betaFeedback

WorksMathematics

Topological aspects of generalized Sklyanin algebras

Abstract

We are concerned with the topology of Casimir foliations of Sklyanin algebras and their certain generalizations. In particular, we give a complete description of the topological types and bifurcations of Casimir leaves of the original Sklyanin algebras, which improves and extends analogous results from the original papers of E. Sklyanin. Our approach is based on the so-called signature formula for the Euler characteristic of real algebraic varieties. Moreover, we show that the same approach yields similar results for certain generalizations of the original Sklyanin algebras known as Jacobian Poisson structures. Keywords: Poisson structure, Sklyanin algebra, Casimir function, Casimir foliation, Nambu bracket, Jacobian Poisson structure, Euler characteristic, algebraic formula for the Euler characteristic, bifurcation diagram, isolated singularity of complete intersection, Milnor number

Introduction

We deal with the so-called Sklyanin algebras introduced in the seminal papers of E. Sklyanin [16], [17] and their generalizations studied in [6], [10], [15]. As explained in [17] the topology of the so-called Casimir foliation of Sklyanin algebra is closely related to the representations of the quantized Sklyanin algebra, which was the central topic in [17]. However the results on the topology of Casimir leaves in [16], [17] were presented without detailed proofs. Moreover, the description of the topology of Casimir levels given in [16] contained a slight error which was corrected in [17] but the proof of the correct result was only sketched. Later on, the present author realized that the topological results given in [16], [17] can be rigorously proved and generalized using standard tools of algebraic geometry and nonlinear analysis. The aim of the present paper is to present some of the results obtained in this way including the rigorous proofs of the mentioned results of E. Sklyanin.

Sklyanin algebras and Jacobian Poisson brackets

We begin by presenting the necessary background on Sklyanin algebras and their generalizations known as Jacobian Poisson structures. The Sklyanin algebras were introduced in [16] as Poisson brackets on the algebra of polynomial functions R4\mathbb{R}_4 on four-dimensional Euclidean space R4\mathbb{R}^4 defined by the following relations

{x0,xi}=Jjkxjxk,{xi,xj}=−2x0xk.(1)\{x_0,x_i\}=J_{jk}x_jx_k,\quad\{x_i,x_j\}=-2x_0x_k. \tag*{(1)}

Here x0,x1,x2,x2x_0,x_1,x_2,x_2 are the coordinate functions, J1,J2,J3J_1,J_2,J_3 are non-zero real numbers, and Jjk=Jj−JkJ_{jk}=J_j-J_k for a cyclic permutation (i,j,k)(i,j,k) of (1,2,3)(1,2,3). For convenience and to comply with the original definition it will be assumed that J1≥J2≥J3J_1\geq J_2\geq J_3. We preserve the original notation of E. Sklyanin to facilitate the comparison of our results with those given in [16], [17].

As was shown in [16], these Poisson brackets possess two quadratic Casimir polynomials K0,K1K_0,K_1 given by

K0=x12+x22+x32,K1=x02+J1x12+J2x22+J3x32.(2)K_0=x_1^2+x_2^2+x_3^2,\quad K_1=x_0^2+J_1x_1^2+J_2x_2^2+J_3x_3^2. \tag*{(2)}

The quantized versions of these algebras are related to various exactly solvable problems of quantum mechanics. As was explained in [17], certain information about the number and types of their representations could be obtained by studying the topology and bifurcations of the so-called Casimir leaves defined as the level surfaces Xab={K0=a,K1=b}X_{ab}=\{K_0=a,K_1=b\} of Casimir functions K0,K1K_0,K_1. For this reason, several results on the topology of Casimir leaves were given in [16] but their proofs were only outlined. Later on, it turned out that the description of the topological structure of Casimir leaves presented on page 30 of [16] contains an error. This error was corrected in [17] but again without a detailed proof. Moreover, the description of bifurcations in the Casimir foliation remained incomplete. One of the aims of this paper is to give rigorous proofs of the mentioned results and give a more detailed description of bifurcations in the Casimir foliation of a Sklyanin algebra.

For the sake of clarity, we outline now our approach and the structure of the paper. We consider the mapping K:R4→R2K:\mathbb{R}^4\to\mathbb{R}^2 defined by the Casimir polynomials K0,K1K_0,K_1, show its properness and study its fibers using methods of nonlinear analysis and real algebraic geometry. Specifically, we compute bifurcation diagram B(K)B(K) and compute the Euler characteristic of non-singular fibers over each chamber (connected component of the complement of B(F)B(F) in R2\mathbb{R}^2) using the so-called signature formula for the Euler characteristic developed in our earlier works (see, e.g., [6], [7]). A computer-assisted computation showed that the Euler characteristic of non-singular Casimir leave can only be equal to zero or four. Since non-singular fibers are compact two-dimensional surfaces it follows that they are diffeomorphic either to a two-dimensional torus T2T^2 or to a disjoint union of two two-dimensional spheres S2S^2. This, in particular, shows that there are no fibers diffeomorphic to sphere S2S^2 as was erroneously claimed on the page 30 of [16].

Then we compute in the same way the Euler characteristic of fibers over each component of B(K)B(K) and find out that it is always equal to two. In order to study bifurcations of Casimir leaves we invoke a well known fact that quadratic mapping into the plane has only singularities of quadratic cone type [1]. This implies that all singular fibers are homeomorphic to a torus T2T^2 with two pinched meridians, which can be informally described as the union of two bananas with identified tips. These results are summed up in Theorem 8 below. In the rest of the paper we present similar results for generalized Sklyanin algebras defined by Jacobian Poisson brackets (Theorem 9) and outline further possible developments in the same spirit.

We conclude this section by recalling the definition of generalized Sklyanin algebras in terms of Jacobian Poisson structures which play the main role in § 1.0.3. To this end we recall first the definition of Jacobian Poisson structures which were introduced in [8] and studied in some detail [10] and [15]. Given a collection FF of n−2n-2 functionally independent real polynomials F={f1,…,fn−2}F=\{f_1,\ldots,f_{n-2}\} of nn variables the Jacobian Poisson brackets (JPB) of two smooth functions g,hg,h on Rn\mathbb{R}^n are defined by the formula

{g,h}F=J(f,g,f1,…,fn−2),(3)\{g,h\}_F=J(f,g,f_1,\ldots,f_{n-2}), \tag*{(3)}

where JJ stands for the jacobian of the functions f,g,f1,…,fn−2f,g,f_1,\ldots,f_{n-2}. Functions fjf_j are called the generators of JPB (3). Formula (3) is a particular case of the so-called Nambu brackets, which implies that {g,h}F\{g,h\}_F defines a genuine Poisson structure on Rn\mathbb{R}^n called a Jacobian Poisson structure (JPS). The basic properties of JPS have been established in [8] and [10]. The generators {f1,…,fn−2}\{f_1,\ldots,f_{n-2}\} are obviously Casimir functions of JPS (3), which define a foliation called the Casimir foliation of JPS (3). The non-singular leaves of this foliation are two-dimensional as in the case of Sklyanin algebras. The following proposition from [8] shows that Sklyanin algebras are a special case of JPS and suggests several natural generalizations of Sklyanin algebras some of which are discussed in the sequel.

Proposition 1. Sklyanin algebra is a JPS defined by Jacobian Poisson brackets (3) with generators K0,K1K_0,K_1:

{f,g}S=J(f,g,K0,K1).(4)\{f,g\}_S=J(f,g,K_0,K_1). \tag*{(4)}

If all generators of JPS (3) are homogeneous quadratic polynomials (quadratic forms) and all leaves of its Casimir foliation are compact then such a JPS will be called a generalized Sklyanin algebra (GSA). There also exist other generalizations of Sklyanin algebras but in the sequel this term is only used in the above sense. In the rest of the paper we are basically concerned with the topology of Casimir foliations of Sklyanin algebras and generalized Sklyanin algebras. A few of other generalizations will be briefly discussed from the same perspective in the last section of the paper.

From the definition of GSA given above follows that its non-singular Casimir leaves are smooth compact two-dimensional surfaces. Hence important information on the topology of Casimir leaves and Casimir foliation is given by the values of Euler characteristic of its fibers. For this reason, the main tools in our approach are algebraic formulas for the Euler characteristic given in the next section.

Algebraic formulas for the Euler characteristic

We now present the formulas for the Euler characteristic used in the sequel. The first one is an algebraic formula for the Euler characteristic of the so-called real Milnor fibre {f=0}∩Srn−1\{f = 0\} \cap S_{r}^{n-1} of an isolated singularity of real algebraic hypersurface {f=0}\{f = 0\} in Rn\mathbb{R}^{n}, where Srn−1S_{r}^{n-1} is a hypersphere of small radius rr centered at the origin. It is known that, for sufficiently small r>0r > 0 this intersection is a smooth algebraic surface of dimension n−2n - 2. It follows that, for an odd nn, the Euler characteristics of this set vanishes while, for an even nn, we have the following formula which was proved in [6].

Theorem 2 ([6]). For even nn,

χ({f=0}∩Srn−1)=2−2deg⁡0∇f,(5)\chi(\{f = 0\} \cap S_{r}^{n-1}) = 2 - 2\deg_{0}\nabla f, \tag*{(5)}

where deg⁡0∇f\deg_{0}\nabla f is the local topological degree of the gradient ∇f\nabla f (aka the index of the gradient vector field).

The above formula will enable us to compute the Euler characteristics of Casimir leaves of Sklyanin algebra. In the case of generalized Sklyanin algebras we need a more general formula for the Euler characteristic of compact real algebraic set XX which was proven in [7], [3]. Suppose that compact real algebraic set XX is given as the set of common zeroes of kk polynomials fj(x1,…,xn)f_j(x_1,\ldots,x_n) of degree dd in Rn\mathbb{R}^{n}. Let us introduce the auxiliary polynomials hjh_j Rn+1\mathbb{R}^{n+1} by the formulas

hj(x0,x1,…,xn)=x0d+1fj(x1x0,…,xnx0),j=1,…,k.h_j(x_0,x_1,\ldots,x_n)=x_0^{d+1}f_j\left(\frac{x_1}{x_0},\ldots,\frac{x_n}{x_0}\right),\quad j=1,\ldots,k.

After that we define the "master polynomial" HH by

H=∑j=1nhj2−∑j=0nxj2d+4.(6)H=\sum_{j=1}^{n}h_j^2-\sum_{j=0}^{n}x_j^{2d+4}. \tag*{(6)}

Theorem 3 ([7], [3]). With the above assumptions and notations, polynomial HH has an isolated critical point at the origin and

2χ(X)=1−deg⁡0∇H,(7)2\chi(X)=1-\deg_{0}\nabla H, \tag*{(7)}

where deg⁡0∇H\deg_{0}\nabla H is the local topological degree of the gradient ∇H\nabla H.

The above formulas being conceptually important are moreover practically applicable due to the so-called signature formula for the topological degree established in [6], [4].

The latter formula yields an algorithmic way of computing the local gradient degree participating in formulas (2, 3). In the next section, we use formulas (2, 3) to compute the Euler characteristics of Casimir leaves of Sklyanin algebras. To do that in a rigorous way we need to ensure certain topological properties of the Casimir mapping, which is done in the next section.

Casimir foliation of Sklyanin algebra

We give now the necessary topological properties of Casimir mappings and outlines of their proofs. These properties are basically corollaries of the following three properties of quadratic mappings into the plane which are well known (see, e.g., [7]).

  • P1. Quadratic mapping is proper if and only if the pencil of quadrics generated by its components contains a positively determined quadratic form.

  • P2. The image of quadratic mapping into the plane is convex.

Proposition 4. For any Sklyanin algebra, its Casimir mapping is proper.

This follows from P1 by noticing that the sum MK0+K1M K_0 + K_1 is a positively determined quadratic form for sufficiently big M>0M > 0.

Proposition 5. For any Sklyanin algebra, the image ot its Casimir map is convex.

This is a direct corollary of P2. Thus the above two properties are special cases of general results on quadratic mappings into the plane. The next two properties follow from the explicit formulas of Casimir polynomials of Sklyanin algebra.

Proposition 6. For any Sklyanin algebra, the image ot its Casimir mapping is a half-plane.

Indeed, it is easy to verify that, for any pair of numbers (u,v)(u, v) with u>0u > 0, the system K0=u,K1=vK_0 = u, K_1 = v has real solutions.

Proposition 7. For any Sklyanin algebra, the discriminant of its Casimir mapping consists of three rays in the upper half-plane.

This follows by explicit computation of the singular set S(K)S(K) of Casimir mapping and its discriminant B(K)=K(S(K))B(K) = K(S(K)). More precisely, using the notations of § 1.0.1 it can be verified that B(K)B(K) consists of three rays v=Jjuv = J_j u, j=1,2,3j = 1,2,3, in the upper half-plane u≥0u \ge0.

This done we can describe the topology of Casimir foliation as follows. Let us denote by UjU_j the four sectors which are the components of the complement of B(K)B(K) in the half-plane u≥0u \ge0. By Ehresmann theorem the topological type of the preimage K−1(P)K^{-1}(P) does not change within each UjU_j. Taking a point PjP_j in each UjU_j and computing the Euler characteristic χ(K−1(Pj))\chi(K^{-1}(P_j)) by formula (3) one finds that it only takes two values - zero and four. Since the Casimir leaves are smooth compact two-dimensional surfaces it follows that each Casimir leave is diffeomorphic either to the torus T2T^2 or to a disjoint union of two two-dimensional spheres S2S^2. The same computation shows that the leaves of the second type correspond to the outer sectors adjacent to the axis u=0u = 0, while the toric leaves correspond to the two inner sectors. It remains to describe the topological type of the leaves over each of the rays u=Jjvu = J_jv. This is again possible using formula (3), which yields that the Euler characteristic of singular leaves is equal to two. Recalling that the singular points of the fibers of proper quadratic mappings into the plane are of quadratic cone type we conclude that the bifurcation happens by pinching two meridians of the toric leave as was claimed in the introduction. One can now describe the Reeb graph of the Casimir mapping in the spirit of [19] and give a more detailed picture of bifurcations by explicit formulas. The results obtained are summed up in the following theorem.

Theorem 8. The complete list of homeomorphic types of Casimir leaves is: disjoint union of two two-dimensional spheres, two-dimensional torus, and torus with two meridians pinched. There is only type of bifurcation in Casimir foliation which corresponds to pinching two meridians of a two-torus.

In the next section we present a partial generalization of Theorem (8) to the case of GSA. We add that the same results can be obtained using the so-called Agrachev spectral sequence (see, e.g., [1]).

Topology of generalized Sklyanin algebras

The discussion in this section follows the same way of reasoning. We establish first a characterization of GSA in terms of properties of its generators.

Theorem 9. A collection QQ of n−2n - 2 real nn-ary quadratic forms qjq_j can serve as generators of a GSA Rn\mathbb{R}^n if and only if their Jacobi determinant has rank n−2n - 2 almost everywhere, and there exists a positively definite linear combination of the forms qjq_j. The image of Casimir mapping KQ:Rn→R2K_Q : \mathbb{R}^n \to\mathbb{R}^2 is a halfplane and its discriminant B(KQ)B(K_Q) is a collection of smooth curves emanating from the origin of R2\mathbb{R}^2.

This follows from the above discussion in the same way as Propositions (4), (5), (6) combined with a direct computation of the discriminant. In this setting the number of discriminant branches depends on the generators qjq_j and we are only able to obtain an upper estimate of this number. The emerging estimate of Bezout type is far from being exact and therefore we omit it. An upper estimate for the number of possible topological types of Casimir leaves can be given in terms of the so-called Petrovsky numbers introduced by V. Arnol’d in [2].

Recall that Petrovsky number P(M,m)P(M,m) was defined in [2] as the number of nn-ary monomials of degree m(M−2)m(M - 2). According to [2] the number of possible values of the gradient degree of polynomial of degree MM in mm variables does not exceed 2P(M,m)2P(M,m). Moreover, formula (3) implies that the number of possible values of the Euler characteristic of Casimir leaves of GSA does not exceed the number of possible values of the gradient degree of the master polynomial HH given by formula (6). In case of GSA, the master polynomial depends on n+1n + 1 variables and has degree 8. Hence the relevant Petrovsky number is P(8,n+1)P(8,n + 1), which implies that the number of non-diffeomorphic Casimir leaves does not exceed

(n+6)!8!(n−2)!\frac{(n+6)!}{8!(n-2)!}

This estimate is obviously far from being exact but it can be improved if the generators qjq_j are simultaneously diagonalizable in a certain coordinate system as happens in the case of original Sklyanin algebra. An interesting problem is to improve the above estimate for various special classes of GSA but this is beyond the aims of the present paper.

We proceed by noting that in addition to the above direct generalizations of the results on Sklyanin algebras, one can consider several further topological aspects of GSA involving more refined concepts and technical tools. As an example, we notice that the definition of GSA given above implies that the collection of its quadratic Casimirs defines a so-called isolated singularity of complete intersection X={F=0}X = \{F = 0\} (ICIS) in the sense of singularity theory. One of the most important topological invariants of such a singularity is its Milnor number (see, e.g., [12]). Using it we can obtain certain information on the topology of Poisson structure induced by GSA on the common zero-set XX of its Casimirs.

Theorem 10. For any generalized Sklyanin algebra on Rn\mathbb{R}^n with n≥4n \ge4, the dimension of zero-th Poisson cohomology of the induced Poisson structure on the zero-set of Casimirs is equal to

∑j=0n−3(−1)n−j−3n!j!(n−j)!2j.(8)\sum_{j=0}^{n-3}(-1)^{n-j-3}\frac{n!}{j!(n-j)!}2^j. \tag*{(8)}

This follows from the following two general results. First, as was proved in [8], for any JPS, the dimension of zero-th Poisson cohomology of the induced Poisson structure on the zero-set of its Casimirs is equal to the Milnor number of the ICIS defined by the generators. The second general fact is the following formula for the Milnor number of an ICIS defined by homogeneous polynomials of arbitrary degrees, which was given by G.Greuel and H.Hamm in [5] (see [12] for a detailed and simplified proof).

Theorem 11. ([5])

μ(f)=∑j=0k−1(−1)k−j−1n!j!(n−j)!2j.(9)\mu(f)=\sum_{j=0}^{k-1}(-1)^{k-j-1}\frac{n!}{j!(n-j)!}2^j. \tag*{(9)}

In view of the said above, formula (8) is obtained by merely combining the above observations with formula (9) and taking into account that k=n−2k = n - 2 in the case of a generalized Sklyanin algebra on Rn\mathbb{R}^n.

Corollary 12. For an original Sklyanin algebra, the dimension of zero-th Poisson cohomology of the induced Poisson structure on the zero-set of Casimirs is equal to seven.

It would be interesting to give explicit formulas for the generators of the zero-th Poisson cohomology of Sklyanin algebras and generalized Sklyanin algebras.

Concluding remarks

The approach and results presented above suggest several further developments along the same lines. For example, the upper estimate for the number of possible topological types of the Casimir leaves of GSA in terms of Petrovsky number can be improved for special classes of GSA. In particular, if n=5n = 5 and all Casimir polynomials admit simultaneous diagonalization, all possible topological types have been listed by V. Krasnov [11].

Next, on more sophisticated level, an obvious research prospect is to compute the Poisson (co)homology of generalized Sklyanin algebras. As a paradigm one can use the comprehensive results on the Poisson cohomology of the original Sklyanin algebras given by S.R.T. Pelap [14]. A natural next step would be to extend the results of [14] to all generalized Sklyanin algebras in dimension four.

Finally, analogs of Theorem 8 can be obtained for Jacobian Poisson structures with Casimirs of arbitrary degree using the general formula (9).

Developments suggested above require separate discussion and will (hopefully) be considered elsewhere.

References

  1. [1]A. Agrachev, Quadratic cohomology, Arnold Math. J. 1, No.1, 2015, 37-58.
  2. [2]V. Arnold’s, Index of singular point of a vector field, the Petrovsky-Oleynik inequality, and mixed Hodge structures, Func. Anal. Applic. 12, No.1, 1978, 1-12.
  3. [3]W. Bruce, Euler characteristics of real varieties, Bull. Lond. Math. Soc. 22, No.6, 1990, 547-552.DOI
  4. [4]D. Eisenbud, H. Levine, An algebraic formula for the mapping degree of a C∞-map germ, Ann. Math. 108, 1977, ??
  5. [5]G.-M. Greuel, H. Hamm, Invarianten quasihomogener vollständiger Durchschnitte, Invent. Math. 49, No.1, 1978, 67-86.
  6. [6]G. Khimshiashvili, On the local degree of a smooth mapping, Bull. Acad. Sci. Georgian SSR 85, No.2, 1977, 309-312.
  7. [7]G. Khimshiashvili, On the cardinality of a semi-algebraic subset, Georgian Math. J. 1, No.4, 1994, 511-521.
  8. [8]G. Khimshiashvili, On certain exact Poisson structures, Proc. A.Razmadze Math. Inst. 119, 1999.
  9. [10]G. Khimshiashvili, R. Przybysz, On generalized Sklyanin algebras, Georgian Math. J. 7, No.4, 2000, 789-800.DOI
  10. [11]V. Krasnov, On intersections of two real quadrics, Izvestiya: Mathematics 82, No.1, 2018, 91-139.DOI
  11. [12]E. Looijenga, Isolated singular points on complete intersections, Cambridge University Press, 2010.
  12. [14]S.R.T. Pelap, Poisson (co)homology of polynomials Poisson brackets in dimension four: Sklyanin’s case, J. Algebra 322, 2009, 1151-1169.
  13. [15]R. Przybysz, On one class of exact Poisson structures, J. Math. Phys. 42, 2001, 1913-1920.DOI
  14. [16]E. Sklyanin, On some algebraic structures connected with Yang-Baxter equation, Func. Anal. Applic. 1, No.1, 1982, 27-34.
  15. [17]E. Sklyanin, On some algebraic structures connected with Yang-Baxter equation. Representations of quantum algebra, Func. Anal.Applic. 17, No.1, 1983, 34-48.
  16. [19]P. Vanhaecke, Integrable systems in the realm of algebraic geometry, Lecture Notes Math. 1638, 2001.

Paper details

vlnc:W_arkw6e3x
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents