The Poisson Bracket
The Poisson bracket is how functions talk to each other on a symplectic manifold. Given f and g, {f, g} is a new function, and this operation satisfies three properties: it's bilinear, skew-symmetric, and satisfies the Jacobi identity. In that sense, the space of smooth functions C^\infty(M) with the bracket is a Lie algebra.
But it's more than just a Lie algebra. The bracket also satisfies the Leibniz rule: {f, gh} = {f, g}h + g{f, h}. This makes it a derivation in each argument. A Lie algebra with a compatible Leibniz rule — that's a Poisson algebra, and it's the algebraic structure underlying classical mechanics.
Concretely, {f, g} = X_f(g), where X_f is the Hamiltonian vector field defined by i_{X_f} omega = df. In Darboux coordinates, this works out to:
$${f, g} = \sum_{i=1}^{n} \left( \frac{\partial f}{\partial q^i} \frac{\partial g}{\partial p_i} - \frac{\partial f}{\partial p_i} \frac{\partial g}{\partial q^i} \right)$$
The fundamental brackets are {q^i, p_j} = delta^i_j, {q^i, q^j} = 0, {p_i, p_j} = 0. These are the relations that all the rest follow from. They encode the canonical commutation relations of classical mechanics, and when you replace {,} with (1/i\hbar)[,] and let \hbar go to zero, you get the transition to quantum mechanics.
The Poisson bracket tells you how observables evolve. If H is the Hamiltonian, then df/dt = {f, H}. This is the Heisenberg equation of motion in classical guise. A function f is conserved (a constant of motion) if and only if {f, H} = 0. Noether's theorem is encoded in this: symmetries of H produce conserved quantities, and conserved quantities commute with H under the bracket.
A symplectomorphism (a coordinate change preserving omega) preserves the Poisson bracket. If phi is a symplectomorphism, then {f o phi, g o phi} = {f, g} o phi. This means the bracket is an intrinsic geometric object, not dependent on coordinates. The bracket knows about omega's structure directly.
Poisson manifolds generalize symplectic manifolds by allowing omega to be degenerate. The bracket structure remains, but the inverse of omega is no longer a true bivector — it's a bivector field that may have a kernel. This is where the theory gets interesting, because the kernel can be non-trivial and carries its own geometry.
The Schouten-Nijenhuis bracket extends the Poisson bracket to multivector fields. The Poisson bivector omega^{ij} is a bivector field satisfying [omega^{#}, omega^{#}] = 0, where this bracket is the Schouten-Nijenhuis bracket. This is the geometric way of expressing the Jacobi identity.