Do Classical Observables Form a Lie Algebra Under the Poisson Bracket?

Yes. In Hamiltonian mechanics, smooth real-valued functions on phase space are classical observables, and the Poisson bracket gives them a Lie algebra structure. The key point is that the bracket—not ordinary multiplication—is the operation used as the Lie bracket.

For a canonical phase space with coordinates $(q_1,\ldots,q_n,p_1,\ldots,p_n)$, the standard convention is

$$\{f,g\}=\sum_{i=1}^n\left(\frac{\partial f}{\partial q_i}\frac{\partial g}{\partial p_i}-\frac{\partial f}{\partial p_i}\frac{\partial g}{\partial q_i}\right).$$

This bracket captures how observables interact under Hamiltonian motion. The three defining Lie-algebra properties follow from the properties of the Poisson bracket.

What makes it a Lie algebra?

A Lie algebra is a vector space with a bilinear operation that is antisymmetric and satisfies the Jacobi identity. For smooth observables $f,g,h$, the Poisson bracket has exactly these properties:

• Bilinearity: $\{af+bg,h\}=a\{f,h\}+b\{g,h\}$ for real numbers $a,b$, and likewise in the second argument.
• Antisymmetry: $\{f,g\}=-\{g,f\}$.
• Jacobi identity:

$$\{f,\{g,h\}\}+\{g,\{h,f\}\}+\{h,\{f,g\}\}=0.$$

Thus, if the observables are taken to be smooth functions on a Poisson manifold (in particular, a symplectic phase space), they form a Lie algebra over $\mathbb{R}$ under the Poisson bracket. The Jacobi identity is essential: antisymmetry and bilinearity alone would not be enough.

A canonical-coordinate example

For one degree of freedom, the bracket is

$$\{f,g\}=\frac{\partial f}{\partial q}\frac{\partial g}{\partial p}-\frac{\partial f}{\partial p}\frac{\partial g}{\partial q}.$$

Take $f=q^2p$ and $g=p^2$. Their derivatives are

$$\frac{\partial f}{\partial q}=2qp,\qquad \frac{\partial f}{\partial p}=q^2, \qquad \frac{\partial g}{\partial q}=0,\qquad \frac{\partial g}{\partial p}=2p.$$

Substitution gives

$$\{q^2p,p^2\}=(2qp)(2p)-(q^2)(0)=4qp^2.$$

The result is another smooth function on phase space, as expected. In particular, the coordinate functions satisfy $\{q,p\}=1$ and $\{q,q\}=\{p,p\}=0$.

For a Hamiltonian $H$, one common convention writes the time evolution of an observable $f$ as $\dot f=\{f,H\}$. For example, if $H=p^2/(2m)+V(q)$, then $\dot q=\{q,H\}=p/m$. Some texts define the Hamiltonian vector field or order the bracket arguments differently, which can change signs in related formulas; it does not change the Lie-algebra properties.

Constants and the center

A constant function has zero derivatives, so

$$\{c,f\}=0$$

for every constant $c$ and every observable $f$. Constants therefore belong to the center of this Lie algebra: they have zero bracket with all its elements.

On a connected symplectic phase space, these are the only functions that bracket to zero with every smooth function. Consequently, the map from an observable to its Hamiltonian vector field identifies observables that differ by a constant, up to the sign convention used to define that vector field. This is why Hamiltonian vector fields are often described as corresponding to functions modulo constants. If phase space is disconnected, the corresponding functions may be constant separately on each connected component.

On a general Poisson manifold, there can be additional functions that bracket to zero with every function. Such functions are called Casimir functions; they are central too.

Lie algebra versus Poisson algebra

The observables also have the ordinary pointwise product $fg$. That product is commutative and associative, but it is not the Lie bracket. When a space of functions has both pointwise multiplication and a Poisson bracket, with the bracket obeying the product rule

$$\{f,gh\}=\{f,g\}h+g\{f,h\},$$

the combined structure is called a Poisson algebra. So it is accurate to say that classical observables form a Lie algebra under the Poisson bracket, and more fully that they form a Poisson algebra when their usual product is included.

Common mistakes

• Confusing the two operations: The Lie bracket is $\{f,g\}$, not the pointwise product $fg$.
• Leaving out the Jacobi identity: It is one of the defining conditions for a Lie algebra.
• Assuming only constants are central in every setting: That is true for a connected symplectic phase space, but a general Poisson manifold can have nonconstant Casimir functions.
• Treating sign conventions as contradictions: Authors may define Hamiltonian vector fields or the order of time-evolution brackets differently. Check the stated convention before comparing signs.