2000 Mathematics Subject Classification. Primary 53C15, 53C30; Secondary 53C10, 53C35.
<ph f="cmbx">Invariant </ph> <math xmlns="http://www.w3.org/1998/Math/MathML"> <mi>f</mi> </math> <ph f="cmbx">-structures in the generalized Hermitian geometry</ph>

Vitaly V. Balashchenko

Contents

1 Introduction

Invariant structures on homogeneous manifolds are traditionally one of the most important objects in differential geometry, specifically, in Hermitian geometry.
Some remarkable classes of almost Hermitian structures such as Kähler, nearly Kähler, Hermitian structures etc. are well known and intensively used in geometry and a number of applications. In particular, a special role is played by a significant class of invariant nearly Kähler structures based on the canonical almost complex structure on homogeneous 3   -symmetric spaces (see [S2] , [WG] , [G2] , [Ki1] ). It should be mentioned that the canonical almost complex structure on such spaces became an effective tool and a remarkable example in some deep constructions of differential geometry and global analysis such as homogeneous structures ([TV] , [Sat] , [Ki4] , [GV] , [LV] , [AG] etc.), Einstein metrics ([SW] , [SY] ), holomorphic and minimal submanifolds ([Sal1] , [Sal2] ), real Killing spinors ([Gru] , [BFGK] , [Ka] ).
The concept of generalized Hermitian geometry created in the 1980s (see, for example, [Ki2] , [Ki7] ) is a natural consequence of the development of Hermitian geometry and the theory of almost contact structures with many applications.
One of its central objects is the metric f   -structures of the classical type ( f 3 + f = 0 )   , which include the class of almost Hermitian structures. Many important classes of metric f   -structures such as Kähler, Killing, nearly Kähler, Hermitian f   -structures and some others were introduced and intensively investigated in various aspects (see [Ki2] , [Ki5] , [Ki7] , [KL] etc.). Specifically, Killing and nearly Kähler f   -structures became natural generalizations of classical nearly Kähler structures in Hermitian geometry. However, this theory had not provided new invariant examples of its own up to the recent period, and so the lack of these examples was becoming all the more noticeable.
There has recently been a qualitative change in the situation, related to the complete solution of the problem of describing canonical structures of classical type on regular Φ   -spaces [BS2] . A rich collection of canonical f   -structures has been discovered (including almost complex structures) leading to the presentation of wide classes of invariant examples in generalized Hermitian geometry (see [B4] -[B7] , [C3] and others). In particular, nearly Kähler f   -structures were provided with a remarkable class of their own invariant examples (see [B6] , [B7] ). This has ensured a continuation of the classical results of J.A.Wolf, A.Gray, V.F.Kirichenko and others. As to Killing f   -structures, it is really an essential problem to find proper non-trivial invariant examples of these structures. Moreover, the possibilities for constructing such examples are fairly limited (see [B4] ).
The main goals of this paper are (i) to give a brief survey on invariant structures in generalized Hermitian geometry and (ii) to characterize all invariant f   -structures on the flag manifold S U ( 3 ) / T m a x   in the sense of generalized Hermitian geometry, in particular, to present first invariant examples of Killing f   -structures.
Sections 2-4 are mostly of survey character. In Section 2, we collect some basic notions and results on homogeneous regular Φ   -spaces and canonical affinor structures. In particular, a precise description of all canonical structures of classical types on homogeneous k   -symmetric spaces is included. Besides, the exact formulae for these structures and the relationship between them on 4and 5-symmetric spaces are presented.
In Section 3, we recall the main classes of almost Hermitian structures following the Gray-Hervella division of almost Hermitian manifolds into sixteen classes (see[GH] ). Besides, we select particular results related to invariant almost Hermitian structures.
Further, in Section 4, we describe main classes of metric f   -structures in generalized Hermitian geometry. Here we also formulate the recent results on invariant nearly Kähler, G 1 f   -, Hermitian, and Killing f   -structures. In this consideration, the canonical f   -structures on homogeneous 4and 5-symmetric spaces are especially important.
Finally, in Section 5, we examine in detail all invariant f   -structures on the complex flag manifold S U ( 3 ) / T m a x   with respect to all invariant Riemannian metrics. We discuss belonging these structures to the main classes of metric f   -structures above mentioned. In particular, invariant non-trivial Killing f   -structures together with the corresponding Riemannian metrics are first presented.

2 Homogeneous regular Φ   -spaces and canonical affinor structures

Here we briefly formulate some basic definitions and results related to regular Φ   -spaces and canonical affinor structures on them. More detailed information can be found in [BS2] , [B10] , [WG] , [Ko] , [F] , [S1] , [S2] .
Let G   be a connected Lie group, Φ   its (analytic) automorphism. Denote by G Φ   the subgroup of all fixed points of Φ   and G o Φ   the identity component of G Φ   .
Suppose a closed subgroup H   of G   satisfies the condition G o Φ ⊂ H ⊂ G Φ .   Then G / H   is called a homogeneous Φ   -space.
Homogeneous Φ   -spaces include homogeneous symmetric spaces ( Φ 2 = i d )   and, more general, homogeneous Φ   -spaces of order k   ( Φ k = i d )   or, in the other terminology, homogeneous k   -symmetric spaces (see [Ko] ).
For any homogeneous Φ   -space G / H   one can define the mapping S o = D : G / H → G / H , x H → Φ ( x ) H .   It is known [S1] that S o   is an analytic diffeomorphism of G / H   . S o   is usually called a ”symmetry” of G / H   at the point o = H   . It is evident that in view of homogeneity the ”symmetry” S p   can be defined at any point p ∈ G / H   . More exactly, for any p = τ ( x ) o = x H , q = τ ( y ) o = y H   we put S p = τ ( x ) ∘ S o ∘ τ ( x − 1 ) .   It is easy to show that S p ( y H ) = x Φ ( x − 1 ) Φ ( y ) H .   Thus any homogeneous Φ   -space is equipped with the set of symmetries { S p | p ∈ G / H }   . Moreover, each S p   is an analytic diffeomorphism of the manifold G / H   (see [S1] ).
Note that there exist homogeneous Φ   -spaces that are not reductive. That is why so-called regular Φ   -spaces first introduced by N.A.Stepanov [S1] are of fundamental importance.
Let G / H   be a homogeneous Φ   -space, g   and h   the corresponding Lie algebras for G   and H   , φ = d Φ e   the automorphism of g   . Consider the linear operator A = φ − i d   and the Fitting decomposition g = g 0 ⊕ g 1   with respect to A   , where g 0   and g 1   denote 0   and 1   -component of the decomposition respectively. Further, let φ = φ s φ u   be the Jordan decomposition, where φ s   and φ u   is a semisimple and unipotent component of φ   respectively, φ s φ u = φ u φ s   . Denote by g γ   a subspace of all fixed points for a linear endomorphism γ   in g   . It is clear that h = g φ = K e r A   , h ⊂ g 0   , h ⊂ g φ s   .
Definition 1 ([S1] , [BS2] , [B10] , [F] ). A homogeneous Φ   -space G / H   is called a regular Φ   -space if one of the following equivalent conditions is satisfied:
  • (1) h = g 0   ;
  • (2) g = h ⊕ A g   ;
  • (3) The restriction of the operator A   to A g   is non-singular;
  • (4) A 2 X = 0 ⟹ A X = 0   for all X ∈ g   .
  • (5) The matrix of the automorphism φ   can be represented in the form ( E 0 0 B ) ,   where the matrix B   does not admit the eigenvalue 1   .
  • (6) h = g φ s   .
We recall two basic facts:
Theorem 1. ([S1] )
  • ∙   Any homogeneous Φ   -space of order k   ( Φ k = i d )   is a regular Φ   -space.
  • ∙   Any regular Φ   -space is reductive. More exactly, the Fitting decomposition
    g = h ⊕ m , m = A g (1)
    is a reductive one.
Decomposition ( 1 ) is the canonical reductive decomposition corresponding to a regular Φ   -space G / H   , and m   is the canonical reductive complement.
We also note that for any regular Φ   -space G / H   each point p = x H ∈ G / H   is an isolated fixed point of the symmetry S p   (see [S1] ).
Decomposition ( 1 ) is obviously φ   -invariant. Denote by θ   the restriction of φ   to m   . As usual, we identify m   with the tangent space T o ( G / H )   at the point o = H   .
It is important to note that the operator θ   commutes with any element of the linear isotropy group A d ( H )   (see [S1] ). It also should be noted (see [S1] ) that ( d S o ) o = θ .   An affinor structure on a manifold is known to be a tensor field of type ( 1 , 1 )   or, equivalently, a field of endomorphisms acting on its tangent bundle. Suppose F   is an invariant affinor structure on a homogeneous manifold G / H   . Then F   is completely determined by its value F o   at the point o   , where F o   is invariant with respect to A d ( H )   . For simplicity, we will denote by the same manner both any invariant structure on G / H   and its value at o   throughout the rest of the paper.
Definition 2 ([BS1] ,[BS2] ). An invariant affinor structure F   on a regular Φ   -space G / H   is called canonical if its value at the point o = H   is a polynomial in θ   .
Denote by A ( θ )   the set of all canonical affinor structures on a regular Φ   -space G / H   . It is easy to see that A ( θ )   is a commutative subalgebra of the algebra A   of all invariant affinor structures on G / H   . Moreover, d i m A ( θ ) = d e g ν ≤ d i m G / H ,   where ν   is a minimal polynomial of the operator θ   . It is evident that the algebra A ( θ )   for any symmetric Φ   -space ( Φ 2 = i d )   consists of scalar structures only, i.e.
it is isomorphic to R   . As to arbitrary regular Φ   -space ( G / H , Φ )   , the algebraic structure of its commutative algebra A ( θ )   has been recently completely described (see [B9] ).
It should be mentioned that all canonical structures are, in addition, invariant with respect to the ”symmetries” { S p }   of G / H   (see [S1] ). Moreover, from ( d S o ) o = θ   it follows that the invariant affinor structure F p = ( d S p ) p , p ∈ G / H   generated by the symmetries { S p }   belongs to the algebra A ( θ )   .
The most remarkable example of canonical structures is the canonical almost complex structure J = 1 3 ( θ − θ 2 )   on a homogeneous 3   -symmetric space (see [S2] , [WG] , [G2] ). It turns out that it is not an exception. In other words, the algebra A ( θ )   contains many affinor structures of classical types. In the sequel we will concentrate on the following affinor structures of classical types:
almost complex structures J   ( J 2 = − 1 )   ; almost product structures P   ( P 2 = 1 )   ; f   -structures ( f 3 + f = 0 )   [Y] ; f   -structures of hyperbolic type or, briefly, h   -structures ( h 3 − h = 0 )   [Ki2] .
Clearly, f   -structures and h   -structures are generalizations of structures J   and P   respectively.
All the canonical structures of classical type on regular Φ   -spaces were completely described [BS1] ,[BS2] ,[B8] . In particular, for homogeneous k   -symmetric spaces, precise computational formulae were indicated. For future reference we select here some results.
Denote by s ~   (respectively, s   ) the number of all irreducible factors (respectively, all irreducible quadratic factors) over R   of a minimal polynomial ν   .
Theorem 2. ([BS1] ,[BS2] ,[B8] ) Let G / H   be a regular Φ   -space.
  • (1) The algebra A ( θ )   contains precisely 2 s ~   structures P   .
  • (2) G / H   admits a canonical structure J   if and only if s = s ~   . In this case A ( θ )   contains 2 s   different structures J   .
  • (3) G / H   admits a canonical f   -structure if and only if s ≠ 0   . In this case A ( θ )   contains 3 s − 1   different f   -structures. Suppose s = s ~   . Then 2 s   f   -structures are almost complex and the remaining 3 s − 2 s − 1   have non-trivial kernels.
  • (4) The algebra A ( θ )   contains 3 s ~   different h   -structures. All these structures form a (commutative) semigroup in A ( θ )   and include a subgroup of order 2 s ~   of canonical structures P   .
Further, let G / H   be a homogeneous k   -symmetric space. Then s ~ = s + 1   if − 1 ∈ s p e c θ   , and s ~ = s   in the opposite case. We indicate explicit formulae enabling us to compute all canonical f   -structures and h   -structures. We shall also use the notation u = { n i f k = 2 n + 1 n − 1 i f k = 2 n .  
Theorem 3. ([BS1] ,[BS2] ,[B8] ) Let G / H   be a homogeneous Φ   -space of order k   .
  • (1) All non-trivial canonical f   -structures on G / H   can be given by the operators f = 2 k ∑ m = 1 u ( ∑ j = 1 u ζ j sin 2 π m j k ) ( θ m − θ k − m ) ,   where ζ j ∈ { − 1 ; 0 ; 1 } , j = 1 , 2 , … , u   , and not all coefficients ζ j   are zero. In particular, suppose that − 1 / ∈ s p e c θ   . Then the polynomials f   define canonical almost complex structures J   iff all ζ j ∈ { − 1 ; 1 }   .
  • (2) All canonical h   -structures on G / H   can be given by the polynomials h = ∑ m = 0 k − 1 a m θ m   , where:
    • (a) if k = 2 n + 1   , then a m = a k − m = 2 k ∑ j = 1 u ξ j cos 2 π m j k ;  
    • (b) if k = 2 n   , then a m = a k − m = 1 k ( 2 ∑ j = 1 u ξ j cos 2 π m j k + ( − 1 ) m ξ n )  
    Here the numbers ξ j   take their values from the set { − 1 ; 0 ; 1 }   and the polynomials h   define canonical structures P   iff all ξ j ∈ { − 1 ; 1 }   .
We now particularize the results above mentioned for homogeneous Φ   -spaces of orders 3   , 4   , and 5   only. Note that there are no fundamental obstructions to considering of higher orders k   .
Corollary 1. ([BS2] ,[B8] ) Let G / H   be a homogeneous Φ   -space of order 3   . There are (up to sign) only the following canonical structures of classical type on G / H   :
J = 1 3 ( θ − θ 2 ) , P = 1 .  
We note that the existence of the structure J   and its properties are well known (see [S2] ,[WG] ,[G2] ,[Ki1] ).
Corollary 2. ([BS2] ,[B8] ) On a homogeneous Φ   -space of order 4   there are (up to sign) the following canonical classical structures:
P = θ 2 , f = 1 2 ( θ − θ 3 ) , h 1 = 1 2 ( 1 − θ 2 ) , h 2 = 1 2 ( 1 + θ 2 ) .   The operators h 1   and h 2   form a pair of complementary projectors:
h 1 + h 2 = 1   , h 1 2 = h 1   , h 2 2 = h 2   . Moreover, the following conditions are equivalent:
  • (1) − 1 / ∈ s p e c θ   ;
  • (2) the structure P   is trivial P = − 1   ;
  • (3) the f   -structure is an almost complex structure;
  • (4) the structure h 1   is trivial ( h 1 = 1 )   ;
  • (5) the structure h 2   is null.
General properties of the canonical structures P   and f   on homogeneous 4   -symmetric spaces were investigated in [BD] .
Corollary 3. ([BS1] ,[BS2] ,[B8] ) There exist (up to sign) only the following canonical structures of classical type on any homogeneous Φ   -space of order 5   :
P = 1 5 ( θ − θ 2 − θ 3 + θ 4 ) ; J 1 = α ( θ − θ 4 ) − β ( θ 2 − θ 3 ) ; J 2 = β ( θ − θ 4 ) + α ( θ 2 − θ 3 ) ; f 1 = γ ( θ − θ 4 ) + δ ( θ 2 − θ 3 ) ; f 2 = δ ( θ − θ 4 ) − γ ( θ 2 − θ 3 ) ; h 1 = 1 2 ( 1 + P ) ; h 2 = 1 2 ( 1 − P ) ;  
where α = 5 + 2 5 5   ; β = 5 − 2 5 5   ; γ = 10 + 2 5 10   ; δ = 10 − 2 5 10   .
Besides, the following relations are satisfied:
J 1 P = J 2 ; f 1 P = J 1 h 1 = J 2 h 1 = f 1 ; h 1 P = h 1 ; h 2 P = − h 2 ; f 2 P = J 2 h 2 = − J 1 h 2 = − f 2 ; f 1 f 2 = h 1 h 2 = 0 ; h 1 + h 2 = P .  
In addition, the following conditions are equivalent:
  • (1) s p e c θ   consists of two elements;
  • (2) the structure P   is trivial;
  • (3) the structures J 1   and J 2   coincide (up to sign);
  • (4) one of the structures f 1   and f 2   is null, while the other is an almost complex structure coinciding with one of the structures J 1   and J 2   ;
  • (5) one of the structures h 1   and h 2   is trivial, while the other is null.
We note that for the first time the canonical structure P   on homogeneous 5   -symmetric spaces was introduced and studied in [BC] . Other canonical structures on these spaces were later studied in [C1] -[C3] .
It should be also mentioned that in the particular case of homogeneous Φ   -spaces of any odd order k = 2 n + 1   the method of constructing invariant almost complex structures was described in [Ko] . It can be easily seen that all these structures are canonical in the above sense.

3 Almost Hermitian structures

We briefly recall some notions of Hermitian geometry including the main classes of almost Hermitian structures.
Let M   be a smooth manifold, X ( M )   the Lie algebra of all smooth vector fields on M   , d   the exterior differentiation operator. An almost Hermitian structure on M   (briefly, A H   -structure) is a pair ( g , J )   , where g = 〈 ⋅ , ⋅ 〉   is a (pseudo)Riemannian metric on M   , J   an almost complex structure such that 〈 J X , J Y 〉 = 〈 X , Y 〉   for any X , Y ∈ X ( M )   . It follows immediately that the tensor field Ω ( X , Y ) = 〈 X , J Y 〉   is skew-symmetric, i.e. ( M , Ω )   is an almost symplectic manifold. Ω   is usually called a fundamental form (the Kähler form) of an A H   -structure ( g , J )   .
Further, denote by ∇   the Levi-Civita connection of the metric g   on M   . We recall below some main classes of A H   -structures together with their defining properties (see, for example, [GH] ):
K Kähler structure: ∇ J = 0   ; H Hermitian structure: ∇ X ( J ) Y − ∇ J X ( J ) J Y = 0   ; G 1   A H   -structure of class G 1   , or ∇ X ( J ) X − ∇ J X ( J ) J X = 0   ; G 1   -structure:
QK quasi-Kähler structure: ∇ X ( J ) Y + ∇ J X ( J ) J Y = 0   ; AK almost Kähler structure: d Ω = 0   ; NK nearly Kähler structure, ∇ X ( J ) X = 0   .
or N K   -structure:
It is well known (see, for example, [GH] ) that K ⊂ H ⊂ G 1 ; K ⊂ N K ⊂ G 1 ; N K = G 1 ∩ Q K ; K = H ∩ Q K .   As usual, we will denote by N   the Nijenhuis tensor of an almost complex structure J   , that is, N ( X , Y ) = 1 4 ( [ J X , J Y ] − J [ J X , Y ] − J [ X , J Y ] − [ X , Y ] )   for any X , Y ∈ X ( M )   . Then the condition N = 0   is equivalent to the integrability of J   (see, for example, [KN] ). Moreover, an almost Hermitian structure ( g , J )   belongs to the class H if and only if N = 0   (see, for example, [GH] ).
As was already mentioned, the role of homogeneous almost Hermitian manifolds is particularly important ”because they are the model spaces to which all other almost Hermitian manifolds can be compared” (see [G3] ). A wealth of examples for the most classes above noted, both of general and specific character, can be found in [WG] , [G2] , [G3] , [Ki1] and others. In particular, after the detailed investigation of the 6-dimensional homogeneous nearly Kähler manifolds V.F.Kirichenko proved [Ki1] that naturally reductive strictly nearly Kähler manifolds S O ( 5 ) / U ( 2 )   and S U ( 3 ) / T m a x   are not isometric even locally to the 6-dimensional sphere S 6   . These examples gave a negative answer to the conjecture of S.Sawaki and Y.Yamanoue (see [SaYa] ) claimed that any 6   -dimensional strictly N K   -manifold was a space of constant curvature. It should be noted that the canonical almost complex structure J = 1 3 ( θ − θ 2 )   on homogeneous 3   -symmetric spaces plays a key role in these and other examples of homogeneous A H   -manifolds.
Let G   be a connected Lie group, H   its closed subgroup, g   an invariant (pseudo-)Riemannian metric on the homogeneous space G / H   . Denote by g   and h   the Lie algebrascorresponding to G   and H   respectively. Suppose that G / H   is a reductive homogeneous space, g = h ⊕ m   the reductive decomposition of the Lie algebra g   . As usual, we identify m   with the tangent space T o ( G / H )   at the point o = H   . Then the invariant metric g   is completely defined by its value at the point o   . For convenience we denote by the same manner both any invariant metric g   on G / H   and its value at o   .
Recall that ( G / H , g )   is naturally reductive with respect to a reductive decomposition g = h ⊕ m   [KN] if g ( [ X , Y ] m , Z ) = g ( X , [ Y , Z ] m )   for all X , Y , Z ∈ m   . Here the subscript m   denotes the projection of g   onto m   with respect to the reductive decomposition.
We select here some known results closely related to the main subject of our future consideration.
Theorem 4. ([AG] ) Any invariant almost Hermitian structure on a naturally reductive space ( G / H , g )   belongs to the class G 1   .
Theorem 5. ([WG] , [G2] ) A homogeneous 3   -symmetric space G / H   with the canonical almost complex structure J   and an invariant compatible metric g   is a quasi-Kähler manifold. Moreover, ( G / H , J , g )   belongs to the class NK if and only if g   is naturally reductive.
Theorem 6. ([Ma] , [G4] , [Ki6] ) A 6   -dimensional strictly nearly Kähler manifold is Einstein.
Note that the latter result was obtained in [Ki6] as a particular case of the general approach based on investigating nearly Kähler manifolds of constant type.

4 Metric f   -structures and homogeneous manifolds

The concept of the generalized Hermitian geometry created in the 1980s (see, for example, [Ki2] , [Ki7] , [Ki8] ) was a natural consequence of the development of Hermitian geometry and the theory of almost contact metric structures with many applications. One of its central objects is the metric f   -structures of classical type, which include the class of almost Hermitian structures. We recall here some basic notions.
An f   -structure on a manifold M   is known to be a field of endomorphisms f   acting on its tangent bundle and satisfying the condition f 3 + f = 0   (see [Y] ). The number r = d i m I m f   is constant at any point of M   (see [St] ) and called a rank of the f   -structure. Besides, the number d i m K e r f = d i m M − r   is usually said to be a deficiency of the f   -structure and denoted by d e f f   .
Recall that an f   -structure on a (pseudo)Riemannian manifold ( M , g = 〈 ⋅ , ⋅ 〉 )   is called a metric f   -structure, if 〈 f X , Y 〉 + 〈 X , f Y 〉 = 0   , X , Y ∈ X ( M )   (see [Ki2] ).
In the case the triple ( M , g , f )   is called a metric f   -manifold. It is clear that the tensor field Ω ( X , Y ) = 〈 X , f Y 〉   is skew-symmetric, i.e. Ω   is a 2   -form on M   . Ω   is called a fundamental form of a metric f   -structure [Ki7] , [Ki2] . It is easy to see that the particular cases d e f f = 0   and d e f f = 1   of metric f   -structures lead to almost Hermitian structures and almost contact metric structures respectively.
Let M   be a metric f   -manifold. Then X ( M ) = ℒ ⊕ ℳ   , where ℒ = I m f   and ℳ = K e r f   are mutually orthogonal distributions, which are usually called the first and the second fundamental distributions of the f   -structure respectively.
Obviously, the endomorphisms l = − f 2   and m = i d + f 2   are mutually complementary projections on the distributions ℒ   and ℳ   respectively. We note that in the case when the restriction of g   to ℒ   is non-degenerate the restriction ( F , g )   of a metric f   -structure to ℒ   is an almost Hermitian structure, i.e. F 2 = − i d , 〈 F X , F Y 〉 = 〈 X , Y 〉 , X , Y ∈ ℒ   .
A fundamental role in the geometry of metric f   -manifolds is played by the composition tensor T   , which was explicitly evaluated in [Ki7] :
T ( X , Y ) = 1 4 f ( ∇ f X ( f ) f Y − ∇ f 2 X ( f ) f 2 Y ) , (2)
where ∇   is the Levi-Civita connection of a (pseudo)Riemannian manifold ( M , g )   , X , Y ∈ X ( M )   .
Using this tensor T   , the algebraic structure of a so-called adjoint Q   -algebra in X ( M )   can be defined by the formula:
X * Y = T ( X , Y ) .   It gives the opportunity to introduce some classes of metric f   -structures in terms of natural properties of the adjoint Q   -algebra (see [Ki2] , [Ki7] ).
We enumerate below the main classes of metric f   -structures together with their defining properties:
Kf Kähler f   –structure: ∇ f = 0   ; Hf Hermitian f   –structure: T ( X , Y ) = 0 ,   i.e. X ( M )   is an abelian Q   -algebra; G 1   f f   -structure of class G 1   , or T ( X , X ) = 0 ,   i.e. X ( M )   is G 1 f   -structure: an anticommutative Q   -algebra; QKf quasi-Kähler f   –structure: ∇ X f + T X f = 0   ; Kill f Killing f   -structure: ∇ X ( f ) X = 0   ;NKf nearly Kähler f   -structure, ∇ f X ( f ) f X = 0   .
or N K f   -structure:
The classes Kf, Hf, G 1   f, QKf (in more general situation) were introduced in [Ki2] (see also [Si] ). Killing f   -manifolds Kill f were defined and studied in [Gr1] , [Gr2] . The class NKf was first determined in [B1] (see also [B6] , [B7] ).
The following relationships between the classes mentioned are evident:
K f = H f ∩ Q K f ; K f ⊂ H f ⊂ G 1 f ; K f ⊂ K i l l f ⊂ N K f ⊂ G 1 f .   It is important to note that in the special case f = J   we obtain the corresponding classes of almost Hermitian structures (see [GH] ). In particular, for f = J   the classes Kill f and NKf coincide with the well-known class NK of nearly Kähler structures.
Remark 1. Killing f   -manifolds are often defined by requiring the fundamental form Ω   to be a Killing form, i.e. d Ω = ∇ Ω   (see [Gr1] , [KL] ). It is not hard to prove that the definition is equivalent to the above condition ∇ X ( f ) X = 0   .
Indeed, in accordance with [YB] , Ω   is a Killing 2   -form if and only if ∇ Ω   is a 3   -form. Further, using [KN] , we have
( ∇ Ω ) ( X , Y ; Z ) = Z Ω ( X , Y ) − Ω ( ∇ Z X , Y ) − Ω ( X , ∇ Z Y ) . (3)
It follows that ∇ Ω   is always skew-symmetric in arguments X   and Y   . Using the fact, it is easy to see that ∇ Ω   is a 3   -form if and only if ∇ Ω   is skew-symmetric in Y   and Z   :
( ∇ Ω ) ( X , Y ; Z ) = − ( ∇ Ω ) ( X , Z ; Y ) . (4)
Taking into account formula ( 3 ) and the definition of Ω   , condition ( 4 ) can be written in the form:
Z 〈 X , f Y 〉 − 〈 ∇ Z X , f Y 〉 − 〈 X , f ∇ Z Y 〉 + Y 〈 X , f Z 〉 − 〈 ∇ Y X , f Y Z 〉 − 〈 X , f ∇ Y Z 〉 = 0 .   Since ∇   is the Levi-Civita connection, in particular, we have:
Z 〈 X , Y 〉 = 〈 ∇ Z X , Y 〉 + 〈 X , ∇ Z Y 〉 .   It follows that the previous formula can be written in the form:
〈 X , ∇ Z f Y − f ∇ Z Y 〉 + 〈 X , ∇ Y f Z − f ∇ Y Z 〉 = 0 .   Using the formula ∇ X ( f ) Y = ∇ X f Y − f ∇ X Y   , we get:
〈 X , ∇ Z ( f ) Y + ∇ Y ( f ) Z 〉 = 0 .   It implies that ∇ Z ( f ) Y + ∇ Y ( f ) Z = 0   for any Y , Z ∈ X ( M )   . This is obviously equivalent to the condition ∇ X ( f ) X = 0 , X ∈ X ( M )   . □   Now we dwell on invariant metric f   -structures on homogeneous spaces.
Any invariant metric f   -structure on a reductive homogeneous space G / H   determines the orthogonal decomposition m = m 1 ⊕ m 2   such that m 1 = I m f   , m 2 = K e r f   . As it was already noted (see Section 3), the main classes of almost Hermitian structures are provided with the remarkable set of invariant examples. It turns out that there is also a wealth of invariant examples for the basic classes of metric f   -structures. These invariant metric f   -structures can be realized on homogeneous k   -symmetric spaces with canonical f   -structures. We select here some results in this direction. More detailed information can be found in [B1] -[B7] , [C3] , [Li] .
Theorem 7. ([B5] ) Any invariant metric f   -structure on a naturally reductive space ( G / H , g )   is a G 1 f   -structure.
As a special case ( K e r f = 0 )   , it follows Theorem  4 .
Theorem 8. ([B5] ) Let ( G / H , g , f )   be a naturally reductive space with an invariant metric f   -structure that satisfies the condition [ m 1 , m 1 ] ⊂ m 2 ⊕ h   . Then G / H   is a Hermitian f   -manifold.
We note that Theorem  8 is also valid for arbitrary invariant (pseudo)Riemannian metric g   compatible with an invariant f   -structure on a reductive homogeneous space G / H   (see [BV] ).
Theorems  7 and  8 can be effectively provided with a large class of examples. In particular, for a semi-simple group G   , the invariant (pseudo)Riemannian metric g   generated by the Killing form on any regular Φ   -space G / H   is naturally reductive with respect to the canonical reductive decomposition g = h ⊕ m   (see [S1] ).
Moreover, all canonical structures f   and J   on homogeneous naturally reductive k   -symmetric spaces are compatible with such a metric, i.e. f   is a metric f   -structure, J   is an almost Hermitian structure (see [B1] , [B10] ). In what follows, we mean by a naturally reductive decomposition the canonical reductive decomposition for a regular Φ   -space G / H   . To sum up, we have
Theorem 9. ([B5] ) Let ( G / H , g )   be a naturally reductive k   -symmetric space. Any canonical metric f   -structure on G / H   is a G 1 f   -structure, and any canonical almost Hermitian structure J   is a G 1   -structure.
Theorem 10. ([B6] ,[B7] ) Let G / H   be a regular Φ   -space, g   a naturally reductive metric on G / H   with respect to the canonical reductive decomposition g = h ⊕ m   , f   a metric canonical f   -structure on G / H   . Suppose the f   -structure satisfies the condition f 2 = ± θ f   . Then ( G / H , g , f )   is a nearly Kähler f   -manifold.
Corollary 4. ([B6] ,[B7] ) Let ( G / H , g )   be a naturally reductive homogeneous Φ   -space of order k = 4 n , n ≥ 1   . If { i , − i } ⊂ s p e c θ   , then there exists a non-trivial canonical N K f   -structure on G / H   .
We stress the particular role of homogeneous 4   and 5   -symmetric spaces.
Theorem 11. ([B3] -[B7] ) The canonical f   -structure f = 1 2 ( θ − θ 3 )   on any naturally reductive 4   -symmetric space ( G / H , g )   is both a Hermitian f   -structure and a nearly Kähler f   -structure. Moreover, the following conditions are equivalent:
1) f   is a Kähler f   -structure; 2) f   is a Killing f   -structure; 3) f   is a quasi-Kähler f   -structure; 4) f   is an integrable f   -structure; 5) [ m 1 , m 1 ] ⊂ h   ; 6) [ m 1 , m 2 ] = 0   ; 7) G / H   is a locally symmetric space:
[ m , m ] ⊂ h   .
Theorem 12. ([B2] -[B5] , [B7] , [C3] ) Let ( G / H , g )   be a naturally reductive 5   -symmetric space, f 1   and f 2   , J 1   and J 2   the canonical structures on this space. Then f 1   and f 2   belong to both classes Hf and NKf.
Moreover, the following conditions are equivalent: 1) f 1   is a Kähler f   -structure; 2) f 2   is a Kähler f   -structure; 3) f 1   is a Killing f   -structure; 4) f 2   is a Killing f   -structure; 5) f 1   is a quasi-Kähler f   -structure; 6) f 2   is a quasi-Kähler f   -structure; 7) f 1   is an integrable f   -structure; 8) f 2   is an integrable f   -structure; 9) J 1   and J 2   are N K   -structures; 10) [ m 1 , m 2 ] = 0   (here m 1 = I m f 1 = K e r f 2 , m 2 = I m f 2 = K e r f 1   ); 11) G / H   is a locally symmetric space: [ m , m ] ⊂ h   .
It should be mentioned that Riemannian homogeneous 4   -symmetric spaces of classical compact Lie groups were classified and geometrically described in [J] . The similar problem for homogeneous 5   -symmetric spaces was considered in [TX] . By Theorem  11 and Theorem  12 , it presents a collection of homogeneous f   -manifolds in the classes NKf and Hf. Note that the canonical f   -structures under consideration are generally non-integrable.
Besides, there are invariant N K f   -structures and H f   -structures on homogeneous spaces ( G / H , g )   , where the metric g   is not naturally reductive. The example of such a kind can be realized on the 6   -dimensional Heisenberg group ( N , g )   . As to details related to this group, we refer to [Ka1] , [Ka2] , [TV] .
Theorem 13. ([B5] -[B7] ) The 6-dimensional generalized Heisenberg group ( N , g )   with respect to the canonical f   -structure f = 1 2 ( θ − θ 3 )   of a homogeneous Φ   -space of order 4   is both H f   and N K f   -manifold.
This f   -structure is neither Killing nor integrable on ( N , g )   .
Remark 2. Theorems  11 and  13 , in particular, illustrate simultaneously the analogy and the difference between the canonical almost complex structure J   on homogeneous 3   -symmetric spaces ( G / H , g , J )   and the canonical f   -structure on homogeneous 4   -symmetric spaces ( G / H , g , f )   (see Theorem  5 ).
Let us also remark that the 6-dimensional generalized Heisenberg group ( N , g )   is an example of solvable type. In Section 5, we present N K f   -structures with non-naturally reductive metrics in the case of semi-simple type.
Finally, we briefly discuss the existence problem for invariant Killing f   -structures.
By Theorems  11 and  12 , the canonical f   -structures on naturally reductive 4   and 5   -symmetric spaces are never strictly (i.e. non-Kähler) Killing f   -structures.
Moreover, we recall the following general result:
Theorem 14. ([B4] ) Let ( G / H , g , f )   be a naturally reductive Killing f   -manifold. Then the following relations hold:
[ m 1 , m 1 ] ⊂ m 1 ⊕ h , [ m 2 , m 2 ] ⊂ m 2 ⊕ h , [ m 1 , m 2 ] ⊂ h .   In particular, both the fundamental distributions of the Killing f   -structure generate invariant totally geodesic foliations on G / H   .
By the results in [Gr1] and Theorem  14 , it follows
Corollary 5. ([B4] ) There are no non-trivial (i.e. d e f f > 0   ) invariant Killing f   -structures of the so-called fundamental type (see [Gr1] ) on naturally reductive homogeneous spaces ( G / H , g )   .
This fact is a wide generalization of the similar result of A.Gritsans obtained for Riemannian globally symmetric spaces. Besides, it shows a substantial difference between invariant Killing f   -structures and invariant N K   -structures.
In Section 5, we will indicate, in particular, first examples of invariant Killing f   -structures.

5 Invariant f   -structures on the complex flag manifold M = S U ( 3 ) / T m a x  

In this Section, we will consider all invariant f   -structures on the flag manifold M = S U ( 3 ) / T m a x   . Note that invariant almost complex structures (i.e. f   -structures of maximal rank 6   ) on this space were investigated in [G3] , [AGI1] , [AGI2] and many other papers.
The homogeneous manifold S U ( 3 ) / T m a x   is known to be an important example in many branches of differential geometry and beyond. In particular, M = S U ( 3 ) / T m a x   is a Riemannian homogeneous 3   -symmetric space not homeomorphic with the underlying manifold M   of any Riemannian symmetric space (see [LP1] ).
Further, M   is a homogeneous k   -symmetric space for any k ≥ 3   . Moreover, M   is a naturally reductive Riemannian homogeneous space that is non-commutative (see [J2] ). It means that the algebra of invariant differential operators D ( S U ( 3 ) / T m a x )   is not commutative (see [H1] ). It follows that M = S U ( 3 ) / T m a x   is not even a weakly symmetric space (see, for example, [Vi] ).
Besides, M   is the twistor space for the projective space C P 2   (see, for example, [Be] , Chapter 13). It was a key point for constructing the first examples of 6   -dimensional Riemannian manifolds admitting a real Killing spinor (see [BFGK] ).
More exactly, the flag manifold M = S U ( 3 ) / T m a x   with the nearly Kähler structure ( g , J )   just possesses a real Killing spinor (see [BFGK] , [Gru] ). Moreover, using the duality procedure for this space S U ( 3 ) / T m a x   , one can effectively construct pseudo-Riemannian homogeneous manifolds with the real Killing spinors (see [Ka] ).
Let Φ = I ( s )   be an inner automorphism of the Lie group S U ( 3 )   defined by the element s = d i a g ( ɛ , ɛ ¯ , 1 )   , where ɛ   is a primitive third root of unity. Then the subgroup H = G Φ   of all fixed points of Φ   is of the form:
G Φ = { d i a g ( e i β 1 , e i β 2 , e i β 3 ) | β 1 + β 2 + β 3 = 0 , β j ∈ R } .   Obviously, G Φ   is isomorphic to T 2 = T m a x   diagonally imbedded into S U ( 3 )   . It means that the flag manifold M = S U ( 3 ) / T m a x   is a homogeneous 3   -symmetric space defined by the automorphism Φ   .
Consider the canonical reductive decomposition g = h ⊕ m   of the Lie algebra g = s u ( 3 )   for the homogeneous Φ   -space M   . Using the notations in [R1] , we obtain:
g = s u ( 3 ) = { ( α 1 a c ¯ − a ¯ α 2 b − c − b ¯ α 3 ) | α 1 , α 2 , α 3 ∈ I m C , a , b , c ∈ C , α 1 + α 2 + α 3 = 0 } =   = E ( α 1 , α 2 , α 3 ) ⊕ D ( a , b , c ) = h ⊕ m .   If we put X = D ( a , b , c ) , Y = D ( a 1 , b 1 , c 1 ) , Z = E ( α 1 , α 2 , α 3 )   , then the Lie brackets can be briefly indicated (see [R2] ):
[ X , Y ] = D ( b c 1 − b 1 c ¯ , c a 1 − c 1 a ¯ , a b 1 − a 1 b ¯ ) −   2 E ( I m ( a a 1 ¯ + c ¯ c 1 ) , I m ( a ¯ a 1 + b b 1 ¯ ) , I m ( c c 1 ¯ + b ¯ b 1 ) ) ,   [ Z , X ] = D ( α 1 a − a α 2 , α 2 b − b α 3 , α 3 c − c α 1 ) .   Further, we put m = m 1 ⊕ m 2 ⊕ m 3 ,   where m 1 = { X ∈ s u ( 3 ) | X = D ( a , 0 , 0 ) , a ∈ C } ,   m 2 = { X ∈ s u ( 3 ) | X = D ( 0 , b , 0 ) , b ∈ C } ,   m 3 = { X ∈ s u ( 3 ) | X = D ( 0 , 0 , c ) , c ∈ C } .   Using the Killing form of the Lie algebra s u ( 3 )   , we define an invariant inner product on m   :
g o ( X , Y ) = 〈 X , Y 〉 o = − 1 2 R e t r X Y .   Then (see [R1] ) g = h ⊕ m 1 ⊕ m 2 ⊕ m 3   is 〈 ⋅ , ⋅ 〉 o   -orthogonal decomposition satisfying the following relations:
[ h , m j ] ⊂ m j , [ m j , m j ] ⊂ h , [ m j , m j + 1 ] ⊂ m j + 2 ,   where j = 1 , 2 , 3   and the index j   should be reduced by modulo 3   . Besides, the H   -modules m j   are pairwise non-isomorphic.
Now we turn to invariant Riemannian metrics on M   . Taking into account the well-known one-to-one correspondence between G   -invariant Riemannian metrics on G / H   and A d ( H )   -invariant inner products on m   (see [KN] ), we will make use of the following fact:
Lemma 1. ([R1] ) Any S U ( 3 )   -invariant Riemannian metric g = 〈 ⋅ , ⋅ 〉   on the flag manifold M = S U ( 3 ) / T m a x   can be written in the form g = 〈 ⋅ , ⋅ 〉 = λ 1 〈 ⋅ , ⋅ 〉 o | m 1 × m 1 + λ 2 〈 ⋅ , ⋅ 〉 o | m 2 × m 2 + λ 3 〈 ⋅ , ⋅ 〉 o | m 3 × m 3 ,   where λ j > 0 , j = 1 , 2 , 3 .  
A triple ( λ 1 , λ 2 , λ 3 )   is called [R1] a characteristic collection of a Riemannian metric g   above mentioned . Considering Riemannian metrics up to homothety, one can assume that ( λ 1 , λ 2 , λ 3 ) = ( 1 , t , s ) , t > 0 , s > 0 .   For convenience we will denote this correspondence in the following way: g = ( λ 1 , λ 2 , λ 3 )   or g = ( 1 , t , s ) .   We also recall the following result:
Theorem 15. ([ZW] ,[AN] ,[R1] ) There are exactly (up to homothety) the following invariant Einstein metrics on the flag manifold S U ( 3 ) / T m a x   :   ( 1 , 1 , 1 ) , ( 1 , 2 , 1 ) , ( 1 , 1 , 2 ) , ( 2 , 1 , 1 ) .  
Let α   be the Nomizu function (see [N] ) of the Levi-Civita connection ∇   for an invariant Riemannian metric g = 〈 ⋅ , ⋅ 〉   on a reductive homogeneous space G / H   .
Then
α ( X , Y ) = 1 2 [ X , Y ] m + U ( X , Y ) , X , Y ∈ m , (5)
where U : m × m → m   is a symmetric bilinear mapping determined by the formula (see[KN] ):
2 〈 U ( X , Y ) , Z 〉 = 〈 X , [ Z , Y ] m 〉 + 〈 [ Z , X ] m , Y 〉 .   For our case in these notations we have
Lemma 2. ([W] ,[R2] ) For the Levi-Civita connection of a Riemannian metric g = ( λ 1 , λ 2 , λ 3 )   on the flag manifold S U ( 3 ) / T m a x   the following conditions are satisfied :   U ( X , Y ) = 0 , i f X , Y ∈ m j , j ∈ { 1 , 2 , 3 } ;   U ( X , Y ) = − ( 2 λ j ) − 1 ( λ j + 1 − λ j + 2 ) [ X , Y ] , i f X ∈ m j + 1 , Y ∈ m j + 2 ,   where j = 1 , 2 , 3   and the numbers j   are reduced by modulo 3   .
Let us now turn to invariant f   -structures on M = S U ( 3 ) / T m a x   . Keeping the above notations, any invariant f   -structure on M   can be expressed by the mapping
f : D ( a , b , c ) → D ( ζ 1 i a , ζ 2 i b , ζ 3 i c ) , (6)
where ζ j ∈ { 1 , 0 , − 1 } , j = 1 , 2 , 3   , i   is the imaginary unit. We will call the collection ( ζ 1 , ζ 2 , ζ 3 )   a characteristic collection of the invariant f   -structure and for convenience denote f = ( ζ 1 , ζ 2 , ζ 3 ) .   Obviously, all invariant f   -structures on M   pairwise commute.
If we agree to consider f   -structures up to sign, then there are the following invariant f   -structures on M = S U ( 3 ) / T m a x   :
1) invariant f   -structures of rank 6   (invariant almost complex structures): J 1 = ( 1 , 1 , 1 ) , J 2 = ( 1 , − 1 , 1 ) , J 3 = ( 1 , 1 , − 1 ) , J 4 = ( 1 , − 1 , − 1 ) .   2) invariant f   -structures of rank 4   :
f 1 = ( 1 , 1 , 0 ) , f 2 = ( 1 , 0 , 1 ) , f 3 = ( 0 , 1 , 1 ) ,
f 4 = ( 1 , − 1 , 0 ) , f 5 = ( 1 , 0 , − 1 ) , f 6 = ( 0 , 1 , − 1 ) .
3) invariant f   -structures of rank 2   :
f 7 = ( 1 , 0 , 0 ) , f 8 = ( 0 , 1 , 0 ) , f 9 = ( 0 , 0 , 1 ) .   Our description of all invariant f   -structures and all invariant Riemannian metrics evidently implies that any invariant f   -structure f = ( ζ 1 , ζ 2 , ζ 3 )   is a metric f   -structure with respect to any invariant Riemannian metric g = ( λ 1 , λ 2 , λ 3 )   . In particular, J j , j = 1 , 2 , 3 , 4   are invariant almost Hermitian structures with respect to all invariant Riemannian metrics g = ( λ 1 , λ 2 , λ 3 )   .
Now we are able to investigate all invariant f   -structures in the sense of generalized Hermitian geometry, i.e. the special classes Kf, NKf, Kill f, Hf, G 1   f.
A key point of our consideration belongs to the expression ∇ X ( f ) Y   . Using formula ( 5 ), we get:
∇ X ( f ) Y = ∇ X f Y − f ∇ X Y = α ( X , f Y ) − f α ( X , Y )
= 1 2 ( [ X , f Y ] m − f [ X , Y ] m ) + U ( X , f Y ) − f U ( X , Y ) .
As a result, we can obtain:
(7) ∇ X ( f ) Y = 1 2 D ( A , B , C ) , where A = i ( ( ζ 1 + ζ 3 ) ( 1 + s − t ) b c 1 + ( ζ 1 + ζ 2 ) ( s − t − 1 ) b 1 c ) ¯ , B = i ( ( ζ 2 + ζ 1 ) ( 1 + 1 − s t ) c a 1 + ( ζ 2 + ζ 3 ) ( 1 − s t − 1 ) c 1 a ) ¯ , C = i ( ( ζ 3 + ζ 2 ) ( t − 1 s + 1 ) a b 1 + ( ζ 3 + ζ 1 ) ( t − 1 s − 1 ) a 1 b ) ¯ .

5.1 Kähler f   -structures

Kähler f   -structures are defined by the condition ∇ X ( f ) Y = 0   (see Section 4). Using formula ( 7 ), this condition is equivalent to the following system of equations:
{ ( ζ 1 + ζ 3 ) ( s − t + 1 ) = 0 ( ζ 1 + ζ 2 ) ( s − t − 1 ) = 0 ( ζ 2 + ζ 3 ) ( s + t − 1 ) = 0 (8)
Solving system ( 8 ) for all invariant f   -structures, we obtain the following result:
Proposition 1. The flag manifold M = S U ( 3 ) / T m a x   admits the following invariant Kähler f   -structures with respect to the corresponding invariant Riemannian metrics only:
J 2 = ( 1 , − 1 , 1 ) , g t = ( 1 , t , t − 1 ) , t > 1 ;
J 3 = ( 1 , 1 , − 1 ) , g t = ( 1 , t , t + 1 ) , t > 0 ;
J 4 = ( 1 , − 1 , − 1 ) , g t = ( 1 , t , 1 − t ) , 0 < t < 1 .
In particular, there are no invariant Kähler f   -structures of rank 2   and 4   on M   .
We note that the result is known for invariant almost complex structures (see [G3] ,[AGI2] ). We can also observe that for each of Kähler f   -structures J 2 , J 3 , J 4   the corresponding 1   -parameter set g t   of invariant Riemannian metrics contains exactly one Einstein metric excluding the naturally reductive metric g = ( 1 , 1 , 1 )   (see Theorem  15 ). Taking into account Theorem  5 , the latter fact implies that the structures J 2 , J 3 , J 4   cannot be realized as the canonical almost complex structures J = 1 3 ( θ − θ 2 )   for some homogeneous Φ   -spaces of order 3   .
In addition, Lie brackets relations for the subspaces m j , j = 1 , 2 , 3   imply that all invariant f   -structures of rank 2   and 4   are non-integrable. It immediately follows that these f   -structures cannot be Kähler f   -structures.

5.2 Killing f   -structures

The defining condition for Killing f   -structures can be written in the form ∇ X ( f ) X = 0   (see Section 4). From ( 7 ), it follows
∇ X ( f ) X = 1 2 D ( A 0 , B 0 , C 0 ) , where A 0 = i b c ( ( ζ 1 + ζ 3 ) ( 1 + s − t ) + ( ζ 1 + ζ 2 ) ( s − t − 1 ) ) ¯ , B 0 = i c a ( ( ζ 2 + ζ 1 ) ( 1 + 1 − s t ) + ( ζ 2 + ζ 3 ) ( 1 − s t − 1 ) ) ¯ , C 0 = i a b ( ( ζ 3 + ζ 2 ) ( t − 1 s + 1 ) + ( ζ 3 + ζ 1 ) ( t − 1 s − 1 ) ) ¯ .  
It easy to show that the condition ∇ X ( f ) X = 0   is equivalent to the following system of equations:
{ ( ζ 1 + ζ 3 ) ( s − t + 1 ) + ( ζ 1 + ζ 2 ) ( s − t − 1 ) = 0 ( ζ 1 + ζ 2 ) ( s − t − 1 ) + ( ζ 2 + ζ 3 ) ( s + t − 1 ) = 0   Analyzing this system for all invariant f   -structures, we obtain the following result:
Proposition 2. All invariant strictly Killing (i.e. non-Kähler) f   -structures on the flag manifold M = S U ( 3 ) / T m a x   and the corresponding invariant Riemannian metrics (up to homothety) are indicated below:
J 1 = ( 1 , 1 , 1 ) , g = ( 1 , 1 , 1 ) ;
f 1 = ( 1 , 1 , 0 ) , g = ( 3 , 3 , 4 ) ;
f 2 = ( 1 , 0 , 1 ) , g = ( 3 , 4 , 3 ) ;
f 3 = ( 0 , 1 , 1 ) , g = ( 4 , 3 , 3 ) .
In particular, there are no invariant Killing f   -structures of rank 2   on M   .
Note the structure J 1   is a well-known non-integrable nearly Kähler structure on a naturally reductive space M   (see [G2] , [G3] , [Ki1] , [AGI2] and others). The structures f 1 , f 2 , f 3   present first invariant non-trivial Killing f   -structures [B11] .
The important feature of these structures is that the corresponding invariant Riemannian metrics are not Einstein (see Theorem  15 ). It illustrates a substantial difference between non-trivial strictly Killing f   -structures and strictly N K   -structures at least in the 6   -dimensional case (see Theorem  6 ).
Remark 3. It is interesting to note that all strictly Killing f   -structures above indicated are canonical f   -structures for suitable homogeneous Φ   -spaces of the Lie group S U ( 3 )   . We already mentioned that M = S U ( 3 ) / T m a x   is a homogeneous k   -symmetric space for any k ≥ 3   . It means M   as an underlying manifold could be generated by various automorphisms Φ   of the Lie group S U ( 3 )   . In particular, J 1   is the canonical almost complex structure J = 1 3 ( θ − θ 2 )   for the homogeneous Φ   -space of order 3   , where Φ = I ( s ) , s = d i a g ( ɛ , ɛ ¯ , 1 ) , ɛ = 1 3   (see the beginning of this Section). Further, if we consider the automorphism Φ 1 = I ( s 1 ) , s 1 = d i a g ( i , − i , 1 )   , where i = 1 4   is the imaginary unit, then M   is a homogeneous Φ 1   -space of order 4   . The corresponding canonical f   -structure f = 1 2 ( θ 1 − θ 1 3 )   for this Φ 1   -space just coincides (up to sign) with the f   -structure f 3 = ( 0 , 1 , 1 )   .
The structures f 1   and f 2   can be obtained in the similar way. Moreover, all the structures f 1 , f 2 , f 3   and f 7 , f 8 , f 9   can be realized as canonical f   -structures for suitable homogeneous Φ   -spaces of order 5   .
We also note that all f   -structures f 1 , f 2 , f 3   are just the restrictions of the structure J 1   onto the corresponding distributions m p ⊕ m q , p , q ∈ { 1 , 2 , 3 } .  

5.3 Nearly Kähler f   -structures

Using ( 7 ), we can easily obtain:
∇ f X ( f ) f X = 1 2 D ( A ^ , B ^ , C ^ ) , where A ^ = − i ζ 2 ζ 3 b c ( ( ζ 1 + ζ 3 ) ( 1 + s − t ) + ( ζ 1 + ζ 2 ) ( s − t − 1 ) ) ¯ , B ^ = − i ζ 1 ζ 3 c a ( ( ζ 2 + ζ 1 ) ( 1 + 1 − s t ) + ( ζ 2 + ζ 3 ) ( 1 − s t − 1 ) ) ¯ , C ^ = − i ζ 1 ζ 2 a b ( ( ζ 3 + ζ 2 ) ( t − 1 s + 1 ) + ( ζ 3 + ζ 1 ) ( t − 1 s − 1 ) ) ¯ .  
It follows that the condition ∇ f X ( f ) f X = 0   is reduced to the following system of equations:
{ ζ 2 ζ 3 ( ( ζ 1 + ζ 3 ) ( s − t + 1 ) + ( ζ 1 + ζ 2 ) ( s − t − 1 ) ) = 0 ζ 1 ζ 3 ( ( ζ 2 + ζ 1 ) ( 1 + t − s ) + ( ζ 2 + ζ 3 ) ( 1 − s − t ) ) = 0 ζ 1 ζ 2 ( ( ζ 3 + ζ 2 ) ( t + s − 1 ) + ( ζ 3 + ζ 1 ) ( t − s − 1 ) ) = 0   Consideration of this system implies
Proposition 3. The only invariant strictly nearly Kähler (i.e.
non-Kähler) f   -structure of rank 6   on the flag manifold M = S U ( 3 ) / T m a x   is the nearly Kähler structure J 1 = ( 1 , 1 , 1 )   with respect to the naturally reductive metric g = ( 1 , 1 , 1 )   .
Invariant strictly nearly Kähler f   -structures of rank 4   and the corresponding invariant Riemannian metrics (up to homothety) on M   are:
f 1 = ( 1 , 1 , 0 ) , g s = ( 1 , 1 , s ) , s > 0 ;
f 2 = ( 1 , 0 , 1 ) , g t = ( 1 , t , 1 ) , t > 0 ;
f 3 = ( 0 , 1 , 1 ) , g t = ( 1 , t , t ) , t > 0 .
The invariant f   -structures f 7 , f 8 , f 9   of rank 2   on M   are strictly N K f   -structures with respect to all invariant Riemannian metrics g = ( 1 , t , s ) , t , s > 0   .
First we notice that the structures f 1 , f 2 , f 3   and f 7 , f 8 , f 9   provide invariant examples of N K f   -structures with non-naturally reductive metrics on the homogeneous space M = S U ( 3 ) / T m a x   , which belongs to a semi-simple type.
We can also observe that for any invariant strictly N K f   -structure on M   there exists at least one (up to homothety) corresponding Einstein metric. More exactly, for these N K f   -structures of rank 6   , 4   , and 2   there are (up to homothety) 1   , 2   , and 4   Einstein metrics respectively (see Theorem  15 ). In a certain degree, it is a particular analogy with the result of Theorem  6 . This particular fact and some related general results lead to the following conjecture, which seems to be plausible:
Conjecture. For any strictly nearly Kähler f   -structure on a 6   -dimensional manifold there exists at least one corresponding Einstein metric.
Remark 4. The invariant f   -structures f 4 , f 5 , f 6   on the flag manifold M = S U ( 3 ) / T m a x   cannot be canonical f   -structures for all homogeneous Φ   -spaces of orders 4   and 5   of the Lie group S U ( 3 )   . It evidently follows by comparing the results in Theorem  11 , Theorem  12 , and Proposition  3 .

5.4 Hermitian f   -structures

First let us calculate the composition tensor T   (see formula ( 2 )) for arbitrary invariant f   -structure on ( M = S U ( 3 ) / T m a x , g = ( 1 , t , s ) )   . Combining ( 7 ) and ( 6 ), we can obtain:
(9) T ( X , Y ) = 1 8 D ( A ˇ , B ˇ , C ˇ ) , where A ˇ = − ζ 1 ζ 2 ζ 3 ( 1 + ζ 2 ζ 3 ) ( ( ζ 1 + ζ 3 ) ( 1 + s − t ) b c 1 ¯ + ( ζ 1 + ζ 2 ) ( s − t − 1 ) b 1 c ¯ ) , B ˇ = − ζ 1 ζ 2 ζ 3 ( 1 + ζ 1 ζ 3 ) ( ( ζ 2 + ζ 1 ) ( 1 + 1 − s t ) c a 1 ¯ + ( ζ 2 + ζ 3 ) ( 1 − s t − 1 ) c 1 a ¯ ) , C ˇ = − ζ 1 ζ 2 ζ 3 ( 1 + ζ 1 ζ 2 ) ( ( ζ 3 + ζ 2 ) ( t − 1 s + 1 ) a b 1 ¯ + ( ζ 3 + ζ 1 ) ( t − 1 s − 1 ) a 1 b ¯ ) .
We recall that the defining property for a Hermitian f   -structure is the condition T ( X , Y ) = 0   . Now from ( 9 ), we get the following result:
Proposition 4. The invariant f   -structures J 2 , J 3 , J 4   and f 1 , . . . , f 9   are Hermitian f   -structures with respect to all invariant Riemannian metrics g = ( 1 , t , s )   , t , s > 0   on the flag manifold M = S U ( 3 ) / T m a x   .
Notice that the almost complex structure J 1 = ( 1 , 1 , 1 )   is non-integrable. It agrees with the fact that J 1   is not a Hermitian f   -structure for each Riemannian metric. While we stress that all f   -structures f 1 , . . . , f 9   of rank 4   and 2   are non-integrable, but they are Hermitian f   -structures.

5.5 G 1   f-structures

Finally, we consider the condition T ( X , X ) = 0   , which is the defining property for G 1 f   -structures. Using ( 9 ) and taking into account Propositions  3 and  4 , we get
Proposition 5. The flag manifold M = S U ( 3 ) / T m a x   does not admit invariant strictly G 1 f   -structures (i.e. neither N K f   -structures nor H f   -structures). In particular, there are no invariant strictly G 1   -structures J   (i.e. neither nearly Kähler nor Hermitian) on M   .
References

  1. Abbena E., Garbiero S., Almost Hermitian homogeneous manifolds and Lie groups, Nihonkai Math. J., 1993, V.4, 1-15.
  2. Apostolov V., Grantcharov G., Ivanov S., Hermitian structures on twistor spaces, Ann. Glob. Anal. Geom., 1998, V.16, 291-308.
  3. Apostolov V., Grantcharov G., Ivanov S., Orthogonal complex structures on certain Riemannian 6   -manifolds, Diff. Geom. Appl., 1999, V.11, 279-296.
  4. D'Atri, Nickerson, Geodesic symmetries in spaces with special curvature tensors, J. Diff. Geom., 1974, V.9, 251-262.
  5. Balashchenko V.V., Riemannian geometry of canonical structures on regular Φ   -spaces, Preprint No.174/1994, Fakultät für Mathematik der Ruhr-Universität Bochum, 1994, 1-19.
  6. Balashchenko V.V., Invariant nearly Kähler f   -structures on homogeneous spaces, SFB 288 Preprint no.499, Berlin, 2001, 1-14.
  7. Balashchenko V.V.,Extending an idea of A.Gray: homogeneous k   -symmetric spaces and generalized Hermitian geometry, International Congress on Differential geometry in memory of Alfred Gray, September 18-23, 2000, Bilbao (Spain), Abstracts, 5-7.
  8. Balashchenko V.V., Naturally reductive Killing f   -manifolds, Uspekhi Mat. Nauk, 1999, V.54, no.3, 151-152, (in Russian); English translation: Russian Math. Surveys, 1999, V.54, no.3, 623-625.
  9. Balashchenko V.V., Homogeneous Hermitian f   -manifolds, Uspekhi Mat. Nauk, 2001, V.56, no.3, 159-160, (in Russian); English translation: Russian Math. Surveys, 2001, V.56, no.3, 575-577.
  10. Balashchenko V.V., Invariant nearly Kähler f   -structures on homogeneous spaces, Contemporary Mathematics, 2001, V.288, 263-267.
  11. Balashchenko V.V., Homogeneous nearly Kähler f   -manifolds, Doklady Akademii Nauk, Ross. Akad. Nauk, 2001, V.376, no.4, 439-441, (in Russian); English translation: Doklady Mathematics, 2001, V.63, no.1, 56-58.
  12. Balashchenko V.V., Canonical f   -structures of hyperbolic type on regular Φ   -spaces, Uspekhi Mat. Nauk, 1998, V.53, no.4, 213-214, (in Russian); English translation: Russian Math. Surveys, V.53, no.4, 861-863.
  13. Balashchenko V.V., The algebra of canonical affinor structures and classes of regular Φ   -spaces, Doklady Akademii Nauk, Ross. Akad. Nauk, 2002, V.385, no.6, 727-730, (in Russian); English translation: Doklady Mathematics, 2002, V.66, no.1, 111-114.
  14. Balashchenko V.V., Invariant structures generated by Lie group automorphisms on homogeneous spaces, Proceedings of the Workshop ”Contemporary Geometry and Related Topics” (Belgrade, Yugoslavia, 15-21 May, 2002). Editors: N.Bokan, M.Djoric, A.T.Fomenko, Z.Rakic, J.Wess. World Scientific, 2004, 1-32.
  15. Balashchenko V.V., Invariant structures generated by Lie group automorphisms, Workshop ”Contemporary Geometry and Related Topics”, Belgrade, May 15-21, 2002, Abstracts, 3-4.
  16. Besse A.L., Einstein Manifolds, Springer-Verlag, Berlin, Heidelberg, 1987.
  17. Balashchenko V.V., Churbanov Yu.D., Invariant structures on homogeneous Φ   -spaces of order 5   , Uspekhi Mat. Nauk, 1990, V.45, no.1, 169-170, (in Russian); English translation: Russian Math. Surveys, 1990, V.45, no.1, 195-197.
  18. Balashchenko V.V., Dashevich O.V., Geometry of canonical structures on homogeneous Φ   -spaces of order 4   , Uspekhi Mat. Nauk, 1994, V.49, no.4, 153-154, (in Russian); English translation: Russian Math. Surveys, 1994, V.49, no.4, 149-150.
  19. Baum H., Friedrich T., Grunewald R., Kath I., Twistor and Killing Spinors on Riemannian Manifolds, Teubner-Texte zur Mathematik, V.124, Teubner-Verlag, Stuttgart/Leipzig, 1991.
  20. Balashchenko V.V., Stepanov N.A., Canonical affinor structures on regular Φ   -spaces, Uspekhi Mat. Nauk, 1991, V.46, no.1, 205-206, (in Russian); English translation: Russian Math. Surveys, 1991, V.46, no.1, 247-248.
  21. Balashchenko V.V., Stepanov N.A., Canonical affinor structures of classical type on regular Φ   -spaces, Matematicheskii Sbornik, 1995, V.186, no.11, 3-34, (in Russian); English translation: Sbornik: Mathematics, 1995, V.186, no.11, 1551-1580.
  22. Balashchenko V.V., Vylegzhanin D.V., Generalized Hermitian geometry on homogeneous Φ   -spaces of finite order, Izv. Vyssh. Uchebn. Zaved. Mat., 2004, no.10, 31-42, (in Russian); English translation: Russian Math. (Iz. VUZ), 2004, V.48, no.10.
  23. Churbanov Yu.D., On some classes of homogeneous Φ   -spaces of order 5   , Izv. Vyssh. Uchebn. Zaved. Mat., 1992, no.2, 88-90, (in Russian); English translation: Russian Math. (Iz. VUZ), 1992, V.36, no.2, 88-90.
  24. Churbanov Yu.D., Canonical f   -structures of homogeneous Φ   -spaces of order 5   , Vestnik BGU. Ser.1: Fiz, Mat, Mech., 1994, no.1, 51-54, (in Russian).
  25. Churbanov Yu.D., The geometry of homogeneous Φ   -spaces of order 5   , Izv. Vyssh. Uchebn. Zaved. Mat., 2002, no.5, 70-81, (in Russian); English translation: Russian Math. (Iz. VUZ), 2002, V.46, no.5, 68-78.
  26. Fedenko A.S., Spaces with symmetries, Belarusian State University, Minsk, 1977, (in Russian).
  27. Gray A., Nearly Kähler manifolds, J. Diff. Geom., 1970, V.4, no.3, 283-309.
  28. Gray A., Riemannian manifolds with geodesic symmetries of order 3   , J. Diff. Geom., 1972, V.7, no.3-4, 343-369.
  29. Gray A., Homogeneous almost Hermitian manifolds, Proceedings of the Conference on Differential Geometry on Homogeneous Spaces, Turin, Italy, 1983; Rendiconti del Seminario Matematico Universita e Politecnico di Torino, 1983, 17-58. (Special Issue.)
  30. Gray A., The structure of nearly Kähler manifolds, Math. Ann., 1976, V.223, no.3, 233-248.
  31. Gritsans A.S., On the geometry of Killing f   -manifolds, Uspekhi Mat. Nauk, 1990, V.45, no.4, 149-150, (in Russian); English translation: Russian Math. Surveys, 1990, V.45, no.4, 168-169.
  32. Gritsans A.S., On the structure of Killing f   -manifolds, Izv. Vyssh. Uchebn. Zaved. Mat., 1992, no.6, 49-57, (in Russian); English translation: Russian Math. (Iz. VUZ), 1992, V.36, no.6, 46-54.
  33. Grunewald R., Six-dimensional Riemannian manifolds with a real Killing spinor, Ann. Global Anal. Geom., 1990, V.8, no.1, 43-59.
  34. Gray A., Hervella L.M., The sixteen classes of almost Hermitian manifolds and their linear invariants, Ann. Mat. Pura ed Appl., 1980, V.123, no.4, 35-58.
  35. Garbiero S., Vanhecke L., A characterization of locally 3   -symmetric spaces, Riv. Mat. Univ. Parma (5), 1993, V.2, 331-335.
  36. Helgason S., Differential Geometry, Lie Groups, and Symmetric Spaces, New York: Academic Press, 1978.
  37. Helgason S., Groups and Geometric Analysis, New York: Academic Press, 1984.
  38. Jimenez J.A., Riemannian 4   -symmetric spaces, Trans. Amer. Math. Soc., 1988, V.306, no.2, 715-734.
  39. Jimenez J.A., Existence of Hermitian n   -symmetric spaces and of non-commutative naturally reductive spaces, Math. Z., 1987, V.196, no.2, 133-139.
  40. Kath I., Pseudo-Riemannian T   -duals of compact Riemannian homogeneous spaces, Transformation Groups, 2000, V.5, no.2, 157-179.
  41. Kaplan A., Riemannian nilmanifolds attached to Clifford modules, Geom. Dedicata, 1981, V.11, 127-136.
  42. Kaplan A., On the geometry of groups of Heisenberg type, Bull. London Math. Soc., 1983, V.15, 35-42.
  43. Kirichenko V.F., On the geometry of homogeneous K   -spaces, Mat. Zametki, 1981, V.30, no.4, 569-582, (in Russian); English translation: Math. Notes, 1981, V.30., 779-785.
  44. Kirichenko V.F., Methods of generalized Hermitian geometry in the theory of almost contact manifolds, Itogi Nauki i Tekhniki: Probl. Geom. V.18, VINITI, Moscow, 1986, 25-71, (in Russian); English translation: J. Soviet Math. 1988, V.42, no.5, 1885-1919.
  45. Kirichenko V.F., Sur la geometrie des varietes approximativement cosymplectiques, C. R. Acad. Sci. Paris, Ser.1, 1982, V.295, no.12, 673-676.
  46. Kirichenko V.F., Hermitian-homogeneous generalized almost Hermitian manifolds, Dokl. Akad. Nauk SSSR, 1984, V.277, no.6, 1310-1315, (in Russian); English translation: Soviet Math. Dokl., 1984, V.30, 267-271.
  47. Kirichenko V.F., Generalized quasi-Kaehlerian manifolds and axioms of C R   -submanifolds in generalized Hermitian geometry, I, Geom. Dedicata, 1994, V.51, 75-104.
  48. Kirichenko V.F., K   -spaces of constant type, Sibirsk. Mat. Zh., 1976, V.17, no.2, 282-289, (in Russian); English translation: Siberian Math. J., 1976, V.17, 220-225.
  49. Kirichenko V.F., Quasi-homogeneous manifolds and generalized almost Hermitian structures, Izv. Akad. Nauk. SSSR. Ser. Mat., 1983, V.47, no.6, 1208-1223, (in Russian); English translation: Math. USSR, Izv., 1984, V.23, 473-486.
  50. Kirichenko V.F., Differential-geometric structures on manifolds, MPGU, Moscow, 2003, (in Russian).
  51. Kirichenko V.F., Lipagina L.V., Killing f   -manifolds of constant type, Izv. Ross. Akad. Nauk, Ser. Mat., 1999, V.63, no.5, 127-146, (in Russian); English translation: Izv. Math., 1999, V.63, no.5, 963-981.
  52. Kobayashi S., Nomizu K., Foundations of differential geometry, V.2, Intersc. Publ. J.Wiley & Sons, New York London, 1969.
  53. Kowalski O., Generalized symmetric spaces, LN in Math., V.805, Berlin, Heidelberg, New York: Springer-Verlag, 1980.
  54. Ledger A.J., Pettitt R.B., Classification of metrisable regular s   -manifolds with integrable symmetry tensor field, J. Math. Soc. Japan, 1976, V.28, no.4, 668-675.
  55. Ledger A.J., Vanhecke L., On a theorem of Kiric̆enko relating to 3   -symmetric spaces, Riv. Mat. Univ. Parma (4), 1987, V.13, 367-372.
  56. Lipagina L.V., On the structure of the algebra of invariant affinor structures on the sphere S 5   , Izv. Vyssh. Uchebn. Zaved. Mat., 1997, no.9, 17-20, (in Russian); English translation: Russian Math. (Iz. VUZ), 1997, V.41, no.9, 15-18.
  57. Matsumoto M., On 6   -dimensional almost Tachibana spaces, Tensor, 1972, V.23, no.2, 250-252.
  58. Nomizu K., Invariant affine connections on homogeneous spaces, Amer. J. Math., 1954, V.76, no.1, 33-65.
  59. Rodionov E.D., Einstein metrics on even-dimensional homogeneous spaces admitting a homogeneous Riemannian metric of positive sectional curvature, Sibirsk. Mat. Zh., 1991, V.32, no.3, 126-131, (in Russian); English translation: Siberian Math. J., 1991, V.32, no.3, 455-459.
  60. Rodionov E.D., Homogeneous Riemannian manifolds with Einstein metrics, Dr. Sci. Dissertation, Novosibirsk, 1994, (in Russian).
  61. Salamon S., Harmonic and holomorphic maps, Geometry Seminar ”Luigi Bianchi” II – 1984, LN in Math., V.1164, Springer–Verlag, 1985, 161-224.
  62. Salamon S., Minimal surfaces and symmetric spaces, Diff. Geometry. Proc. Colloq. Santiago de Compostela, 1985, 103-114.
  63. Sato T., Riemannian 3   -symmetric spaces and homogeneous K   -spaces, Memoirs of the Faculty of Technology, Kanazawa Univ. 1979, V.12, no.2, 137-143.
  64. Singh K.D., Singh Rakeshwar, Some f ( 3 , ɛ )   -structure manifolds, Demonstr. Math., 1977, V.10, no.3-4, 637-645.
  65. Stepanov N.A., Basic facts of the theory of φ   -spaces, Izv. Vyssh. Uchebn. Zaved. Mat., 1967, no.3, 88-95, (in Russian); English translation: Soviet Math. (Iz. VUZ), 1967, V.11, no.3.
  66. Stepanov N.A., Homogeneous 3   -cyclic spaces, Izv. Vyssh. Uchebn. Zaved. Mat., 1967, no.12, 65-74, (in Russian); English translation: Soviet Math. (Iz. VUZ), 1967, V.11, no.12.
  67. Sekigawa K., Watanabe J., On some compact Riemannian 3   -symmetric spaces, Sci. Reports of Niigata Univ. Ser. A., no.19, 1983, 1-17.
  68. Sawaki S., Yamanoue Y., On a 6   -dimensional K   -space, Sci. Reports of Niigata Univ. Ser. A., no.13, 1976, 13-17.
  69. Sekigawa K., Yoshida H., Riemannian 3   -symmetric spaces defined by some outer automorphisms of compact Lie groups, Tensor, 1983, V.40, no.3, 261-268.
  70. Stong R.E. The rank of an f   -structure, Kodai Math. Sem. Rep., 1977, V.29, 207-209.
  71. Tricerri F., Vanhecke L., Homogeneous structures on Riemannian manifolds, London Math. Soc., LN Ser., no.83, 1983.
  72. Tsagas Gr., Xenos Ph., Homogeneous spaces which are defined by diffeomorphisms of order 5   , Bull. Math. Soc. Sci. Math. RSR, 1987, V.31, no.1, 57-77.
  73. Vinberg E.B. Commutative homogeneous spaces and co-isotropic symplectic actions, Uspekhi Mat. Nauk, 2001, V.56, no.1, 3-62, (in Russian); English translation: Russian Math. Surveys, 2001, V.56, no.1, 1-60.
  74. Wallach N., Compact homogeneous Riemannian manifolds with strictly positive curvature, Ann. Math., 1972, V.96, 277-295.
  75. Wolf J.A., Gray A., Homogeneous spaces defined by Lie group automorphisms, J. Diff. Geom., 1968, V.2, no.1-2, 77-159.
  76. Yano K., On a structure defined by a tensor field f   of type ( 1 , 1 )   satisfying f 3 + f = 0   , Tensor, 1963, V.14, 99-109.
  77. Yano K., Bochner S., Curvature and Betti numbers. Princeton, New Jersey: Princeton University Press, 1953.
  78. Yano K., Kon M., Structures on manifolds. Singapore: World Scientific, 1984.
  79. Ziller W., Wang M., On normal homogeneous Einstein manifolds, Ann. sci. Ecole norm. super., 1985, V.18, 563-633.

Vitaly V. Balashchenko Faculty of Mathematics and Mechanics Belarusian State University F.Scorina av. 4 Minsk 220050, BELARUS E-mail: balashchenko@bsu.by vitbal@tut.by