In the present work, being well known that there is an one-to-one correspondence between generalized Lie bialgebroids structures $\left(\right(A,\phi ),(A^{*},W\left)\right)$  over $M$  and Lie bialgebroids structures $(\tilde{A},{\tilde{A}}^{*})$  , $\tilde{A}=A\times \text{IR}$  , ${\tilde{A}}^{*}=A^{*}\times \text{IR}$  , over $\tilde{M}=M\times \text{IR}$  , we establish some basic relations between the Dirac subbundles $L$  for $(A\oplus A^{*},\phi +W)$  and the associated Dirac subbundles $\tilde{L}=\{X+e^t\alpha /X+\alpha \in L\}$  for $\tilde{A}\oplus {\tilde{A}}^{*}$  . We prove : 1) $L$  is a reducible Dirac structure for $(A\oplus A^{*},\phi +W)$  if and only if $\tilde{L}$  is a reducible Dirac structure for $\tilde{A}\oplus {\tilde{A}}^{*}$  . 2) If $\mathcal{ℱ}$  and $\widetilde{\mathcal{ℱ}}$  are the characteristic foliations of $M$  and $\tilde{M}$  defined by $L$  and $\tilde{L}$  , respectively, then $\tilde{L}$  induces an homogeneous Poisson structure on $\tilde{M}/\widetilde{\mathcal{ℱ}}=M/\mathcal{ℱ}\times \text{IR}$  which is the Poissonization of the induced Jacobi structure on $M/\mathcal{ℱ}$  by $L$  .

Notation : In this paper, $M$  is a $C^{\infty }$  -differential manifold of finite dimension.

We denote by $C^{\infty }\left(M\right)$  the space of all real $C^{\infty }$  -differentiable functions on $M$  and by $\delta$  the usual de Rham differential operator. F. Petalidou and J.M. Nunes da Costa

2 Generalized Lie bialgebroids

Let $(A,[,],a)$  be a Lie algebroid over $M$  ([12] ), $A^{*}$  its dual vector bundle over $M$  , $\wedge A^{*}={\oplus }_{k\in \text{ZZ}}{\wedge }^k{A}^{*}$  the graded exterior algebra of $A^{*}$  whose differential sections are called A-forms on $M$  , $d:\Gamma (\wedge A^{*})\to \Gamma (\wedge A^{*})$  the exterior derivative of degree 1 and $\phi \in \Gamma \left(A^{*}\right)$  an 1-cocycle in the Lie algebroid cohomology complex with trivial coefficients ([12] , [6] ), i.e., for all $X,Y\in \Gamma \left(A\right)$  , $\langle \phi ,[X,Y]\rangle =a\left(X\right)(\langle \phi ,Y\rangle )-a\left(Y\right)(\langle \phi ,X\rangle )$  . We modify the usual representation $a$  of the Lie algebra $(\Gamma (A),[,\left]\right)$  on the space $C^{\infty }\left(M\right)$  by defining $a^{\phi }:\Gamma \left(A\right)\times C^{\infty }\left(M\right)\to C^{\infty }\left(M\right)$  , $a^{\phi }(X,f)=a\left(X\right)f+\langle \phi ,X\rangle f$  . The resulting cohomology operator $d^{\phi }:\Gamma (\wedge A^{*})\to \Gamma (\wedge A^{*})$  of the new cohomology complex is called the $\phi$  -differential of $A$  and $d^{\phi }\eta =d\eta +\phi \wedge \eta$  , for all $\eta \in \Gamma \left({\wedge }^k{A}^{*}\right)$  . $d^{\phi }$  allows us to define the $\phi$  -Lie derivative by $X\in \Gamma \left(A\right)$  , ${\mathcal{ℒ}}_X^{\phi }:\Gamma \left({\wedge }^k{A}^{*}\right)\to \Gamma \left({\wedge }^k{A}^{*}\right)$  , as ${\mathcal{ℒ}}_X^{\phi }=d^{\phi }\circ i_X+i_X\circ d^{\phi }$  , where $i_X$  is the contraction by $X$  . Using $\phi$  we can also modify the Schouten bracket $[,]$  on $\Gamma (\wedge A)$  to the $\phi$  -Schouten bracket $[,{]}^{\phi }$  on $\Gamma (\wedge A)$  by setting, for all $P\in \Gamma \left({\wedge }^{p}A\right)$  and $Q\in \Gamma \left({\wedge }^{q}A\right)$  , $[P,Q{]}^{\phi }=[P,Q]+(p-1)P\wedge (i_{\phi }Q)+(-1{)}^p(q-1)\left(i_{\phi }P\right)\wedge Q$  , where $i_{\phi }Q$  can be interpreted as the usual contraction of a multivector field with an 1-form.

For details, see [12] , [6] and [4] .

The notion of generalized Lie bialgebroid has been introduced by D. Iglesias and J.C. Marrero in [6] and independently by J. Grabowski and G. Marmo in [4] under the name of Jacobi bialgebroid, in such a way that a Jacobi manifold ([9] ) has a generalized Lie bialgebroid canonically associated and conversely. We recall that a Jacobi manifold is a smooth manifold $M$  equipped with a bivector field $\Lambda$  and a vector field $E$  such that $[\Lambda ,\Lambda ]=-2E\wedge \Lambda$  and $[E,\Lambda ]=0$  , where $[,]$  denotes the Schouten bracket.

We consider a Lie algebroid $(A,[,],a)$  over $M$  and an 1-cocycle $\phi \in \Gamma \left(A^{*}\right)$  and we assume that the dual vector bundle $A^{*}\to M$  admits a Lie algebroid structure $\left(\right[,{]}_{*},a_{*})$  and that $W\in \Gamma \left(A\right)$  is an 1-cocycle in the Lie algebroid cohomology complex with trivial coefficients of $(A^{*},[,{]}_{*},a_{*})$  . Then, we say that :

Definition 2.1. The pair $\left(\right(A,\phi ),(A^{*},W\left)\right)$  is a generalized Lie bialgebroid over $M$  if, for all $X,Y\in \Gamma \left(A\right)$  and $P\in \Gamma \left({\wedge }^{p}A\right)$  , the following conditions hold :

$$d_{*}^W[X,Y]=[d_{*}^{W}X,Y{]}^{\phi }+[X,d_{*}^{W}Y{]}^{\phi }and{\mathcal{ℒ}}_{*\phi }^{W}P+{\mathcal{ℒ}}_W^{\phi }P=0;$$  $d_{*}^W$  and ${\mathcal{ℒ}}_{*}^W$  are, respectively, the $W$  -differential and the $W$  -Lie derivative of $A^{*}$  .

Obviously, if $\phi =0$  and $W=0$  , we recover the notion of Lie bialgebroid introduced by K. Mackenzie and P. Xu in [13] and its equivalent definition given by Yv. Kosmann-Schwarzbach in [8] .

Given a Lie algebroid $(A,[,],a)$  over $M$  , we can construct a Lie algebroid structure on $\tilde{A}\to \tilde{M}$  , $\tilde{A}=A\times \text{IR}$  and $\tilde{M}=M\times \text{IR}$  . We identify $\Gamma \left(\tilde{A}\right)$  with Dirac structures for generalized Courant and Courant algebroids the set of the time-dependent sections of $A\to M$  , i.e. for any $\tilde{X}\in \Gamma \left(\tilde{A}\right)$  and $(x,t)\in M\times \text{IR}$  , $t$  being the canonical coordinate on $\text{IR}$  , $\tilde{X}(x,t)={\tilde{X}}_t\left(x\right)$  , where ${\tilde{X}}_t\in \Gamma \left(A\right)$  , and we take : i) the Lie bracket $[,{]}^{\stackrel{~}{}}$  on $\Gamma \left(\tilde{A}\right)$  defined, for any $\tilde{X},\tilde{Y}\in \Gamma \left(\tilde{A}\right)$  and $(x,t)\in \tilde{M}$  , by $[\tilde{X},\tilde{Y}{]}^{\stackrel{~}{}}(x,t)=[{\tilde{X}}_t,{\tilde{Y}}_t\left]\right(x)$  , ii) the bundle map $\tilde{a}:\tilde{A}\to T\tilde{M}$  , $\tilde{a}\left(\tilde{X}\right)(x,t)=a\left({\tilde{X}}_t\right)\left(x\right)$  . Then $(\tilde{A},[,{]}^{\stackrel{~}{}},\tilde{a})\to \tilde{M}$  is a Lie algebroid.

Also, taking an 1-cocycle $\phi$  of $A$  , we deform $\left(\right[,{]}^{\stackrel{~}{}},\tilde{a})$  in two different ways and we obtain two new Lie algebroid structures on $\tilde{A}$  , [6] . Precisely, for any $\tilde{X},\tilde{Y}\in \Gamma \left(\tilde{A}\right)$  :

$$\begin{array}{c}[\tilde{X},\tilde{Y}{]}^{\stackrel{~}{}\phi }=[\tilde{X},\tilde{Y}{]}^{\stackrel{~}{}}+i_{\phi }{\tilde{X}}_t\partial \tilde{Y}/\partial t-i_{\phi }\tilde{Y}\partial \tilde{X}/\partial t,{\tilde{a}}^{\phi }\left(\tilde{X}\right)=\tilde{a}\left(\tilde{X}\right)+i_{\phi }\tilde{X}\partial /\partial t;\end{array}$$

(2.1)

F. Petalidou and J.M. Nunes da Costa

$$\begin{array}{ccc}[\tilde{X},\tilde{Y}{]}^{\hat{\phi }}& =& e^{-t}\left(\right[\tilde{X},\tilde{Y}{]}^{\stackrel{~}{}}+\langle \phi ,{\tilde{X}}_t\rangle (\partial \tilde{Y}/\partial t-\tilde{Y})-\langle \phi ,{\tilde{Y}}_t\rangle (\partial \tilde{X}/\partial t-\tilde{X})),\end{array}$$

(2.2)

$$\begin{array}{ccc}{\hat{a}}^{\phi }\left(\tilde{X}\right)& =& e^{-t}\left(\tilde{a}\right(\tilde{X})+\langle \phi ,{\tilde{X}}_t\rangle \partial /\partial t).\end{array}$$

Theorem 2.1 ([6] ). Let $(A,[,],a)$  be a Lie algebroid over $M$  and $\phi \in \Gamma \left(A^{*}\right)$  an 1-cocycle. Suppose that $A^{*}$  has a Lie algebroid structure $\left(\right[,{]}_{*},a_{*})$  and that $W\in \Gamma \left(A\right)$  is an 1-cocycle for this structure. Consider on $\tilde{A}=A\times \text{IR}$  and ${\tilde{A}}^{*}=A^{*}\times \text{IR}$  the Lie algebroid structures $\left(\right[,{]}^{\stackrel{~}{}\phi },{\tilde{a}}^{\phi })$  and $\left(\right[,{]}_{*}^{\hat{W}},{\hat{a}}_{*}^W)$  , respectively. Then $(\tilde{A},{\tilde{A}}^{*})$  is a Lie bialgebroid over $\tilde{M}=M\times \text{IR}$  if and only if $\left(\right(A,\phi ),(A^{*},W\left)\right)$  is a generalized Lie bialgebroid over $M$  . The induced Poisson structure on $\tilde{M}$  is the Poissonization of the induced Jacobi structure on $M$  .

Moreover, the image $Ima$  of the anchor map $a$  of $(A,[,],a)\to M$  is an integrable distribution on $M$  ([3] ) which defines a singular foliation ${\mathcal{ℱ}}_A$  of $M$  , called the Lie algebroid foliation of M associated with A ([7] ). The relation between the leaves of the Lie algebroid foliation ${\mathcal{ℱ}}_{\tilde{A}}$  of $M\times \text{IR}$  associated with $(\tilde{A},[,{]}^{\stackrel{~}{}\phi },{\tilde{a}}^{\phi })$  (given by ( 2.1 )) and the leaves of the Lie algebroid foliation ${\mathcal{ℱ}}_A$  of $M$  associated with $A$  was studied in [7] by D. Iglesias and J.C. Marrero. They have proved :

Theorem 2.2 ([7] ). Under the above considerations, suppose that $(x_0,t_0)\in M\times \text{IR}$  and that $\tilde{F}$  and $F$  are the leaves of the Lie algebroid foliations ${\mathcal{ℱ}}_{\tilde{A}}$  and ${\mathcal{ℱ}}_A$  passing through $(x_0,t_0)\in M\times \text{IR}$  and $x_0\in M$  , respectively, and denote by $A_{x_0}$  the fiber of $A$  over $x_0$  . Then : (1) If $ker\left(a{|}_{A_{x_0}}\right)\mathbb{⊈}\langle \phi \left(x_0\right){\rangle }^{\circ }$  , $\tilde{F}=F\times \text{IR}$  . (2) If $ker\left(a{|}_{A_{x_0}}\right)\subseteq \langle \phi \left(x_0\right){\rangle }^{\circ }$  and ${\pi }_1:M\times \text{IR}\to M$  is the canonical projection onto the first factor, ${\pi }_1\left(\tilde{F}\right)=F$  and ${\pi }_1{|}_{\tilde{F}}:\tilde{F}\to F$  is a covering map.

Dirac structures for generalized Courant and Courant algebroids

3 Generalized Courant algebroids

The notion of generalized Courant algebroid has been introduced by the second author and J. Clemente-Gallardo in [14] and independently, under the name of Courant-Jacobi algebroid, by J. Grabowski and G. Marmo in [5] .

Definition 3.1 ([14] ). Let $E\to M$  to be a vector bundle over a differentiable manifold $M$  equipped with : (i) a nondegenerate symmetric bilinear form $(,)$  on the bundle, (ii) a skew-symmetric bilinear bracket $[,]$  on $\Gamma \left(E\right)$  , (iii) a bundle map $\rho :E\to TM$  and (iv) an $E$  -1-form $\theta$  such that, for any $e_1,e_2\in \Gamma \left(E\right)$  , $\langle \theta ,[e_1,e_2]\rangle =\rho \left(e_1\right)\langle \theta ,e_2\rangle -\rho \left(e_2\right)\langle \theta ,e_1\rangle$  .

We consider : (a) the bundle map ${\rho }^{\theta }:E\to TM\times \text{IR}$  defined, for any $e\in E$  , by ${\rho }^{\theta }\left(e\right)=\rho \left(e\right)+\langle \theta ,e\rangle$  , (b) the applications $\mathcal{D},{\mathcal{D}}^{\theta }:C^{\infty }\left(M\right)\to \Gamma \left(E\right)$  defined, for any $f\in C^{\infty }\left(M\right)$  , respectively, by $\mathcal{D}f=\frac{1}{2}{\beta }^{-1}{\rho }^{*}\delta f$  ^$\text{1}$  and ${\mathcal{D}}^{\theta }f=\mathcal{D}f+\frac{1}{2}f{\beta }^{-1}\left(\theta \right)$  and (c) for any $e_1,e_2,e_3\in \Gamma \left(E\right)$  , the function $T(e_1,e_2,e_3)=\frac{1}{3}\left(\right[e_1,e_2],e_3)+c.p.$  on the base $M$  . Then, we say that $E$  is a generalized Courant algebroid if the following relations are satisfied :

1. $\left[\right[e_1,e_2],e_3]+c.p.={\mathcal{D}}^{\theta }T(e_1,e_2,e_3),\forall e_1,e_2,e_3\in \Gamma \left(E\right)$  ; 2. ${\rho }^{\theta }\left(\right[e_1,e_2\left]\right)=\left[{\rho }^{\theta }\right(e_1),{\rho }^{\theta }(e_2\left)\right],$  ^$\text{2}$  $\forall e_1,e_2\in \Gamma \left(E\right)$  ; 3. $[e_1,f{e}_2]=f[e_1,e_2]+\left(\rho \right(e_1\left)f\right)e_2-(e_1,e_2)\mathcal{D}f,\forall e_1,e_2\in \Gamma \left(E\right),\forall f\in C^{\infty }\left(M\right)$  ; 4. ${\rho }^{\theta }\circ {\mathcal{D}}^{\theta }=0$  , i.e., for any $f,g\in C^{\infty }\left(M\right)$  , $({\mathcal{D}}^{\theta }f,{\mathcal{D}}^{\theta }g)=0$  ; 5. ${\rho }^{\theta }\left(e\right)(e_1,e_2)=\left(\right[e,e_1]+{\mathcal{D}}^{\theta }(e,e_1),e_2)+(e_1,[e,e_2]+{\mathcal{D}}^{\theta }(e,e_2\left)\right),\forall e,e_1,e_2\in \Gamma \left(E\right)$  .

Definition 3.2. A Dirac structure for a generalized Courant algebroid $(E,\theta )$  over $M$  is a subbundle $L\subset E$  that is maximal isotropic under $(,)$  and integrable, i.e. $\Gamma \left(L\right)$  is closed under $[,]$  .

A Dirac subbundle $L$  of $(E,\theta )$  is a Lie algebroid under the restrictions of the bracket $[,]$  and of the anchor $\rho$  to $\Gamma \left(L\right)$  . If $\theta \in \Gamma \left(L^{*}\right)$  , then it is an 1-cocycle for the Lie algebroid cohomology with trivial coefficients of $(L,[,]{|}_L,\rho {|}_L)$  .

The most important example of generalized Courant algebroid is the double $(A\oplus A^{*},\phi +W)$  of a generalized Lie bialgebroid $\left(\right(A,\phi ),(A^{*},W\left)\right)$  over $M$  . On $A\oplus A^{*}$  there exist two natural nondegenerate bilinear forms, one symmetric and another skew-symmetric $(,{)}_{\pm }$  : for any $X_1+{\alpha }_1,X_2+{\alpha }_2\in A\oplus A^{*}$  , $(X_1+{\alpha }_1,X_2+{\alpha }_2{)}_{\pm }=1/2(\langle {\alpha }_1,X_2\rangle \pm \langle {\alpha }_2,X_1\rangle )$  and on $\Gamma (A\oplus A^{*})\sim =\Gamma \left(A\right)\oplus \Gamma \left(A^{*}\right)$  we introduce the bracket $\left[\right[,\left]\right]$  : for all $X_1+{\alpha }_1,X_2+{\alpha }_2\in \Gamma (A\oplus A^{*})$  , F. Petalidou and J.M. Nunes da Costa

$$\begin{array}{ccc}\left[\right[X_1+{\alpha }_1,X_2+{\alpha }_2\left]\right]& =& \left(\right[X_1,X_2{]}^{\phi }+{\mathcal{ℒ}}_{*{\alpha }_1}^W{X}_2-{\mathcal{ℒ}}_{*{\alpha }_2}^W{X}_1-d_{*}^W(e_1,e_2{)}_{-})+\end{array}$$

$$\begin{array}{ccc}& +& \left(\right[{\alpha }_1,{\alpha }_2{]}_{*}^W+{\mathcal{ℒ}}_{X_1}^{\phi }{\alpha }_2-{\mathcal{ℒ}}_{X_2}^{\phi }{\alpha }_1+d^{\phi }(e_1,e_2{)}_{-}).\end{array}$$

Also, we consider the bundle map $\rho :A\oplus A^{*}\to TM$  given by $\rho =a+a_{*}$  , i.e., for any $X+\alpha \in E$  , $\rho (X+\alpha )=a\left(X\right)+a_{*}\left(\alpha \right)$  . We have :

Theorem 3.1 ([14] ). If $\left(\right(A,\phi ),(A^{*},W\left)\right)$  is a generalized Lie bialgebroid over $M$  , then $A\oplus A^{*}$  endowed with $\left(\right[[,]],(,{)}_{+},\rho )$  and $\theta =\phi +W\in \Gamma \left(E^{*}\right)$  is a generalized Courant algebroid over $M$  . The operators $\mathcal{D}$  and ${\mathcal{D}}^{\theta }$  are, respectively, $\mathcal{D}=(d_{*}+d){|}_{C^{\infty }\left(M\right)}$  and ${\mathcal{D}}^{\theta }=(d_{*}^W+d^{\phi }){|}_{C^{\infty }\left(M\right)}$  .

Dirac structures for generalized Courant and Courant algebroids

4 Dirac structures of $\left(\right(A,\phi ),(A^{*},W\left)\right)$  and of $(\tilde{A},{\tilde{A}}^{*})$ 

Let $\left(\right(A,[,],a,\phi ),(A^{*},[,{]}_{*},a_{*},W))$  be a generalized Lie bialgebroid over $M$  and $(A\oplus A^{*},[[,]],(,{)}_{+},a+a_{*},\phi +W)$  the associated generalized Courant algebroid.

Definition 4.1. We say that a Dirac subbundle $L$  of $A\oplus A^{*}$  is reducible if the image $a\left(D\right)$  of its characteristic subbundle $D=L\cap A$  by $a$  defines a simple foliation $\mathcal{ℱ}$  of $M$  . By the term ”simple foliation” we mean that $\mathcal{ℱ}$  is a regular foliation such that the space $M/\mathcal{ℱ}$  is a nice manifold and the canonical projection $M\to M/\mathcal{ℱ}$  is a submersion.

Definition 4.2. Let $L$  be a Dirac subbundle of $A\oplus A^{*}$  . A function $f\in C^{\infty }\left(M\right)$  is called $L$  -admissible if there exists $Y_f\in \Gamma \left(A\right)$  such that $Y_f+d^{\phi }f\in \Gamma \left(L\right)$  . We denote by $C_L^{\infty }(M,\text{IR})$  the set of all $L$  -admissible functions of $C^{\infty }\left(M\right)$  .

Let $\left(\right(\tilde{A},[,{]}^{\stackrel{~}{}\phi },{\tilde{a}}^{\phi }),({\tilde{A}}^{*},[,{]}_{*}^{\hat{W}},{\hat{a}}_{*}^W\left)\right)$  be the Lie bialgebroid over $\tilde{M}$  defined by $\left(\right(A,[,],a,\phi ),(A^{*},[,{]}_{*},a_{*},W))$  as in Theorem  2.1 . Then, $\tilde{A}\oplus {\tilde{A}}^{*}$  endowed with :

(i) the two nondegenerate bilinear forms $(,{)}_{\pm }$  on $\tilde{A}\oplus {\tilde{A}}^{*}$  : for all ${\tilde{X}}_1+{\tilde{\alpha }}_1,{\tilde{X}}_2+{\tilde{\alpha }}_2\in \tilde{A}\oplus {\tilde{A}}^{*}$  , $({\tilde{X}}_1+{\tilde{\alpha }}_1,{\tilde{X}}_2+{\tilde{\alpha }}_2{)}_{\pm }=1/2(\langle {\tilde{\alpha }}_1,{\tilde{X}}_2\rangle \pm \langle {\tilde{\alpha }}_2,{\tilde{X}}_1\rangle )$  , (ii) the bracket $\left[\right[,]{]}^{\stackrel{~}{}}$  on $\Gamma (\tilde{A}\oplus {\tilde{A}}^{*})$  : for all ${\tilde{X}}_1+{\tilde{\alpha }}_1,{\tilde{X}}_2+{\tilde{\alpha }}_2\in \Gamma (\tilde{A}\oplus {\tilde{A}}^{*})$  , F. Petalidou and J.M. Nunes da Costa

$$\begin{array}{ccc}\left[\right[{\tilde{X}}_1+{\tilde{\alpha }}_1,{\tilde{X}}_2+{\tilde{\alpha }}_2]{]}^{\stackrel{~}{}}& =& \left([{\tilde{X}}_1,{\tilde{X}}_2{]}^{\stackrel{~}{}\phi }+{\widehat{\mathcal{ℒ}}}_{{\tilde{\alpha }}_1}^W{\tilde{X}}_2-{\widehat{\mathcal{ℒ}}}_{{\tilde{\alpha }}_2}^W{\tilde{X}}_1-{\hat{d}}_{*}^W(({\tilde{e}}_1,{\tilde{e}}_2{)}_{-})\right)+\end{array}$$

$$\begin{array}{ccc}& & \left([{\tilde{\alpha }}_1,{\tilde{\alpha }}_2{]}_{*}^{\hat{W}}+{\widetilde{\mathcal{ℒ}}}_{{\tilde{X}}_1}^{\phi }{\tilde{\alpha }}_2-{\widetilde{\mathcal{ℒ}}}_{{\tilde{X}}_2}^{\phi }{\tilde{\alpha }}_1+{\tilde{d}}^{\phi }(({\tilde{e}}_1,{\tilde{e}}_2{)}_{-})\right),\end{array}$$

(for any $\tilde{f}\in C^{\infty }\left(\tilde{M}\right)$  , ${\tilde{d}}^{\phi }\tilde{f}=\tilde{d}\tilde{f}+\frac{\partial \tilde{f}}{\partial t}\phi$  and ${\hat{d}}_{*}^W\tilde{f}=e^{-t}(\tilde{d}\tilde{f}+\frac{\partial \tilde{f}}{\partial t}\phi )$  , [6] ), (iii) the bundle map $\tilde{\rho }:\tilde{A}\oplus {\tilde{A}}^{*}\to T\tilde{M}$  , $\tilde{\rho }={\tilde{a}}^{\phi }+{\hat{a}}_{*}^W$  , is a Courant algebroid over $\tilde{M}$  ([10] ).

Let $\mathbf{E}:\Gamma (A\oplus A^{*})\to \Gamma (\tilde{A}\oplus {\tilde{A}}^{*})$  be the embedding of $\Gamma (A\oplus A^{*})$  into $\Gamma (\tilde{A}\oplus {\tilde{A}}^{*})$  defined, for any $X+\alpha \in \Gamma (A\oplus A^{*})$  , by $$\mathbf{E}(X+\alpha )=X+e^t\alpha ,$$  where $X$  and $\alpha$  are regarded as time-independent sections of $\tilde{A}$  and ${\tilde{A}}^{*}$  , respectively.

If $L$  is a subbundle of $A\oplus A^{*}$  , we write $\tilde{L}=\mathbf{E}\left(L\right)$  in order to denote the vector subbundle $\tilde{L}$  of $\tilde{A}\oplus {\tilde{A}}^{*}$  whose space of global cross sections is the image by $\mathbf{E}$  of the space of global cross sections of $L$  , i.e. $\Gamma \left(\tilde{L}\right)=\mathbf{E}(\Gamma (L\left)\right)$  .

Proposition 4.1. Let $L$  be a vector subbundle of $A\oplus A^{*}$  and $\tilde{L}=\mathbf{E}\left(L\right)$  .

Then, $L$  is a Dirac structure for the generalized Courant algebroid $(A\oplus A^{*},\phi +W)$  if and only if $\tilde{L}$  is a Dirac structure for the Courant algebroid $\tilde{A}\oplus {\tilde{A}}^{*}$  .

• Proof. It is easy to check that $\tilde{L}$  is a maximally isotropic subbundle of $(\tilde{A}\oplus {\tilde{A}}^{*},(,{)}_{+})$  if and only if $L$  is a maximally isotropic subbundle of $(A\oplus A^{*},(,{)}_{+})$  .
  Moreover, by a straightforward calculation we get that $$\left[\right[\mathbf{E}(X_1+{\alpha }_1),\mathbf{E}(X_2+{\alpha }_2)]{]}^{\stackrel{~}{}}=\mathbf{E}(\left[\right[X_1+{\alpha }_1,X_2+{\alpha }_2\left]\right]),\forall X_1+{\alpha }_1,X_2+{\alpha }_2\in \Gamma (L),$$  i.e. $\Gamma \left(\tilde{L}\right)$  is closed under $\left[\right[,]{]}^{\stackrel{~}{}}$  if and only if $\Gamma \left(L\right)$  is closed under $\left[\right[,\left]\right]$  .

Proposition 4.2. Let $L$  be a Dirac structure of $(A\oplus A^{*},\phi +W)$  and $\tilde{L}=\mathbf{E}\left(L\right)$  the associated Dirac structure of $\tilde{A}\oplus {\tilde{A}}^{*}$  . Then $\tilde{f}\in C^{\infty }\left(\tilde{M}\right)$  is a $\tilde{L}$  -admissible function if and only if $\tilde{f}=e^{t}f$  and $f\in C_L^{\infty }\left(M\right)$  .

• Dirac structures for generalized Courant and Courant algebroids Proof. Let $\tilde{f}\in C_{\tilde{L}}^{\infty }\left(\tilde{M}\right)$  , i.e. there exists $Y\in \Gamma \left(A\right)$  : $Y+{\tilde{d}}^{\phi }\tilde{f}\in \Gamma \left(\tilde{L}\right)$  . But, $Y+{\tilde{d}}^{\phi }\tilde{f}\in \Gamma \left(\tilde{L}\right)$  implies that there exists $\xi \in \Gamma \left(A^{*}\right)$  : $Y+\xi \in \Gamma \left(L\right)$  and $Y+{\tilde{d}}^{\phi }\tilde{f}=\mathbf{E}(Y+\xi )$  , thus ${\tilde{d}}^{\phi }\tilde{f}=e^t\xi$  . From Theorem of normal forms for Lie algebroids ([3] ) we have that, if the rank of $a\left(D\right)$  , $D=L\cap A$  , at a point $q\in M$  is $k$  , then we can construct on a neighborhood $U$  of $q$  in $M$  a system of local coordinates $(x_1,\dots ,x_k,\dots ,x_n)$  ( $n=dimM$  ) and a basis of sections $(X_1,\dots ,X_k,\dots X_r)$  of $\Gamma \left(A\right)$  ( $r$  is the dimension of the fibres of $A\to M$  ), with $(X_1,\dots ,X_k)$  sections of $\Gamma \left(D\right)$  , such that $a\left(X_i\right)=\frac{\partial }{\partial x_i}$  , for every $i=1,\dots ,k$  . Let $({\alpha }_1,\dots ,{\alpha }_k,\dots ,{\alpha }_r)$  be the basis of $\Gamma \left(A^{*}\right)$  , dual of $(X_1,\dots ,X_k,\dots X_r)$  . Since $\phi ,\xi \in \Gamma \left(A^{*}\right)$  , there exist ${\phi }_i,{\xi }_i\in C^{\infty }\left(U\right)$  , $i=1,\dots ,r$  , such that $\phi ={\sum }_{i=1}^r{\phi }_i{\alpha }_i$  and $\xi ={\sum }_{i=1}^r{\xi }_i{\alpha }_i$  . So, for any $i=1,\dots ,r$  ,

  $$\begin{array}{c}{\tilde{d}}^{\phi }\tilde{f}=e^t\xi \Rightarrow \langle \tilde{d}\tilde{f}+(\partial \tilde{f}/\partial t)\phi ,X_i\rangle =\langle e^t\xi ,X_i\rangle \iff \langle \tilde{d}\tilde{f},X_i\rangle +(\partial \tilde{f}/\partial t){\phi }_i=e^t{\xi }_i.\end{array}$$

  (4.1)

  But, for $i=1,\dots ,k$  , $\langle \tilde{d}\tilde{f},X_i\rangle =\langle \delta \tilde{f},\tilde{a}\left(X_i\right)\rangle =\langle \delta \tilde{f},a\left(X_i\right)\rangle =\langle \delta \tilde{f},\frac{\partial }{\partial x_i}\rangle =\frac{\partial \tilde{f}}{\partial x_i}$  .
  Hence, the last equation of ( 4.1 ) can be written, for any $i=1,\dots ,k$  , as

  $$\begin{array}{c}\partial \tilde{f}/\partial x_i+(\partial \tilde{f}/\partial t){\phi }_i=e^t{\xi }_i.\end{array}$$

  (4.2)

  By resolving the characteristic system $\frac{\delta x_i}{1}=\frac{\delta t}{{\phi }_i}=\frac{\delta \tilde{f}}{e^t{\xi }_i}$  of ( 4.2 ), we obtain that $\tilde{f}$  must be, at least locally, of the form $\tilde{f}=e^{t}f$  with $f\in C^{\infty }\left(U\right)$  . Taking into account Definition  4.2 and that $\tilde{L}=\mathbf{E}\left(L\right)$  , we get $\tilde{f}=e^{t}f\in C_{\tilde{L}}^{\infty }\left(\tilde{M}\right)\iff f\in C_L^{\infty }\left(M\right)$  .

Proposition 4.3. Let $L$  be a Dirac subbundle for $(A\oplus A^{*},\phi +W)$  and $\tilde{L}=\mathbf{E}\left(L\right)$  the associated Dirac subbundle of $\tilde{A}\oplus {\tilde{A}}^{*}$  . Then, $L$  is reducible if and only if $\tilde{L}$  is reducible.

• Proof. Let $D=L\cap A$  and $\tilde{D}=\tilde{L}\cap \tilde{A}$  be the characteristic subbundles of $L$  and $\tilde{L}$  , respectively, $\mathcal{ℱ}$  and $\widetilde{\mathcal{ℱ}}$  the foliations of $M$  and $\tilde{M}$  , respectively, defined by $a\left(D\right)$  and ${\tilde{a}}^{\phi }\left(\tilde{D}\right)$  , respectively. Obviously, $\tilde{D}\sim =D$  and ${\tilde{a}}^{\phi }\left(\tilde{D}\right)=\left\{{\tilde{a}}^{\phi }\right(X)/X\in D\}=\left\{a\right(X)+\langle \phi ,X\rangle \partial /\partial t/X\in D\}$  . Let $(x_0,t_0)$  be a point of $\tilde{M}=M\times \text{IR}$  , $\tilde{F}$  and $F$  the leaves of $\widetilde{\mathcal{ℱ}}$  and $\mathcal{ℱ}$  passing through $(x_0,t_0)\in \tilde{M}$  and $x_0\in M$  , respectively, and $D_{x_0}$  the fibre of $D$  over $x_0$  . By Theorem  2.2 , we have : (i) if $ker\left(a{|}_{D_{x_0}}\right)\mathbb{⊈}\langle \phi \left(x_0\right){\rangle }^{\circ }$  , then $\tilde{F}=F\times \text{IR}$  , so $dim\tilde{F}=dimF+1$  and the vector field $\partial /\partial t$  is tangent to $\tilde{F}$  ; (ii) if $ker\left(a{|}_{D_{x_0}}\right)\subseteq \langle \phi \left(x_0\right){\rangle }^{\circ }$  and ${\pi }_1:M\times \text{IR}\to M$  is the canonical projection, then ${\pi }_1\left(\tilde{F}\right)=F$  and ${\pi }_1{|}_{\tilde{F}}:\tilde{F}\to F$  is a covering map, thus $dim\tilde{F}=dimF$  and the vector field $\partial /\partial t$  is not tangent to $\tilde{F}$  . Since every $\tilde{L}$  -admissible function $\tilde{f}$  is of type $\tilde{f}=e^{t}f$  , $f\in C_L^{\infty }\left(M\right)$  , (Proposition  4.2 ) and also it is constant along the leaves of $\widetilde{\mathcal{ℱ}}$  ([1] ,[11] ), it is not possible the leaves $\tilde{F}$  of $\widetilde{\mathcal{ℱ}}$  to be of type $\tilde{F}=F\times \text{IR}$  (because, in this case, $\partial /\partial t$  is tangent to $\tilde{F}$  and $\tilde{f}=e^{t}f$  is not constant along $\partial /\partial t$  ). Thus, for any leaf $\tilde{F}$  of $\widetilde{\mathcal{ℱ}}$  and for the corresponding leaf $F$  of $\mathcal{ℱ}$  , we have : ${\pi }_1\left(\tilde{F}\right)=F$  and ${\pi }_1{|}_{\tilde{F}}:\tilde{F}\to F$  is a covering map. Hence, we get : (1) Every leaf $\tilde{F}$  of $\widetilde{\mathcal{ℱ}}$  is of the same dimension as the corresponding leaf $F$  of $\mathcal{ℱ}$  , so $\mathcal{ℱ}$  is a F. Petalidou and J.M. Nunes da Costa regular foliation of $M$  if and only if $\widetilde{\mathcal{ℱ}}$  is a regular foliation of $\tilde{M}$  . (2) $\widetilde{\mathcal{ℱ}}\sim =\mathcal{ℱ}$  , so $\tilde{M}/\widetilde{\mathcal{ℱ}}\sim =(M\times \text{IR})/\mathcal{ℱ}\sim =(M/\mathcal{ℱ})\times \text{IR}$  ; thus, $M/\mathcal{ℱ}$  is a nice manifold if and only if $\tilde{M}/\widetilde{\mathcal{ℱ}}$  is a nice manifold and the projection $M\to M/\mathcal{ℱ}$  is a submersion if and only if the projection $M\times \text{IR}=\tilde{M}\to \tilde{M}/\widetilde{\mathcal{ℱ}}\sim =(M/\mathcal{ℱ})\times \text{IR}$  is a submersion.
  Consequently, $L$  is a reducible Dirac subbundle for $A\oplus A^{*}$  if and only if $\tilde{L}=\mathbf{E}\left(L\right)$  is a reducible Dirac subbundle for $\tilde{A}\oplus {\tilde{A}}^{*}$  .

Let $L$  be a Dirac structure of $(A\oplus A^{*},\phi +W)$  and $\tilde{L}$  the associated Dirac structure of $\tilde{A}\oplus {\tilde{A}}^{*}$  . On $C_L^{\infty }\left(M\right)$  we define the bracket $\{,{\}}_L$  by setting, for all $f,g\in C_L^{\infty }\left(M\right)$  , $\{f,g{\}}_L:={\rho }^{\theta }(e_f)g$  , where $e_f=Y_f+d^{\phi }f\in \Gamma \left(L\right)$  . Also, on $C_{\tilde{L}}^{\infty }\left(\tilde{M}\right)$  we define the bracket $\{,{\}}_{\tilde{L}}$  by setting, for all $\tilde{f},\tilde{g}\in C_{\tilde{L}}^{\infty }\left(\tilde{M}\right)$  , $\tilde{f}=e^{t}f$  , $\tilde{g}=e^{t}g$  with $f,g\in C_L^{\infty }\left(M\right)$  , $\{\tilde{f},\tilde{g}{\}}_{\tilde{L}}:=\tilde{\rho }({\tilde{e}}_{\tilde{f}})\tilde{g}$  , where ${\tilde{e}}_{\tilde{f}}=Y_f+{\tilde{d}}^{\phi }\tilde{f}\in \Gamma \left(\tilde{L}\right)$  . By a straightforward calculation we get :

$$\begin{array}{c}\{\tilde{f},\tilde{g}{\}}_{\tilde{L}}=\{e^{t}f,e^{t}g{\}}_{\tilde{L}}=e^t\{f,g{\}}_L.\end{array}$$

(4.3)

Theorem 4.1 ([15] ). 1) If $1\in C_L^{\infty }\left(M\right)$  ^$\text{3}$  , then $\left(C_L^{\infty }\right(M),\{,{\}}_L)$  is a Jacobi algebra. 2) If $L$  is a reducible Dirac subbundle of $(A\oplus A^{*},\phi +W)$  and $1\in C_L^{\infty }\left(M\right)$  , then $L$  induces a Jacobi structure on $M/\mathcal{ℱ}$  defined by the Jacobi bracket $\{,{\}}_L$  .

Theorem 4.2. 1) If $1\in C_L^{\infty }\left(M\right)$  , then $\left(C_{\tilde{L}}^{\infty }\right(\tilde{M}),\{,{\}}_{\tilde{L}})$  is an homogeneous Poisson algebra with respect $\partial /\partial t$  ^$\text{4}$  . 2) If $L$  is a reducible Dirac subbundle of $(A\oplus A^{*},\phi +W)$  and $1\in C_L^{\infty }\left(M\right)$  , then $\tilde{L}$  induces an homogeneous Poisson structure on $\tilde{M}/\widetilde{\mathcal{ℱ}}$  defined by the homogeneous Poisson bracket $\{,{\}}_{\tilde{L}}$  . 3) $\tilde{M}/\widetilde{\mathcal{ℱ}}=(M/\mathcal{ℱ})\times \text{IR}$  and the induced homogeneous Poisson structure on $\tilde{M}/\widetilde{\mathcal{ℱ}}$  by $\tilde{L}$  is the Poissonization of the induced Jacobi structure on $M/\mathcal{ℱ}$  by $L$  .

• Proof. 1) It is checked by taking account ( 4.3 ) and the fact that, if $1\in C_L^{\infty }\left(M\right)$  , then $\left(C_L^{\infty }\right(M),\{,{\}}_L)$  is a Jacobi algebra. 2) By applying the results of [11] to the reducible Dirac subbundle $\tilde{L}$  and the homogeneous Poisson algebra $\left(C_{\tilde{L}}^{\infty }\right(\tilde{M}),\{,{\}}_{\tilde{L}})$  .
  3) We have $\widetilde{\mathcal{ℱ}}=\mathcal{ℱ}\times \left\{0\right\}$  ([15] ), thus $\tilde{M}/\widetilde{\mathcal{ℱ}}=(M/\mathcal{ℱ})\times \text{IR}$  , and by ( 4.3 ) we conclude the announced result.

References

1. T. Courant, Dirac manifolds, Trans. Amer. Math. Soc. 319 (1990), 631-661.
2. Dirac structures for generalized Courant and Courant algebroids P. Dazord, A. Lichnerowicz, C.-M. Marle, Structure locale des variétés de Jacobi, J. Math. Pures Appl. 70 (1991) 101-152.
3. J.-P. Dufour, Normal forms for Lie algebroids, in Lie Algebroids, Banach Center Publications, Vol. 54, Warszawa 2001, pp. 35-41.
4. J. Grabowski and G. Marmo, Jacobi structures revisited, J. Phys. A : Math. Gen. 34 (2001) 10975-10990.
5. J. Grabowski and G. Marmo, The graded Jacobi algebras and (co)homology, J. Phys. A : Math. Gen. 36 (2003) 161-181.
6. D. Iglesias and J.C. Marrero, Generalized Lie bialgebroids and Jacobi structures, J. Geom. Phys. 40 (2001) 176-200.
7. D. Iglesias and J.C. Marrero, Lie algebroid foliations and ${\mathcal{ℰ}}^1\left(M\right)$  -Dirac structures, J. Phys. A : Math. Gen. 35 (2002) 4085-4104.
8. Y. Kosmann-Schwarzbach, Exact Gerstenhaber algebras and Lie bialgebroids, Acta Appl. Math 41 (1995) 153-165.
9. A. Lichnerowicz, Les variétés de Jacobi et leurs de Lie associées, J. Math. pures et appl. 57 (1978) 453-488.
10. Z.-J. Liu, A. Weinstein, P. Xu, Manin triples for Lie bialgebroids, J. Diff. Geom. 45 (1997) 547-574.
11. Z.-J. Liu, A. Weinstein, P. Xu, Dirac Structures and Poisson Homogeneous Spaces, Commun. Math. Phys. 192 (1998) 121-144.
12. K. Mackenzie, Lie groupoids and Lie algebroids in differential geometry, London Math. Soc. Lecture notes series 124, Cambridge University Press, Cambridge 1987.
13. K. Mackenzie and P. Xu, Lie bialgebroids and Poisson groupoids, Duke Math. J. 73 (1994) 415-452.
14. J.M. Nunes da Costa and J. Clemente-Gallardo, Dirac structures for generalized Lie bialgebroids, J. Phys. A : Math. Gen. 37 (2004) 2671-2692.
15. F. Petalidou and J.M. Nunes da Costa, Reduction of Jacobi manifolds via Dirac structures theory, to appear in Diff. Geom. and its Applic.
16. A. Wade, Conformal Dirac structures, Lett. Math. Phys. 53 (2000) 331-348. F. Petalidou and J.M. Nunes da Costa

Fani Petalidou		Joana M. Nunes da Costa
Faculty of Sciences and Technology		Department of Mathematics
University of Peloponnese		University of Coimbra
22100 Tripoli, Greece		Apartado 3008
		3001-454 Coimbra, Portugal
e-mail : petalido@uop.gr		e-mail : jmcosta@mat.uc.pt