This paper explores the combinatorial construction of Calabi-Yau threefolds within Hibi toric varieties, focusing on their singularities, smoothability, and topological invariants, with new examples illustrating these properties.
Contribution
It provides a combinatorial framework for describing Calabi-Yau threefolds in Hibi toric varieties and analyzes their smoothing and topological invariants.
Findings
01
Calabi-Yau threefolds in Hibi toric varieties often have conifold singularities.
02
Many such threefolds are smoothable to non-singular Calabi-Yau threefolds.
03
Explicit calculations of topological invariants for new examples are presented.
Abstract
In this article, we summarize combinatorial description of complete intersection Calabi-Yau threefolds in Hibi toric varieties. Such Calabi-Yau threefolds have at worst conifold singularities, and are often smoothable to non-singular Calabi-Yau threefolds. We focus on such non-singular Calabi-Yau threefolds of Picard number one, and illustrate the calculation of topological invariants, using new motivating examples.
Tables1
Table 1. Table 1: Examples of X 𝑋 X of Picard number one.
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.
Taxonomy
TopicsAlgebraic Geometry and Number Theory · Advanced Combinatorial Mathematics · Commutative Algebra and Its Applications
Full text
Complete intersection Calabi–Yau threefolds in Hibi toric varieties
and their smoothing
Makoto Miura
Abstract
In this article,
we summarize combinatorial description of
complete intersection Calabi–Yau threefolds
in Hibi toric varieties.
Such Calabi–Yau threefolds
have at worst conifold singularities, and
are often smoothable to
non-singular Calabi–Yau threefolds.
We focus on
such non-singular Calabi–Yau threefolds of Picard number one,
and illustrate the calculation of topological invariants,
using new motivating examples.
1 Introduction
A Hibi toric variety is defined as a projective toric variety PΔ(P)
associated with an order polytope
[TABLE]
for a finite poset P=(P,≺).
For example, all products of projective spaces are Hibi toric varieties; hence
at least 2590 topologically distinct non-singular Calabi–Yau threefolds are
obtained as complete intersections [12].
In general, complete intersection Calabi–Yau threefolds in Hibi toric varieties
have finite number of nodes,
and are often smoothable to non-singular Calabi–Yau threefolds by flat deformations.
Complete intersections in Grassmannians
(or more generally in minuscule Schubert varieties)
give basic examples of such smoothing
[5, 17].
The purpose of this article is to
provide a brief summary on
combinatorial descriptions of
complete intersection Calabi–Yau threefolds
in Hibi toric varieties and their smoothing.
Based on [7],
we describe the smoothability in terms of posets
(Proposition 3.6), and survey the calculation of
topological invariants for resulting non-singular simply-connected
Calabi–Yau threefolds (Subsection 4.2),
by focusing on the case of Picard number one for simplicity.
In addition to the summary,
we show the simply-connectedness
as a corollary of the result on small resolutions for Hibi toric varieties
(Proposition 2.6).
To illustrate the calculation,
we introduce several new examples of
such non-singular Calabi–Yau threefolds
of Picard number one (Subsection 4.3, Table 1).
A Calabi–Yau threefold is a complex projective threefold X
with at worst canonical singularities satisfying
ωX≃OX and H1(X,OX)=0.
There are a huge number of such threefolds,
even non-singular.
Mirror symmetry is a conjectural duality
between a non-singular Calabi–Yau threefold X and
another non-singular Calabi–Yau threefold X∗,
called a mirror manifold for X.
Various non-trivial relations between X and X∗
are expected.
For example, Hodge numbers satisfy
[TABLE]
One of the big mysteries of mirror symmetry
is whether
every non-singular Calabi–Yau threefold X has a mirror manifold X∗ or not.
Note an obvious exception in the case with h2,1(X)=0, and that
the mirror manifold X∗ is not unique in general,
even as topological manifolds.
There is an excellent class of non-singular Calabi–Yau threefolds such that
the above question has an affirmative answer;
for crepant resolutions of complete intersection Calabi–Yau threefolds
in Gorenstein toric Fano varieties,
we have mirror manifolds in the same class, called the Batyrev–Borisov mirrors [3, 8].
In order to expand this class,
the conjectural mirror construction via conifold transitions seems to be a
promising direction.
Let X0 be a Calabi–Yau threefold
with finitely many nodes.
Suppose that
X0 admits a smoothing X⇝X0 by a flat deformation,
and a small resolution Y→X0.
The composite operation connecting two non-singular Calabi–Yau threefolds
X and Y is called a conifold transition:
[TABLE]
There is a natural closed immersion of the Kuranishi space Def(Y) to Def(X0)
[19, Proposition 2.3], and hence, it makes sense to put them together
into some giant moduli space.
There is a question, commonly referred to as (a version of) Reid’s fantasy,
which asks whether all simply-connected non-singular Calabi–Yau threefolds
fit together into a single irreducible family via conifold transitions
[20].
Suppose that
X and Y have torsion-free homology
for a conifold transition (3).
Morrison’s conjecture in [18] says that the mirror manifolds
are also connected via a conifold transition of the opposite direction:
[TABLE]
Together with the spirit of Reid’s fantasy,
one may expect a mirror construction
for a large number of non-singular Calabi–Yau threefolds
from the Batyrev–Borisov mirror pairs.
We still do not know the existence of a mirror manifold X∗,
even for the smoothing X⇝X0 of
a complete intersection Calabi–Yau threefold X0 in a Hibi toric variety.
Nevertheless, we can discuss the mirror symmetry
by calculating periods and Picard–Fuchs operators for the conjectural mirror family,
as we see in Remark 4.3 for example.
2 Hibi toric varieties
2.1 Examples
Let us begin with simple examples of Hibi toric varieties.
For the empty poset,
we set the Hibi toric variety PΔ(∅) to be a point.
For a singleton
u:={u} (by abuse of notation),
the order polytope is a line segment Δ(u)=[0,1], and hence,
the Hibi toric variety PΔ(u) is
a projective line P1.
Let P be a finite poset consisting of n:=∣P∣ elements.
If P is a chain, i.e., a totally ordered set,
the order polytope Δ(P) is a regular simplex
defined by the inequalities
0≤x1≤⋯≤xn≤1,
so that the Hibi toric variety PΔ(P) is
a projective space Pn.
It is equally clear the case that P is an anti-chain,
i.e., the poset in which every pair of elements is incomparable.
In this case, the order polytope Δ(P) is a unit hypercube [0,1]n,
and the Hibi toric variety PΔ(P) is
the product of n copies of P1.
Example 2.1**.**
A first non-trivial example is a poset P={u,v,w}
with the partial order defined by u≻w and v≻w.
The defining inequalities of
the order polytope Δ(P) is shown in the left of Figure 1,
also depicted symbolically in the middle.
It becomes a pyramid in RP≃R3
as shown in the right of Figure 1.
Therefore, the associated Hibi toric variety PΔ(P) is
a projective cone over P1×P1 with a general apex in P3.
A disjoint unionP=P1+P2 of finite posets P1 and P2
is a disjoint union as sets
equipped with the partial order ≺ satisfying
(i) u∈P1, v∈P1 and u≺v∈P1 imply u≺v∈P,
(ii) u∈P2, v∈P2 and u≺v∈P2 imply u≺v∈P,
and
(iii) u∈P1 and v∈P2 imply u∼v∈P (i.e., u and v are incomparable
in P).
The corresponding Hibi toric variety is
projectively equivalent to
the product of two Hibi toric varieties,
[TABLE]
A ordinal sumP=P1⊕P2 of P1 and P2
is a disjoint union as sets
equipped with the partial order ≺ satisfying
the same (i) and (ii) as the disjoint union P1+P2 above,
and (iii)′u∈P1 and v∈P2 imply u≺v∈P.
Note that the operation ⊕ is not commutative though it is associative.
The corresponding Hibi toric variety is
a special hyperplane section
of a projective join
of two Hibi toric varieties with general positions,
[TABLE]
These operations generalize the examples,
a chain P=⨁i=1nui,
an anti-chain P=∑i=1nui,
and P=w⊕(u+v)=∅⊕w⊕(u+v) in Example 2.1.
The posets built up
by disjoint unions and ordinal sums
from singletons are sometimes called series-parallel posets.
One of the simplest examples that are not series-parallel is the poset
with the Hasse diagram:
[TABLE]
Recall that, in a Hasse diagram for a poset P,
a vertex represents an element of P
and an oriented edge represents a covering relationu\mathrel{\ooalign{<\cr\hbox{\cdot\mkern 1.0mu}\cr}}v on P, that is,
[TABLE]
For example,
the Hasse diagram (7) represents the poset P={a,b,c,d} with
a≺b≻c≺d.
The associated Hibi toric variety PΔ(P)
is a limit of a toric degeneration of a general linear section fourfold
in a Grassmannian G(2,5).
2.2 Invariant subvarieties and singularities
Invariant subvarieties of Hibi toric varieties are again
(projectively equivalent to) lower dimensional Hibi toric varieties.
We follow the description of invariant subvarieties by Wagner [22].
Let P be a finite poset. The associated
bounded poset is defined as
[TABLE]
where 0^ and 1^ are singletons.
By definition, the elements 0^ and 1^ are
the unique minimal and the maximal elements in P^,
respectively.
Note that the Hasse diagram of P^
can be regarded as the graph describing the defining inequalities of
order polytope Δ(P), as we see in the middle of Figure 1.
We use this identification between inequalities with edges,
and variables with vertices for the Hasse diagram of P^.
Furthermore, by abuse of notation,
we write the same symbol P as the Hasse diagram of P.
For example, we say
that P is connected if the Hasse diagram of P is connected,
and P is a cycle if the Hasse diagram of P is a cycle
as an unoriented graph, and so on.
Definition 2.2**.**
Let P^ be a bounded poset.
A surjective order-preserving map
[TABLE]
with φ(0^)=0^ and
φ(1^)=1^
is called a contraction of P^ if
every fiber is connected and
there exists a covering relation u\mathrel{\ooalign{<\cr\hbox{\cdot\mkern 1.0mu}\cr}}v\in\hat{P}
for all \bar{u}\mathrel{\ooalign{<\cr\hbox{\cdot\mkern 1.0mu}\cr}}\bar{v}\in\hat{P}^{\prime}
such that uˉ=φ(u) and vˉ=φ(v).
There is a one-to-one correspondence between
faces of order polytope Δ(P) and contractions of
the associated bounded poset P^.
More precisely, a face θφ corresponding to a contraction
φ:P^→P′^
is unimodular equivalent to the order polytope Δ(P′).
In other words,
the associated invariant subvariety
of a Hibi toric variety PΔ(P) is
projectively equivalent to the Hibi toric variety PΔ(P′),
as mentioned at the beginning of this subsection.
In particular, we have one-to-one correspondences between
facets of Δ(P) and edges of P^,
and vertices of Δ(P) and order ideals of P.
Here an order ideal is defined as a subset τ⊂P satisfying
[TABLE]
Let us write the set of edges of P^ as E=Edges(P^),
and the set of order ideals of P as J(P).
We illustrate the correspondences
by using the poset P in Example 2.1.
We have five facets corresponding to E, and
five vertices corresponding to J(P).
The defining equalities of a face θφ
can be obtained
by making all variables in a fiber φ−1(uˉ) equal.
In Figure 2, four facets of the order polytope Δ(P)
are meeting at the same vertex circled.
Hence the corresponding point should be singular in PΔ(P).
In general,
a singular locus comes from a contraction
replacing more inequalities to equalities than codimension.
A subposet C⊂P^ is said to be convex if it satisfies
[TABLE]
Furthermore, let us call C⊂P^ a minimal convex cycle
if C is (i) a full subposet not containing both 0^ and 1^, and
(ii) a convex cycle such that all convex full subposets C′⊂C are trees.
Let P be a finite poset.
An irreducible singular locus of PΔ(P)
corresponds to a minimal convex cycle C⊂P^.
For the corresponding contraction,
all the fibers are singletons except one fiber C.
Remark 2.4**.**
Now it is worth noting the homogeneous coordinate rings of Hibi toric varieties.
Since all the lattice points in Δ(P) are vertices,
the homogeneous coordinate ring of PΔ(P) is
a Hibi algebra,
[TABLE]
where
C[J(P)] is the polynomial C-algebra in variables pτ for τ∈J(P),
and IJ(P) is the ideal coming from linear relations of vertices of Δ(P).
In fact, the ideal IJ(P) has
quadratic generators,
[TABLE]
Note here that J(P) is a lattice, i.e., a poset
with the least upper bound α∨β and the greatest lower bound α∧β for each pair of elements.
In fact, J(P) with the partial
order given by set inclusions is equipped with
α∨β=α∪β
and α∧β=α∩β.
Furthermore, J(P) becomes a distributive lattice, i.e.,
the lattice
with distributive laws,
[TABLE]
One may start from
finite distributive lattices instead of finite posets,
which gives another description of Hibi toric varieties in literature.
Example 2.5**.**
Under the notation in Remark 2.4,
the Hibi toric variety PΔ(P) in Example 2.1 is
embedded as a quadric threefold in P4 defined by
[TABLE]
2.3 Divisors
Let P be a finite poset.
For each edge e∈E=Edges(P^),
we have the corresponding invariant prime divisor,
denoted by De.
We write DE′=∑e∈E′De for each subset E′⊂E.
The linear equivalences in the divisor class group Cl(PΔ(P))
are generated by the following relations:
[TABLE]
where we write t(e)\mathrel{\ooalign{<\cr\hbox{\cdot\mkern 1.0mu}\cr}}s(e) for each e=(s(e),t(e))∈E⊂P^×P^.
Let us describe the Picard group of PΔ(P).
First, suppose P is connected.
For an order ideal τ⊂P and a subset E′⊂E,
a set of edges
[TABLE]
defines a Weil divisor
DE′(τ)
on PΔ(P).
We have DE(τ)≃DE(P) for all τ∈J(P).
The divisor DE(P) is, in fact, the Cartier divisor
corresponding to the lattice polytope Δ(P) itself.
One can show that the Picard group is isomorphic to Z
generated by the associated very ample invertible sheaf
O(1)=O(DE(P)).
More generally,
suppose P=∑j=1ρPj with ρ connected components P1,…,Pρ.
Note that there is a natural decomposition as sets
[TABLE]
where Ej:=Edges(P^j)⊂E for each j.
The Picard group Pic(PΔ(P)) becomes a free abelian group of rank ρ
generated by
[TABLE]
for
each connected component Pj⊂P.
We have
[TABLE]
Next, we note a formula for the self-intersection number,
[TABLE]
where cJ(P) denotes the number of maximal chains on J(P).
It follows from a formula for the Hilbert–Poincaré series of Hibi algebra AJ(P)
obtained by [15, Corollary of Lemma 5].
Lastly, let us suppose P is pure.
Recall that a finite poset P is called pure
if every maximal chain on P has the same length.
We define a height h(u) of u∈P^ as the length of
the longest chain bounded above by u in P^, and
write hP=h(1^).
Thus an anti-canonical divisor −KPΔ(P)=DE is written as
[TABLE]
where
τk:={u∈P∣h(u)<k}∈J(P)
for k=1,…,hP.
Together with (21),
it turns out that
the Hibi toric variety PΔ(P)
for a pure poset P is a Gorenstein Fano variety
with ω∨=O(−KPΔ(P))≃O(hP).
Moreover,
one can show that it has at worst terminal singularities,
by looking at the normal fan Σ of Δ(P)
(see [13, Lemma 1.4]).
2.4 Small resolution
Let P be a finite pure poset.
The associated Hibi toric variety PΔ(P)
is a Gorenstein terminal Fano variety with ω∨≃O(hP).
If PΔ(P) is Q-factorial in addition,
it turns out to be non-singular, and even more, a product of projective spaces
by [13, Corollary 2.4].
Even if it is not Q-factorial,
we have the following property indicating the mildness of singularities of PΔ(P).
Proposition 2.6**.**
For a finite pure poset P,
any toric crepant Q-factorialization
of the Hibi toric variety PΔ(P)
is a small resolution.
Proof.
Let P be a finite poset, N=ZP and M=ZP
the free abelian groups of rank n=∣P∣ dual to each other,
and NR=RP and MR=RP the real scalar extensions, respectively.
First, let us see a description of
the normal fan Σ in NR for
the order polytope Δ(P)⊂MR.
By definition, a one-dimensional cone in Σ
is generated by the normal vector of a facet of Δ(P).
Hence it corresponds to an edge of P^.
Let δ(e)∈N denote such primitive vector
associated with e∈E,
The map δ is extended
to be the composite linear map
δ=pr∘∂:ZE→N of
[TABLE]
and a projection pr:N⊕Z0^⊕Z1^→N.
By using the same symbol δ as the real extension,
each maximal dimensional cone in Σ
associated with an order ideal τ∈J(P)
is written as
[TABLE]
where E(τ) is defined by (18).
On the other hand,
Convδ(E) is a Gorenstein terminal Fano polytope by
[13, Lemma 1.3–1.5].
Namely, for any τ∈J(P), all the primitive generators of the cone στ,
i.e., the elements in δ(E∖E(τ)),
lie on an affine hyperplane
with integral distance one from the origin, and it holds
[TABLE]
Suppose P is pure, and
let
XΣ→PΔ(P)
be a toric crepant Q-factorialization.
In other words, Σ is a maximal simplicial refinement
of the normal fan Σ of Δ(P) such that XΣ
denotes the corresponding Q-factorial toric variety.
Since PΔ(P) has at worst terminal singularities,
the crepant birational morphism
XΣ→PΔ(P)
is a small modification by definition.
Hence it is sufficient to show that XΣ is non-singular.
Fix a maximal dimensional cone σ in Σ.
Since (25) and (26),
there exist an order ideal τ⊂P and
a subset
B⊂E∖E(τ)
consisting of n+1 elements
such that
σ=Coneδ(B)⊂στ.
As in the example shown in Figure 3,
the subgraph (P^,E∖E(τ)) of the Hasse diagram of P^
defining στ consists of two connected graphs, and
the subgraph (P^,B)
defining σ consists of two connected tree graphs.
In fact, if (P^,B) contains a cycle,
σ cannot have maximal dimension.
Therefore, we have a unique unoriented path in (P^,B)
from any u∈P to 0^ or 1^,
which attains a value ±u∈Zδ(B) by summing up and mapping by δ.
Hence δ(B) forms a Z-basis of N=ZP.
Since σ is arbitrary, it follows that
XΣ is non-singular.
∎
3 CICY threefolds in Hibi toric varieties
3.1 Examples
We describe Calabi–Yau threefolds
obtained as a complete intersection
of general sections of invertible sheaves
in Hibi toric varieties.
We call such Calabi–Yau threefolds
complete intersection Calabi–Yau (CICY)
threefolds in Hibi toric varieties.
Let P be a finite poset,
and X0 a CICY threefold in PΔ(P).
From the adjunction formula,
PΔ(P)
has at worst Gorenstein singularities.
In other words, all connected components of P are pure.
If P is a disjoint union of several pure connected posets,
we have a number of Calabi–Yau threefolds
as complete intersections of nef divisors in PΔ(P),
e.g.,
complete intersection Calabi–Yau threefolds in
products of projective spaces.
However, we assume in the sequel
that P is pure connected for simplicity.
Under this assumption,
X0 is merely a complete intersection of very ample divisors in PΔ(P).
Let (d1,…,dr)⊂PΔ(P) denote
a complete intersection variety of
very ample divisors defined by
general sections
of O(d1),…,O(dr), respectively.
Then X0=(d1,…,dr)⊂PΔ(P) is
a CICY threefold if and only if
[TABLE]
Example 3.1**.**
The poset in (7) gives an example of
hypersurface Calabi–Yau threefolds in Hibi toric varieties.
Thus X0=(d1=3)⊂PΔ(P) in this case.
Example 3.2**.**
As an example to illustrate calculations,
we introduce a finite pure connected poset P=P1:
[TABLE]
We have ∣P1∣=6 and hP1=3, and hence,
the associated Hibi toric variety PΔ(P1)
is a six-dimensional Gorenstein terminal Fano variety with
ω∨≃O(3).
We have a linear section Calabi–Yau threefold
X0=(13)⊂PΔ(P1).
The first part of J(P1) corresponds to order ideals in the
left of Figure 4.
By continuing while focusing on set inclusions,
we obtain the lattice J(P1) as in the middle of Figure 4,
consisting of ∣J(P1)∣=18 elements.
Moreover, we have cJ(P1)=48, the number of maximal chains on J(P1),
by counting as in the
right in Figure 4.
3.2 Stringy Hodge numbers
Let P be a pure connected poset and X0=(d1,…,dr)⊂PΔ(P).
We have a small resolution Y→X0, by taking the strict transform of X0
for a small toric resolution XΣ→PΔ(P) for example.
In this case,
the stringy Hodge numbers of X0 are nothing but
usual Hodge numbers of Y.
From [4, Proposition 8.6],
the following combinatorial formulas hold.
Proposition 3.3**.**
[TABLE]
where l(θ) and l∗(θ)
denote the number of lattice points in a face θ⊂MR
and in the interior of θ, respectively;
[r]={1,…,r},
dJ=∑j∈Jdj and θe
is the facet of Δ(P) corresponding to an edge e∈E.
Note that a nonzero contribution in the first term of
(30)
comes only from the range of di−dJ≥0,
and in the second term it comes only from the range dJ=hP−1 or hP.
In particular, if (i) dj=1 for all j, and (ii)
P has no ordinal summand of singleton,
i.e., P=P1⊕u⊕P2 for any P1 and P2, we have
[TABLE]
Example 3.4**.**
For the example P=P1
and a complete intersection X0=(13),
we obtain hst1,1(X0)=12−6=6 from (29).
Since the both conditions (i) and (ii) are satisfied,
the simplified formula (31) holds in this case.
By the symmetry S3×S3 of P^1 as a poset
and the order duality,
it suffices to see
the following two types of facets:
[TABLE]
There are six facets for each type, and clearly
l∗(3θ)=1 (resp. [math])
for the former (resp. the latter) type.
Therefore hst1,2(X0)=3(18−3)−6−6=33.
3.3 Numbers of nodes
Recall that three-dimensional Gorenstein terminal toric singularities
are at worst nodes (i.e, ordinary double points),
since they are presented by three-dimensional cones over a unit triangle or a unit square.
Together with the Bertini-type theorem for toroidal singularities,
the singularities of X0 are also at worst nodes.
We count the number of nodes dp(X0) on X0 in the following.
Each node on X0
lies on one singular locus of codimension three
of PΔ(P),
corresponding to
a minimal convex cycle
C⊂P^
with four elements.
There are four types of such minimal convex cycles:
[TABLE]
Let Λ4(P^) denote the set of
such minimal convex cycles consisting of four elements on P^.
For each C∈Λ4(P^),
there is the contraction P^→P^C such that
all the fibers are singleton except one fiber C.
Of course it holds ∣PC∣=∣P∣−3 for all C∈Λ4(P^).
Hence
C defines a singular locus of codimension three and of
degree degΔ(PC)=cJ(PC) from (22).
Therefore, the number of nodes dp(X0)
is calculated by a formula
[TABLE]
Example 3.5**.**
There are six minimal convex cycles on P1^,
each of which consists of four elements.
By symmetry, they are all equivalent to
[TABLE]
Since the locus is a quadric threefold in Example 2.5,
we obtain dp(X0)=6⋅2=12.
3.4 Smoothability
For smoothability, we follow the argument in the case of hypersurfaces in
toric varieties by [7].
Let
{p1,…,pdp} be the set of nodes on X0,
where dp=dp(X0), and
f:Y→X0 be a small resolution.
The exceptional lines Li:=f−1(pi)≃P1 for i=1,…,dp
form a linear subspace of H2(Y,C).
By [19, Theorem 2.5], the Calabi–Yau threefold X0 is smoothable
by a flat deformation
if and only if
the homology classes [Li]∈H2(Y,C) satisfy a relation,
[TABLE]
where αi=0 for all i=1,…,dp.
Note that one can identify
[TABLE]
Under this identification,
the homology class [Li] coincides with
a relation,
[TABLE]
up to signs,
where the corresponding node pi lies on a singular locus
associated with a minimal convex cycle C∈Λ4(P^),
and the cycle C with an orientation
passes through the four edges;
ep,eq in the forward direction and er,es in the opposite direction.
Proposition 3.6**.**
Let P be a pure connected poset,
X0 a CICY threefold
of degree (d1,…,dr)
in the Hibi toric variety PΔ(P).
If ∏j=1rdj>1, X0 is smoothable.
If ∏j=1rdj=1,
X0 is smoothable if and only if
for any C∈Λ4(P^) such that PC is a chain
the element ρC is a linear combination of the
remaining elements ρC′
with C′∈Λ4(P^), C=C′.
Example 3.7**.**
For the example P=P1,
X0=(13) is smoothable since PC is not a chain for all C∈Λ4(P^1)
as we see in (35),
although ∏j=13dj=1.
Example 3.8**.**
There are cases that X0 is not smoothable.
For example, in (39) we present the two cycles on the depicted P^
satisfying the condition that PC is a chain,
[TABLE]
The former cycle is a linear combination of remaining cycles
as expressed by abuse of notation.
However, the latter cycle is linearly independent to other cycles.
Therefore, X0=(14) is not smoothable by Proposition 3.6.
4 Smoothing of CICY threefolds in Hibi toric varieties
4.1 Simply-connectedness
Proposition 4.1** (Corollary of Proposition 2.6).**
Let P be a pure poset, X0 a CICY threefold in PΔ(P),
and Y→X0 a small resolution.
Then Y is simply-connected.
If a smoothing X⇝X0
exists, X is also simply-connected.
Proof.
By Proposition 2.6,
we have a small resolution XΣ→XΣ=PΔ(P).
Since XΣ is a compact toric variety,
it is simply-connected (see for example [10, Theorem 9.1]).
Let Σ(1) denote the subfan
of Σ consisting of cones of dimension less than or equal to one.
Note Σ(1)=Σ(1).
The quasi-projective toric variety XΣ(1)
is simply-connected as well.
In fact, the difference between
XΣ and XΣ(1)
in subsets of real codimension four
does not effect fundamental groups.
By the Lefschetz theorem for non-singular quasi-projective manifolds
[11, 14],
a complete intersection
XΣ(1)∩X0=XΣ(1)∩Y is also simply-connected.
Similarly as above,
the difference between XΣ(1)∩Y and Y
does not effect fundamental groups.
Therefore Y is also simply-connected.
The latter statement follows from the fact that
a conifold transition does not change fundamental groups.
∎
Let X be a smoothing of a CICY threefold in Hibi toric variety.
We do not know whether homology groups of X can have torsion or not.
Suppose that X has torsion-free homology and h1,1(X)=1.
In this case, by Wall’s theorem [23, Theorem 5],
the diffeomorphism class
of X is determined only by the three topological invariants,
[TABLE]
where
H is the hyperplane class,
c2(X) is the second Chern class
and χ(X)=2(h1,1(X)−h2,1(X)) is the topological Euler number
of X.
We summarize the calculation of these topological invariants
in the next subsection.
4.2 Topological invariants
For a conifold transition
X⇝X0←Y,
Hodge numbers satisfy
[TABLE]
where
rk=rk(X0)
is the dimension of
linear subspace of
H2(Y,C)
spanned by classes of
exceptional lines [Li] for i=1,…dp, i.e.,
[TABLE]
and
dp=dp(X0)
is the number of
nodes on X0, which we compute by (34).
In particular, from
h1,1(Y)=∣E∣−∣P∣=b1(P^)+1,
we have h1,1(X)=1 if and only if
all minimal cycles
on P^ are generated by cycles in
Λ4(P^).
Assume h1,1(X)=1.
From (22) and the invariance by a flat deformation, it holds
[TABLE]
Since also the invariance
χ(X,OX(1))=χ(X0,OX0(1)) and a standard
cohomology calculation
for complete intersection varieties in PΔ(P),
we obtain
[TABLE]
where r1=#{j∣dj=1}.
Therefore the Hirzebruch–Riemann–Roch theorem gives
[TABLE]
Example 4.2**.**
For P=P1, h1,1(X)=1 holds since all minimal cycles are in
Λ4(P1^).
Hence rk(X0)=5 by (41).
From h1,2(Y)=33 and dp(X0)=12,
it holds h1,2(X)=40 and χ(X)=2(h1,1(X)−h1,2(X))=−78
by (42).
We also obtain deg(X)=48 and c2(X)⋅H=12(18−3)−2⋅48=84
by (44) and (49), respectively.
4.3 Examples
Let P be a pure poset and X0 a CICY threefold in PΔ(P).
We write X=XP for a smoothing X⇝X0 if it exists.
In spite of the large number of smoothable
CICY threefolds in Hibi toric varieties,
there are few examples of XP with h1,1(XP)=1.
From [17, Proposition 3.1],
we have twelve such threefolds
as complete intersections in minuscule Schubert varieties
up to deformation equivalence;
five in projective spaces,
five in other Grassmannians G(k,n),
one in an orthogonal Grassmannian OG(5,10),
and one in a singular Schubert variety of the Cayley plane OP2.
The latter two threefolds are also regarded as
complete intersections of some homogeneous vector bundles
on Grassmannians, i.e., No. 4 and No. 7 in [16, Table 1], respectively.
Apart from these twelve examples,
we introduce six more examples in Table 1. The topological invariants
are computed by formulas in the previous subsection.
Note that XP4, XP5 and XP6
are deformation equivalent to
complete intersections of some homogeneous vector bundles on Grassmannians, i.e.,
No. 23, No. 20 and No. 5 in [16, Table 1], respectively
(from a private communication with Daisuke Inoue and Atsushi Ito for XP4).
Furthermore, XP5 and XP6 are also
regarded as
a complete intersection (12,2) in a Lagrangian Grassmannian LG(3,6),
and a complete intersection of two Grassmannians G(2,5) in P9, respectively.
Remark 4.3**.**
Let us discuss mirror symmetry for examples in Table 1.
For each conifold transition X⇝X0←Y,
we expect another conifold transition Y∗⇝Y0∗←X∗
in the mirror side, if X and Y have torsion-free homology.
We have a Batyrev–Borisov mirror Y∗ for Y⊂PΣ,
and the degeneration Y∗⇝Y0∗
corresponding to the small resolution Y→X0
based on the argument by
[2].
However, we do not know whether Y0∗ has the same number of nodes as X0
and admits a small resolution X∗→Y0∗ or not.
In spite of that,
periods and the Picard–Fuchs operator vanishing the periods
for the conjectural mirror family
are computable in advance.
The resulting operators in the case of
P2, P3, P4, P5 and P6
coincide with already known operators,
#195, #28, #124, #42, and #101 in [21, 1], respectively.
The operator for P1 seems unknown, thus we write it here.
A formula for the fundamental period ω0(z)=∑m=0∞Amzm
is given in [17, Eq. (5,2)], where
[TABLE]
is read from the (slightly modified) dual graph of P^1,
by associating binomial coefficients with oriented edges and
linear relations with pairs of dashed edges:
[TABLE]
With the aid of numerical method, we obtain the following Picard–Fuchs operator
for a conjectural mirror family for XP1,
[TABLE]
where θ=z∂z and Dω0(z)=0.
We observe that the operator
generates integral BPS numbers
for genus [math] and genus 1 with small degrees,
by standard methods for the computation
[9, 6].
Acknowledgments
This article is an extended version of the author’s talk at
Algebraic and Geometric Combinatorics on Lattice Polytopes 2018.
He would like to thank the organizers for the kind invitation.
He is also grateful to
Tomoyuki Hisamoto, Atsushi Ito and Taro Sano
for helpful discussions.
Bibliography23
The reference list from the paper itself. Each links out to its DOI / PubMed record.
1[1] G. Almkvist, C. van Enckevort, D. van Straten, and W. Zudilin, Tables of Calabi–Yau equations , [ar Xiv:math.AG/0507430].
2[2] V. V. Batyrev, Toric degenerations of Fano varieties and constructing mirror manifolds , in The Fano Conference (Univ. Torino, Turin, 2004), pp. 109–122.
3[3] V. V. Batyrev, Dual polyhedra and mirror symmetry for Calabi–Yau hypersurfaces in toric varieties , J. Algebraic Geom. 3, 493–535 (1994).
4[4] V. V. Batyrev and L. A. Borisov, On Calabi-Yau complete intersections in toric varieties , in Higher-dimensional complex varieties (Trento, 1994) (de Gruyter, Berlin, 1996), pp. 39–65.
5[5] V. V. Batyrev, I. Ciocan-Fontanine, B. Kim, and D. van Straten, Conifold transitions and mirror symmetry for Calabi-Yau complete intersections in Grassmannians , Nuclear Phys. B 514 , 640–666 (1998).
6[6] M. Bershadsky, S. Cecotti, H. Ooguri, and C. Vafa, Holomorphic anomalies in topological field theories , Nuclear Phys. B 405 , 279–304 (1993).
7[7] V. Batyrev and M. Kreuzer, Constructing new Calabi-Yau 3-folds and their mirrors via conifold transitions , Adv. Theor. Math. Phys. 14 , 879–898 (2010).
8[8] L. Borisov, Towards the Mirror Symmetry for Calabi-Yau Complete intersections in Gorenstein Toric Fano Varieties , [arxiv:alg-geom/9310001].