BI-PARACONTACT STRUCTURES AND LEGENDRE ...paracontact struct. and φ3,ξ,η) is normal as almost...

111
XIX FALL WORKSHOP “GEOMETRY AND PHYSICS” Beniamino Cappelletti Montano CMUC, University of Coimbra (visitor) BI-PARACONTACT STRUCTURES AND LEGENDRE FOLIATIONS OPORTO - SEPTEMBER 6, 2010

Transcript of BI-PARACONTACT STRUCTURES AND LEGENDRE ...paracontact struct. and φ3,ξ,η) is normal as almost...

  • XIX FALL WORKSHOP “GEOMETRY AND PHYSICS”

    Beniamino Cappelletti Montano

    CMUC, University of Coimbra (visitor)

    BI-PARACONTACT STRUCTURES AND LEGENDRE FOLIATIONS

    OPORTO - SEPTEMBER 6, 2010

  • A contact manifold is a smooth manifold M endowed with a 1-form η such that η ∧ (dη)n ≠ 0 (1) everywhere on M.

    Contact manifolds

  • A contact manifold is a smooth manifold M endowed with a 1-form η such that η ∧ (dη)n ≠ 0 (1) everywhere on M.

    M is necessarily of odd dimension, 2n+1

    Contact manifolds

  • A contact manifold is a smooth manifold M endowed with a 1-form η such that η ∧ (dη)n ≠ 0 (1) everywhere on M.

    M is necessarily of odd dimension, 2n+1

    There exists a unique vector field ξ, called Reeb vector field, such that

    η(ξ) = 1 and dη(ξ,·) = 0.

    Contact manifolds

  • D:=ker(η) is called the contact distribution.

    Foliation theory ⇒ D is integrable if and only if η ∧ dη = 0. Thus the geometric interpretation of

    η ∧ (dη)n ≠ 0

    is that the contact distribution is as far as possible from being integrable.

    Contact manifolds

  • D:=ker(η) is called the contact distribution.

    Foliation theory ⇒ D is integrable if and only if η ∧ dη = 0. Thus the geometric interpretation of

    η ∧ (dη)n ≠ 0

    is that the contact distribution is as far as possible from being integrable. The maximal dimension of an integrable subbundle of D is n.

    Contact manifolds

  • Definition Let (M2n+1,η) be a contact manifold. A foliation F of M2n+1 is called Legendre foliation if

    dim(F) = n

    T(F) is a subbundle of the contact distribution D

    Legendre foliations

  • More in general, a Legendre distribution is an n-dimensional sub-bundle L of the contact distribution such that

    dη(X,X’) = 0

    for all X,X’∈Г(L). Note that this condition holds if L is involutive, since

    2dη(X,X’) = X(η(X’)) – X’(η(X)) – η([X,X’]).

    Legendre foliations

  • More in general, a Legendre distribution is an n-dimensional sub-bundle L of the contact distribution such that

    dη(X,X’) = 0

    for all X,X’∈Г(L). Note that this condition holds if L is involutive, since

    2dη(X,X’) = X(η(X’)) – X’(η(X)) – η([X,X’]).

    M. Y. Pang, The structure of Legendre foliations, Trans. Amer. Math. Soc. 1991

    P. Libermann, Legendre foliations on contact manifolds, Differential Geom. Appl. 1990

    A. Bejancu, H. Farran, Foliations and geometric structures, Springer, 2006

    Legendre foliations

  • Definition Let (M,η) be a contact manifold.

    Almost contact and paracontact structures

  • Definition Let (M,η) be a contact manifold.

    An almost contact structure on (M,η) is given by a tensor field φ such that

    φ2 = −I + η⊗ξ.

    Almost contact and paracontact structures

  • Definition Let (M,η) be a contact manifold.

    An almost contact structure on (M,η) is given by a tensor field φ such that

    φ2 = −I + η⊗ξ.

    An almost paracontact structure on (M,η) is given by a tensor field φ such that

    � φ2 = I − η⊗ξ

    � φ induces an almost paracomplex structure on the contact distribution (i.e. the ±1 eigendistributions of φ|D have equal dimension n).

    Almost contact and paracontact structures

  • The geometry of an almost contact / paracontact manifold (M2n+1,φ,ξ,η) is related to almost complex / paracomplex structures by considering

    M2n+1 × R and

    ( )

    =

    dt

    dXηξfXφ

    dt

    dfXJ ,, m

    for all X Γ(TM2n+1) and f C∞(M2n+1×R).

    Almost contact and paracontact structures

  • The geometry of an almost contact / paracontact manifold (M2n+1,φ,ξ,η) is related to almost complex / paracomplex structures by considering

    M2n+1 × R and

    ( )

    =

    dt

    dXηξfXφ

    dt

    dfXJ ,, m

    for all X Γ(TM2n+1) and f C∞(M2n+1×R).

    J2 = −I / J2 = I, respectively.

    Almost contact and paracontact structures

  • The geometry of an almost contact / paracontact manifold (M2n+1,φ,ξ,η) is related to almost complex / paracomplex structures by considering

    M2n+1 × R and

    ( )

    =

    dt

    dXηξfXφ

    dt

    dfXJ ,, m

    for all X Γ(TM2n+1) and f C∞(M2n+1×R).

    J2 = −I / J2 = I, respectively.

    When [J,J] ≡ 0,

    the almost contact / paracontact manifold M2n+1 is called normal.

    Almost contact and paracontact structures

  • It suffices to evaluate [J,J] in the couples of vector fields of type

    ((X,0),(Y,0)) and ((X,0),(0,d/dt)), for any X,Y Γ(TM2n+1):

    [ ] ( ) ( )( )0,,0,, YXJJ

    and

    [ ] ( )

    dt

    dXJJ ,0,0,,

    Almost contact and paracontact structures

  • It suffices to evaluate [J,J] in the couples of vector fields of type

    ((X,0),(Y,0)) and ((X,0),(0,d/dt)), for any X,Y Γ(TM2n+1):

    [ ] ( ) ( )( ) ( )( ) ( )( )

    =

    dt

    dYXNYXNYXJJ ,,,0,,0,, 21

    and

    [ ] ( ) ( )( ) ( )( )( )XNXNdt

    dXJJ 43 ,,0,0,, =

    Almost contact and paracontact structures

  • It suffices to evaluate [J,J] in the couples of vector fields of type

    ((X,0),(Y,0)) and ((X,0),(0,d/dt)), for any X,Y Γ(TM2n+1):

    [ ] ( ) ( )( ) ( )( ) ( )( )

    =

    dt

    dYXNYXNYXJJ ,,,0,,0,, 21

    and

    [ ] ( ) ( )( ) ( )( )( )XNXNdt

    dXJJ 43 ,,0,0,, =

    where N(1)(X,Y) := [φ,φ](X,Y) + 2dη(X,Y)ξ

    N(2)(X,Y) := ( φXη)(Y) – ( φYη)(X)

    N(3)(X) := ( ξφ)X

    N(4)(X) := ( ξη)X.

    Almost contact and paracontact structures

  • Then

    (φ, ξ, η) is normal

    N(1) = N(2) = N(3) = N(4) = 0

    Almost contact and paracontact structures

  • Then

    (φ, ξ, η) is normal

    N(1) = N(2) = N(3) = N(4) = 0

    N(1)= 0

    Almost contact and paracontact structures

  • Definition An almost bi-paracontact structure on a contact manifold (M,η) is given by the triplet (φ1,φ2,φ3), where φ1,φ2,φ3 are tensor fields of type (1,1) on M such that

    Bi-paracontact structures

  • Definition An almost bi-paracontact structure on a contact manifold (M,η) is given by the triplet (φ1,φ2,φ3), where φ1,φ2,φ3 are tensor fields of type (1,1) on M such that

    φ12 = φ22 = I – η⊗ξ

    Bi-paracontact structures

  • Definition An almost bi-paracontact structure on a contact manifold (M,η) is given by the triplet (φ1,φ2,φ3), where φ1,φ2,φ3 are tensor fields of type (1,1) on M such that

    φ12 = φ22 = I – η⊗ξ

    φ32 = –I + η⊗ξ

    Bi-paracontact structures

  • Definition An almost bi-paracontact structure on a contact manifold (M,η) is given by the triplet (φ1,φ2,φ3), where φ1,φ2,φ3 are tensor fields of type (1,1) on M such that

    φ12 = φ22 = I – η⊗ξ

    φ32 = –I + η⊗ξ

    φ1φ2 = –φ2φ1 = φ3

    Bi-paracontact structures

  • Definition An almost bi-paracontact structure on a contact manifold (M,η) is given by the triplet (φ1,φ2,φ3), where φ1,φ2,φ3 are tensor fields of type (1,1) on M such that

    φ12 = φ22 = I – η⊗ξ

    φ32 = –I + η⊗ξ

    φ1φ2 = –φ2φ1 = φ3

    For each =1,2 set

    Dα+ := {X∈T(M) │ φ X = X} Dα− := {X∈T(M) │ φ X = −X}

    Bi-paracontact structures

  • Proposition The canonical distributions D1+, D1−, D2+, D2− satisfy the following properties:

    Bi-paracontact structures

  • Proposition The canonical distributions D1+, D1−, D2+, D2− satisfy the following properties:

    1. D1± = {X + φ3X | X D2±}, D2± = {Y + φ3Y | Y D1}.

    Bi-paracontact structures

  • Proposition The canonical distributions D1+, D1−, D2+, D2− satisfy the following properties:

    1. D1± = {X + φ3X | X D2±}, D2± = {Y + φ3Y | Y D1}. 2. For any ≠β,

    T(M) = D + ⊕ D − ⊕ Rξ = D ± ⊕ Dβ± ⊕ Rξ.

    Bi-paracontact structures

  • Proposition The canonical distributions D1+, D1−, D2+, D2− satisfy the following properties:

    1. D1± = {X + φ3X | X D2±}, D2± = {Y + φ3Y | Y D1}. 2. For any ≠β,

    T(M) = D + ⊕ D − ⊕ Rξ = D ± ⊕ Dβ± ⊕ Rξ.

    3. dim(D1+) = dim(D1−) = dim(D2+) = dim(D2−) = n.

    It follows that φ1 and φ2 are almost paracontact structures.

    Bi-paracontact structures

  • Proposition The canonical distributions D1+, D1−, D2+, D2− satisfy the following properties:

    1. D1± = {X + φ3X | X D2±}, D2± = {Y + φ3Y | Y D1}. 2. For any ≠β,

    T(M) = D + ⊕ D − ⊕ Rξ = D ± ⊕ Dβ± ⊕ Rξ.

    3. dim(D1+) = dim(D1−) = dim(D2+) = dim(D2−) = n.

    It follows that φ1 and φ2 are almost paracontact structures. Are D + and D − Legendre distributions?

    Bi-paracontact structures

  • Theorem Let (φ1,φ2,φ3) be an almost bi-paracontact structure on (M,η). Then, for each =1,2,

    D + and D − are Legendre distributions

    dη is of type (1,1) with respect to φ

    N (2) = 0 where

    N (2)(X,Y) := ( X

    η)(Y) – ( Y

    η)(X).

    Bi-paracontact structures

  • .

    Theorem Let (φ1,φ2,φ3) be an almost bi-paracontact structure on (M,η). Then, for each =1,2,

    D + and D − are Legendre foliations

    N (1) vanishes on the contact distribution where

    N (1) := [φ ,φ ] + 2dη⊗ξ.

    Bi-paracontact structures

  • .

    Theorem Let (φ1,φ2,φ3) be an almost bi-paracontact structure on (M,η). Then, for each =1,2,

    D + and D − are Legendre foliations

    N (1) vanishes on the contact distribution where

    N (1) := [φ ,φ ] + 2dη⊗ξ.

    Definition When, for all =1,2, one among the above two conditions holds, the almost bi-paracontact structure is called integrable.

    Bi-paracontact structures

  • Theorem Let (φ1,φ2,φ3) be an almost bi-paracontact structure on (M,η). Then, for each =1,2,

    D + and D − are flat (= ξ is basic) Legendre foliations

    N (1) = 0

    Bi-paracontact structures

  • Theorem Let (φ1,φ2,φ3) be an almost bi-paracontact structure on (M,η). Then, for each =1,2,

    D + and D − are flat (= ξ is basic) Legendre foliations

    N (1) = 0

    Definition When, for all =1,2, one among the above two conditions holds, the almost bi-paracontact structure is called normal.

    Bi-paracontact structures

  • Theorem Let (φ1,φ2,φ3) be an almost bi-paracontact structure on (M,η). Then, for each =1,2,

    D + and D − are flat (= ξ is basic) Legendre foliations

    N (1) = 0

    Definition When, for all =1,2, one among the above two conditions holds, the almost bi-paracontact structure is called normal.

    (φ1,φ2,φ3) is normal ⇔ (φ1,ξ,η), (φ2,ξ,η) are normal as almost paracontact struct. and (φ3,ξ,η) is normal as almost contact struct.

    Bi-paracontact structures

  • Theorem Let (φ1,φ2,φ3) be an almost bi-paracontact structure on (M,η). Then, for each =1,2,

    D + and D − are flat (= ξ is basic) Legendre foliations

    N (1) = 0

    Definition When, for all =1,2, one among the above two conditions holds, the almost bi-paracontact structure is called normal.

    (φ1,φ2,φ3) is normal ⇔ (φ1,ξ,η), (φ2,ξ,η) are normal as almost paracontact struct. and (φ3,ξ,η) is normal as almost contact struct.

    normality ⇒⇒⇒⇒ integrability

    Bi-paracontact structures

  • Odd dimension Even dimension

    almost bi-paracontact

    structure

    Bi-paracontact structures

  • Odd dimension Even dimension

    almost bi-paracontact

    structure

    � Almost quaternion structure of the II kind (Yano, Ako 1973)

    Bi-paracontact structures

  • Odd dimension Even dimension

    almost bi-paracontact

    structure

    � Almost quaternion structure of the II kind (Yano, Ako 1973)

    � 3-web (Nagy 1988)

    Bi-paracontact structures

  • Odd dimension Even dimension

    almost bi-paracontact

    structure

    � Almost quaternion structure of the II kind (Yano, Ako 1973)

    � 3-web (Nagy 1988) � Bi-paracomplex structure (Santamaría

    1999; Etayo, Santamaría, Trías 2006)

    Bi-paracontact structures

  • Odd dimension Even dimension

    almost bi-paracontact

    structure

    � Almost quaternion structure of the II kind (Yano, Ako 1973)

    � 3-web (Nagy 1988) � Bi-paracomplex structure (Santamaría

    1999; Etayo, Santamaría, Trías 2006) � Anti-hypercomplex structure

    (Marchiafava, Nagy 2003)

    Bi-paracontact structures

  • Odd dimension Even dimension

    almost bi-paracontact

    structure

    � Almost quaternion structure of the II kind (Yano, Ako 1973)

    � 3-web (Nagy 1988) � Bi-paracomplex structure (Santamaría

    1999; Etayo, Santamaría, Trías 2006) � Anti-hypercomplex structure

    (Marchiafava, Nagy 2003) � Complex-product structure (Andrada

    2005)

    Bi-paracontact structures

  • Odd dimension Even dimension

    almost bi-paracontact

    structure

    � Almost quaternion structure of the II kind (Yano, Ako 1973)

    � 3-web (Nagy 1988) � Bi-paracomplex structure (Santamaría

    1999; Etayo, Santamaría, Trías 2006) � Anti-hypercomplex structure

    (Marchiafava, Nagy 2003) � Complex-product structure (Andrada

    2005) � Para-hypercomplex structure (Ivanov,

    Zamkovoy 2005)

    Bi-paracontact structures

  • Odd dimension Even dimension

    almost bi-paracontact

    structure

    � Almost quaternion structure of the II kind (Yano, Ako 1973)

    � 3-web (Nagy 1988) � Bi-paracomplex structure (Santamaría

    1999; Etayo, Santamaría, Trías 2006) � Anti-hypercomplex structure

    (Marchiafava, Nagy 2003) � Complex-product structure (Andrada

    2005) � Para-hypercomplex structure (Ivanov,

    Zamkovoy 2005)

    Obata connection Chern connection

    Bi-Lagrangian connection

    Bi-paracontact structures

  • Odd dimension Even dimension

    almost bi-paracontact

    structure

    � Almost quaternion structure of the II kind (Yano, Ako 1973)

    � 3-web (Nagy 1988) � Bi-paracomplex structure (Santamaría

    1999; Etayo, Santamaría, Trías 2006) � Anti-hypercomplex structure

    (Marchiafava, Nagy 2003) � Complex-product structure (Andrada

    2005) � Para-hypercomplex structure (Ivanov,

    Zamkovoy 2005)

    Obata connection Chern connection

    Bi-Lagrangian connection ?

    Bi-paracontact structures

  • Theorem Let (φ1,φ2,φ3) be an almost bi-paracontact structure on (M,η). For each = 1,2,3 there exists a unique connection � such that

    Bi-paracontact structures

  • Theorem Let (φ1,φ2,φ3) be an almost bi-paracontact structure on (M,η). For each = 1,2,3 there exists a unique connection � such that i) � ξ = 0

    Bi-paracontact structures

  • Theorem Let (φ1,φ2,φ3) be an almost bi-paracontact structure on (M,η). For each = 1,2,3 there exists a unique connection � such that i) � ξ = 0

    ii) �1φ1 = 0, �1φ2 = η⊗(2h2–h1φ3+φ3h1), �1φ3 = η⊗(2h3–h1φ2+φ2h1)

    Bi-paracontact structures

  • Theorem Let (φ1,φ2,φ3) be an almost bi-paracontact structure on (M,η). For each = 1,2,3 there exists a unique connection � such that i) � ξ = 0

    ii) �1φ1 = 0, �1φ2 = η⊗(2h2–h1φ3+φ3h1), �1φ3 = η⊗(2h3–h1φ2+φ2h1)

    �2φ1 = η⊗(2h1+h2φ3–φ3h2), �2φ2 = 0, �2φ3 = η⊗(2h3–h2φ1+φ1h2)

    Bi-paracontact structures

  • Theorem Let (φ1,φ2,φ3) be an almost bi-paracontact structure on (M,η). For each = 1,2,3 there exists a unique connection � such that i) � ξ = 0

    ii) �1φ1 = 0, �1φ2 = η⊗(2h2–h1φ3+φ3h1), �1φ3 = η⊗(2h3–h1φ2+φ2h1)

    �2φ1 = η⊗(2h1+h2φ3–φ3h2), �2φ2 = 0, �2φ3 = η⊗(2h3–h2φ1+φ1h2)

    �3φ1 = η⊗(2h1–h3φ2+φ2h3), �3φ2 = η⊗(2h2+h3φ1–φ1h3), �3φ3 = 0

    Bi-paracontact structures

  • Theorem Let (φ1,φ2,φ3) be an almost bi-paracontact structure on (M,η). For each = 1,2,3 there exists a unique connection � such that i) � ξ = 0

    ii) ����1φ1 = 0, �1φ2 = η⊗(2h2–h1φ3+φ3h1), �1φ3 = η⊗(2h3–h1φ2+φ2h1)

    �2φ1 = η⊗(2h1+h2φ3–φ3h2), ����2φ2 = 0, �2φ3 = η⊗(2h3–h2φ1+φ1h2)

    �3φ1 = η⊗(2h1–h3φ2+φ2h3), �3φ2 = η⊗(2h2+h3φ1–φ1h3), ����3φ3 = 0

    Bi-paracontact structures

  • Theorem Let (φ1,φ2,φ3) be an almost bi-paracontact structure on (M,η). For each = 1,2,3 there exists a unique connection � such that i) � ξ = 0

    ii) ����1φ1 = 0, �1φ2 = η⊗(2h2–h1φ3+φ3h1), �1φ3 = η⊗(2h3–h1φ2+φ2h1)

    �2φ1 = η⊗(2h1+h2φ3–φ3h2), ����2φ2 = 0, �2φ3 = η⊗(2h3–h2φ1+φ1h2)

    �3φ1 = η⊗(2h1–h3φ2+φ2h3), �3φ2 = η⊗(2h2+h3φ1–φ1h3), ����3φ3 = 0

    iii) T (φ X,Y) – T (X,φ Y) = 2(dη(φ X,Y) – dη(X,φ Y)) ξ + η(Y)h X + η(X)h Y where

    hα := 21

    ξ φ

    Bi-paracontact structures

  • Theorem Any almost bi-paracontact manifold admits a unique connection c∇ (called the canonical connection) such that

    Bi-paracontact structures

  • Theorem Any almost bi-paracontact manifold admits a unique connection c∇ (called the canonical connection) such that

    i) c∇ ξ = 0

    Bi-paracontact structures

  • Theorem Any almost bi-paracontact manifold admits a unique connection c∇ (called the canonical connection) such that

    i) c∇ ξ = 0

    ii) c∇ φ = 32

    η⊗h for each =1,2,3

    Bi-paracontact structures

  • Theorem Any almost bi-paracontact manifold admits a unique connection c∇ (called the canonical connection) such that

    i) c∇ ξ = 0

    ii) c∇ φ = 32

    η⊗h for each =1,2,3

    iii) cT = dη − 31(dη(φ1∙,φ1∙) + dη(φ2∙,φ2∙) − dη(φ3∙,φ3∙))

    − 61( ( ) ( ) ( )111

    321NNN −+ ).

    Bi-paracontact structures

  • Theorem Any almost bi-paracontact manifold admits a unique connection c∇ (called the canonical connection) such that

    i) c∇ ξ = 0

    ii) c∇ φ = 32

    η⊗h for each =1,2,3

    iii) cT = dη − 31(dη(φ1∙,φ1∙) + dη(φ2∙,φ2∙) − dη(φ3∙,φ3∙))

    − 61( ( ) ( ) ( )111

    321NNN −+ ).

    Furthermore, c∇ is explicitly given by

    ( )32131

    ∇∇∇∇ ++=c .

    Bi-paracontact structures

  • Theorem If the almost bi-paracontact manifold is normal then c∇ satisfies the following properties:

    Bi-paracontact structures

  • Theorem If the almost bi-paracontact manifold is normal then c∇ satisfies the following properties:

    1) c∇ φ1 = c∇ φ2 = c∇ φ3 = 0

    Bi-paracontact structures

  • Theorem If the almost bi-paracontact manifold is normal then c∇ satisfies the following properties:

    1) c∇ φ1 = c∇ φ2 = c∇ φ3 = 0

    2) cT = 2dη⊗ξ

    Bi-paracontact structures

  • Theorem If the almost bi-paracontact manifold is normal then c∇ satisfies the following properties:

    1) c∇ φ1 = c∇ φ2 = c∇ φ3 = 0

    2) cT = 2dη⊗ξ

    3) cR (φ1∙,φ1∙) = cR (φ2∙,φ2∙) = − cR (φ3∙,φ3∙) = − cR

    Bi-paracontact structures

  • Theorem If the almost bi-paracontact manifold is normal then c∇ satisfies the following properties:

    1) c∇ φ1 = c∇ φ2 = c∇ φ3 = 0

    2) cT = 2dη⊗ξ

    3) cR (φ1∙,φ1∙) = cR (φ2∙,φ2∙) = − cR (φ3∙,φ3∙) = − cR

    4) The Ricci tensor is given by

    Ricc(X,Y) = −21

    trace(Rc(X,Y))

    Hence, Ricc is skew-symmetric and

    Ricc(φ1∙,φ1∙) = Ricc(φ2∙,φ2∙) = − Ricc(φ3∙,φ3∙) = Ricc

    Bi-paracontact structures

  • 5) c∇∇∇∇ === 321

    Bi-paracontact structures

  • 5) c∇∇∇∇ === 321 6) M is foliated by four mutually transverse Legendre foliations

    whose leaves are totally geodesic and admit a flat affine structure

    Bi-paracontact structures

  • 5) c∇∇∇∇ === 321 6) M is foliated by four mutually transverse Legendre foliations

    whose leaves are totally geodesic and admit a flat affine structure

    7) The 1-dimensional foliation defined by the Reeb vector field ξ is

    transversely para-hypercomplex. c∇ is (locally) projectable to the Obata connection on the leaf

    space.

    Bi-paracontact structures

  • The model in the normal case Consider R2n+1 with coordinates (x1,…,xn,y1,…,yn,z) and contact form

    ∑=

    −=n

    iiidxydz

    1

    : .

    Examples

  • The model in the normal case Consider R2n+1 with coordinates (x1,…,xn,y1,…,yn,z) and contact form

    ∑=

    −=n

    iiidxydz

    1

    : .

    Set, for each i∈{1,…,n}

    ii

    yX

    ∂∂

    = , z

    yx

    Y ii

    i ∂∂

    +∂∂

    = .

    Define φ1 Xi := Xi φ1Yi := ‒Yi φ1ξ := 0 φ2 Xi := ‒Yi φ2Yi := ‒Xi φ2ξ := 0 φ3 Xi := Yi φ3Yi := ‒Xi φ3ξ := 0

    Then (φ1,φ2,φ3) is a normal almost bi-paracontact structure on R2n+1.

    Examples

  • The model in the normal case Consider R2n+1 with coordinates (x1,…,xn,y1,…,yn,z) and contact form

    ∑=

    −=n

    iiidxydz

    1

    : .

    Set, for each i∈{1,…,n}

    ii

    yX

    ∂∂

    = , z

    yx

    Y ii

    i ∂∂

    +∂∂

    = .

    Define φ1 Xi := Xi φ1Yi := ‒Yi φ1ξ := 0 φ2 Xi := ‒Yi φ2Yi := ‒Xi φ2ξ := 0 φ3 Xi := Yi φ3Yi := ‒Xi φ3ξ := 0

    Then (φ1,φ2,φ3) is a normal almost bi-paracontact structure on R2n+1.

    The canonical distributions D1+, D1−, D2+, D2− are given by

    D1+ = span{X1,…,Xn}, D1‒ = span{Y1,…,Yn}

    D2+ = span{X1‒Y1,…,Xn‒Yn}, D2‒ = span{X1+Y1,…,Xn+Yn}

    Examples

  • Further examples Theorem Every contact Riemannian manifold endowed with a Legendre distribution admits an almost bi-paracontact structure.

    Examples

  • Further examples Theorem Every contact Riemannian manifold endowed with a Legendre distribution admits an almost bi-paracontact structure. Therefore we have many examples of almost bi-paracontact manifolds:

    - The cotangent sphere bundle of a Riemannian manifold

    - Every contact Riemannian manifold such that ξ is an Anosov flow (e.g. the tangent sphere bundle of a negatively curved manifold)

    - Any contact metric (κ,µ)-space

    Examples

  • Definition A contact metric ( , )-space is a contact Riemannian manifold (M,φ,ξ,η,g) such that

    Rg(X,Y)ξ = κ(η(Y)X – η(X)Y) + µ(η(Y)hX – η(X)hY), ( )

    for some κ,µ R, where

    h := 21

    ξ φ.

    ( ) is called “(κ,µ)-nullity condition”. If in ( ) κ = 1, then h 0 and the manifold is Sasakian.

    Contact metric ( , )-spaces

  • Definition A contact metric ( , )-space is a contact Riemannian manifold (M,φ,ξ,η,g) such that

    Rg(X,Y)ξ = κ(η(Y)X – η(X)Y) + µ(η(Y)hX – η(X)hY), ( )

    for some κ,µ R, where

    h := 21

    ξ φ.

    ( ) is called “(κ,µ)-nullity condition”. If in ( ) κ = 1, then h 0 and the manifold is Sasakian.

    D.E. Blair, T. Koufogiorgos, B.J. Papantoniou, Contact metric manifolds satisfying a nullity condition, Israel J. Math. 1995

    Contact metric ( , )-spaces

  • 1. The (κ,µ)-nullity condition determines the curvature completely.

    Motivations

  • 1. The (κ,µ)-nullity condition determines the curvature completely.

    2. There are non-trivial examples of contact metric (κ,µ)-spaces, the most important being the tangent sphere bundle of a Riemannian manifold of constant sectional curvature c (κ = c(2 – c), µ = –2c).

    Motivations

  • 1. The (κ,µ)-nullity condition determines the curvature completely.

    2. There are non-trivial examples of contact metric (κ,µ)-spaces, the most important being the tangent sphere bundle of a Riemannian manifold of constant sectional curvature c (κ = c(2 – c), µ = –2c).

    3. Any contact metric (κ,µ)-space is a strongly pseudo-convex CR-manifold.

    Motivations

  • 1. The (κ,µ)-nullity condition determines the curvature completely.

    2. There are non-trivial examples of contact metric (κ,µ)-spaces, the most important being the tangent sphere bundle of a Riemannian manifold of constant sectional curvature c (κ = c(2 – c), µ = –2c).

    3. Any contact metric (κ,µ)-space is a strongly pseudo-convex CR-manifold.

    4. In the non-Sasakian case (κ≠1) the contact metric structure is not “projectable”. Thus contact metric ( , )-spaces have no analogues in even dimension.

    Motivations

  • 1. The (κ,µ)-nullity condition determines the curvature completely.

    2. There are non-trivial examples of contact metric (κ,µ)-spaces, the most important being the tangent sphere bundle of a Riemannian manifold of constant sectional curvature c (κ = c(2 – c), µ = –2c).

    3. Any contact metric (κ,µ)-space is a strongly pseudo-convex CR-manifold.

    4. In the non-Sasakian case (κ≠1) the contact metric structure is not “projectable”. Thus contact metric ( , )-spaces have no analogues in even dimension.

    5. A classification is known

    E. Boeckx, A full classification of contact (κ,µ)-spaces, Illinois J. Math. (2000)

    Motivations

  • Theorem (Blair et alt. 1995; C.M. 2010) Let (M,φ,ξ,η,g) be a contact (κ,µ)-manifold. Then κ ≤ 1.

    If κ = 1 (M,φ,ξ,η,g) is Sasakian.

    If κ < 1 the operators h and φh admit 3 eigenvalues 0, λ, –λ, where λ = κ−1 .

    Contact metric ( , )-spaces

  • Theorem (Blair et alt. 1995; C.M. 2010) Let (M,φ,ξ,η,g) be a contact (κ,µ)-manifold. Then κ ≤ 1.

    If κ = 1 (M,φ,ξ,η,g) is Sasakian.

    If κ < 1 the operators h and φh admit 3 eigenvalues 0, λ, –λ, where λ = κ−1 .

    The corresponding eigendistributions are Legendre foliations of M and satisfies

    T(M) = Dh(λ) ⊕ Dh(-λ) ⊕ Rξ (orthogonal) = Dh(λ) ⊕ Dφh(λ) ⊕ Rξ = Dh(λ) ⊕ Dφh(-λ) ⊕ Rξ = Dh(-λ) ⊕ Dφh(λ) ⊕ Rξ = Dh(-λ) ⊕ Dφh(-λ) ⊕ Rξ = Dφh(λ) ⊕ Dφh(-λ) ⊕ Rξ (orthogonal)

    Contact metric ( , )-spaces

  • Corollary Every (non-Sasakian) contact metric (κ,µ)-space admits a canonical almost bi-paracontact structure (φ1,φ2,φ3) given by

    hφκ

    φ−

    =11

    :1 , hκ

    φ−

    =11

    :2 , φφ =:3 .

    Such almost bi-paracontact structure is integrable, non-normal.

    Contact metric ( , )-spaces

  • Theorem The canonical connection c∇ is a contact connection (i.e. c∇ =

    c∇ d = 0), and satisfies

    23

    312

    21

    132

    132

    2

    132

    21

    32

    φφ

    φφφ

    φφ

    c

    c

    c

    ⊗−=

    ⊗−+⊗

    −=

    −−=

    Contact metric ( , )-spaces

  • Theorem The canonical connection c∇ is a contact connection (i.e. c∇ =

    c∇ d = 0), and satisfies

    23

    312

    21

    132

    132

    2

    132

    21

    32

    φφ

    φφφ

    φφ

    c

    c

    c

    ⊗−=

    ⊗−+⊗

    −=

    −−=

    Conversely …

    Contact metric ( , )-spaces

  • Theorem Let (φ1,φ2,φ3) be an integrable almost bi-paracontact structure on a contact manifold (M, ) such that c∇ = c∇ d = 0 and

    23

    312

    21

    φbφ

    φbφaφ

    φaφ

    c

    c

    c

    ⊗=

    ⊗+⊗=

    ⊗−=

    for some a, b > 0.

    Contact metric ( , )-spaces

  • Theorem Let (φ1,φ2,φ3) be an integrable almost bi-paracontact structure on a contact manifold (M, ) such that c∇ = c∇ d = 0 and

    23

    312

    21

    φbφ

    φbφaφ

    φaφ

    c

    c

    c

    ⊗=

    ⊗+⊗=

    ⊗−=

    for some a, b > 0.

    Set g1 := dη(∙,φ1∙) + η⊗η

    g2 := dη(∙,φ2∙) + η⊗η

    g3 := −dη(∙,φ3∙) + η⊗η and assume that g3 is positive definite.

    Contact metric ( , )-spaces

  • Then

    (φ1,ξ,η,g1), (φ2,ξ,η,g2) are paracontact metric structures

    (φ3,ξ,η,g3) is a contact metric structure Moreover, the Levi Civita connections of g1, g2, g3 satisfy a ( , )-nullity condition, where

    κ1 = 9/4 a2 − 1, µ1 = 2 − 3b

    κ2 = 9/4 (a2 − b2), µ2 = 2

    κ3 = 1 − 9/4 b2, µ3 = 2 + 3a

    Contact metric ( , )-spaces

  • Let us recall that Boeckx introduced the number

    −=

    1

    21MI

    in order to classify contact metric (κ,µ)-spaces.

    Contact metric ( , )-spaces

  • Let us recall that Boeckx introduced the number

    −=

    1

    21MI

    in order to classify contact metric (κ,µ)-spaces.

    Theorem Let (M,φ,ξ,η,g) be a contact metric (κ,µ)-space such that |IM|≠1.

    Contact metric ( , )-spaces

  • Let us recall that Boeckx introduced the number

    −=

    1

    21MI

    in order to classify contact metric (κ,µ)-spaces.

    Theorem Let (M,φ,ξ,η,g) be a contact metric (κ,µ)-space such that |IM|≠1. Then M admits a further integrable almost bi-paracontact structure (φ’1,φ’2,φ’3) given by

    Contact metric ( , )-spaces

  • Let us recall that Boeckx introduced the number

    −=

    1

    21MI

    in order to classify contact metric (κ,µ)-spaces.

    Theorem Let (M,φ,ξ,η,g) be a contact metric (κ,µ)-space such that |IM|≠1. Then M admits a further integrable almost bi-paracontact structure (φ’1,φ’2,φ’3) given by

    i) If |IM|>1

    ( )( )φhφIφ M −+=′

    +−−1:

    121

    121

    , hφ−

    =′1

    1:2 , 213 : φφφ ′′=′

    Contact metric ( , )-spaces

  • Let us recall that Boeckx introduced the number

    −=

    1

    21MI

    in order to classify contact metric (κ,µ)-spaces.

    Theorem Let (M,φ,ξ,η,g) be a contact metric (κ,µ)-space such that |IM|≠1. Then M admits a further integrable almost bi-paracontact structure (φ’1,φ’2,φ’3) given by

    i) If |IM|>1

    ( )( )φhφIφ M −+=′

    +−−1:

    121

    121

    , hφ−

    =′1

    1:2 , 213 : φφφ ′′=′

    ii) If |IM|

  • Let us consider the above almost contact structure

    φ'3 := φ’1φ’2.

    Contact metric ( , )-spaces

  • Let us consider the above almost contact structure

    φ'3 := φ’1φ’2.

    Then one proves that

    g'3 := −dη(∙,φ’3∙) + η⊗η

    is a Riemannian metric, which is compatible with the almost contact structure.

    Contact metric ( , )-spaces

  • Let us consider the above almost contact structure

    φ'3 := φ’1φ’2.

    Then one proves that

    g'3 := −dη(∙,φ’3∙) + η⊗η

    is a Riemannian metric, which is compatible with the almost contact structure. Theorem (C.M. 2010) g'3 is a Sasakian metric!

    Contact metric ( , )-spaces

  • Let us consider the above almost contact structure

    φ'3 := φ’1φ’2.

    Then one proves that

    g'3 := −dη(∙,φ’3∙) + η⊗η

    is a Riemannian metric, which is compatible with the almost contact structure. Theorem (C.M. 2010) g'3 is a Sasakian metric! Corollary Any contact metric (κ,µ)-space such that IM ≠ ±1 admits a compatible Sasakian structure. Consequently, if M is compact, the Betti numbers bi, 1 ≤ i ≤ 2n are even.

    Contact metric ( , )-spaces

  • 1. B. Cappelletti Montano, Bi-Legendrian structures and paracontact

    geometry, Int. J. Geom. Meth. Mod. Phys. 2009

    2. B. Cappelletti Montano, Bi-paracontact structures and Legendre foliations, Kodai Math. J. 2010.

    3. B. Cappelletti Montano, The foliated structure of contact metric (κ,µ)-spaces, Illinois J. Math. (to appear)

    4. B. Cappelletti Montano, L. Di Terlizzi, Geometric structures associated to a contact metric (κ,µ)-space, Pacific J. Math. 2010

    References

  • 1. B. Cappelletti Montano, Bi-Legendrian structures and paracontact

    geometry, Int. J. Geom. Meth. Mod. Phys. 2009

    2. B. Cappelletti Montano, Bi-paracontact structures and Legendre foliations, Kodai Math. J. 2010.

    3. B. Cappelletti Montano, The foliated structure of contact metric (κ,µ)-spaces, Illinois J. Math. (to appear)

    4. B. Cappelletti Montano, L. Di Terlizzi, Geometric structures associated to a contact metric (κ,µ)-space, Pacific J. Math. 2010

    OBRIGADO !

    References

  • Bi-paracontact structures on contact manifolds Let (M,φ,ξ,η,g) be a contact metric manifold endowed with a Legendre distribution L. Define Q:=φ(L), Legendre distribution. Set

    ψ│L = I, ψ│Q = −I, ψξ = 0.

    Then φ1 := φψ, φ2 := ψ, φ3 := φ

    is an almost bi-paracontact structure.

  • Bi-paracontact structures on contact manifolds Every contact manifold (M,η) endowed with a Legendre distribution admits infinitely many bi-Legendrian structures. Infact let L be a Legendre distribution, (φ,ξ,η,g) be an associated contact Riemannian structure and Q=φL. Let ψ be the paracontact structure associated with (L,Q). For any , such that 2 + 2 = 1 we set

    ψ , := ψ + φψ.

    Then ψ ,

    2 = I - η⊗ξ The eingendistributions + ,L and

    −,L of ψ , corresponding to 1 and

    -1 defines transversal Legendre distributions. Thus ( )−+ ,, ,LL is an almost bi-Legendrian structure on (M,η).

  • Legendre foliations Example M - smooth manifold, dim(M)=n. T*(M) - the cotangent bundle of M. ω = yidxi - the Liouville form on T*(M), (xi) being coordinates on M, (yi) fiber coordinates. Suppose there exists a function F: T*(M) → [0,+∞[ such that

    F(tv) = tF(v), for all t ≥ 0, v∈T*(M).

    The unit cotangent bundle with respect to F is the (2n+1)-dimensional hypersurface of T*(M)

    SF*(M) = {v∈T*(M) │ F(v) = 1}.

    Typical case: (M,g) a Riemannian manifold and take F = ǁ∙ǁ. ( ) is denoted also 1

    *( ) and called of ( , ).

    Then (S*F(M),η) is a contact manifold, where η = i*(ω) and the foliation FF by fibers of the projection map : SF*(M) → M is a Legendre foliation of (SF*(M),η).

  • Legendre foliations Theorem (Pang) Any Legendre foliation F on (M,η) is locally equivalent with one of the form FF. Moreover, the equivalence is global provided that the leaves of F are compact and simply connected; in this case F defines a Finsler metric on M.

  • Finsler metrics A Finsler metric is a function F: T *M → [0,+∞[ such that

    F is positively homogeneous: F(tv) = t F(v) for any t > 0 F(v1+v2) ≤ F(v1) + F(v2).

    The triangle inequality is equivalent to the convexity of the function F2, and this is again equivalent to the condition that (F2)ij is positive definite on T *M.

  • Contact Riemannian manifolds Theorem

    Let (M,η) be a contact manifold. Then there exist

    a Riemannian metric g such that

    g(X,ξ) = η(X),

    i.e. g(ξ,ξ) = 1 and ξ ⊥ D

    a tensor field φ of type (1,1) such that

    φ2 = –I + η⊗ξ, g(X,φY) = dη(X,Y).

    The geometric structure (φ,ξ,η,g) is called contact Riemannian (or metric) structure.

  • Anosov flows

    Example (Anosov flows) Let (M,φ,ξ,η,g) be a contact Riemannian manifold for which ξ is Anosov, i.e. there exist subbundles Es and Eu (stable and unstable subbundles), invariant along the flow {ωt} of ξ, such that

    T(M) = Es ⊕ Eu ⊕ Rξ

    and

    ( )vctavωt −≤∗ exp for all t≥0 and v∈Esx

    ( )wctawωt exp≤∗ for all t≤0 and w∈Eux,

    where a, c > 0 are constants independent of x∈M and v∈Esx, w∈Eux.

    Then (Es,Eu) is a flat bi-Legendrian structure on M.

    The most notable example of contact manifold for which ξ is Anosov is the tangent sphere bundle of a negatively curved manifold.

  • Tangent sphere bunlde Let π’: TM → M be the tangent bundle of (M,g). The tangent space to TM at a point (x,u) splits into the direct sum of the vertical subspace VTM(x,u) = ker(π’*|(x,u)) and the horizontal subspace HTM(x,u) with respect to the Levi Civita connection ∇g on M:

    T(x,u)TM = VTM(x,u) ⊕ HTM(x,u).

    If X∈Γ(TM), Xh and Xv are the horizontal and vertical lift of X on TM. The Sasaki metric gS on TM is defined by

    gS(Xh,Yh) = gS(Xv,Yv) = g(X,Y) ◦ π’, gS(Xh,Xv) = 0.

    TM admits an almost complex structure J defined by

    JXh = Xv, JXv = –Xh.

  • Tangent sphere bunlde The tangent sphere bundle π: T1M → M is the hypersurface of TM defined by

    T1M = {(x,u)∈TM | gx(u,u)=1}.

    The geodesic flow of (M,g) is the horizontal vector field of TM defined by

    h

    i

    iu

    xuJN

    ∂=−=′

    where (x,u)∈TM and i

    i

    uuN

    ∂∂

    = . If (x,z)∈T1M then z′ is tangent to

    T1M. Let η’ be the 1-form on T1M dual to ′ with respect to gS and φ’ be the (1,1) tensor field given by

    φ’X = JX – η’(X)N. Then

    (φ’, 2 ′, η’, gS)

    is the standard contact metric structure on T1M.

  • D-homotethic deformations Let (M,φ,ξ,η,g) be a contact metric manifold. By a Da-homothetic deformation we mean the change of structure tensor of the form

    a= , a

    1= , = , ⊗−+= )1(aaagg

    where a > 0. Then

    ),,,,( gM is also a contact metric manifold.

    A ),( -manifold is transformed in a ),( -manifold where

    2

    2 1a

    a −+= ,

    a

    a 22 −+= .

  • Let (M,F) be a foliated space and � be a linear connection on M. � is F-projectable if it projects to connections of the local slice space of F. The conditions for this are: a. if σ is a projectable cross section of N(F):=TM/TF, then �Xσ = 0

    for all X∈Γ(TF), b. if σ is a projectable cross section of N(F) and X is a basic vector

    field, then �Xσ is a projectable cross section of N(F). Here by “projectable cross section” we mean that the vector field that represents σ is basic: σ = [Y]F , Y∈Γ(TM), [Y1]=[Y2] ⇔ Y1–Y2 ∈Γ(TF).

  • A 3-web is a triple of foliations (F1,F2,F3) on M2n such that for

    any , ∈{1,2,3}, ≠ , T(M) = T(F )⊕T(F ).

    A para-hypercomplex structure on M2n is given by two anti-commuting product structures I, J and a complex structure K such that IJ = K. Then M2n admits a unique connection, called Obata connection, defined as the unique torsion-free connection preserving the para-hypercomplex structure.

  • Let N be an n-dimensional manifold and TCN its complexifed tangent bundle. Let H be a complex subbundle of complex dimension l. A CR-manifold of real dimension n and CR-dimension l is a pair (N,H) such that - { }0=∩ pp HH - H is involutive.

    Then there exists an unique subbundle D of TN such that DC = HH ⊕ and an unique bundle map J: D → D such that J2 = -I and

    H = {X – iJX | X∈D}.

    Since φ2 = –I + η⊗ξ and φξ = 0, the eigenvalues of φ are 0, ± i. Thus

    DC = D’ ⊕ D’’ where

    D’ = {X – iφX | X∈D} D’’ = {X + iφX | X∈D}

    Theorem (Tanno) (M2n+1,D’) is (strongly pseudo-convex) CR-manifold if and only if

    (∇gXφ)Y = g(X+hX,Y)ξ - η(Y)(X+hX).