Schouten-Nijenhuis Bracket

1. Definition & Axioms

The Schouten–Nijenhuis bracket is the unique extension of the Lie bracket to the graded algebra of multivector fields $X^{∙} (M) = ⨁_{k = 0}^{n} X^{k} (M)$ .

For multivectors $A \in X^{p} (M)$ and $B \in X^{q} (M)$ , the bracket $[A, B]$ results in a multivector of degree $p + q - 1$ . It is defined by the following axioms:

Graded Commutativity: $[A, B] = - (- 1)^{(p - 1) (q - 1)} [B, A]$
Graded Leibniz Rule: $[A, B \land C] = [A, B] \land C + (- 1)^{(p - 1) q} B \land [A, C]$
Jacobi Identity: $(- 1)^{(p - 1) (r - 1)} [A, [B, C]] + cyclic = 0$

2. Geometric Intuition

While the Lie bracket $[X, Y]$ measures the failure of two flows to commute, the Schouten–Nijenhuis bracket measures the "compatibility" of higher-dimensional "flows" (volumes or planes) generated by multivectors.
If you view a $p$ -vector as a local $p$ -dimensional subspace, the bracket tells you how one subspace is pushed along the flow of the other (think of "multidimensional flows").

Particular cases:

Generalized Lie Derivative: If $X$ is a vector field ( $p = 1$ ), then $[X, B] = L_{X} B$ . It measures how the multivector $B$ "evolves" along the flow of $X$ .
Poisson Consistency: A bivector $π$ is Poisson if $[π, π] = 0$ . Geometrically, this ensures that the "local planes" defined by $π$ satisfy the Jacobi identity, allowing the manifold to be foliated by symplectic leaves.
Frobenius & Integrability. A $k$ -dimensional distribution $D$ can be locally represented by a $k$ -vector $P = X_{1} \land \dots \land X_{k}$ .
- Frobenius Condition: Recall, $D$ is integrable iff $[X_{i}, X_{j}] \in span {X_{1}, \dots, X_{k}}$ .
- SN-Bracket Version: $D$ is integrable iff $[P, P]_{S N} = Q \land P$ for some multivector $Q$ .
  This generalizes the Lie bracket test by evaluating the "self-compatibility" of the $k$ -dimensional volume element $P$ .

3. Local Coordinate Formula

In a local coordinate system $(x^{1}, \dots, x^{n})$ , let:

A = \frac{1}{p!} A^{i_{1} \dots i_{p}} \partial_{i_{1}} \land \dots \land \partial_{i_{p}}

B = \frac{1}{q!} B^{j_{1} \dots j_{q}} \partial_{j_{1}} \land \dots \land \partial_{j_{q}}

The components of the bracket $[A, B]$ are given by:

[A, B]^{k_{1} \dots k_{p + q - 1}} = \sum_{σ} sgn (σ) (\frac{1}{(p - 1)! q!} A^{i k_{1} \dots k_{p - 1}} \partial_{i} B^{k_{p} \dots k_{p + q - 1}} - \frac{1}{p! (q - 1)!} B^{i k_{1} \dots k_{q - 1}} \partial_{i} A^{k_{q} \dots k_{p + q - 1}})

Simple Case: Bivector with itself
For a bivector $π = \frac{1}{2} π^{i j} \partial_{i} \land \partial_{j}$ , the condition $[π, π] = 0$ reduces to:

π^{i l} \partial_{l} π^{j k} + π^{j l} \partial_{l} π^{k i} + π^{k l} \partial_{l} π^{i j} = 0