Continuum mechanics — the leap to infinite degrees of freedom

Send a chain of $N$ spring-coupled particles to the limit $N \to \infty$ and one line of Lagrangian density and one line of the wave equation drop out — the same variational principle, now extended to fields with infinitely many degrees of freedom.

Opening

So far we have handled systems with finite degrees of freedom (DOF) — a handful of particles, rigid bodies, coupled oscillators. This chapter climbs one more rung on that ladder. Push the particle count $N$ on a lattice to infinity, and the set of coordinates $\{q_n(t)\}$ turns into a field $\phi(t, x)$ over spacetime, while the Lagrangian generalizes to a spacetime integral. By the end of the chapter, the question “why do electromagnetism, general relativity, and quantum field theory all use the same variational principle?” has its first concrete answer.

Main 1 — From $N$ particles to a continuum

Take $N$ particles of mass $m$ on a line, joined by springs of stiffness $K$. Let $q_n(t)$ be the displacement of particle $n$ from its equilibrium position. The Lagrangian is

$$L = \sum_n \tfrac{1}{2} m \dot q_n^2 - \sum_n \tfrac{1}{2} K (q_{n+1} - q_n)^2.$$

Now perform two limits at once — let the inter-particle spacing $a \to 0$ and $N \to \infty$ — while fixing three quantities: the total length $L_0 = N a$, the mass per unit length $\rho = m/a$ (rho, linear density, units kg/m), and the tension $T = K a$ (units N). The displacement $q_n(t)$ is reinterpreted as the value $\phi(t, x)$ of a continuous field at the position $x = n a$. Finite differences become derivatives — $q_{n+1} - q_n \to a\, \partial_x \phi$. Sums become integrals — $\sum_n \to \int dx / a$. Rewriting both the kinetic and potential terms gives

$$L \to \int dx \left[ \tfrac{1}{2}\rho \dot\phi^2 - \tfrac{1}{2} T (\partial_x \phi)^2 \right].$$

The integrand is the Lagrangian density $\mathcal{L}$ (calligraphic L):

$$\mathcal{L} = \tfrac{1}{2}\rho \dot\phi^2 - \tfrac{1}{2} T (\partial_x \phi)^2.$$

The action is now an integral over both time and space,

$$S = \int dt\, dx\, \mathcal{L}\!\left(\phi, \partial_t \phi, \partial_x \phi\right),$$

which in 3+1 dimensions is written compactly as $S = \int \mathcal{L}\, d^4 x$. The field $\phi(t, x)$ carries one independent DOF per point $x$ — a dynamical system with infinitely many DOF.

Main 2 — Euler–Lagrange in field form

Apply the same variational procedure used for discrete systems. Perturb $\phi \to \phi + \delta\phi$ inside a spacetime region, with $\delta\phi$ vanishing on the boundary of that region. Requiring $\delta S = 0$ yields

$$\frac{\partial}{\partial t}\!\left(\frac{\partial \mathcal{L}}{\partial \dot\phi}\right) + \partial_x\!\left(\frac{\partial \mathcal{L}}{\partial(\partial_x \phi)}\right) - \frac{\partial \mathcal{L}}{\partial \phi} = 0.$$

The shape mirrors the discrete Euler–Lagrange equation — one time derivative has been replaced by a sum of a time derivative and a space derivative. For the elastic-string $\mathcal{L}$ above, $\partial \mathcal{L}/\partial \dot\phi = \rho \dot\phi$, $\partial \mathcal{L}/\partial(\partial_x \phi) = -T\, \partial_x \phi$, and $\partial \mathcal{L}/\partial \phi = 0$, so

$$\rho \ddot\phi - T\, \phi'' = 0,$$

i.e. the wave equation

$$\ddot\phi = c^2\, \phi'', \qquad c = \sqrt{T/\rho}.$$

For a string with fixed ends ($\phi(0, t) = \phi(L_0, t) = 0$), separation of variables gives the standing-wave modes

$$\phi_n(t, x) = \sin(n\pi x/L_0)\cos(\omega_n t), \qquad \omega_n = n\pi c / L_0.$$

The normal modes of the discrete lattice merge smoothly into an infinite family of standing waves, labelled by the integer $n$, as $N \to \infty$.

Main 3 — Why this single step is a big deal

The same variational principle, the same Noether logic, now applies verbatim to a field with infinitely many DOF. As the next chapter (classical field theory) makes precise: write down a single object — the Lagrangian density — and the equations of motion, the conserved currents, and the symmetries fall out automatically. The power of the framework is that swapping $\mathcal{L}$ alone gives entirely different physics on the same underlying machine.

Electromagnetism uses $\mathcal{L} = -\tfrac{1}{4} F_{\mu\nu} F^{\mu\nu}$ (where $F_{\mu\nu}$ is the electromagnetic field-strength tensor, an antisymmetric rank-2 tensor on 4D spacetime that packages the electric and magnetic fields together); gauge theory, general relativity, and quantum field theory all sit on top of this single template. Chapter 6 extends the framework to relativistic scalar fields and the Klein–Gordon equation.

In Python

# 1D elastic string with fixed ends, integrated by finite-difference leapfrog.
# An initial Gaussian pulse splits left/right and reflects off the boundaries.
import numpy as np
import matplotlib.pyplot as plt

L, c = 1.0, 1.0               # string length and wave speed
N = 200                       # grid points
dx = L / (N - 1)
dt = 0.5 * dx / c             # CFL stability
Nt = 2000
x = np.linspace(0, L, N)
r2 = (c * dt / dx) ** 2       # squared Courant number

# Initial condition: centred Gaussian, zero initial velocity
phi_prev = np.exp(-((x - L / 2) ** 2) / (2 * 0.05 ** 2))
phi_prev[0] = phi_prev[-1] = 0.0
phi = phi_prev.copy()         # zero initial velocity, so one step ahead matches

snap_steps = np.linspace(0, Nt - 1, 5, dtype=int)
snapshots = []
for n in range(Nt):
    phi_next = np.zeros_like(phi)
    phi_next[1:-1] = (2 * phi[1:-1] - phi_prev[1:-1]
                      + r2 * (phi[2:] - 2 * phi[1:-1] + phi[:-2]))
    phi_prev, phi = phi, phi_next
    if n in snap_steps:
        snapshots.append((n, phi.copy()))

fig, ax = plt.subplots(figsize=(7, 3))
for n, snap in snapshots:
    ax.plot(x, snap, label=f"t = {n*dt:.2f}")
ax.set(xlabel="x", ylabel="phi(t, x)", title="Wave on a string: five snapshots")
ax.legend(fontsize=8)
plt.tight_layout()

When the pulse splits into two counter-propagating halves that reflect with inverted sign at the fixed ends, the equation of motion of an infinite-DOF system has been solved by hand.

To the next chapter

Chapter 6: classical field theory extends the scalar field of this chapter to the relativistic setting and treats the Klein–Gordon equation, the Lorentz invariance of $\mathcal{L}$, and Noether’s theorem for fields in one sweep. The string equation written here is exactly the non-relativistic shadow of what we’ll meet under rotations of spacetime.