A complexanalytic proof of a criterion for isomorphism of Artinian Gorenstein algebras
 Alexander Isaev^{1}Email author
https://doi.org/10.1186/2197120X11
© Isaev; licensee Springer. 2015
Received: 5 March 2014
Accepted: 9 March 2014
Published: 3 June 2015
Abstract
Let A be an Artinian Gorenstein algebra over a field k of characteristic zero with dimk A>1. To every such algebra and a linear projection π on its maximal ideal with range equal to the socle Soc(A) of A, one can associate a certain algebraic hypersurface , which is the graph of a polynomial map P _{ π }: kerπ→Soc(A)≃k. Recently, the following surprising criterion has been obtained: two Artinian Gorenstein algebras A and are isomorphic if and only if any two hypersurfaces S _{ π } and arising from A and , respectively, are affinely equivalent. In the present paper, we focus on the cases and and explain how in these situations the above criterion can be derived by complexanalytic methods. The complexanalytic proof for has not previously appeared in print but is foundational for the general result. The purpose of our paper is to present this proof and compare it with that for , thus highlighting a curious connection between complex analysis and commutative algebra.
Mathematics Subject Classification
13H10; 32V40
Keywords
Artinian Gorenstein algebras CRautomorphisms of real quadrics in complex space1 Introduction
This paper concerns a result in commutative algebra but is intended mainly for experts in complex analysis and CRgeometry. The background required for understanding the algebraic content is rather minimal and is given in Section 2.
We consider Artinian local commutative associative algebras over a field k. Such an algebra A is Gorenstein if and only if the socle Soc(A) of A is a onedimensional vector space over k. Gorenstein algebras frequently occur in various areas of mathematics and its applications to physics (see, e.g., [1],[12]). In the case when the field k has characteristic zero, in articles [6],[11], a surprising criterion for isomorphism of Artinian Gorenstein algebras was found. The criterion is given in terms of a certain algebraic hypersurface S _{ π } in the maximal ideal of A associated to a linear projection π on with range Soc(A), where we assume that dimk A>1. The hypersurface S _{ π } passes through the origin and is the graph of a polynomial map P _{ π }: kerπ→Soc(A)≃k. It is shown in [6],[11] that two Artinian Gorenstein algebras A and are isomorphic if and only if any two hypersurfaces S _{ π } and arising from A and , respectively, are affinely equivalent, that is, there exists a bijective affine map such that .
Currently, there are two different proofs of the above criterion. The one given in [11] is purely algebraic, whereas the one proposed in [6] reduces the case of an arbitrary field to that of . A proof of the criterion in the latter case is contained in our earlier article [7] and, quite surprisingly, is based on a complexanalytic argument. It turns out that one can give an independent complexanalytic proof of the criterion for as well. This proof has not been previously published and is derived from ideas developed in paper [5], which never appeared in print (see Remark 4.1). However, the argument utilized in the case is rather important as it inspired our proof for the case in [7] and therefore laid the foundation of the general result for an arbitrary field. Thus, the purpose of the present article is to provide a complexanalytic proof of the criterion for and to compare it with that for given in [7]. These proofs emphasize an intriguing connection between complex analysis and commutative algebra. In each of the two cases, the idea is to consider certain tube submanifolds in complex space associated to the algebras in question and utilize their CRautomorphisms. In fact, as stated in Remark 5.1, there is a deep relationship between Artinian Gorenstein algebras for and tube hypersurfaces locally CRequivalent to Levi nondegenerate hyperquadrics. We believe that this relationship has much to offer if fully understood.
The paper is organized as follows. Section 2 contains algebraic preliminaries and the precise statement of the criterion in Theorem 2.1. The proof of the necessity implication of Theorem 2.1 is given in Section 3. This part of the proof has no relation to complex analysis and is only included for the completeness of our exposition. Further, Sections 4 and 5 contain the complexanalytic proofs of the sufficiency implications for and , respectively. Finally, in Section 6, we demonstrate how powerful our criterion can be in applications. Namely, we apply it to a oneparameter family A _{ t } of 15dimensional Gorenstein algebras. While directly finding all pairwise isomorphic algebras in the family A _{ t } seems to be quite hard, this problem is easily solved with the help of Theorem 2.1
2 Preliminaries
Let A be a commutative associative algebra over a field k. We assume that A is unital and denote by 1 its multiplicative identity element. Further, we assume that A is local, that is, (i) A has a unique maximal ideal (which we denote by ), and (ii) the natural injective homomorphism is in fact an isomorphism. In this case, one has the vector space decomposition , where 〈·〉 denotes linear span. Furthermore, A is isomorphic to the unital extension of its maximal ideal , which is the direct sum endowed with an operation of multiplication in the obvious way. For example, the algebra of germs of holomorphic functions at the origin in is a complex local algebra whose maximal ideal consists of all germs vanishing at 0. Clearly, every element of is the sum of the germ of a constant function and a germ vanishing at the origin.
Next, we suppose that dimk A>1 and that A is Artinian, i.e., dimk A<∞. In addition, we let A to be Gorenstein, which means that the socle of A, defined as , is a onedimensional vector space over k (see, e.g., [9]). For example, if I is the ideal in generated by the germs of , then is a complex Artinian Gorenstein algebra with and spanned by the element of represented by the germ of z _{1} z _{2}.
is nondegenerate on A (see, e.g., p. 11 in [8]). Numerous examples of hypersurfaces S _{ π } explicitly computed for particular algebras can be found in [2],[6],[7] (see also Section 6 below).
We will now state the criterion for isomorphism of Artinian Gorenstein algebras obtained in [6],[11].
Theorem 2.1.
Let A and be Gorenstein algebras of finite vector space dimension greater than 1 over a field of characteristic zero and π and admissible projections on A and , respectively. Then A and are isomorphic if and only if the hypersurfaces S _{ π } and are affinely equivalent.
Remark 2.2.
Clearly, S _{ π } and are affinely equivalent if and only if and are affinely equivalent. Therefore, in order to prove Theorem 2.1, it is sufficient to obtain its statement with and in place of S _{ π } and , respectively. The hypersurfaces and are easier to work with, and we utilize them instead of S _{ π } and in our proofs below.
3 Proof of the necessity implication
First of all, we explain how the necessity implication of Theorem 2.1 is derived. As stated in the introduction, this part of the proof has no relation to complex analysis and is only included in the paper for the completeness of our exposition. The proof below works for any field of characteristic zero. The idea is to show that if π _{1} and π _{2} are admissible projections on A, then for some . Clearly, the necessity implication is a consequence of this fact (cf. [7]).
For every , let M _{ y } be the multiplication operator from A to defined by a↦y a and set . The correspondence defines a linear map from into the space of linear maps from to Soc(A). Since for every admissible projection π the form b _{ π } defined in (2.3) is nondegenerate on A and since , it follows that is an isomorphism.
4 Proof of the sufficiency implication for
By assumption, , and we denote this common dimension by N. If N=2, then A and are clearly isomorphic, and thus, from now on, we suppose that N>2.
Let be the complexification of the real algebra A. Then , and is a complex Artinian Gorenstein algebra with maximal ideal . Next, we denote by the complexlinear extension of π to and by the conjugation on defining the real form A, for all , with x,y∈A. Then is a valued Hermitian form on that coincides with b _{ π } on A (see (2.3)). Since the bilinear form b _{ π } is nondegenerate on A, the Hermitian form h is nondegenerate on .
where w:=(w _{1},…,w _{ N2}) and H is a nondegenerate Hermitian form on .
Analogously, for the algebra , we obtain a Hermitian valued form on , a real Levi nondegenerate hyperquadric in , the corresponding affine hyperquadric in , and a tube hypersurface in .
where and is the exponential map associated to . Observe that Φ maps onto .
where satisfies H(U w,U w)≡σ H(w,w), , , λ>0, σ=±1, and σ may be equal to 1 only if the numbers of positive and negative eigenvalues of H are equal. In particular, Φ is an affine map as claimed.
where x _{0}:=f(0), g:=fx _{0} is the linear part of f, and is the exponential map associated to . Since Φ is affine, formula (4.4) implies g(x)^{2}=g(x ^{2}) for all , which is equivalent to the statement that is an algebra isomorphism. Therefore, and are isomorphic, hence A and are also isomorphic as required. □
Remark 4.1.
The method utilized in the above proof can be extracted, in principle, from ideas contained in paper [5] (which should be read in conjunction with [4]), but is by no means explicitly articulated there.
5 Proof of the sufficiency implication for
The argument that follows is contained in [7] and is included in this paper for comparison with the proof in the case given in Section 4.
By assumption, , and we denote this common dimension by N. If N=2, then A and are isomorphic, and thus, from now on, we suppose that N>2.
Consider the maximal ideal of A. We will now forget the complex structure on and treat it as a real algebra. Let be the (real) unital extension of (see Section 2) and 1 the multiplicative identity element in . Observe that is not Gorenstein since . We now consider the complexification of . Then , and is a complex Artinian local algebra with maximal ideal . The complex algebra is not Gorenstein since .
Next, let be the extension of to defined by the condition and denote by the complexlinear extension of to . Further, denote by the conjugation on defining the real form , for all , with . Then is a valued Hermitian form on that coincides on with the valued bilinear form defined analogously to (2.3).
This can be seen by choosing coordinates in (regarded as a complex algebra) in which the restriction of the Soc(A)valued bilinear form b _{ π } to is given by the identity matrix.
Analogously, for the algebra , we obtain algebras and , a Hermitian valued form on , a real Levi nondegenerate codimension 2 affine quadric in , the corresponding coordinates in , and a tube hypersurface in .
where and is the exponential map associated to . Observe that Φ maps onto .
We claim that, upon identification of with and with , the map Φ is affine. Indeed, when written in the coordinates w _{1},…,w _{2N2}, , the map Φ becomes an automorphism of preserving quadric (5.1). The fact that Φ is affine now follows from a description of CRautomorphism of this quadric (see the elliptic case on pp. 37–38 in [3]).
where x _{0}:=f(0), g:=fx _{0} is the linear part of f, and is the exponential map associated to . Since Φ is affine, formula (5.3) implies g(x)^{2}=g(x ^{2}) for all , i.e., is an algebra isomorphism. Therefore, and are isomorphic, hence A and are also isomorphic as required. □
Remark 5.1.
The proofs of Theorem 2.1 for the cases presented in this paper are based on considering real tube submanifolds in complex space CRequivalent to Levi nondegenerate affine quadrics. For , we utilized hypersurfaces, whereas for , codimension 2 submanifolds were required. The former are called spherical tube hypersurfaces (see [10]), and there is in fact an intriguing relationship between them and real and complex Artinian Gorenstein algebras. This relationship was outlined in [5] (see also Section 9.2 in [10] for a brief survey). It turns out that

to every real Artinian Gorenstein algebra of dimension greater than 2 and to every complex Artinian Gorenstein algebra one can associate a (closed) spherical tube hypersurface, where in the real case one obtains exactly hypersurfaces with bases (2.4) as in (4.2);

any two such hypersurfaces are affinely equivalent if and only if the corresponding algebras are isomorphic (one way to obtain the necessity implication in the real case is to proceed as in Section 4);

in a certain sense, all spherical tube hypersurfaces can be obtained by combining hypersurfaces of these two types.
We note that no highercodimensional analogues of spherical tube hypersurfaces were considered in [5]. However, as the proof of Theorem 2.1 for suggests, analogues of this kind are related to Artinian Gorenstein algebras as well. We believe that the curious connection between complex analysis and commutative algebra manifested through tube submanifolds CRequivalent to affine quadrics deserves further investigation.
6 Example of application of Theorem 2.1
Theorem 2.1 is particularly useful when at least one of the hypersurfaces S _{ π } and is affinely homogeneous (recall that a subset S of a vector space V is called affinely homogeneous if for every pair of points p,q∈S there exists a bijective affine map g of V such that g(S)=S and g(p)=q). In this case, the hypersurfaces S _{ π } and are affinely equivalent if and only if they are linearly equivalent. Indeed, if, for instance, S _{ π } is affinely homogeneous and is an affine equivalence between S _{ π } and , then f∘g is a linear equivalence between S _{ π } and , where g is an affine automorphism of S _{ π } such that g(0)=f ^{1}(0). Clearly, in this case, is affinely homogeneous as well.
where and are the homogeneous components of degree m of P _{ π } and , respectively.
Thus, Theorem 2.1 yields the following corollary (cf. Theorem 2.11 in [7]).
Corollary 6.1
 (i)
If A and are isomorphic and at least one of S _{ π } and is affinely homogeneous, then for some linear isomorphisms and identity (6.1) holds. In this case, both S _{ π } and are affinely homogeneous.
 (ii)
If for some linear isomorphisms and identity (6.1) holds, then the hypersurfaces S _{ π } and are linearly equivalent, and therefore the algebras A and are isomorphic.
where A ^{ j } are linear subspaces of A, with A ^{0}=k (in this case and Soc(A)=A ^{ d } for d:= max{j:A ^{ j }≠0}). It then follows that part (i) of Corollary 6.1 applies in the situation when one (hence the other) of the algebras A and is nonnegatively graded (see also [7] for the case ). Note, however, that the existence of a nonnegative grading on A is not a necessary condition for the affine homogeneity of S _{ π } (see, e.g., Remark 2.6 in [7]). Also, as shown in Section 8.2 in [6], the hypersurface S _{ π } need not be affinely homogeneous in general.
It is straightforward to verify that every A _{ t } is a Gorenstein algebra of dimension 15. We will prove the following proposition.
Proposition 6.2
A _{ r } and A _{ s } are isomorphic if and only if r=±s.
Proof.
where, as before, 〈·〉 denotes linear span. It is straightforward to check that the subspaces form a nonnegative grading on A _{ t }.
Now, (6.3), (6.4), and (6.5) yield r ^{2}=s ^{2} as required. □
Declarations
Acknowledgements
This work is supported by the Australian Research Council.
Authors’ Affiliations
References
 Bass H: On the ubiquity of Gorenstein rings. Math. Z 1963, 82: 8–28. 10.1007/BF01112819MATHMathSciNetView ArticleGoogle Scholar
 Eastwood MG, Isaev AV: Extracting invariants of isolated hypersurface singularities from their moduli algebras. Math. Ann 2013, 356: 73–98. 10.1007/s0020801208367MATHMathSciNetView ArticleGoogle Scholar
 Ezhov V, Schmalz G: A matrix Poincaré formula for holomorphic automorphisms of quadrics of higher codimension. Real associative quadrics. J. Geom. Anal 1998, 8: 27–41. 10.1007/BF02922107MATHMathSciNetView ArticleGoogle Scholar
 Fels G, Kaup W: Local tube realizations of CRmanifolds and maximal Abelian subalgebras. Ann. Sc. Norm. Super. Pisa Cl. Sci. (5) 2011, 10: 99–128.MATHMathSciNetGoogle Scholar
 Fels G, Kaup W: Classification of commutative algebras and tube realizations of hyperquadrics. preprint, available from http://arxiv.org/pdf/0906.5549v2
 Fels G, Kaup W: Nilpotent algebras and affinely homogeneous surfaces. Math. Ann 2012, 353: 1315–1350. 10.1007/s0020801107184MATHMathSciNetView ArticleGoogle Scholar
 Fels G, Isaev A, Kaup W, Kruzhilin N: Isolated hypersurface singularities and special polynomial realizations of affine quadrics. J. Geom. Analysis 2011, 21: 767–782. 10.1007/s122200119223yMATHMathSciNetView ArticleGoogle Scholar
 Hertling C: Frobenius Manifolds and Moduli Spaces for Singularities. Cambridge: Cambridge Tracts in Mathematics 151, Cambridge University Press; 2002.MATHView ArticleGoogle Scholar
 Huneke C: Hyman Bass and ubiquity: Gorenstein rings. Algebra, Ktheory, Groups, and Education (New York, 1997), Contemp. Math. 243, Amer. Math. Soc., Providence, RI (1999)Google Scholar
 Isaev AV: Spherical Tube Hypersurfaces. Lecture Notes in Mathematics 2020. Berlin: Springer; 2011.Google Scholar
 Isaev AV: On the affine homogeneity of algebraic hypersurfaces arising from Gorenstein algebras. Asian J. Math 2011, 15: 631–640. 10.4310/AJM.2011.v15.n4.a7MATHMathSciNetView ArticleGoogle Scholar
 Lurie J: On the classification of topological field theories. In Current Developments in Mathematics, 2008. Somerville, MA: Int. Press; 2009.Google Scholar
 Matsumura H: Commutative Ring Theory. Cambridge: Cambridge Studies in Advanced Mathematics 8, Cambridge University Press; 1986.MATHGoogle Scholar
Copyright
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited.