next up previous contents
Next: Compatible Poisson tensors Up: Schouten-Nijenhuis bracket Previous: Lichnerowicz definition of the   Contents

Jacobi condition and Schouten-Nijenhuis bracket

Let $\Pi$ be a bivector on $M$, so $[\Pi, \Pi]\in\gX^3 M$. Let $\om$ be a 3-form on $M$. Then $\bracket{\om}{[\Pi, \Pi]}$ is a function and

\begin{displaymath}
\bracket{\om}{[\Pi, \Pi]}=-\bracket{d(i_{\Pi}\om)}{\Pi}-
\bracket{d(i_{\Pi}\om)}{\Pi} + \bracket{d\om}{\Pi\wedge\Pi}.
\end{displaymath}

Let $\om=df\wedge dg\wedge dh$. Put $\{f,g\}=\bracket{df\wedge dg}{\Pi}$. Remark that

\begin{displaymath}
\bracket{i_{\Pi}\om}{X}=\bracket{df\wedge dg\wedge dh}{\Pi\wedge X}
\end{displaymath}

for all $X\in\gX^1 M$. Since $d(df\wedge dg\wedge dh)=0$ we have

\begin{displaymath}
\bracket{\om}{[\Pi, \Pi]}=-2\bracket{d(i_{\Pi}\om)}{\Pi}.
\end{displaymath}


\begin{lem}
\begin{displaymath}
\bracket{df\wedge dg\wedge dh}{[\Pi, \Pi]}=\pm 2\Jac(f,g,h)
\end{displaymath}\end{lem}

\begin{proof}
\begin{displaymath}
\bracket{df\wedge dg\wedge dh}{[\Pi, \Pi]}=
-...
...end{displaymath}\begin{displaymath}
= 2\Jac(f,g,h).
\end{displaymath}\end{proof}

\begin{cor}
A bivector $\Pi\in\gX^2 M$\ is Poisson if and only if $[\Pi, \Pi]=0$.
\end{cor}
As an application of corollary ([*]) consider a Lie algebra $\gerg$, manifold $M$, $\xi\:\gerg\to\gX(M)$ an infinitesimal map (a Lie algebra homomorphism). Then $\xi$ extends uniquely to a degree 0 map
\begin{align*}
\wedge \xi\: \La^{\bullet}\gerg &\to \gX^{\bullet}(M) \\
x_1\wyw x_n &\mapsto \xi(x_1)\wyw \xi(x_n)
\end{align*}
which preserves the graded brackets

\begin{displaymath}
\wedge \xi[\al,\bt]=[\wedge \xi(\al), \wedge\xi(\bt)]_{SN}
\end{displaymath}

This is a simple consequence of the fact that both brackets are determined by their values in degree 0 and 1.

Another easy consequence of characterization $[\Pi, \Pi]=0$ is that if $\dim M=2$, then any bivector $\Pi\in\gX^2 M$ is Poisson. Indeed $[\Pi, \Pi]\in\gX^3 M=0$.

Let now $G$ be a connected Lie group such that $\Lie(G)=\gerg$ (not necessarily simply connected). Let us denote the translation operators by
\begin{align*}
L_g\: G\to G, &\quad h\mapsto gh, \\
R_g\: G\to G, &\quad h\mapsto hg.
\end{align*}
We will use the following notations for the tangent maps $T_eG\to T_gG$.
\begin{align*}
L_{g,*} &\: T_h G\to T_{gh} G & R_{g,*} &\: T_h G\to T_{hg} G\\
...
...
& R_{g,*}^{\wedge} &\: \La^{\bullet}T_e G\to \La^{\bullet}T_{g} G
\end{align*}

Let now $\al\in\La^{\bullet}\gerg$. We will denote with $\al^L$ (resp. $\al^R$) the left (resp. right) invariant multivector field on $G$ whose value at $e\in G$ (identity of $G$) is $\al$ i.e.

\begin{displaymath}
\al^L(g):=L^{\wedge}_{g, *}\al,\word{(resp.} \al^R(g):=R^{\wedge}_{g, *}\al)
\end{displaymath}

In the same way we consider $\al^R$ to be the right invariant multivector field on $G$ whose value at $e$ is $\al$.
\begin{prop}
For any $\ga\in\La^2 \gerg$\ the following are equivalent
\begin{en...
...item $[\ga, \ga]=0$\ (bracket in $\La^{\bullet}\gerg$)
\end{enumerate}\end{prop}

\begin{proof}
$L_{X, g}\:\gerg\to \gX(G)$\ is an infinitesimal action.
\begin{di...
...he computation for right invariant
multivectors is exactly the same.
\end{proof}
The condition $[\ga, \ga]=0$ is called classical Yang-Baxter equation. The above proposition can be stated as:
\begin{cor}
There is a one to one correspondence between left (resp. right)
inva...
...oup $G$\ and solutions
of classical Yang-Baxter equation on $\Lie(G)$.
\end{cor}

next up previous contents
Next: Compatible Poisson tensors Up: Schouten-Nijenhuis bracket Previous: Lichnerowicz definition of the   Contents
Pawel Witkowski 2006-06-26