First of all one comment: I know there is a question titled ''Lie bracket is not a vector field'' which proofs Lie bracket is not $C^\infty(M)$-linear.

Now, I show my question. In the paper Poisson-Nijenhuis Structures, F. Magri and Y. Kosmann-Schwarzbach (1990) (avaible on http://www.numdam.org/item?id=AIHPA_1990__53_1_35_0) one can find:

Let $\mu$ an $TM$-valued 2-form [i.e. a (1-2) tensor skew-symmetric in its covariant indexes] that defines a Lie algebra-structure on $TM$.

But they write the Lie bracket as


instead of $\mu(X,Y)$. In addition they distinguish between

$$N.\mu(X,Y) = \mu(N(X),Y)) + \mu(X,N(Y)) - N(\mu(X,Y))$$


$$[X,Y]_{N.\mu} = [N(X),Y] + [X,N(Y)] - N([X,Y])$$

when they define the ''deformed Lie bracket''.

So, there is a (''canonical'') way to induce a Lie bracket (being not $C^\infty(M)$-lienar) by means of a vectorial valued 2-form? For example, it is possible with Poisson bracket (setting Schouten-Nijenhuis bracket of the bivector with itself be zero).

Thank you


The solution can be found in the Lie algebroids.

Basically, it is a quatern $(M,E,[\cdot,\cdot],\Phi)$ where

  • $M$ is a manifold,

  • $E$ is a vector bundle over $M$,

  • $[\cdot,\cdot]$ is a Lie bracket on the $\mathbb R$-algebra of vector fields over $M$ and

  • $\Phi:E\rightarrow TM$ is a bundle morphism that also satisfy the Leibniz rule,

$$ [X,fY] = (\Phi(X))(f) Y + f[X,Y] , $$ for all vector fields $X,Y$ and for all differentiable function $f$ over the manifold.

This structure allows us to generalize the usual case of vector fields.

In addition, with this formalism, every vectorial-valued 2-form which satisfies Jacobi identity can define a Lie bracket (as an operation of a $\mathbb R$-algebra) setting $E=TM$ and $\Phi$ the identity map. In the paper, the authors set $\Phi=N$ the Nijenhuis operator associated to $\mu$ and as Lie bracket the "deformed" bracket $[\cdot,\cdot]_{N.\mu}$. It is $C^{\infty}$-linear, that is true, but the Leibniz rule is also satisfy because we have the morphism.

PD: My definition is essentially the same as in https://en.wikipedia.org/wiki/Lie_algebroid but including the manifold explicitly.

| cite | improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.