Representation theory of (modified) Reflection Equation Algebra of GL m n ( | ) type 7 0 0 2 Dimitri Gurevich∗ n USTV, Universit´e de Valenciennes, 59304 Valenciennes, France a J Pavel Pyatov† 0 Bogoliubov Laboratory of Theoretical Physics, JINR, 141980 Dubna, Moscow region, Russia 1 Pavel Saponov‡ ] A Division of Theoretical Physics, IHEP, 142284 Protvino, Russia Q . February 2, 2008 h t a m [ Abstract 2 v Let R : V⊗2 → V⊗2 be a Hecke type solution of the quantum Yang-Baxter equation (a 5 Hecke symmetry). Then, the Hilbert-Poincr´e series of the associated R-exterior algebra of the 1 space V is a ratio of two polynomials of degree m (numerator) and n (denominator). 8 Assuming R to be skew-invertible, we define a rigid quasitensor category SW(V ) of 2 (m|n) 1 vector spaces, generated by the space V and its dual V∗, and compute certain numerical char- 6 acteristics of its objects. Besides, we introduce a braided bialgebra structure in the modified 0 Reflection Equation Algebra, associated with R, and equip objects of the category SW(V ) (m|n) / h with an action of this algebra. In the case related to the quantum group Uq(sl(m)), we con- t sider the Poisson counterpart of the modified Reflection Equation Algebra and compute the a m semiclassical term of the pairing, defined via the categorical (or quantum) trace. : v AMS Mathematics Subject Classification, 1991: 17B37, 81R50 i Key words: (modified) reflection equation algebra, braiding, Hecke symmetry, Poincar´e- X Hilbert series, bi-rank, Schur-Weyl category, (quantum) trace, (quantum) dimension, braided bial- r a gebra 1 Introduction Reflection Equation Algebra (REA) is a very useful tool of the theory of integrable systems with boundaries. It derives its name from an equation describing the factorized scattering on a half-line (cf. [C], where the REA depending on a spectral parameter was first introduced). Nowadays, different types of the REA are known (cf. [KS]), which have applications in mathe- matical physics and non-commutative geometry. TheREA related to the Drinfeld-JimboQuantum Group (QG) U (sl(m)) appears in construct- q ing a q-analog of differential calculus on the groups GL(m) and SL(m), where it was treated to be a q-analog of the exponential of vector fields (cf. [FP]). ∗[email protected] †[email protected] ‡[email protected] 1 In the case related to the QG U (g), an appropriate quotient of the REA can be treated as a q deformation of the coordinate ring K[G] where G is the Lie group, corresponding to a classical Lie algebra g. The Poisson bracket corresponding to this deformation was introduced by M.Semenov- Tian-Shansky1. Though the best known REA is related to the QG U (g), such an algebra can be associated to q any braiding R : V⊗2 → V⊗2, where V is a finite dimensional linear space over the ground field2 K and R is an invertible solution of the quantum Yang-Baxter equation R R R = R R R . (1.1) 12 23 12 23 12 23 Here the indices of R relate to the space (or spaces) in which the operator is applied. Thus, R 12 and R are the following operators in the space V⊗3: R = R⊗I, R = I ⊗R. 23 12 23 In the present paper we deal with Hecke type solutions of the Yang-Baxter equation (1.1) which satisfy the following condition (R−qI)(R+q−1I) = 0, (1.2) where the nonzero parameter q ∈ K is assumed to be generic. By definition, this means, that the values of q do not belong to a countable set of the roots of unity: qk 6= 1, k = 2,3,... (whereas the value q = 1 is not excluded). Consequently, qk −q−k k := 6= 0, ∀k ∈ N, q q−q−1 k beingaa q-analog aninteger k. Inwhatfollows, abraidingsatisfyingrelation (1.2)willbecalled q a Hecke symmetry. Especially, we are interested in families of Hecke symmetries R analytically depending on the q parameter q in a neighbourhood of 1 ∈ K in such a way, that for q = 1 the symmetry R = R is 1 involutive: R2 =I. The well known example of such a family is the U (sl(m)) Drinfeld-Jimbo braidings q m m R = qδij hj ⊗hi + (q−q−1)hi⊗hj (1.3) q i j i j i,j=1 i<j X X j where the elements h form the natural basis in the space of left endomorphisms of V, that is i j j h (x ) = δ x in afixed basis {x } of the space V. Note that for q = 1 theabove braidingR equals i k k i k the usual flip P. The Hecke symmetry (1.3) and all related objects will be called standard. However, a large number of Hecke symmetries different from the standard one are known, even those which are not deformations of the usual flip (cf. [G3]). Let us consider the REA corresponding to the standard U (sl(m)) Hecke symmetry (1.3) in q more detail. This algebra possesses some very important properties, in contrast with the REA related to other quantum groups U (g), g 6= sl(m). q First of all, it is a q-deformation of the commutative algebra Sym(gl(m)) = K[gl(m)∗] (so, we getadeformationalgebrawithouttakinganyadditionalquotient). Second,byalinearshiftofREA generators (proportional to a parameter ~), we come to quadratic-linear commutation relations for theshiftedgenerators. InthisbasistheREAcanbetreated as a”doubledeformation”oftheinitial commutative algebra K[gl(m)∗]. We refer to this form of the REA as modified Reflection Equation 1Note that on any classical Lie group G there exists another Poisson bracket due to E.Sklyanin. Its quantum analog isanappropriatequotientoftheso-called RTTalgebra (cf. [FRT]). Thesetwoquantumanalogs ofthespace K[G] are related by a transmutation procedure introduced by S. Majid (cf. [M] and references therein). Nowadays, there exists theiruniversal treatment based on pairs of so-called compatible braidings (cf. [IOP, GPS1, GPS2]). 2Mainly we are dealing with K=C but sometimes K=R is allowed. 2 Algebra (mREA) and we denote it L(R ,~). By specializing ~ = 0 we return to the (non-modified) q REA L(R ). q The specialization of the algebra L(Rq,~) at q = 1 gives the enveloping algebra U(gl(m)~) where the notation g means that the bracket [, ] of a Lie algebra g is replaced by ~[, ]. (Note, ~ that this fact was observed in [IP].) The commutative algebra K[gl(m)∗] is obtained by the double specialization of the algebra L(R ,~) at ~ =0 and q = 1. q Being equipped with the U (sl(m))-module structure, the algebra L(R ,~) (as well as L(R )) q q q is U (sl(m))-equivariant (or covariant). This means that q M(x·y)= M (x)·M (y), ∀M ∈ U (sl(m)), ∀x,y ∈ L(R ,~) (1) (2) q q where we use the Sweedler’s notation for the quantum group coproduct ∆(M) = M ⊗M . (1) (2) ThePoissoncounterpartoftheabovedoubledeformationofthealgebraK[gl(m)∗]isthePoisson pencil {,} = a{,} +b{,} , a,b ∈ K (1.4) PL,r PL r where{,} is thelinear Poisson-Lie bracket related totheLiealgebra gl(m) and{,} is anatural PL r extension of the Semenov-Tian-Shansky bracket on the linear space gl(m)∗. WeconsiderthesePoissonstructuresandbrieflydiscusstheirroleindefininga”quantumorbit” O ⊂ gl(m)∗ in Section 7. Taking a two-dimensional sphere as an example, we suggest a method of constructing such quantum orbits. In contrast with other definitions of quantum homogeneous spaces, our quantum orbits are some quotients of the algebra L(R ,~). They look like the ”fuzzy q sphere” SLc(~) = U(su(2)~)/hC −ci where C is the quadratic Casimir element. As is known, there exists a discrete series of numbers c ∈ K such that any algebra SLck(~) has a finite dimensional representation in a linear space V k k and the corresponding map SLck(~) → End(V ) is an su(2)-morphism. k A similar statement is valid for the aforementioned quotients of the algebra L(R ,~). However, q the corresponding spaces V become objects of a quasitensor category. In such a category, an k object is characterized by its quantum dimension which is defined via the categorical (quantum) trace. A deformation of the usual trace is one of the main features of our approach to the quantum homogeneous spaces. In Section 7 we describe the semiclassical term of the paring defined via the quantum trace in the case of the standard Hecke symmetry. In a similar way we treat other quasitensor categories generated by skew-invertible Hecke sym- metries. Roughly speaking, we are dealing with three problems in the present paper. The first problem is the classification of all (skew-invertible) Hecke symmetries R. One of the main tools for studying this problem is the Hilbert-Poincar´e (HP) series P (t) corresponding to the ”R-exterior − algebra” of the space V (its definition is presented in Section 3). Though a classification of all possible forms of the HP series P (t) has not been found yet, it is known that the HP series P (t) − − of any Hecke symmetry is a rational function3 [H, D]. The ordered pair of integers (m|n), where m (resp., n) is the degree of the numerator N(t) (resp., denominator D(t)) of P (t), plays an − important role in the sequel and will be called the bi-rank of the Hecke symmetry R (or of the corresponding space V). This pair enters our notation of the quasitensor Schur-Weyl category SW(V ) generated by V. (m|n) Constructing the category SW(V ) is the second problem we are dealing with in this paper. (m|n) The objects of the category are direct sums of vector spaces V ⊗V∗. Here V is the basic vector λ µ 3The HP series corresponding to a skew-invertible Hecke symmetry is described in Section 3. When P−(t) is a polynomial (in this case we say that R is even) it can drastically differ from the classical one (1+t)n, n = dimV. Thus,in[G3]allskew-invertibleHeckesymmetrieswithP−(t)=1+nt+t2 wereclassified. Besides,suggestedin[G3] wasawayof”gluing”suchsymmetrieswhichgivesrisetoskew-invertibleHeckesymmetrieswithothernon-standard HP series. 3 space equipped with a skew-invertible Hecke symmetry R, V∗ is its dual, and λ and µ stand for arbitrary partitions (Young diagrams) of positive integers. The map V → V is nothing but the λ Schur functor corresponding to the Hecke symmetry R (for its classical version cf. [FH]). The map V∗ → V∗ can be defined in a similar way. Note, that SW(V ) is a monoidal quasitensor rigid µ (m|n) category (as defined in [CP]) but it is not abelian. Wecomputesomenumericalcharacteristicsofobjectsofthiscategory. Namely,weareinterested in their dimensions (classical and quantum). In contrast with the classical dimensions, which essentially depend on a concrete form of the initial Hecke symmetry and is expressed via the roots of above polynomials N(t) and D(t), the quantum dimensions depend only on the bi-rank (m|n). Moreover, inasense,thecategorySW(V )lookslikethetensorcategoryofU(gl(m|n))-modules. (m|n) ThethirdproblemelaboratedbelowisinconstructingtherepresentationsofthemREAL(R ,~) q in the category SW(V ). Since for q 6= 1 the algebra L(R ,~) is isomorphic to the non-modified (m|n) q REA L(R ) (in fact, we have thesame algebra written in two different bases), we automatically get q arepresentation category ofthelatter algebra4. Note, thatcertain representations oftheREAhave alreadybeenknown,mainlyfortheevencase(thebi-rank(m|0))[K,Mu1,GS2,S]. Incontrastwith those papers, we here consider the mREA L(R ,~) connected with a general type skew-invertible q Hecke symmetry R of the bi-rank (m|n) and equip objects of the category SW(V ) with the (m|n) L(R ,~)-module structure. Note, that all the corresponding representations are equivariant (see q Section 6). A particular example we are interested in is the ”adjoint” representation. By this we mean a representation ρ of the mREA L(R ,~) in the linear span of its generators. In the case, when a ad q Hecke symmetry is a super-flip in a Z -graded linear space V 2 R : V⊗2 → V⊗2, R(x⊗y)= (−1)xyy⊗x, where x and y are homogenous elements of V and z¯denotes the parity (grading) of a homogeneous element z, the mREA becomes the enveloping algebra U(gl(m|n)) and the representation ρ ad coincides with the usual adjoint one. This is one of the reasons why we treat the mREA L(R ,~) q as a suitable analog of the enveloping algebra. Moreover, in the case of involutive skew-invertible Hecke symmetry, the corresponding mREA becomes the enveloping algebra of a generalized Lie algebra End(V) as is explained in Section 5. Such algebras were introduced in [G1]. The other property that makes the mREA similar to the enveloping algebra of a generalized Lie algebra (in particular, a super Lie algebra) is its braided bialgebra structure. Such a structure is determined by a coproduct ∆ and a counit ε. On the generators of mREA (organized into a matrix L (see Section 6)) the coproduct reads ∆(L) = L⊗1+1⊗L−(q−q−1)L⊗L andcoincides withthecoproductoftheenvelopingalgebraofthe(generalized) Liealgebraatq = 1. Note, that though we do not define an antipode in the algebra L(R ,~), the category SW(V ) q (m|n) of its representations is closed. In addition to the L(R ,~)-module structure, the objects of the Schur-Weyl category, corre- q sponding to the standard Hecke symmetry (1.3), can be equipped with the action of the QG U (sl(m)). Besides, the q-analogs of super-groups (cf. [KT]) can also be represented in the corre- q spondingSchur-Weyl category. (Suggested in[Z] is anotherway of constructingthe representations of q-deformed algebras U(gl(m|n) which is based on the triangular decomposition.) Nevertheless, in general we know no explicit construction of the QG type algebra for a skew-invertible Hecke symmetry5 whereas the mREA can be defined for any skew-invertible Hecke symmetry. 4Since for q = 1 the isomorphism L(Rq,~) ∼= L(Rq) breaks, we prefer to consider these algebras separately and use different names for them. 5An attempt of explicit description of such an object for some even non-quasiclassical Hecke symmetries was undertakenin [AG]. 4 The mREA has one more advantage compared with the QG or their super-analogs. It is a more convenient tool for the explicit construction of projective modules over quantum orbits in the frameworks of approach suggested in [GS1, GS3]. We plan to turn to these objects in a general (not necessarily even) case in our subsequent publications. To complete the Introduction, we would like to emphasize a difference between the Hecke type braidings and the Birman-Murakami-Wenzl ones (in particular, those coming from the QG of B , n C and D series). In thelatter case it is not difficult to definea ”braided Lie bracket” in the space n n End(V) (cf. [DGG]) and introduce the corresponding ”enveloping algebra”. But this ”enveloping algebra” is not a deformation of its classical counterpart and therefore is not an interesting object from our viewpoint. The paper is organized as follows. In the next Section we reproduce some elements of R- technique which form the base of subsequent computations of some interesting numerical charac- teristics ofobjectsinvolved (themostcumbersomepartofthecomputationsisplacedinAppendix). Section 3 is devoted to the classification of (skew-invertible) Hecke symmetries. In Section 4 we construct the Schur-Weyl category SW(V ) generated by the space V. Our main object, the (m|n) mREA L(R ,~), is introduced in Section 5 where we also study its deformation properties. In q Section 6 we equip the mREA with a braided bialgebra structure which allows us to define an equivariant action of the algebra L(R ,~) on each object of the category SW(V ). There we q (m|n) also present our viewpoint on definition of braided (quantum) Lie algebras. Section 7 is devoted to study of some semiclassical structures. Acknowledgement. We would like to thank the Max-Planck-Institut fu¨r Mathematik, wherethis work was written, for the warm hospitality and stimulating atmosphere. The work of D.G. was partially supported by the grant ANR-05-BLAN-0029-01, the work of P.P. and P.S. was partially supported by the RFBR grant 05-01-01086. 2 Elements of R-technique By R-technique we mean computational methods based on general properties of braidings (in particular, Hecke symmetries) regardless of their concrete form. We are mostly interested in the so-called skew-invertible braidings since they enable us to define numerical characteristics of Hecke symmetries and related objects. AbraidingR(see(1.1))iscalledskew-invertibleifthereexistsanendomorphismΨ :V⊗2 → V⊗2 such that Tr R Ψ = P = Tr Ψ R (2.1) (2) 12 23 13 (2) 12 23 where the symbol Tr means calculating trace in the second factor of the tensor product V⊗3. (2) Hereafter P stands for the usual flip P(x⊗y) = y⊗x. Fixing bases {x } and {x ⊗ x } in V and V⊗2 respectively, we identify R (resp., Ψ) with a i i j matrix kRklk (resp., kΨklk): ij ij R(x ⊗x ) = x ⊗x Rkl (2.2) i j k l ij where the upper indices mark the rows of the matrix and from now on the summation over the repeated indices is understood. Being written in terms of matrices, relation (2.1) reads Ria Ψbl = δiδl = Ψia Rbl . jb ak k j jb ak Using Ψ we define two endomorphisms B and C of the space V j j B(x ) = x B , C(x ) = x C , i j i i j i 5 where j kj j jk B := Ψ , C := Ψ , (2.3) i ki i ik that is B := Tr Ψ, C := Tr Ψ. (1) (2) If the operator B (or C) is invertible, then the corresponding braiding R is called strictly skew- invertible. As was shown in [O], R is strictly skew-invertible iff R−1 is skew-invertible and, besides, the invertibility of B leads to the invertibility of C and vice versa. A well known important example of a strictly skew-invertible braiding is the super-flip R on a super-space V = V ⊕V , where V and V are respectively the even and odd components of V. In 0 1 0 1 this case the operators B and C are called the parity operators and their explicit form is as follows B(z) = C(z) =z −z , ∀z ∈V, 0 1 where z (z ) is the even (odd) component of z = z +z . 0 1 0 1 LetRbeaskew-invertiblebraiding. Listedbelowaresomeusefulpropertiesofthecorresponding endomorphisms Ψ, B and C. 1. TrB = TrC, Tr B R = Tr C R = I , (2.4) (2) 2 21 (2) 2 12 1 where I is the identical automorphism of V. These relations directly follow from definitions (2.1) and (2.3). 2. The endomorphisms B and C commute and their product is a scalar operator BC = CB = νI, (2.5) where the numeric factor ν is nonzero iff the braiding R is strictly skew-invertible (in partic- ular, if R is a skew-invertible Hecke symmetry). 3. The matrix elements of B and C realize a one-dimensional representation of the so-called RTT algebra, associated with R (cf. [FRT]), that is R B B = B B R , R C C = C C R . (2.6) 12 1 2 1 2 12 12 1 2 1 2 12 As a direct consequence of the above relations, we have Tr (B B R X R−1) = Tr (B B R−1X R )= Tr (B B X ), (12) 1 2 12 12 12 (12) 1 2 12 12 12 (12) 1 2 12 Tr (C C R X R−1)= Tr (C C R−1X R ) = Tr (C C X ) (12) 1 2 12 12 12 (12) 1 2 12 12 12 (12) 1 2 12 where X ∈ End(V⊗2) is an arbitrary endomorphism and Tr (...) = Tr (Tr (...)). (12) (1) (2) 4. The following important relations were proved in [OP2, S] B Ψ = R−1B , Ψ B = B R−1, 1 12 21 2 12 1 2 21 (2.7) C Ψ = R−1C , Ψ C = C R−1, 2 12 21 1 12 2 1 21 whereR = PR P. Incase ν 6= 0, only one of thelines above is independentdueto relation 21 12 (2.5). Therefore, for an arbitrary endomorphism X ∈ End(V) we obtain Tr (B R X R−1) = Tr (B R−1X R ) = Tr(BX)I , (1) 1 12 2 12 (1) 1 12 2 12 2 Tr (C R X R−1) = Tr (C R−1X R ) = Tr(CX)I . (2.8) (2) 2 12 1 12 (2) 2 12 1 12 1 This completes the list of technical facts to be used in the text below. 6 3 The general form of a Hecke symmetry In this Section we study the classification problem of (skew-invertible) Hecke symmetries. Our presentation is based on the theory of the A series Hecke algebras and their R-matrix represen- k−1 tations. As a review of the subject we can recommend the work [OP1]. Some necessary facts of the mentioned theory are given in Appendix for the reader’s convenience. Given a Hecke symmetry R : V⊗2 → V⊗2, we consider the R-symmetric Λ (V) and the R- + skew-symmetric Λ (V) algebras of the space V, which by definition are the following quotients − Λ (V) := T(V)/h(Im(q±1I ∓R )i, I = I ⊗I. (3.1) ± 12 12 12 Hereafter T(V) stands for the free tensor algebra of the space V and hJi denotes the two-sided ideal generated in this algebra by a subset J ⊂ T(V). Then, we consider the Hilbert-Poincar´e (HP) series of the algebras Λ (V) ± P (t):= tk dim Λk(V), (3.2) ± ± k≥0 X where Λk(V)⊂ Λ (V) is the homogenous component of degree k. ± ± The following proposition plays a decisive role in the classification of all possible forms of the Hecke symmetries. Proposition 1 Consider an arbitrary Hecke symmetry R, satisfying (1.1) and (1.2) at a generic value of the parameter q. Then the following properties hold true. 1. The HP series P (t) obey the relation ± P (t)P (−t) = 1. + − 2. The HP series P (t) (and hence P (t)) is a rational function of the form: − + N(t) 1+a t+...+a tm m (1+x t) P (t) = = 1 m = i=1 i , (3.3) − D(t) 1−b t+...+(−1)nb tn n (1−y t) 1 n Qj=1 j Q where the coefficients a and b are positive integers, the polynomials N(t) and D(t) are i i mutually prime, and all real numbers x and y are positive. i i 3. If, in addition, the Hecke symmetry is skew-invertible, then the polynomials N(t) and D(−t) are reciprocal6. The first item of the above list was proved in [G2], the second and the third ones — in [H, Da] and [DH]. Definition 2 LetR :V⊗2 → V⊗2 beaskew-invertible Hecke symmetryandlet m(resp., n)bethe degree of the numerator N(t) (resp., the denominator D(t)) of the HP series P (t). The ordered − pair of integers (m|n) will be called the bi-rank of R. If n= 0 (resp., m = 0), the Hecke symmetry will be called even (resp., odd). Otherwise we say that R is of the general type. Remark 3 In the sense of the above definition, any skew-invertible Hecke symmetry is a general- ization of the super-flip for which P (t) = (1+t)m(1−t)−n, where m = dim V , n= dim V . Such − 0 1 a treatment of Hecke symmetries is also motivated by similarity of the corresponding Schur-Weyl categories (see below). 6Recall,thatapolynomialp(t)=c0+c1t+...+cntn withrealcoefficientsci iscalledreciprocal ifp(t)=tnp(t−1) or, equivalently,ci =cn−i, 0≤i≤n. 7 Now we obtain some important consequences of Proposition 1. Let R be a Hecke symmetry of the bi-rank (m|n). As is known, the Hecke symmetry R allows to define a representations ρ of R the A series Hecke algebras H (q), k ≥ 2, in homogeneous components V⊗p ⊂ T(V), ∀p ≥ k k−1 k ρ : H (q) → End(V⊗p), p ≥ k. R k Explicitly, these representations are given in (A.3) of Appendix. Under the presentation ρ , the primitive idempotents eλ ∈ H (q), λ ⊢ k, convert to the R a k projection operators Eλ(R) = ρ (eλ)∈ End(V⊗p), p ≥ k, (3.4) a R a where the index a enumerates the standard Young tableaux (λ,a), which can be constructed for a given partition λ ⊢ k. The total number of the standard Young tableaux corresponding to the partition λ is denoted as d . λ Under the action of these projectors the spaces V⊗p, p ≥ 2, are expanded into the direct sum d λ V⊗p = V , V = Im(Eλ). (3.5) (λ,a) (λ,a) a λ⊢pa=1 MM Duetorelation(A.2),theprojectorsEλ withdifferentaareconnectedbyinvertibletransformations a and, therefore, all spaces V with fixed λ and different a are isomorphic. (λ,a) At a generic value of q, theHecke algebra H (q) is known to beisomorphic to the group algebra k K[S ] [We]. Basing on this fact, we can prove the following result [GLS1, H] k V ⊗V = V ∼= cν V , λ⊢ p, µ ⊢ k,ν ⊢ (p+k), (3.6) (λ,a) (µ,b) (ν,dab) λµ (ν,d0) Mν daMb∈Iab Mν where the integers cν are the Littlwood-Richardson coefficients, the tableau index d takes the λµ ab values form a subset I ⊂ {1,2,...,d }, which depends on the values of the indices a and b. The ab ν number d in the last equality stands for the index of an arbitrary fixed tableau from the set (ν,d), 0 1 ≤ d ≤ d . This equality has the following meaning. Though the summands V do depend ν (ν,dab) on the values of a and b, the total number of these summands (the cardinality of I ) depends only ab on the partitions λ, µ and ν and is equal to the Littlewood-Richardson coefficient cν . Therefore, λµ due to isomorphism V ∼= V , we can replace the sum over d by the space V with the (ν,dab) (ν,d0) ab (ν,d0) corresponding multiplicity cν (cf. [GLS1]). λµ A particular example of the spaces V is the homogeneous components Λk(V) and Λk(V) of (λ,a) + − the algebras Λ (V) (3.1). They are images of the projectors E(k) and E(1k), corresponding to one- ± row and one-column partitions (k) and (1k) respectively. This important fact allows us to calculate the dimensions (over the ground field K) of all spaces V , provided that the Poincar´e series (λ,a) P (t) is known. Since all the spaces V corresponding to the same partition λ are isomorphic, − (λ,a) we denote their K-dimensions by the symbol dimV . λ In the sequel, the following corollary of Proposition 1 will be useful. Corollary 4 Let R be a Hecke symmetry of the bi-rank (m|n), the Poincar´e series of Λ (V) being − given by (3.3). Then for the partitions (k) and (1k), k ∈ N, the dimensions of the spaces V and (k) V is determined by the formulae (1k) k dimV = s (x|y) := h (x)e (y), (3.7) (k) (k) i k−i i=0 X k dimV = s (x|y) := e (x)h (y), (3.8) (1k) (1k) i k−i i=0 X where h and e are respectively the complete symmetric and elementary symmetric functions of i i their arguments. 8 Proof. We prove only the first of the above formulae since the second one can be proved in the same way. Since V = Λk(V), the dimension of V can be found as an appropriate derivative of (k) + (k) the Poincar´e series P (t) + 1 dk dimV = P (t) . (k) k! dtk + |t=0 Using P (t)P (−t)= 1 (see Proposition 1) and relation (3.3) we present P (t) in the form + − + n m 1 P (t) = (1+y t) = E(y|t)H(x|t), + i (1−x t) j i=1 j=1 Y Y where E(·) and H(·) stands for the generating functions of the elementary and complete symmetric functions in the finite set of variables [Mac]: 1 dk e (y) = y ...y = E(y|t) k i1 ik k! dtk |t=0 1≤i1<X...<ik≤n 1 dk h (x) = x ...x = H(x|t) . k j1 jk k! dtk |t=0 1≤j1≤X...≤jk≤m Calculating the k-th derivative of P (t) at t = 0 we get (3.7). + Note, that polynomials s (x|y) and s (x|y) defined in (3.7) and (3.8) belong to the class (k) (1k) of super-symmetric polynomials in {x } and {y }. By definition [St], a polynomial p(u|v) in two i j sets of variables is called super-symmetric if it is symmetric with respect to any permutation of arguments {u } as well as of arguments {v } and, additionally, on setting u = v = t in p(u|v) one i j 1 1 gets the result independent of t. Evidently, the polynomials in question satisfy this definition if we set, for example, u= x, v = −y. Actually, the set of polynomials s (x|y) (respectively s (x|y)), k ∈ N, are super-symmetric (k) (1k) analogs of complete symmetric (respectively elementary symmetric) functions in finite numbers of variables. In particular, they generate the whole ring of super-symmetric polynomials in variables {x } and {y }. The Z-basis of this ring is formed by the Schur super-symmetric functions s (x|y) i j λ which can be expressed in terms of s (or s ) through Jacobi-Trudi relations [Mac]. The Schur (k) (1k) super-symmetric functions determine the value of dimensions dimV . In order to formulate the λ corresponding result we need one more definition. Definition 5 ([BR]) Given two arbitrary integers m ≥ 0 and n ≥ 0, consider a partition λ = (λ ,λ ,...), satisfying the following restriction λ ≤ n. The (infinite) set of all such partitions 1 2 m+1 are denoted as H(m,n) and any partition λ ∈ H(m,n) will be called a hook partition of the type H(m,n). Proposition 6 ([H]) Let R be a Hecke symmetry of the bi-rank (m|n). Then the dimensions dimV of spaces in decomposition (3.5) are determined by the rules: λ 1. For any λ = (λ ,...,λ ) ∈ H(m,n) the dimension dimV 6= 0 and is given by the formula 1 k λ dimV = s (x|y). (3.9) λ λ Here s (x|y) = detks (x|y)k . λ (λi−i+j) 1≤i,j≤k where s (x|y) is defined in (3.7) for k ≥ 0 and s := 0 for k < 0. (k) (k) 2. For arbitrary partition λ we have dimV = 0 ⇔ λ 6∈ H(m,n). λ 9 Proof. Taking into account that dim(U ⊗W)= dimUdimW, dim(U ⊕W) =dimU +dimW and calculating dimensions of the spaces in the both sides of (3.6) we find dimV dimV = cν dimV . λ µ λµ ν ν X Now the result (3.9) is a direct consequence of an inductive procedure based on Corollary 4 (cf., for example, [GPS2]). The second claim can be deduced from the properties of the Schur functionss (x|y) established λ in [BR] (also cf. [H]). To end the Section, we present one more important numerical characteristic of the Hecke sym- metry which can be expressed in terms of its bi-rank. Proposition 7 Let R be a skew-invertible Hecke symmetry with the bi-rank (m|n). Then TrB = TrC = qn−m(m−n) . (3.10) q The proof of the theorem is rather technical and is placed in the Appendix. Corollary 8 For a skew-invertible Hecke symmetry with the bi-rank (m|n), the factor ν in (2.5) equals q2(n−m) that is BC = CB = q2(n−m)I. Proof. First, observe, that if R is a skew-invertible Hecke symmetry, then the same is true for the operator R = PR P, and therefore 21 12 R−1 = R −(q−q−1)I . 21 21 21 Applying Tr to the first formula in (2.7), we have (2) B C = Tr (B Ψ )= Tr (R−1B ) 1 1 (2) 1 12 (2) 21 2 = Tr ((R −(q−q−1)I )B ) = I −(q−q−1)I Tr(B) = q2(n−m)I . (2) 21 21 2 1 1 1 4 The quasitensor category SW(V ) (m|n) Our next goal is to construct the quasitensor Schur-Weyl category SW(V ) of vector spaces, (m|n) generated by the space V equipped with a skew-invertible Hecke symmetry R of the bi-rank (m|n). Theobjectsofthiscategory possessthemodulestructureoverthereflection equation algebra, which will be considered in detail in the next Sections. In constructing the above mentioned category we proceed analogously to the paper [GLS1], where such a category was constructed for an even Hecke symmetry of the bi-rank (m|0). A peculiarity of the even case is that the space V∗, dual to V, can be identified with a specific object V(1m−1) (see (3.5) for the definition of Vλ) of the category. This property ensures the category constructed in [GLS1] to be rigid7. It is not so in the case of general bi-rank (m|n), and we have to properly enlarge the category by adding the dual spaces to all objects. This requires, in turn, a consistent extending of the 7Recall, that a (quasi)tensor category of vector spaces is rigid if to any of its objects U there corresponds a dual object U∗ such that themaps U ⊗U∗ →K and U∗⊗U →K are categorical morphisms. 10