Symmetry and conservation — Noether's theorem
Symmetry and conservation — Noether’s theorem
If the Lagrangian is invariant under a continuous transformation, exactly one conserved quantity drops out along every trajectory for each such transformation.
Opening
So far we have extracted the Euler–Lagrange equations from a Lagrangian and seen them as the flow of a vector field on a manifold. This chapter is about the one-line theorem — Emmy Noether’s 1918 result — that turns every symmetry of into a conserved quantity. After this chapter the reader should be able to explain in a single paragraph that conservation of linear momentum, angular momentum, and energy all fall out of the same machine, and should experience how a fact like “central-force motion is planar” — previously hard to motivate — is summed up by one word: symmetry.
Main 1 — Noether’s theorem
On the configuration space with coordinates , take a one-parameter transformation
where (epsilon) is a small real parameter and (xi) are smooth component functions. We call this transformation a symmetry of the Lagrangian when, to first order in ,
with .
Noether’s theorem. Under this hypothesis the quantity
is constant in time along any solution of the Euler–Lagrange equations: .
The proof fits in one paragraph. Expand to first order:
The left side vanishes by hypothesis. On the right, substitute the Euler–Lagrange equation on-shell (i.e. along a solution) into the first term; the two terms then collapse into a single total time derivative, namely . Hence . The entire symmetry conservation machine is contained in that line.
Main 2 — Three classical examples
The same theorem produces all the familiar conservation laws in one stroke.
(a) Spatial translation . When the Lagrangian does not depend on absolute position — as for a free particle or a closed many-body system — we have . Noether’s becomes , so the component of linear momentum along is conserved. Choosing three independent directions gives all of .
(b) Rotation about the axis . An infinitesimal rotation is , so . Noether’s is
where is the angular momentum. If is invariant under rotations about , then is conserved.
(c) Time translation . When has no explicit time dependence (), the same argument cannot be applied verbatim, but the same spirit yields conservation of
the Hamiltonian — in the standard case the energy of the system. This example is slightly different from (a) and (b): the transformation moves the time parameter rather than the coordinates, so it is cleaner to treat it as a symmetry of the action functional rather than of evaluated at a fixed time. The proper treatment is deferred to chapter 10.
Main 3 — Central force and symmetry
Consider a 3D particle in a central potential, with Lagrangian
Because depends only on , rotating by any leaves and unchanged. In other words, is invariant under every rotation. Applying example (b) to three independent choices of conserves all three components of angular momentum:
From this one line the planarity of Kepler orbits follows. Since is a fixed vector, at every time we have — so the motion is trapped in the plane perpendicular to . Even before the orbit is shown to be an ellipse, the fact that the motion lies in a plane is the direct corollary of rotational symmetry.
In Python
# Kepler problem (m=1, U=-1/r): verify L = r × v is conserved
# by integrating with hand-rolled RK4. Pass if relative drift < 1e-6.
import numpy as np
def accel(r): # gravitational acceleration (GM=1)
return -r / np.linalg.norm(r)**3
def rhs(state): # state = (r, v) in 6D
r, v = state[:3], state[3:]
return np.concatenate([v, accel(r)])
def rk4_step(s, dt): # standard 4th-order Runge-Kutta
k1 = rhs(s)
k2 = rhs(s + 0.5 * dt * k1)
k3 = rhs(s + 0.5 * dt * k2)
k4 = rhs(s + dt * k3)
return s + dt * (k1 + 2*k2 + 2*k3 + k4) / 6
r0 = np.array([1.0, 0.0, 0.0]) # initial position
v0 = np.array([0.0, 0.9, 0.0]) # initial velocity -> elliptic orbit
state = np.concatenate([r0, v0])
dt, T = 0.005, 10.0
N = int(round(T / dt))
checkpoints = {int(round(t / dt)): t for t in (0.0, 2.5, 5.0, 7.5, 10.0)}
L0 = np.linalg.norm(np.cross(r0, v0)) # initial |L|
for n in range(N + 1):
if n in checkpoints: # print |L| at five times
r, v = state[:3], state[3:]
L = np.linalg.norm(np.cross(r, v))
drift = abs(L - L0) / L0
print(f"t={checkpoints[n]:4.1f} |L|={L:.10f} rel.drift={drift:.2e}")
state = rk4_step(state, dt)
If matches its initial value to six decimals or better at all five checkpoints, then the Noether machine that turns rotational symmetry into a conservation law produces the same answer on a computer that it does on paper.
To the next chapter
Chapter 10: Hamiltonian mechanics treats the time-translation symmetry and energy conservation that we deferred in Main 2 (c). Introducing the Legendre transform from the Lagrangian to the Hamiltonian cleans up the picture of flows and conserved quantities on phase space in a single stroke. Noether’s theorem returns there in the company of the Poisson bracket.