functorial in A and converges to Hp+q(A).
The statement and proof of Theorem 2.4 can be found both in [26, Corollary 3.3.2], in [18, Section 2.7.3] and in [5, Theorem 2.1.14].
Let us assume next that the full exceptional collection ⟨E0,…,En⟩ contains only pure objects of type Ei=Ei∗[−ki] with Ei a vector bundle for each i, and moreover the right dual collection ⟨E0∨,…,En∨⟩ consists of coherent sheaves. Then the Beilinson spectral sequence is much simpler since
[TABLE]
Note however that the grading in this spectral sequence applied for the projective space is slightly different from the grading of the usual Beilison spectral sequence, due to the existence of shifts by n in the index p,q. Indeed, the E1-terms of the usual spectral sequence are Hq(A(p))⊗Ω−p(−p) which are zero for positive p. To restore the order, one needs to change slightly the gradings of the spectral sequence from Theorem 2.4. If we replace, in the expression
[TABLE]
u=−n+p and v=n+q so that the fourth quadrant is mapped to the second quadrant, we obtain the following version (see [2]) of the Beilinson spectral sequence:
Theorem 2.5**.**
Let X be a smooth projective variety with a full exceptional collection
⟨E0,…,En⟩
where Ei=Ei∗[−ki] with each Ei a vector bundle and (k0,…,kn)∈Z⊕n+1 such that there exists a sequence ⟨Fn=Fn,…,F0=F0⟩ of vector bundles satisfying
[TABLE]
i.e. the collection ⟨Fn,…,F0⟩ labelled in the reverse order is the right dual collection of ⟨E0,…,En⟩.
Then for any coherent sheaf A on X there is a spectral sequence in the square −n≤p≤0, 0≤q≤n with the E1-term
[TABLE]
which is functorial in A and converges to
[TABLE]
3. Cohomology of Ulrich bundles on P2×P2
Let X=P2×P2. We will often use the following exact sequences.
[TABLE]
[TABLE]
[TABLE]
and
[TABLE]
On X we have two different notions of Castelnuovo-Mumford regularity
(see [4] and [19]):
Definition 3.1**.**
A coherent sheaf F on X is
(BM)-regular if:
[TABLE]
[TABLE]
Definition 3.2**.**
A coherent sheaf F on X is
(HW)-regular if:
[TABLE]
[TABLE]
[TABLE]
Remark 3.3**.**
If F is a (BM)-regular coherent on X then it is globally generated and F(p,p′) is (BM)-regular for p,p′≥0 by [4] Proposition 2.2.
If F is a (HW)-regular coherent on X then it is globally generated and F(p,p′) is (HW)-regular for p,p′≥0 by [19] Proposition 2.7 and Proposition 2.8.
Lemma 3.4**.**
Let V be an Ulrich bundle on X.
(i)
H0(V(j1,j2))=0* for j1≤−1, \j2≤−1 and H4(V(j1,j2))=0 for j1≥−4, \j2≥−4.*
2. (ii)
V* is (BM)-regular and (HW)-regular.*
3. (iii)
For i=1,2,3, Hi(V(j1,j2))=0 if j1≥−i and \j2≥−i or if j1≤−i−1 and \j2≤−i−1.
4. (iV)
For i=1,2,3, Hi(V⊗Ω(j1+2)⊠OP2(j2))=Hi(V⊗OP2(j1)⊠Ω(j2+2))=0 if j1≥−i and \j2≥−i or if j1≤−i−1 and \j2≤−i−1.
Proof.
Hi(V(j,j))=0 for any i and j=−1,−2,−3,−4. Since H0(V(−1,−1))=0, from (4), (5), (6) and (7), we get H0(V(j1,j2))=0 for j1≤−1, \j2≤−1, H0(V⊗Ω(j1+1)⊠OP2(j2))=H0(V⊗OP2(j1)⊠Ω(j2+1))=0 for j1≤−1, \j2≤−1.
Since H4(V(−4,−4))=0, from (4), (5), (6) and (7) we get H4(V(j1,j2))=0 for j1≥−4, \j2≥−4, H4(V⊗Ω(j1+2)⊠OP2(j2))=H4(V⊗OP2(j1)⊠Ω(j2+2))=0 for j1≥−4, \j2≥−4.
In particular we get (i).
Since H3(V(−3,−3))=0, from (5) tensored by V(−3,−2) and (7) tensored by V(−2,−3), H3(V(−3,−2))=H3(V(−2,−3))=0.
From (5) tensored by V(−3,−1) and (7) tensored by V(−1,−3), H3(V(−3,−1))=H3(V(−1,−3))=0.
Since H3(V(−2,−2))=0, from (5) tensored by V(−2,−1) and (7) tensored by V(−1,−2), H3(V(−2,−1))=H3(V(−1,−2))=0.
From (4) tensored by V(−3,−1) and (6) tensored by V(−1,−3) we get H3(V⊗Ω(−1)⊠OP2(−3))=H3(V⊗OP2(−3)⊠Ω(−1))=0.
Since H3(V(−3,−2))=H3(V(−2,−3))=0, from (4) tensored by V(−2,−1) and (6) tensored by V(−1,−2) we get H3(V⊗Ω(−1)⊠OP2(−2))=H3(V⊗OP2(−2)⊠Ω(−1))=0.
Since H3(V(−2,−2))=0, from (4) tensored by V(−2,0) and (6) tensored by V(0,−2) we get H3(V⊗Ω⊠OP2(−2))=H3(V⊗OP2(−2)⊠Ω)=0.
Since H2(V(−2,−2))=0 and H3(V⊗Ω(−1)⊠OP2(−2))=H3(V⊗OP2(−2)⊠Ω(−1))=0, from (5) tensored by V(−2,−1) and (7) tensored by V(−1,−2), H2(V(−2,−1))=H2(V(−1,−2))=0.
Since H2(V(−2,−1))=H2(V(−2,−1))=0 and H3(V⊗Ω⊠OP2(−2))=H3(V⊗OP2(−2)⊠Ω)=0, from (5) tensored by V(−2,0) and (7) tensored by V(0,−2)H2(V(−2,0))=H2(V(0,−2))=0.
Since H3(V(−3,−1))=H3(V(−1,−3))=H2(V(−2,−1))=H2(V(−2,−1))=0, from (4) tensored by V(−1,0) and (6) tensored by V(0,−1) we get H2(V⊗Ω⊠OP2(−1))=H2(V⊗OP2(−1)⊠Ω)=0.
Since H1(V(−1,−1))=0 and H2(V⊗Ω⊠OP2(−1))=H2(V⊗OP2(−1)⊠Ω)=0, from (5) tensored by V(−1,0) and (7) tensored by V(0,−1)H1(V(−1,0))=H1(V(0,−1))=0.
We have hence proved the (BM)-regularity and (HW)-regularity so (ii).
Since V(p,p′) is (BM)-regular and (HW)-regular for p,p′≥0, we obtain Hi(V(j1,j2))=0 if j1≥−i and \j2≥−i for i>0.
Now since V∨(2,2) is Ulrich we have Hj(V∨(k1+2,k2+2))=0 if k1≥−j and k2≥−j, so by Serre duality
H4−j(V(−k1−5,−k2−5))=0 if k1≥−j and k2≥−j. Let i=4−j, j1=−k1−5, j2=−k2−5, we get Hi(V(j1,j2))=0 if j1≤−i−1 and j2≤−i−1 for i=1,2,3. So also (iii) is proved.
Now from
[TABLE]
we get Hi(V⊗OP2(j1)⊠Ω(j2+2))=0 if j1≥−i and \j2≥−i.
From
[TABLE]
we get Hi(V⊗OP2(j1)⊠Ω(j2+2))=0 if j1≤−i−1 and \j2≤−i−1 for i=1,2,3,. In the same way we obtain Hi(V⊗Ω(j1+2)⊠OP2(j2))=0 if j1≥−i and \j2≥−i or if j1≤−i−1 and \j2≤−i−1 for i=1,2,3, and proof of (iv) is completed.
∎
Remark 3.5**.**
Let call for i=0,…4
[TABLE]
An helpful tool for summarizing the part (iii) of the above Lemma are the following pictures dealing with the subsets of the plane j1,j2 whose points correspond to some intermediate cohomology group of the sheaf V(j1,j2).
In the following pictures we show the vanishing regions and the positions of ai,bi for h1(V(j1,j2)),h2(V(j1,j2)) and h3(V(j1,j2)):
h1(V⊗Ω⊠OP2(−1))=b0, h1(V⊗OP2(−1)⊠Ω))=a0 and hi(V⊗Ω⊠OP2(−1))=hi(V⊗OP2(−1)⊠Ω))=0 for any i except for i=1.
2. (b)
h1(V⊗Ω⊠OP2(−2))=b2, h1(V⊗OP2(−2)⊠Ω)=a2 and hi(V⊗Ω⊠OP2(−2))=hi(V⊗OP2(−2)⊠Ω))=0 for any i except for i=1.
3. (c)
h1(V⊗Ω⊠Ω)=3a2+3b2* and hi(V⊗Ω⊠Ω)=0 for any i except for i=1.*
4. (d)
h2(V⊗Ω(−1)⊠Ω(−1))=3a3+3b3=3a1+3b1* and hi(V⊗Ω(−1)⊠Ω(−1))=0 for any i except for i=2.*
Proof.
From (5) tensored by V(−1,0) and (7) tensored by V(0,−1) we get h1(V⊗OP2(−1)⊠Ω)=a0, h1(V⊗Ω⊠OP2(−3))=b0 and hi(V⊗OP2(−1)⊠Ω))=hi(V⊗Ω⊠OP2(−1))=0 for any i except for i=1. We have proved (a).
From (4) tensored by V(−2,0) and (6) tensored by V(0,−2) we get h1(V⊗Ω⊠OP2(−2))=b2, h1(V⊗OP2(−2)⊠Ω)=a2 and hi(V⊗Ω⊠OP2(−2))=hi(V⊗OP2(−2)⊠Ω)=0 for any i except for i=1. We have proved (b).
From (4) tensored by V(−3,0) and (6) tensored by V(0,−3) we get h2(V⊗OP2(−3)⊠Ω)=3b2, h2(V⊗Ω⊠OP2(−3))=3a2 and hi(V⊗OP2(−3)⊠Ω))=hi(V⊗Ω⊠OP2(−3))=0 for any i except for i=2.
Now let us consider the sequence
[TABLE]
tensored by V. Since we have computed the cohomology of V⊗OP2(−3)⊠Ω and V⊗OP2(−2)⊠Ω, we may deduce that h1(V⊗Ω⊠Ω)=3a2+3b2 and hi(V⊗Ω⊠Ω)=0 for any i except for i=1. We have proved (c).
From (4) tensored by V(−4,−1) and (6) tensored by V(−1,−4) we get h3(V⊗OP2(−4)⊠Ω(−1))=3b3, h3(V⊗Ω(−1)⊠OP2(−4))=3a3 and hi(V⊗OP2(−4)⊠Ω(−1))=hi(V⊗Ω(−1)⊠OP2(−4))=0 for any i except for i=3.
From (4) tensored by V(−3,−1) and (6) tensored by V(−1,−3) we get h2(V⊗Ω(−1)⊠OP2(−3))=b3, h2(V⊗OP2(−3)⊠Ω(−1))=a3 and hi(V⊗Ω(−1)⊠OP2(−3))=hi(V⊗OP2(−3)⊠Ω(−1))=0 for any i except for i=2.
Now let us consider the sequence
[TABLE]
tensored by V. Since we have computed the cohomology of V⊗OP2(−4)⊠Ω(−1) and V⊗OP2(−3)⊠Ω(−1), we may deduce that h2(V⊗Ω⊠Ω)=3a3+3b3 and hi(V⊗Ω(−1)⊠Ω(−1))=0 for any i except for i=2.
From (5) and (7) tensored by V(−1,−1) we get h1(V⊗OP2(−1)⊠Ω(−1))=3b1, h1(V⊗Ω(−1)⊠OP2(−1))=3a1 and hi(V⊗OP2(−1)⊠Ω(−1))=hi(V⊗Ω(−1)⊠OP2(−1))=0 for any i except for i=1.
From (5) tensored by V(−2,−1) and (7) tensored by V(−1,−2) we get h2(V⊗Ω(−1)⊠OP2(−2))=a1, h2(V⊗OP2(−2)⊠Ω(−1))=b1 and hi(V⊗Ω(−1)⊠OP2(−2))=hi(V⊗OP2(−2)⊠Ω(−1))=0 for any i except for i=2.
Now let us consider the sequence
[TABLE]
tensored by V. Since we have computed the cohomology of V⊗OP2(−2)⊠Ω(−1) and V⊗OP2(−1)⊠Ω(−1), we may deduce that h2(V⊗Ω⊠Ω)=3a1+3b1 and hi(V⊗Ω(−1)⊠Ω(−1))=0 for any i except for i=2. So also (d) is proved.
∎
4. Families of Ulrich bundles on P2×P2
We start this section with examples of Ulrich bundles:
Remark 4.1**.**
The only rank one Ulrich bundles on X are OX(2,0) and OX(0,2).
Example 4.2**.**
Let a≥1 and b≥a+2 then the set of elements
[TABLE]
such that H0(Φ(1)) are surjective forms a non-empty dense open subset (see [14] Proposition 4.1). Using the vector bundles obtained as the kernel of these maps when b=2a and a>1 in [12] Theorem 3.6. has been constructed families of Ulrich of even rank. The construction works also for odd rank. So for any r>1 we have families of rank r Ulrich bundles arising from the following exact sequences.
[TABLE]
or
[TABLE]
The same families may be obtained from the exact sequences
[TABLE]
or
[TABLE]
The same families may be given by the exact sequences
[TABLE]
or
[TABLE]
Lemma 4.3**.**
Let a,b,c,d be integers. Let V1 and V1 be indecomposable coherent sheaves on X arising from exact sequences
[TABLE]
or
[TABLE]
Then we have:
(1)
Ext1(V1,V2)=Ext1(V2,V1)=0.
2. (2)
If V1V2 are Ulrich bundles we must have a=2c and b=2d.
Proof.
(1) If we apply the functor Hom(−,V2) to (17) we obtain Ext1(V1,V2)=0 because Hom(OX(−1,1),V2)=H0(V2(1,−1))=0 and Ext1(OX(0,1),V2)=H1(V2(0,−1))=0. Similarly we prove that Ext1(V2,V1)=0.
(2) The rank of V1 is b−d so if V1 is Ulrich we must have h0(V1)=6b−6d. From (17) we get h0(V1)=3b, hence 3b=6b−6d if and only if b=2d.
∎
Now we construct the full exceptional collections that we will use in the next theorems:
Let us consider on both copies of P2 the full exceptional collection {OP2(−2),OP2(−1),OP2}. We may obtain (see [25]):
[TABLE]
The associated full exceptional collection ⟨F8=Fn,…,F0=F0⟩ of Theorem 2.5 is
The associated full exceptional collection ⟨F8=Fn,…,F0=F0⟩ of Theorem 2.5 is
[TABLE]
Let call G1=OP2⊠Ω(1) and G2=Ω(1)⊠OP2.
Theorem 4.4**.**
Let V be an Ulrich bundle on X. Then V arises from an exact sequence of the form:
[TABLE]
[TABLE]
or
[TABLE]
[TABLE]
Proof.
We consider the Beilinson type spectral sequence associated to A:=V(−1,−1) and identify the members of the graded sheaf associated to the induced filtration as the sheaves mentioned in the statement. We assume due to [13, Proposition 2.1] that
[TABLE]
and consider the full exceptional collection E∙ given in (21) and collection F∙ given in (22).
We construct a Beilinson complex, quasi-isomorphic to A, by calculating Hi+kj(A⊗Fj)⊗Ej with i,j∈{0,…,8} to get the following table. Here we use several vanishing in the intermediate cohomology of A,A(−1,−1),A(−2,−2),A(−3,−3) together with vanishing of Lemma 3.6:
[TABLE]
From this table, since Extk(Fi,Fj)=0 for k>0 and any i,j, we have that the full exceptional collection (31) is strong. So we get the claimed resolution.
∎
Remark 4.5**.**
From (23) we deduce that we must have a0=0 or b0=0. From (24) we deduce that we must have a4=0 or b4=0.
Corollary 4.6**.**
Let V be an Ulrich bundle on X with H1(V⊗Ω⊠Ω)=0. Then V arises from an exact sequence of the form:
[TABLE]
or
[TABLE]
with a1,b1>1.
Proof.
We consider the Beilinson type spectral sequence associated to A:=V(−1,−1) with the full exceptional collection E∙ and left dual collection F∙ as in the above Theorem.
Since H1(V⊗Ω⊠Ω)=0 and H1(V⊗Ω⊠Ω)=3a2+3b2 we get
[TABLE]
so we obtain the following table:
[TABLE]
We call α and β the maps arising from the table which are defined from OX(−2,0)a1 to OX(−1,0)a0 and from OX(0,−2)b1 to OX(0,−1)b0. So we get the following exact sequence
[TABLE]
and
[TABLE]
where kerα≅B⊠OP2 and cokerβ≅OP2⊠C with B a vector bundle on P2 and C a coherent sheaf on P2. Moreover kerβ=0 since the spectral sequence converges to an object in degree [math], so we obtain that h0(C)=0, hence Hom(kerα,cokerβ)≅H0(B∨⊠C)=0. This implies that also kerα=0 and A is given by an extension of cokerα with cokerβ But, by Lemma4.3, Ext1(cokerβ,cokerα)=0 and a0=2a1 and b0=2b1. Then we have the claimed result.
∎
Theorem 4.7**.**
Let V be an Ulrich bundle on X with H2(V⊗Ω(−1)⊠Ω(−1))=0. Then V≅OX(2,0) or V≅OX(0,2).
Proof.
We consider the Beilinson type spectral sequence associated to A:=V(−1,−1) and identify the members of the graded sheaf associated to the induced filtration as the sheaves mentioned in the statement. We assume due to [13, Proposition 2.1] that
[TABLE]
and consider the full exceptional collection E∙ given in (21) and collection F∙ given in (22).
We construct a Beilinson complex, quasi-isomorphic to A, by calculating Hi+kj(A⊗Ej)⊗Fj with i,j∈{0,…,8} to get the following table. Here we use several vanishing in the intermediate cohomology of A,A(−1,−1),A(−2,−2),A(−3,−3) together with vanishing of Lemma 3.6:
[TABLE]
Since h2(V⊗Ω(−1)⊠Ω(−1))=3a3+3b3=3a1+3b1=0 we get
[TABLE]
so we obtain the following table:
[TABLE]
Finally, since Ext1(OX(−1,1),OX(1,−1))=Ext1(OX(1,−1),OX(−1,1))=0 we get the claimed result.
∎
Theorem 4.8**.**
Let V be an Ulrich bundle on X, then
(1)
if a0=a4=0, then V≅Ω(2)⊠Ω(3) or V≅OX(0,2);
2. (2)
if b0=b4=0, then V≅Ω(3)⊠Ω(2) or V≅OX(2,0);
3. (3)
if V is a twist of a pullback from P2 then V≅OX(2,0), or V≅OX(0,2) or V arises from sequences (25), (26).
Proof.
We prove only (2) and (3).
Since an Ulrich bundles V arising from
[TABLE]
or
[TABLE]
do not satisfy the condition b4=0 or b0=0 we may assume H1(V⊗Ω⊠Ω)=3a2+3b2=0 so a2=h2(V∨(0,−1))=h2(V(−3,−2))=0 or a2=h2(V∨(0,−1))=h2(V(−3,−2))=0.
If a2=0, from the exact sequence
[TABLE]
tensored by V, since h2(V(−2,−2))=0 and h1(V⊗Ω⊠O(−1))=b0=0 we get a surjection from H0(V⊗Ω⊠Ω(1))=H0(V⊗(Ω(2)⊠Ω(3))∨) to H2(V(−3,−2))=0. Moreover from the exact sequence
[TABLE]
tensored by V∨, since h2(V∨)=0 and h1(V∨⊗O(1)⊠Ω(2))=h3(V⊗O(1)⊠Ω(−2))=b4=0 we get a surjection from H0(V∨⊗Ω(3)⊠Ω(2)) to H2(V∨(0,−1))=0.
So we may conclude that V≅Ω(3)⊠Ω(2) by arguing as in [4].
Finally let assume b2=0 and a1=a3=0, from the exact sequence
[TABLE]
tensored by V, since h2(V(−2,−2))=0 and h1(V(−2,−1))=a1=0 we get a surjection from H0(V(−2,0))=H0(V⊗(O(2,0))∨) to H2(V(−2,−3))=0. Moreover from the exact sequence
[TABLE]
tensored by V∨, since h2(V∨)=0 and h1(V∨(1,0))=h3(V(−4,−3))=a3=0 we get a surjection from H0(V∨(2,0)) to H2(V∨(−1,0))=0.
So we may conclude that V≅O(2,0) (see [4]) and (2) is proven.
In order to prove (3) let assume that V is a twist of a pullback from the second copy of P2.
First let us consider the case V=OP2⊠B where B is a vector bundle on P2. Since hi(V(−3))=hi(V(−4))=0 for any i but h2(OP2(−3))=0 and h2(OP2(−4))=0 we must have hi(B(−3))=hi(B(−4))=0 for any i. So we may deduce that B(−2) is Ulrich on P2 hence V=OX(0,2).
Now let us consider the case V=OP2(1)⊠B where B is a vector bundle on P2. Since hi(V(−1))=hi(V(−4))=0 for any i but h0(OP2)=0 and h2(OP2(−3))=0 we must have hi(B(−1))=0 and hi(B(−4))=0 for any i. In particular h2(B(t))=0 for any t≥−4. We consider the Beilinson type spectral sequence associated to B(−1)
with E∙={OP2(−1),Ω(1),OP2} and F∙={OP2(−2),OP2(−1),OP2} given in (22).
Finally let us consider the case V=OP2(t)⊠B where B is a vector bundle on P2 and t≥2. Since h0(V(−1))=0 we must have hi(B(−1))=0 and we may deduce that b0=0. Since h2(OP2(t−4))=0 we deduce that b4=0. So by (2) we obtain V=OX(2,0).
∎
Remark 4.9**.**
We have just proved that the Ulrich bundles obtained as a pullback from P2 are rigid or they are in the high dimensional families defined in [12]. For instance the closure of the family of rank two bundles M2 in the associated moduli space is a generically smooth component of dimension 5 (see [12] Theorem 3.9.).
Ω(3)⊠Ω(2) and Ω(2)⊠Ω(3) are the only Ulrich bundles characterized so far which are not pullbacks.
Another cohomological characterization of Ω(3)⊠Ω(2) and Ω(2)⊠Ω(3) can be found in [21].
So far we have seen and characterized families of Ulrich bundles with a0=0 or b0=0. Now we construct three interesting families with both a0=0 and b0=0:
Example 4.10**.**
Since Ext1(O(2,0),Ω(2)⊠Ω(3))≅H1(Ω⊠Ω(3))≅C8 we get a 7-dimensional familiy of rank 5 Ulrich bundles arising from the following extension
From these rank 8 simple Ulrich bundles we may obtain higher rank Ulrich bundles by other extension with Ω(3)⊠Ω(2) or with Ω(2)⊠Ω(3).
Now we have a more complicated example arising from the following proposition:
Proposition 4.12**.**
The generic element ψ∈Ext1(O⊠Ω(2),O(1)⊠Ω) gives an extension
[TABLE]
where Eψ(1) is a simple Ulrich bundle of rank 4 for which a0=0 and b0=0.
Proof.
Let W=Ext1(Ω(2),Ω)≅C3. Let us notice that any nonzero η∈W gives an exact sequence
[TABLE]
and the map in cohomology
[TABLE]
is nonzero.
On P(W∨)×P2 we have Ext1(O⊠Ω(2),O(1)⊠Ω)≅H1(O(1)⊠(Ω(2))∨)≅H0(P2,O(1))⊗H1(P2,Ω⊗(Ω(2))∨)≅W∨×W. Now let us consider the identity element I∈W∨×W (I restricts to η on P(W∨)) and we obtain
[TABLE]
In cohomology we get the map
[TABLE]
where
[TABLE]
and
[TABLE]
We may conclude that δI is equivalent to
[TABLE]
Since φ(η)=δη=0, φ must be an inclusion. But dim(H0(P2,Ω(2))∨×H1(P2,Ω))=dim(W)=3 so φ is an isomorphism.
Then we have showed that the generic element ψ∈Ext1(O⊠Ω(2),O(1)⊠Ω) gives an extension (29) with
[TABLE]
So we obtain H0(Eψ)=H1(Eψ)=0. Moreover we compute
[TABLE]
for any i. Then Eψ(1) is Ulrich with both a0=0 and b0=0.
From the dual of (29) it is easy to check that h1(Eψ⊗Eψ∨)=1, hence Eψ is simple.
∎
5. Ulrich bundles on the flag variety F(0,1,2)
Let F⊆P7 be the del Pezzo threefold of degree 6 and Picard number two. Let us consider F as an hyperplane section of P2×P2 with the two natural projections pi:F⊂P2×P2P2 and the following rank two vector bundles:
[TABLE]
We may consider the full exceptional collection
[TABLE]
[TABLE]
and
[TABLE]
Theorem 5.1**.**
Let V be an Ulrich bundle on F. Then V arises from an exact sequence of the form:
[TABLE]
Proof.
We consider the Beilinson type spectral sequence associated to A:=V(−1,−1) and identify the members of the graded sheaf associated to the induced filtration as the sheaves mentioned in the statement. We consider the full exceptional collection E∙ and left dual collection F∙ in (30) and (31).
We construct a Beilinson complex, quasi-isomorphic to A, by calculating Hi+kj(A⊗Fj)⊗Ej with i,j∈{0,…,6} to get the following table:
since H3(A⊗G1(−2,−1))=0 we get H2(A(0,−1))=0. In a similar way we get H2(A(−1,0))=0. So the table become
[TABLE]
Since the spectral sequence converges to an object in degree [math] we get e=f=0, so
[TABLE]
Finally since Exti(OF(−1,0),OF(0,−1))=0, Exti(OF(−1,0),G1)=0, Exti(OF(−1,0),G2)=0, Exti(OF(0,−1),G1)=0, Exti(OF(0,−1),G2)=0 and Exti(G1,G2)=0 for any i>0 we have that the full exceptional collection (31) is strong. So we get the claimed resolution.
∎
Remark 5.2**.**
Let V an indecomposable Ulrich bundle on F.
(1)
If c=h1(V⊗G2(0,1))=0 and b=h1(V(0,1))=0 then V is the restriction of a bundle arising from sequence (15).
If d=h1(V⊗G1(1,0))=0 and a=h1(V(1,0))=0 then V is the restriction of a bundle arising from sequence (16).
3. (2)
If d=b=0 we obtain the exact sequence
[TABLE]
so V≅OF(2,0).
If c=a=0 we obtain V≅OF(0,2).
Bibliography26
The reference list from the paper itself. Each links out to its DOI / PubMed record.
1[1] M. Aprodu, L. Costa and R. M. Miró-Roig, Ulrich bundles on ruled surfaces , J. Pure Appl. Alg. 222 issue 1 (2018) 131–138.
2[2] M. Aprodu, S. Huh, F. Malaspina and J. Pons-Llopis, Ulrich bundles on smooth projective varieties of minimal degree , Proc. AMS 147 (2019), 5117–5129.
3[3] V. Antonelli, Characterization of Ulrich bundles on Hirzebruch surfaces , ar Xiv:1806.10380 to appear on Rev. Mat. Complutense, doi.org/10.1007/s 13163-019-00346-7.
4[4] E. Ballico, F. Malaspina, Regularity and Cohomological Splitting Conditions for Vector Bundles on Multiprojectives Spaces , J. of Algebra 345, 137-149 (2011).
5[5] C. Böhning, Derived categories of coherent sheaves on rational homogeneous manifold . Doc. Math. 11 (2006), 261–331.
6[6] G. Casnati, D. Faenzi and F. Malaspina, Rank two a CM bundles on the del Pezzo fourfold of degree 6 and its general hyperplane section , J. Pure Appl. Alg. 222 issue 3 (2018) 585–609.
7[7] E. Coskun and ö. Genc Ulrich bundles on Veronese surfaces Proc. AMS 145 (2017), 4687–4701.
8[8] I. Coskun, L. Costa, J. Huizenga, R. M. Miró-Roig and M. Woolf, Ulrich Schur bundles on flag varieties , J. Algebra 474 (2017), 49–96.