Integral generalized equivariant cohomologies of weighted Grassmann orbifolds

We consider the $n$-dimensional complex vector space $\mathbb{C}^{n}$ and a positive integer $d$ satisfying $1\leq d<n$. Then the set of all $d$-dimensional vector subspaces of $\mathbb{C}^{n}$ is called a (complex) Grassmann manifold and denoted by $\mbox{Gr}(d,n)$. In particular, the space $\mbox{Gr}(1,n)$ is called the $(n-1)$-dimensional complex projective space. The space $\mbox{Gr}(d,n)$ has a manifold structure of dimension ${d(n-d)}$, see [Muk15, Chapter 1]. This is a projective variety via the Plücker embedding. The natural $(\mathbb{C}^{*})^{n}$-action on $\mathbb{C}^{n}$ induces a $(\mathbb{C}^{*})^{n}$-action on $\mbox{Gr}(d,n)$. Grassmann manifolds are central objects of study in algebraic geometry, algebraic topology and differential geometry. Several interesting topological and geometrical properties of Grassmann manifolds can be found in [Lak72, KT03, JP03].

The orbifold version of a complex projective space was introduced in [Kaw73] and was called a twisted projective space. Orbifolds, a generalization of manifolds, were introduced by Satake [Sat56, Sat57] with the name $V$-manifolds. Later, Thurston [Thu80] used the terminology orbifolds instead. In the past two decades, several development have been appeared to study orbifolds arising in algebraic geometry, differential geometry and string topology. Some cohomology theories such as the de Rham cohomology [ALR07, Chapter 2], the singular cohomology [Hat02], the Dolbeault cohomology [Bai56], Chen-Ruan cohomology ring [CR04] and orbifold $K$-theory [ALR07, Chapter 3] for a class of orbifolds were studied either with rational, real or, complex coefficients. One can construct a CW-complex structure on an effective orbifold following [Gor78]. However, in general, the computation of the singular integral cohomology of an orbifold is considerably difficult.

Let $G$ be a topological group and $X$ a $G$-space. Then the equivariant map $X\longrightarrow\{pt\}$ induces a graded $\mathcal{E}_{G}^{*}(pt)$-algebra structure on $\mathcal{E}_{G}^{*}(X)$. The readers are referred to [May96] for the definitions and several results on the $G$-equivariant generalized cohomology theory $\mathcal{E}_{G}^{*}$. If $\mathcal{E}_{G}^{*}=H_{G}^{*}$, then it is known as the equivariant cohomology theory defined by

The ring $H_{G}^{*}(X)$ is called the Borel equivariant cohomology of $X$. If $\mathcal{E}_{G}^{*}=K_{G}^{*}$, then it is known as the equivariant $K$-theory. If $X$ is compact then $K_{G}^{0}(X)$ is the equivalence classes of $G$-equivariant complex vector bundles on $X$ [Seg68]. If $X$ is a point with trivial action, then $K_{G}^{*}(pt)$ is isomorphic to $R(G)[z,z^{-1}]$ where $R(G)$ is complex representation ring of $G$ and $z$ is the Bott element of cohomological dimension $-2$. The $G$-equivariant ring $MU_{G}^{*}(X)$ is known as equivariant complex cobordism ring see [tD70]. Sinha [Sin01] and Hanke [Han05] have shown several development on $MU_{G}^{*}$. However, many interesting questions on $MU_{G}^{*}(X)$ remain undetermined. For example, $MU_{G}^{*}(pt)$ is not completely known for non-trivial groups $G$.

Corti and Reid [CR02] introduced the weighted projective analogs of Grassmann manifolds and called them weighted Grassmannians. Then Abe and Matsumura [AM15] defined them explicitly and studied the equivariant cohomology ring of weighted Grassmannians with rational coefficients. The weighted Grassmannians are projective varieties with orbifold singularities. The simplest weighted Grassmannians are the weighted projective spaces. Kawasaki [Kaw73] proved that the integral cohomology of weighted projective spaces have no torsion and is concentrated in even degrees. The equivariant cohomology ring of a weighted projective space has been studied in [BFR09] in terms of piecewise polynomials. The equivariant $K$-theory and equivariant cobordism rings of divisive weighted projective spaces have been discussed in [HHRW16] in terms of piecewise Laurent polynomials and piecewise cobordism forms respectively.

Inspired by the above works, we introduce a different definition of weighted Grassmann orbifolds and study their several topological properties such as torsion in the integral cohomology, equivariant cohomology ring, equivariant $K$-theory ring and equivariant cobordism ring with integer coefficients. We note that [CR02, AM15] used the name ‘weighted Grassmannians’. However, keeping other naming in mind like Milnor manifolds and Seifert manifolds, we prefer to use Grassmann manifolds and weighted Grassmann orbifolds.

The paper is organized as follows. In Section 2, analogously to the definition of Grassmann manifold discussed in [Muk15], we introduce another definition of a weighted Grassmann orbifold $\mbox{WGr}(d,n)$ for $d<n$, a ‘weight vector’ $W:=(w_{1},\ldots,w_{n})\in(\mathbb{Z}_{\geq 0})^{n}$ and $a\in\mathbb{Z}_{\geq 1}$. Interestingly, this definition is equivalent to the previous one appeared in [AM15]. We recall the definition of Schubert symbols for $d<n$ and discuss how to get a total ordering on the Schubert symbols. Using this total order we show that there is a ‘weighted Plücker embedding’ from a weighted Grassmann orbifold to a weighted projective space see Lemma 2.5. We describe a $q$-cell structure of $\mbox{WGr}(d,n)$ in Proposition 2.6. Then we discuss a $(\mathbb{C}^{*})^{n}$-invariant filtration

of $\mbox{WGr}(d,n)$ using the $q$-cell decomposition, where $m:={n\choose d}-1$. Here, we consider $q$-cell structure in the sense of [PS10, Section 4]. We note that one may get different $q$-cell structures depending on the choice of the total orderings on the set of all Schubert symbols for $d<n$. Accordingly, one may obtain different $(\mathbb{C}^{*})^{n}$-invariant filtration of $\mbox{WGr}(d,n)$.

In Section 3, first we recall that there is an equivariant homeomorphism from $\mathbb{W}P(rc_{0},rc_{1},\ldots,rc_{m})$ to $\mathbb{W}P(c_{0},c_{1},\ldots,c_{m})$ for any $1\leq r\in\mathbb{N}$. Using this technique, we show how the orbifold singularity on a $q$-cell of some subcomplexes of $\mbox{WGr}(d,n)$ can be reduced, see Lemma 3.3. Consequently, we get a new $q$-cell structure of these subcomplexes including $\mbox{WGr}(d,n)$ possibly with less singularity on each $q$-cell, see Theorem 3.4. We show in Theorem 3.5 that two weighted Grassmann orbifolds are weakly equivariantly homeomorphic if their weight vectors differ by a permutation $\sigma\in S_{n}$. We define ‘admissible permutation’ $\sigma\in S_{n}$ for a prime $p$ and $\mbox{WGr}(d,n)$, see Definition 3.8. The following result says when $H^{*}({\rm WGr}(d,n);\mathbb{Z})$ has no $p$-torsion.

We introduce ‘divisive’ weighted Grassmann orbifolds. We note that this definition coincides with the concept of divisive weighted projective space of [HHRW16] when $1=d<n$. We prove the following.

This result implies that the integral cohomology of a divisive weighted Grassmann orbifold has no torsion and is concentrated in even degrees. We discuss a class of non-trivial examples of divisive weighted Grassmann orbifolds. We remark that the weighted Grassmann orbifold in Example 3.12 is not divisive. However, its integral cohomology has no torsion.

In Section 4, we show that the $(\mathbb{C}^{*})^{n}$-invariant stratification

has the following property. The quotient $X_{i}/X_{i-1}$ is homeomorphic to the Thom space of an orbifold $(\mathbb{C}^{*})^{n}$-bundle

for some $\ell(\lambda^{i})\in(\mathbb{Z}_{\geq 1})$ and finite groups $G_{i}$ for $i=1,\ldots,m$, see Proposition 4.1. We compute the equivariant $K$-theory ring of any weighted Grassmann orbifolds with rational coefficients, see Theorem 4.4. If $\mbox{WGr}(d,n)$ is divisive then $G_{i}$ is trivial for $i=1,\ldots,m$. Then considering $T^{n}:=(S^{1})^{n}\subset(\mathbb{C}^{*})^{n}$, the following result describes the integral equivariant cohomology of certain weighted Grassmann orbifolds.

The computation of $e_{T^{n}}(\xi^{ij})$ is discussed in (4.4). We compute the equivariant cohomology ring of some weighted Grassmann orbifold with integer coefficients which are not divisive, see Theorem 4.10. For $m\geq 2$, corresponding to each pair of positive integers $(n,d)$ such that $d<n$ and $m+1={n\choose d}$ we have a $T^{n}$-action on $\mathbb{W}P(c_{0},c_{1},\dots,c_{m})$. For each pair $(n,d)$, we discuss the generalized $T^{n}$-equivariant cohomology of a divisive $\mathbb{W}P(c_{0},c_{1},\dots,c_{m})$ with integer coefficients, see Theorem 4.11.

In Section 5, we show that there exist equivariant Schubert classes $\{w\widetilde{S}_{\lambda^{i}}\}_{i=0}^{m}$ which form a basis for the integral $T^{n}$-equivariant cohomology of a divisive weighted Grassmann orbifold, see Proposition 5.3. We study some properties of weighted structure constant, see Lemma 5.5. Then we show the following multiplication rule.

In this section, we introduce another definition of weighted Grassmann orbifold $\mbox{WGr}(d,n)$ where $d<n$. We recall the definition of a Schubert symbol for ${d<n}$ and discuss some (total) ordering on the set of Schubert symbols. We show that there is an equivariant embedding from a weighted Grassmann orbifold to a weighted projective space. We show that our definition of weighted Grassmann orbifold is equivalent to the previous one appeared in [AM15]. We study the orbifold and $q$-cell structures of weighted Grassmann orbifolds generalizing the manifolds counter part discussed in [MS74].

A Schubert symbol $\lambda$ for $d<n$ is a sequence of $d$ integers $(\lambda_{1},\lambda_{2},\dots,\lambda_{d})$ such that $1\leq\lambda_{1}<\lambda_{2}<\cdots<\lambda_{d}\leq n$. The length $\ell(\lambda)$ of a Schubert symbol $\lambda:=(\lambda_{1},\lambda_{2},\dots,\lambda_{d})$ is defined by $\ell(\lambda):=(\lambda_{1}-1)+(\lambda_{2}-2)+\cdots+(\lambda_{d}-d)$. There are ${n\choose d}$ many Schubert symbols for $d<n$. One can define a partial order ‘$\preceq$’ on the Schubert symbols for $d<n$ by

Then the set of all Schubert symbols for $d<n$ form a poset with respect to this partial order ‘$\preceq$’.

Let $M_{d}(n,d)$ be the set of all complex $n\times d$ matrix of rank $d$ and $\mbox{GL}(d,\mathbb{C})$ the set of all non-singular complex matrix of order $d$. We denote a matrix $A\in M_{d}(n,d)$ as follows

where ${\bf a}_{i}\in\mathbb{C}^{d}$ for $i=1,\ldots,n$.

This gives a total order on the set of all Schubert symbols. Note that the total order ‘$<$’ in Definition 2.4 preserves the partial order ‘$\preceq$’ in (2.1). That is, for two Schubert symbols $\lambda$ and $\mu$, $\lambda\preceq\mu\implies\lambda\leq\mu$, but the converse may not be true in general. Observe that there may exist several other total orders on the set of all Schubert symbols which preserve the partial order ‘$\preceq$’. For example, the dictionary order also gives a total order on the Schubert symbols. By a total order on the set of all Schubert symbols for $d<n$, we mean one of these total orders on it. For $m:={n\choose d}-1$, let

be a total order on the Schubert symbols for $d<n$.

For $W=(w_{1},w_{2},\dots,w_{n})\in(\mathbb{Z}_{\geq 0})^{n}$, $a\in\mathbb{Z}_{\geq 1}$ and $i\in\{0,1,\dots,m\}$, let

where $\lambda^{i}=(\lambda_{1}^{i},\lambda_{2}^{i},\dots,\lambda_{d}^{i})$ is the $i$-th Schubert symbol given in (2.4). Then $c_{i}\geq 1$ for any $i\in\{0,\ldots,m\}$. Therefore, one can define the weighted projective space $\mathbb{W}P(c_{0},c_{1},\dots,c_{m})$ from Remark 2.3. We denote the associated orbit map $\mathbb{C}^{m+1}-\{0\}\to\mathbb{W}P(c_{0},c_{1},\dots,c_{m})$ by $\pi^{\prime}_{c}$ which can be written as

Note that when $c_{0}=c_{1}=\cdots=c_{m}=1$, then the corresponding orbit map is denoted by

Let $(t_{1},t_{2},\dots,t_{n})\in(\mathbb{C}^{*})^{n}$ and $A=({\bf a}_{1},{\bf a}_{2},\dots,{\bf a}_{n})^{tr}\in M_{d}(n,d)$. Then $(\mathbb{C}^{*})^{n}$ acts on $M_{d}(n,d)$ defined by

This induces a natural $(\mathbb{C}^{*})^{n}$-action on $\mbox{WGr}(d,n)$ such that the orbit map $\pi_{w}$ of (2.2) is $(\mathbb{C}^{*})^{n}$-equivariant.

We remark that the standard ordered basis $\{e_{1},e_{2},\dots,e_{n}\}$ of $\mathbb{C}^{n}$ induces an ordered basis $\{e_{\lambda^{0}},e_{\lambda^{1}},\dots,e_{\lambda^{m}}\}$ of $\Lambda^{d}(\mathbb{C}^{n})$. Therefore, we can identify $\Lambda^{d}(\mathbb{C}^{n})$ with $\mathbb{C}^{m+1}(=\mathbb{C}\{e_{\lambda^{0}},e_{\lambda^{1}},\dots,e_{\lambda^{m}}\})$. The standard action of $(\mathbb{C}^{*})^{n}$ on $\mathbb{C}^{n}$ induces an action of $(\mathbb{C}^{*})^{n}$ on $\mathbb{C}^{m+1}-\{0\}$ which is defined by

where $t_{\lambda}=t_{\lambda_{1}}\cdots t_{\lambda_{d}}$ and $e_{\lambda}=e_{\lambda_{1}}\wedge\cdots\wedge e_{\lambda_{d}}$ for the Schubert symbol $\lambda=(\lambda_{1},\lambda_{2},\dots,\lambda_{d})$. This induces a $(\mathbb{C}^{*})^{n}$-action on the weighted projective space $\mathbb{W}P(c_{0},c_{1},\ldots,c_{m})$ such that the orbit map $\pi^{{}^{\prime}}_{c}$ in (2.6) is $(\mathbb{C}^{*})^{n}$-equivariant.

For each Schubert symbol $\lambda=(\lambda_{1},\lambda_{2},\dots,\lambda_{d})$, let $A_{\lambda}$ be the matrix with row vectors ${\bf a}_{\lambda_{1}},{\bf a}_{\lambda_{2}},\dots,{\bf a}_{\lambda_{d}}$. Consider the map $P\colon M_{d}(n,d)\to\mathbb{C}^{m+1}$ defined by

where ${\bf v}_{1},{\bf v}_{2},\ldots,{\bf v}_{d}\in\mathbb{C}^{n}$ are the columns of $A$. Observe that $P(A)\neq 0$ because $A\in M_{d}(n,d)$ has rank $d$.

Note that the actions of $(\mathbb{C}^{*})^{n}$ on $\mbox{WGr}(d,n)$ and $\mathbb{W}P(c_{0},c_{1},\ldots,c_{m})$ implies that the weighted Plücker embedding $Pl_{w}$ in (2.10) is $(\mathbb{C}^{*})^{n}$-equivariant, and $Pl_{w}(\mbox{WGr}(d,n))$ is a $(\mathbb{C}^{*})^{n}$-invariant subset of $\mathbb{W}P(c_{0},c_{1},\ldots,c_{m})$. Thus all the maps in the diagram (2.12) are $(\mathbb{C}^{*})^{n}$-equivariant.

Now we show that Definition 2.1 is equivalent to the definition of a weighted Grassmannian studied in [AM15]. The algebraic torus $(\mathbb{C}^{*})^{n+1}$ acts on $\Lambda^{d}(\mathbb{C}^{n})$ by

where $t_{\lambda}=t_{\lambda_{1}}\cdots t_{\lambda_{d}}$ for $\lambda=(\lambda_{1},\ldots,\lambda_{d})$. Consider the subgroup $WD$ of $(\mathbb{C}^{*})^{n+1}$ defined by

Then the restricted action of $WD$ on $\Lambda^{d}(\mathbb{C}^{n})-\{0\}$ is given by

Observe that this action of $WD$ is same as the weighted $\mathbb{C}^{*}$-action in Remark 2.3. Then we have $\dfrac{\Lambda^{d}(\mathbb{C}^{n})-\{0\}}{WD}=\mathbb{W}P(c_{0},\dots,c_{m})$ and by the commutativity of the diagram (2.12) we have

Therefore the topologies on $\mbox{WGr}(d,n)$ and $\dfrac{P(M_{d}(n,d))}{WD}$ are equivalent. Abe and Matsumura [AM15] called the quotient $\dfrac{P(M_{d}(n,d))}{WD}$ a weighted Grassmannian and showed that it has an orbifold structure. We call $\mbox{WGr}(d,n)$ a weighted Grassmann orbifold associated to the pair $(W,a)$.

Next, we recall the Schubert cell decomposition of $\mbox{Gr}(d,n)$ following [MS74]. For $k\leq n$, we identify

For the Schubert symbol $\lambda=(\lambda_{1},\lambda_{2},\dots,\lambda_{d})$, the Schubert cell $E(\lambda)$ is defined by

where $[d]:=\{1,2,\dots,d\}$. We have the following homeomorphism from [MS74, Chapter-6].

Note that $j$-th column in the matrices in (2.13) has $\lambda_{j}$-th entry $1$ and all subsequent entries of this column are zero for $j\in[d]$. Then $E(\lambda)$ is an open cell of dimension $\ell(\lambda)=(\lambda_{1}-1)+(\lambda_{2}-2)+\dots+(\lambda_{d}-d)$.