Koszul duality and the Hochschild cohomology of Artin-Schelter regular algebras
Leilei Liu

TL;DR
This paper explores the relationship between Koszul duality and Hochschild cohomology of Artin-Schelter regular algebras, revealing connections between BV algebra structures in dual algebra pairs.
Contribution
It establishes a link between two BV algebra structures on Hochschild cohomology for Koszul dual algebras with semisimple Nakayama automorphisms.
Findings
Identification of two BV algebra structures on Hochschild cohomology
Equivalence of these structures for Koszul dual pairs
Extension of known structures to Artin-Schelter regular algebras
Abstract
We identify two Batalin-Vilkovisky algebra structures, one obtained by Kowalzig and Krahmer on the Hochschild cohomology of an Artin-Schelter regular algebra with semisimple Nakayama automorphism and the other obtained by Lambre, Zhou and Zimmermann on the Hochschild cohomology of a Frobenius algebra also with semisimple Nakayama automorphism, provided that these two algebras are Koszul dual to each other.
Peer Reviews
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 structures and combinatorial models · Advanced Topics in Algebra · Homotopy and Cohomology in Algebraic Topology
Koszul duality and the
Hochschild cohomology of Artin-Schelter regular algebras
Leilei Liu
School of Mathematics, Sichuan University, Chengdu, Sichuan Province 610064, People’s Republic of China
Abstract.
We identify two Batalin-Vilkovisky algebra structures, one obtained by Kowalzig and Krahmer on the Hochschild cohomology of an Artin-Schelter regular algebra with semisimple Nakayama automorphism and the other obtained by Lambre, Zhou and Zimmermann on the Hochschild cohomology of a Frobenius algebra also with semisimple Nakayama automorphism, provided that these two algebras are Koszul dual to each other.
Key words and phrases:
Artin-Schelter algebras, Koszul duality, cohomology, Batalin-Vilkovisky algebras
Contents
- 1 Introduction
- 2 Preliminaries on Hochschild homology
- 3 Artin-Schelter regular algebras
- 4 Frobenius algebras
- 5 Koszul duality of AS-regular algebras
1. Introduction
In 2014, Kowalzig and Krahmer showed in [13] that the Hochschild cohomology of an Artin-Schelter (AS for short) regular algebra with semisimple Nakayama automorphism has a Batalin-Vilkovisky algebra structure. Soon after that, Lambre, Zhou and Zimmerman proved in [15] that the Hochschild cohomology of a Frobenius algebra with semisimple Nakayama automorphism also admits a Batalin-Vilkovisky structure. In this paper, we identify these two Batalin-Vilkovisky algebra structures, provided that these two algebras are Koszul dual to each other. Let us start with some backgrounds.
In 2008 Brown and Zhang made a progress on the understanding the twisted Poincaré duality of AS-regular algebras. They proved in [4] that for an AS-regular algebra of global dimension , there is an isomorphism between the Hochschild cohomology of and the Hochschild homology of with coefficients in
[TABLE]
where is Nakayama automorphism of . Now if we assume is semisimple, then Kowalzig and Krahmer proved in [13] that can be computed by a subcomplex of the corresponding Hochschild complex on which the Connes cyclic operator exists. Therefore we may pull back the Connes cyclic operator to via (1), which is usually denoted by . Kowalzig and Krahmer showed that in fact genetates the Gerstenhaber bracket on , which means is a Batalin-Vilkovisky algebra.
Analogously, in 2016, Lambre, Zhou and Zimmermann proved in [15] that for a Frobenius algebra with semisimple Nakayama automorphism, say , there also exists a Batalin-Vilkovisky algebra structure on .
To relate these two Batalin-Vilkovisky algebra structures, let us recall a result of P. Smith. In [19, Proposition 5.10] he showed that for a graded connected Koszul algebra , is AS-regular if and only if its Koszul dual is Frobenius. Based on the fact that for a Koszul algebra ,
[TABLE]
as Gerstenhaber algebras, it is natural to ask that for an AS-regular algebra with semisimple Nakayama automorphism whether this above isomorphism is an isomorphism of Batalin-Vilkovisky algebras. In this paper we give an affirmative answer to this question:
Theorem 1.1**.**
Suppose is a Koszul AS-regular algebra with semisimple Nakayama automorphism. Denote by its Koszul dual algebra. Then
[TABLE]
as Batalin-Vilkovisky algebras.
This paper may be viewed as a sequel to [5], where the isomorphism of Batalin-Vilkovisky algebras on two Hochschild cohomology groups are proved for Koszul Calabi-Yau algebras, verifying a conjecture of Rouquier given in the preprint [8] of Ginzburg.
Note that Calabi-Yau algebras and AS-regular algebras are highly related: in [20], Reyes, Rogalski and Zhang introduced notion of twisted Calabi-Yau algebras (a Calabi-Yau algebra is twisted with trivial twisting), and proved that an algebra is twisted Calabi-Yau if and only if it is AS-regular (see also [26] for some partial result). Thus the isomorphism of Batalin-Vilkovisky algebras for Koszul Calabi-Yau algebras, proved in [5], is a special case of Theorem 1.1. In other words, we may view Theorem 1.1 as a twisted version of Rouquier’s conjecture.
Notations**.**
Throughout this paper, denotes a field of character [math]. All tensors and Homs are over unless otherwise specified. All algebras (resp. coalgebras) are unital and augmented, (resp. co-unital and co-augmented) over . If is an associative algebra, then is its opposite and is its envelope. Suppose is a graded vector space, then the shift of the gradings of down by is denoted by or , i.e., .
2. Preliminaries on Hochschild homology
In this section, we recall the Hochschild homology and cohomology of an associative algebra . These two homology groups, together with the algebraic operations on them, form the so-called differential calculus, a notion introduced by Tamarkin and Tsygan in [21].
2.1. Hochschild homology and cohomology of algebras
For an associative -algebra , let be its augmentation ideal. The reduced Hochschild chain complex of with coefficients in an -bimodule , denoted by , is
[TABLE]
with boundary given by
[TABLE]
for any and , . The associated homology is called the Hochschild homology of with coefficients in , and is denoted by .
The reduced Hochschild cochain complex of with values in , denoted by , is the complex
[TABLE]
with coboundary given by
[TABLE]
for any and , . The associated cohomology is called the Hochschild cohomology of with values in , and is denoted by .
Later we will use the fact that and (c.f. [25], Lemma 9.1.3). Let us recall the Connes cyclic operator on the Hochschild chain complex.
Definition 2.1** (Connes cyclic operator).**
Assume is an associative algebra. The Connes cyclic operator
[TABLE]
is given by
[TABLE]
It is direct to check , and therefore is a mixed complex in the sense of Kassel [9].
Remark 2.2**.**
The Hochschild homology and cohomology can also be defined for differential graded algebras. It is better to view these two homology groups as follows: Suppose is a possibly differential graded algebra and is a differential graded -bimodule. Let be the bar construction of (see [7, 17, 25] for more details). Considering the following total complex
[TABLE]
with the total degree of the tensor product, and the differential is given by
[TABLE]
and
[TABLE]
where , for any homogeneous elements and , . Here we denote by the degree of . For and taking the gradings into account, the cyclic operator can also be defined in the same way. Similarly,
[TABLE]
with the differential on the right-hand side analogously defined.
2.2. Differential calculus with duality
Let us recall the Gerstenhaber cup product and bracket on the Hochschild cohomology of an associative algebra , and its actions on the Hochschild homology of . Let us start with the notion of Gerstenhaber algebras:
Definition 2.3** (Gerstenhaber).**
A Gerstenhaber algebra is a graded -vector space endowed with two bilinear operators and such that:
- (1)
() is a graded commutative associative algebra, i.e.
[TABLE]
with associativity for any homogeneous elements ; 2. (2)
() is a graded Lie algebra with the bracket of degree , i.e.
[TABLE]
and
[TABLE]
or any homogeneous elements ; 3. (3)
the cup product and the Lie bracket are compatible in the sense that
[TABLE]
for any homogeneous elements .
Definition 2.4** (Tamarkin-Tsygan [21], Definition 3.2.1).**
Let and be two graded vector spaces. A differential calculus is a sextuple
[TABLE]
satisfying the following conditions
- (1)
is a Gerstenhaber algebra; 2. (2)
is a graded module over by the “cap action”
[TABLE]
i.e. for any , , ; 3. (3)
there exists a linear operator such that and moreover, if we set , then
[TABLE]
and
[TABLE]
In the above definition, is called the Lie derivative of on . It is shown by Daletskii, Gelfand and Tsygan in [6] that the Hochschild cohomology and homology of an associative algebra form a differential calculus. Let us give some details:
- (1)
The Gerstenhaber cup product is given by
[TABLE] 2. (2)
The Gerstenhaber Lie bracket is given by
[TABLE]
where
[TABLE] 3. (3)
The cap product is given by
[TABLE] 4. (4)
The differential operator on is nothing but the Connes cyclic operator.
One can show that the above operations respect the boundary operators up to homotopy (which we will not address), and hence are well-defined on the homology level:
Proposition 2.5** (Daletskii-Gelfand-Tsygan [6]).**
Let be an associative algebra, then the data forms a differential calculus.
2.3. Another example of differential calculus
Assume is a finite dimensional associative algebra. Denote . Then has a -bimodule structure induced by the natural -bimodule structure of . There are two operators:
[TABLE]
given by
[TABLE]
and
[TABLE]
given by
[TABLE]
Here is the Connes cyclic operator on the Hochschild complex , and is viewed as the linear dual space of the latter via the following identification
[TABLE]
Theorem 2.6**.**
Assume is a finite dimensional associative algebra. Denote . Then
[TABLE]
is a differential calculus.
Proof.
We know is a Gerstenhaber algebra. We need to show the following:
(1) is a graded -module. In fact,
[TABLE]
Moreover, since is Gerstenhaber, we have . Thus combining the above two identities, we have
[TABLE]
i.e., is a graded -module.
(2) Observe that
[TABLE]
we have that
[TABLE]
and moreover,
[TABLE]
This completes the proof. ∎
2.4. The Batalin-Vilkovisky algebra structure
In this paper we are more concerned with the Batalin-Vilkovisky algebra structure on the Hochschild cohomology. It naturally comes from the “noncommutative Poincaré duality”([22]) of the corresponding algebra. Let us first recall the following notion of differential calculus with duality, introduced by Lambre in [14].
Definition 2.7** (Lambre [14]).**
A differential calculus is called a differential calculus with duality if there exists an integer and an isomorphism of -modules
[TABLE]
Lemma 2.8** (Lambre [14] Theorem 1.6 and Lemma 1.5).**
Assume is a differential calculus with duality. Let . Then
[TABLE]
Proof.
Denote the imagine of by , which is also called the volume form of . Then we have
[TABLE]
and therefore
[TABLE]
and similarly,
[TABLE]
and
[TABLE]
Hence we have the identity
[TABLE]
Since is an isomorphism, we have
[TABLE]
This completes the proof. ∎
The above lemma in fact says that is a Batalin-Vilkovisky algebra. Let us recall its definition:
Definition 2.9**.**
A Batalin-Vilkovisky algebra is a Gerstenhaber algebra together with an operator of degree such that , and
[TABLE]
for any homogeneous elements .
3. Artin-Schelter regular algebras
In this section, we briefly recall the construction of the Batalin-Vilkovisky algebra on the Hochschild cohomology of Artin-Schelter regular algebras with semisimple Nakayama automorphism, obtained by Kowalzig and Krahmer in [13]. In this section, will present a connected graded algebra over an algebraically closed field . A graded algebra is said to be connected if for and .
Definition 3.1** (Artin-Schelter [1]).**
A connected graded algebra is called Artin-Schelter regular (or AS-regular for short) of dimension if
- (1)
has finite global dimension , and 2. (2)
is Gorenstein, that is, for and .
Later in 2014 Reyes, Rogalski and Zhang proved in [20] that AS-regular algebras are in fact twisted Calabi-Yau algebras (see also Yekutieli and Zhang [26] for some partial results):
Theorem 3.2** ([20], Lemma 1.2).**
Suppose is a connected graded algebra. Then is AS-regular if and only if it is skew Calabi-Yau (in the graded sense), namely, satisfies the following two conditions:
- (1)
* is homologically smooth, that is, , viewed as an -module, has a bounded, finitely generated projective resolution, and* 2. (2)
there exists an integer and an algebra automorphism of such that
[TABLE]
as -modules.
In the above theorem, the automorphism is called Nakayama automorphism. The -module structure of is induced by the inner module structure on : . The module is a vector space equipped with the -bimodule structure , for any . We say is semisimple if it is diagonalizable.
In the following, we will always use the above equivalent definition of AS-regular algebras, rather than its original definition.
3.1. Results of Kowalzig and Krahmer
Assume is an AS-regular algebra with semi-simple Nakayama automorphism . In [12, 13], Kowalzig and Krahmer constructed a differential calculus with duality on . Thus as a corollary, they obtained a Batalin-Vilkovisky algebra structure on .
First, let us observe that, compared to the differential calculus structure given in §2, there is no Connes operator on . What Kowalzig and Krahmer did is to consider a subcomplex of , whose homology is and on which the Connes operator is well-defined. Let us briefly recall their results.
Let
[TABLE]
be given by
[TABLE]
Since is semisimple, there is a decomposition of as follows. Let be the set of eigenvalues of acting on and be the eigenvalue space corresponding to . Denote
[TABLE]
The restriction of makes to be a subcomplex of and we denote its homology by . A key observation is that, the restriction of on the subcomplex is exactly the Connes operator. Hence is a mixed complex.
Let
[TABLE]
be given by
[TABLE]
Lemma 3.3** ([13],(2.19)).**
Let and be as above, then there exists
[TABLE]
on the complex .
As an immediate corollary, we have
[TABLE]
Via this isomorphism, we obtain the Connes operator on .
Similarly, there is a decomposition of the Hochschild cochain complex . Let
[TABLE]
The restriction coboundary makes to be a subcomplex of and we denote its cohomology by . In a similar fashion, Kowalzig and Krahmer proved in [12] that the cohomology is concentrated in the subcomplex corresponding to the eigenvalue 1, namely
[TABLE]
It is direct to check that the Gerstenhaber cup product, bracket and cap action restrict to the following maps: for ,
[TABLE]
Considering the case of eigenvalue , we have the following theorem.
Theorem 3.4** ([12], Theorem 1; [13] Theorem 1.5).**
Let , and be the restrictions of the cup product, cap product and Gerstenhaber bracket to the homology and cohomology spaces associated with the eigenvalue . Then together with the Connes operator , they give on
[TABLE]
a differential calculus structure.
In 2008, Brown and Zhang obtained the following
Theorem 3.5** ([4], Corrollary 0.4).**
Suppose is an AS-regular algebra of finite global dimension . Then we have the following isomorphism
[TABLE]
Proof.
Since is homologically smooth, we have the following
[TABLE]
which completes the proof. ∎
Thus combining (2)-(5) we obtain the following:
Theorem 3.6** ([13], Theorem 4.25).**
Suppose is an AS-regular algebra with semisimple Nakayama automorphism . Then
[TABLE]
forms a differential calculus with duality.
Combining the above theorem with Lemma 2.8, we obtain:
Theorem 3.7** ([13], Theorem 1.5).**
If is an AS-regular algebra with semisimple Nakayama automorphism, then the Hochschild cohomology of is a Batalin-Vilkovisky algebra.
4. Frobenius algebras
In this section, we rephrase the construction of the Batalin-Vilkovisky algebra on the Hochschild cohomology of a Frobenius algebra with semisimple Nakayama automorphism, obtained by Lambre, Zhou and Zimmermann in [15].
Definition 4.1**.**
A graded associative -algebra of finite dimension is called Frobenius of degree if there exists a nondegenerate bilinear pairing
[TABLE]
such that , for all .
By the nondegeneracy of the pairing, there exists an automorphism such that . Such is also called Nakayama automorphism of . The nondegenerate pairing given by (6) is equivalent to saying that
[TABLE]
is an isomorphism of -bimodules.
Let be the linear dual space of , which is a graded coalgebra. Nakayama automorphism induces an automorphism on . Here is the adjoint of . We have a left co-module structure on :
[TABLE]
for all . (The coproduct on is viewed as a right co-module structure of , and is denoted by .) To distinguish, let us denote this new co-bimodule structure on by .
Recall that in §2.3 we obtain a differential calculus structure on . In the following we explore this structure in more detail, for being a Frobenius algebra.
4.1. Hochschild homology of coalgebras
Suppose is a coassociative (possibly graded) coalgebra with coproduct given by
[TABLE]
Let , and let by recursion. From the coassociativity of , we have .
Definition 4.2**.**
Suppose is a coassociative (possibly graded) coalgebra and is a co-bimodule over . The Hochschild chain complex of with coefficients in , denoted by , is the complex
[TABLE]
with the boundary given by
[TABLE]
where
[TABLE]
and for any homogeneous elements . The associated homology is called the Hochschild homology of with coefficients in , and is denoted by .
Theorem 4.3** ([15], Proposition 3.3).**
Let be a Frobenius algebra of degree with Nakayama automorphism . Then there is an isomorphism
[TABLE]
Proof.
Given a Frobenius algebra with Nakayama automorphism , we have as -bimodules. The is given by . Therefore we have
[TABLE]
The isomorphisms above are all compatible with boundary maps, and hence we obtain
[TABLE]
which completes the proof. ∎
4.2. Frobenius algebra with semisimple Nakayama automorphism
In this subsection, we go over the Batalin-Vilkovisky structure on the Hochschild cohomology of a Frobenius algebra with semisimple Nakayama automropshim.
Firstly, we consider the Hochschild chain complex of with coefficients in . Similar to the AS-regular case, there is a decomposition on the chain complex of a Frobenius coalgebra .
Since is semisimple, there is a decomposition of as follows. Let be the set of eigenvalues of and be the eigenvalue space corresponding to . Let
[TABLE]
The restriction of makes to be a subcomplex of and we denote its homology by .
Secondly, we define a map
[TABLE]
given by
[TABLE]
where is the imagine of the counit map . The restriction map on the subcomplex is exactly the Connes operator. Hence is a mixed complex.
Let
[TABLE]
be
[TABLE]
Then we have the following
Lemma 4.4** ([15], Proposition 2.1, [13],(2.19)).**
On the space , there exists the identity
[TABLE]
The above lemma implies that
[TABLE]
Similarly, there is a decomposition of the Hochschild cochain complex . Let
[TABLE]
The restriction coboundary makes into be a subcomplex of and we denote its homology by . Similarly to (3) we have
[TABLE]
It is direct to check that the Gerstenhaber cup product, bracket and cap product restrict to the following maps: for ,
[TABLE]
Consider the case of eigenvalue ; we have the following theorem.
Theorem 4.5** ([15], Theorem 2.3).**
Let , and be the restrictions of the cup product, cap product and Gerstenhaber bracket to the homology and cohomology spaces associated with the eigenvalue . Then the Connes operator gives on
[TABLE]
a differential calculus structure.
Together with Theorem 4.3, we obtain the following.
Theorem 4.6** ([15], Theorem 2.3 and Proposition 3.4).**
Suppose is a Frobenius algebra of dimension with semisimple Nakayama automorphism. Then
[TABLE]
forms a differential calculus with duality.
Combining the above theorem with Lemma 2.8, we obtain:
Theorem 4.7** ([15], Theorem 4.1).**
If is a Frobenius algebra with semisimple Nakayama automorphism, then the Hochschild cohomology of is a Batalin-Vilkovisky algebra.
5. Koszul duality of AS-regular algebras
In this section, we study Koszul AS-regular algebras, and then relate the two differential calculus structures in previous two sections by means of Koszul duality. We begin with Koszul algebras, which was introduced by Priddy in [18].
Assume is a -vector space generated by a basis of degree . The free algebra generated by is denoted by . Let be a subspace, and let be the bi-sided ideal generated by in . The quotient algebra is called a quadratic algebra.
Definition 5.1**.**
Given a quadratic algebra , the linear dual of is denoted by and let . Then is called the quadratic dual of .
Let , then
[TABLE]
which is a coalgebra. Its coproduct is the restriction of the coproduct on the co-free coalgebra given by
[TABLE]
Here the summand for is and the summand for is .
The Koszul complex associated to is the complex
[TABLE]
with the differential given by
[TABLE]
where is the dual basis of in . It is direct to check .
Definition 5.2**.**
A quadratic algebra is called Koszul if the complex (9) is exact. In this case, is called the Koszul dual algebra of , and is called the Koszul dual coalgebra of .
One of the advantages of Koszul algebras is that has a much smaller free resolution, which is described as follows. Recall that the cobar construction of is the free tensor differential graded algebra generated by with the differential given by
[TABLE]
for any , .
Consider the composition of the following maps
[TABLE]
where is the projection map and is the natural surjective map. The composition map is denoted by
[TABLE]
For any , , let
[TABLE]
where is the image of the projection .
Proposition 5.3** (c.f. [17] Theorem 3.4.4).**
Suppose is a Koszul algebra. Then
[TABLE]
is an quasi-isomorphism.
Similarly, recall that is a subset of . Let be the restriction map of the natural inclusion , which extends to be a differential graded(DG) coalgebra map . Then is also a quasi-isomorphism.
5.1. Homology of Koszul algebras with algebraic automorphisms
Suppose is a Koszul algebra of global dimension . Let be an algebra automorphism of preserving the grading. Since , we have . Extending to be an algebra map on , we thus have . This also means , by restriction on , is an automorphism of vector spaces.
Lemma 5.4**.**
* is a coalgebra automorphism of .*
Proof.
Recall that . Now we prove
[TABLE]
For any homogeneous element , it has the form . So
[TABLE]
We have
[TABLE]
and
[TABLE]
This implies that is a coalgebra map. ∎
Consider the following complex
[TABLE]
with differential given by
[TABLE]
for any . It is direct to check by the following
[TABLE]
We denote this complex by .
Proposition 5.5**.**
There is a quasi-isomorphism
[TABLE]
Proof.
Consider the complex with the differential , where and are given by
[TABLE]
for any . It is direct to check that . Hence . The Koszul property of (c.f. [11], Proposition 19) implies that is a resolution of as -bimodules.
Now we have two -bimodules free resolutions of , as above and the two sided bar resolution (recall that it is with extra twisted differential). Recall that is a subset of . Let be the extension of , which then commutes with the differentials on both sides. Then is a quasi-isomorphism (see [23], Proposition 3.3).
It is direct to see that , and . Let . Then is the desired quasi-isomorphism. ∎
5.2. Two quasi-isomorphisms
Suppose is a Koszul AS-regular algebra. The following result is nowadays well-known.
Theorem 5.6** (Smith [19], Proposition 5.10).**
Suppose is a Koszul algebra. Then is AS-regular if and only if is Frobenius.
Now suppose admits semisimple Nakayama automorphism . By Van den Bergh (see [24], Theorem 9.2), the adjoint of is the Nakayama automorphism of . Since is semisimple, is also semisimple. Recall from previous subsection that also gives semisimple Nakayama automorphism of . The purpose of this subsection is to prove the following
[TABLE]
which commutes with on the left and on the right.
Recall that the cobar construction is a DG free algebra, we may extend the coalgebra automorphism to be a DG algebra automorphism of . Consider the complex (sometimes also denoted by ). Assume is semisimple, then we set . Let us denote
[TABLE]
and
[TABLE]
The restriction map of makes to be a complex. Denote its homology group by . Again by Kowlzig and Krahmer ([13], Proposition 2.7, Lemma 7.1) we have the following:
Lemma 5.7**.**
On the complex
[TABLE]
we have the identity
[TABLE]
The above lemma implies
[TABLE]
and the sub complex is a mixed complex. Now we have the following two lemmas.
Lemma 5.8**.**
Let be a Koszul algebra with an semi-simple automorphism , and be its Koszul dual coalgebra. Then we have a commutative diagram of quasi-isomorphisms of complexes up to homotopy
[TABLE]
Proof.
The map is defined by
[TABLE]
here has the form with , . And is the imagine of , that is, . The map is given by
[TABLE]
for , . At the bottom of the diagram, the map is given by
[TABLE]
for and . And the map is given by
[TABLE]
for , . In the vertical direction, is the injective map and is the projective map. They are quasi-isomophisms up to homotopy. All these maps are all morphisms of complexes.
By a spectral sequence argument, all these morphisms are quasi-isomorphic. For example, let us consider . There exist filtrations on these two complexes given by
[TABLE]
and
[TABLE]
The boundary maps are compatible with the filtrations respectively. Then the comparison theorem for spectral sequences guarantees the quasi-isomorphism. Similarly we can prove other maps are quasi-isomorphisms. ∎
Lemma 5.9**.**
Let be a Koszul algebra with an semi-simple automorphism and be its Koszul dual coalgebra. Then we have the following quasi-isomorphisms of mixed complexes
[TABLE]
where is a homotopy inverse of .
Proof.
Since is a quasi-isomorphism of differential graded algebras, the map given in the previous lemma is a quasi-isomorphism of mixed complexes([16], Proposition 2.5.15). We next construct the quasi-isomorphism of the mixed complexes, which is a homotopy inverse of .
From now on, let us denote any homogeneous element in the bar construction of by with , and any homogeneous element in the cobar construction of by or . The morphism is defined by
[TABLE]
where .
It is direct to check is a morphism of complexes, that is, it commutes with the Hochschild differential. Now we show commutes with . In fact,
(1) For Hochschild chains like the left hand side of (10), we have:
[TABLE]
and
[TABLE]
This means in this case.
(2) For Hochschild chains like the left hand side of (11), we have:
[TABLE]
and
[TABLE]
for some . This means in this case.
(3) For Hochschild chains like the left hand side of (12), is automatic since both sides are always zero.
In summary, the above calculation implies that is a morphism of mixed complexes. Next, we show that and are homotopy inverse to each other.
First, since
[TABLE]
we get on .
Second, we show
[TABLE]
The homotopy map, denoted by , is given as follows: First, let
[TABLE]
and for ,
[TABLE]
and
[TABLE]
where for and . Here we use the Sweedler notation . The signs are given by
[TABLE]
where , for any with , . Now let and we claim that
[TABLE]
Assuming this identity, we obtain that is a quasi-isomorphism of chain complexes, and thus a quasi-isomorphism of mixed complexes. Then the proof is completed. ∎
Proof of (13).
For any element , where and , and for , we have
[TABLE]
where
[TABLE]
and
[TABLE]
for , and where we write . We have the following three cases to check :
(1) For Hochschild chains like the left hand side of (10), it is direct to see
[TABLE]
(2) For Hochschild chains like the left hand side of (11), we have
[TABLE]
with , where and , is equal to
[TABLE]
The second summand
[TABLE]
has the following terms (we omit the sign)
[TABLE]
Recall the Construction of bar and cobar construction and the corresponding differentials, we obtain that is equal to
[TABLE]
The above identities implies that
[TABLE]
in this case. For elements with , it is direct to obtain the identity
[TABLE]
by similar computations as above.
(3) For Hochschild chains like the left hand side of (12) like
[TABLE]
where , and , and for , we have that equals
[TABLE]
and the second summand equals
[TABLE]
where if has the form with , . From the coassociativity of the coproduct on , we see that the sum of the above two expressions
[TABLE]
Since for , we have the identity
[TABLE]
In summary, we proved the desired identity. ∎
Theorem 5.10**.**
Suppose is a Koszul AS-regular algebra with semisimple Nakayama automorphism . Then we have isomorphism
[TABLE]
and such isomorphisms respect the Connes operator on both sides.
Proof.
Combining the above two lemmas and the following fact
[TABLE]
where we use the -bimodule structure of in and the -cobimodule structure of in , we get the proof. ∎
Now consider the following complex
[TABLE]
with coboundary given by
[TABLE]
Theorem 5.11**.**
Let be a Koszul algebra and denoted its Koszul dual algebra by . Then there are natural isomorphisms
[TABLE]
of graded commutative algebras. The products on both sides are the Gerstenhaber cup product.
Proof.
See Buchweitz [2], or Beilinson-Ginzburg-Soergel [3], or Keller [10]. ∎
5.3. Proof of the main theorem
We are now ready to prove the main theorem of this paper.
Proof of Theorem 1.1.
By Theorems 3.5 and 4.3 we have the following commutative diagram
[TABLE]
which gives the following commutative diagram
[TABLE]
Theorem 5.11 says that the left vertical map is an isomorphism of graded algebras, and Theorem 5.10 says that the right vertical map respects the Connes operators. Thus by Lemma 2.8 we see that two Batalin-Vilkovisky algebras are isomorphic. ∎
Acknowledgements**.**
I would like to thank Xiaojun Chen and Farkhod Eshmatov for many helpful conversations and encouragements. This work is partially supported by NSFC (No. 11521061 and 11671281).
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1[1] M. Artin, W. F. Schelter, Graded algebras of global dimension 3 , Adv. Math 66, 171-216(1987).
- 2[2] R.-O. Buchweitz, Hochschild cohomology of Koszul algebras , Talk at the Conference on Representation theory, Canberra, July 2003.
- 3[3] A. Beilinson, V. Ginzburg, W. Soergel, Koszul duality patterns in representation theory , J. Amer. Math. Soc.9 (1996) 473-527.
- 4[4] K. A. Brown, J. J. Zhang, Dualizing complexes and twisted Hoschild (co)homology for Noetherian Hopf algebras , Journal of Algebra 320(5): 1814-1850.
- 5[5] X. Chen, S. Yang, G. D. Zhou, Batalin-Vilkovisky algebras and noncommutative Poincare duality of Koszul Calabi-Yau algebras , Journal of Pure and Applied Algebra 220(2016) 2500-2532.
- 6[6] Yu. Daletskii, I. Gelfand, B. Tsygan, On a variant of noncommutative differential geometry , Dokl. Akad. Nauk SSSR 308 (1989) 1239–1297 (in Russian), translation in Soviet Math. Dokl. 40 (1990) 422–426.
- 7[7] E. Getzler, J. Jones, S. Petrack, Differential forms on loop spaces and the cyclic bar complex , Volume 30, Issue 3, 3 November 1991, Pages 339-371.
- 8[8] V. Ginzburg, Calabi Yau algebras , ar Xiv: 0612139 v 3.
