Field Lagrangians and the Continuum Limit: Elastic Media and the Wave Equation as a Field Theory

Intuition Beginner

Picture a guitar string. At rest it is a straight line. Pluck it and a wave travels along the string, bounces off the fixed ends, and produces the note you hear. A guitar string is not a single particle -- it is a continuous object, a one-dimensional medium with mass spread along its length. To describe its motion you need a field: a quantity defined at every point along the string at every moment in time. The field is the transverse displacement $u(x,t)$ -- how far the string has moved from its rest position at location $x$ and time $t$.

The Lagrangian framework extends naturally from particles to fields. For a single particle the Lagrangian $L=T-V$ is a function of position, velocity, and time. For a field you replace the particle Lagrangian with a Lagrangian density -- the Lagrangian per unit length (or per unit area, or per unit volume, depending on the dimension). The total Lagrangian is the integral of the density over the whole system.

But how do you get from particles to fields? The bridge is a chain of coupled oscillators. Imagine $N$ point masses $m$ connected by springs of constant $k$, spaced a distance $a$ apart, fixed at both ends. Each mass can move up and down (transversely). The displacement of the $i$-th mass from equilibrium is $q_i(t)$. This is a system with $N$ degrees of freedom, and its Lagrangian is obtained by adding up each mass's kinetic energy $\frac{1}{2}m{\dot{q}}_i^2$ minus the spring potential energy $\frac{1}{2}k(q_{i+1}-q_i{)}^2$ between neighbours, for all $N$ masses.

The Euler-Lagrange equations for this system give $N$ coupled ODEs. When $N=3$ or $N=5$ you can solve them directly -- this is the normal-mode problem studied in 09.02.04. But real strings have $N\sim 10^{23}$ atoms. The only tractable description treats the string as a continuum.

Take the limit $N\to \infty$ while keeping the total mass $M=Nm$ and total length $\ell =Na$ fixed. The spacing $a\to 0$, each individual mass $m\to 0$, but the mass density $\mu =M/\ell =m/a$ stays constant. The discrete index $i$ becomes a continuous position $x=ia$. The displacement $q_i(t)$ becomes the field $u(x,t)$. The finite difference $q_{i+1}-q_i$ becomes $a$ times the spatial rate of change of $u$.

Adding up over all masses becomes summing over the whole length of the string. The total Lagrangian $L$ is obtained by summing the Lagrangian density $\mathcal{L}$ over the whole string from $x=0$ to $x=\ell$. The density depends on the displacement $u$, its spatial rate of change, its velocity $\dot{u}$, position $x$, and time $t$.

where $\mathcal{L}$ is the Lagrangian density. For the vibrating string, this density is $\mathcal{L}=\frac{1}{2}\mu {\dot{u}}^2-\frac{1}{2}\kappa (u^{\prime }{)}^2$, where $\kappa =ka$ is the elastic modulus and $u^{\prime }$ is the spatial rate of change of $u$.

The Euler-Lagrange equations then become partial differential equations instead of ordinary ones. For the vibrating string the field equation is the wave equation:

$$\mu \ddot{u}=\kappa u^{\prime \prime },$$

where $\ddot{u}$ is the acceleration of the string at each point and $u^{\prime \prime }$ is its curvature in space. This single equation describes all wave motion on the string. The wave speed is $c=\sqrt{\kappa /\mu }$.

Visual Beginner

Figure: A vibrating string fixed at both ends. The displacement field $u(x,t)$ is shown as a transverse wave with a single antinode at the centre. Below the string, a row of $N$ point masses connected by springs illustrates the discrete-to-continuum transition: as $N$ grows and the spacing shrinks, the chain becomes the continuous string.

Worked example Beginner

An elastic string of length $\ell$ and mass per unit length $\mu$ is fixed at both ends and stretched with tension $\tau$. At rest the string lies along the $x$-axis. When plucked, the transverse displacement is $u(x,t)$.

The kinetic energy of a small piece of string of length $dx$ is $\frac{1}{2}(\mu dx){\dot{u}}^2$. The potential energy stored in the stretching of that piece is approximately $\frac{1}{2}\tau (u^{\prime }{)}^{2}dx$ for small displacements (the extra length due to the slope is $\sqrt{1+(u^{\prime }{)}^2}dx-dx\approx \frac{1}{2}(u^{\prime }{)}^{2}dx$, multiplied by the tension $\tau$). Adding up all the pieces along the string gives the total kinetic energy $T$ (summing $\frac{1}{2}\mu {\dot{u}}^2$ over the string) and total potential energy $V$ (summing $\frac{1}{2}\tau (u^{\prime }{)}^2$ over the string).

The Lagrangian is $L=T-V$, which is the integral of the Lagrangian density:

$$\mathcal{L}=\frac{1}{2}\mu {\dot{u}}^2-\frac{1}{2}\tau (u^{\prime }{)}^2.$$

For a string of length $\ell =1$ m, mass density $\mu =0.01$ kg/m, and tension $\tau =100$ N, the wave speed is $c=\sqrt{\tau /\mu }=100$ m/s. The fundamental frequency is $c/(2\ell )=50$ Hz -- a low A note. The general solution is $u(x,t)=f(x-ct)+g(x+ct)$ for arbitrary waveforms $f$ and $g$ travelling right and left.

Check your understanding Beginner

Formal definition Intermediate+

The discrete chain and the continuum limit

Consider $N$ masses $m$ connected by springs of constant $k$ and equilibrium spacing $a$. The displacement of the $i$-th mass from equilibrium is $q_i(t)$. The Lagrangian is

$$L=\sum _{i=1}^N\left[\frac{1}{2}m{\dot{q}}_i^2-\frac{1}{2}k(q_{i+1}-q_i{)}^2\right].$$

The Euler-Lagrange equations give the equations of motion $m{\ddot{q}}_i=k(q_{i+1}-2q_i+q_{i-1})$, a system of $N$ coupled ODEs. This is exactly the normal-mode system of 09.02.04.

Now pass to the continuum. Define the mass density $\mu =m/a$, the elastic modulus $\kappa =ka$, the displacement field $u(x_i,t)=q_i(t)$, and the spatial coordinate $x_i=ia$. The finite difference becomes $q_{i+1}-q_i\approx a\partial u/\partial x$. Substituting into the Lagrangian:

$$L=\sum _{i=1}^{N}a\left[\frac{1}{2}\mu {\dot{u}}^2-\frac{1}{2}\kappa {\left(\frac{\partial u}{\partial x}\right)}^2\right]\stackrel{a\to 0}{\to }{\int }_0^{\ell }\mathcal{L}\left(u,\frac{\partial u}{\partial x},\dot{u},x,t\right)dx,$$

where the Lagrangian density is

$$\boxed{\mathcal{L}=\frac{1}{2}\mu {\dot{u}}^2-\frac{1}{2}\kappa {\left(\frac{\partial u}{\partial x}\right)}^2.}$$

The action is the integral of the Lagrangian density over space and time:

$$S[u]={\int }_{t_0}^{t_1}{\int }_0^{\ell }\mathcal{L}dxdt.$$

The field $u(x,t)$ replaces the finite set of coordinates $q_i(t)$. Where particle mechanics has a sum over particles, field theory has an integral over space. Where particle mechanics has ordinary derivatives ${\dot{q}}_i$, field theory has partial derivatives $\partial u/\partial t$ and $\partial u/\partial x$.

Euler-Lagrange equations for fields

For a general Lagrangian density $\mathcal{L}(\varphi ,{\partial }_{\mu }\varphi ,x^{\mu })$ depending on a field $\varphi$ and its first derivatives (using the spacetime index notation $\mu =0,1,\dots ,n$ with $x^0=t$), the action is $S[\varphi ]=\int \mathcal{L}{d}^{n+1}x$ (integrating over space and time). Requiring $\delta S=0$ for arbitrary compactly-supported variations $\delta \varphi$ gives the Euler-Lagrange field equations:

$$\boxed{\frac{\partial \mathcal{L}}{\partial \varphi }-\frac{\partial }{\partial x^{\mu }}\frac{\partial \mathcal{L}}{\partial ({\partial }_{\mu }\varphi )}=0.}$$

The sum over $\mu$ is implied. In 1+1 dimensions with coordinates $(t,x)$ and a single scalar field $u$, this reads

$$\frac{\partial \mathcal{L}}{\partial u}-\frac{\partial }{\partial t}\frac{\partial \mathcal{L}}{\partial \dot{u}}-\frac{\partial }{\partial x}\frac{\partial \mathcal{L}}{\partial u^{\prime }}=0.$$

The derivation follows the same integration-by-parts argument as the particle Euler-Lagrange equations 09.02.02, but the single time integration by parts is supplemented by an integration by parts in each spatial direction.

For the string Lagrangian density $\mathcal{L}=\frac{1}{2}\mu {\dot{u}}^2-\frac{1}{2}\kappa (u^{\prime }{)}^2$: $\partial \mathcal{L}/\partial u=0$, $\partial \mathcal{L}/\partial \dot{u}=\mu \dot{u}$, $\partial \mathcal{L}/\partial u^{\prime }=-\kappa u^{\prime }$. The Euler-Lagrange equation gives

$$\mu \frac{{\partial }^{2}u}{\partial t^2}-\kappa \frac{{\partial }^{2}u}{\partial x^2}=0,$$

the wave equation with speed $c=\sqrt{\kappa /\mu }$.

The Klein-Gordon Lagrangian

Adding a restoring force (a potential per unit length $\frac{1}{2}\alpha u^2$) to the string modifies the Lagrangian density:

$$\mathcal{L}=\frac{1}{2}\mu {\dot{u}}^2-\frac{1}{2}\kappa (u^{\prime }{)}^2-\frac{1}{2}\alpha u^2.$$

The Euler-Lagrange equation becomes

$$\mu \ddot{u}-\kappa u^{\prime \prime }+\alpha u=0.$$

Setting $\mu =1$, $\kappa =c^2$, and $\alpha =m^2$ and switching to relativistic notation with ${\partial }_{\mu }=({\partial }_t,{\partial }_x)$:

$$\ddot{u}-c^2{u}^{\prime \prime }+m^{2}u=0⟷({\partial }_{\mu }{\partial }^{\mu }+m^2)\varphi =0.$$

This is the Klein-Gordon equation -- the relativistic wave equation for a scalar field of mass $m$. The Lagrangian density that produces it is

$${\mathcal{L}}_{\mathrm{KG}}=\frac{1}{2}{\partial }_{\mu }\varphi {\partial }^{\mu }\varphi -\frac{1}{2}{m}^2{\varphi }^2.$$

The elastic string is the massless case $m=0$. The massive case introduces a characteristic frequency ${\omega }_0=\sqrt{\alpha /\mu }=m$ and a dispersion relation ${\omega }^2=c^2{k}^2+m^2$ that departs from the linear (non-dispersive) behaviour of the plain string.

Higher-dimensional fields

In $d$ spatial dimensions the Lagrangian density depends on $\varphi$ and $\nabla \varphi$, and the Euler-Lagrange equation becomes

$$\frac{\partial \mathcal{L}}{\partial \varphi }-\nabla \cdot \frac{\partial \mathcal{L}}{\partial (\nabla \varphi )}-\frac{\partial }{\partial t}\frac{\partial \mathcal{L}}{\partial \dot{\varphi }}=0.$$

For an elastic membrane (2D drumhead) with displacement $u(x,y,t)$, surface tension $\sigma$, and mass per unit area $\mu$:

$$\mathcal{L}=\frac{1}{2}\mu {\dot{u}}^2-\frac{1}{2}\sigma |\nabla u{|}^2,$$

giving the two-dimensional wave equation $\mu \ddot{u}=\sigma {\nabla }^{2}u$.

The elastic rod: longitudinal vibrations

An elastic rod of cross-section $A$, Young's modulus $E$, and density $\rho$ undergoes longitudinal displacement $u(x,t)$. The strain is $ϵ=\partial u/\partial x$ and the stress is $\sigma =Eϵ$. The Lagrangian density is

$$\mathcal{L}=\frac{1}{2}\rho A{\dot{u}}^2-\frac{1}{2}EA{\left(\frac{\partial u}{\partial x}\right)}^2,$$

giving the wave equation $\rho \ddot{u}=E{\partial }^{2}u/\partial x^2$ with longitudinal wave speed $c_L=\sqrt{E/\rho }$. For steel ($E\approx 200$ GPa, $\rho \approx 8000$ kg/m${}^3$), $c_L\approx 5000$ m/s.

Noether current preview

If the Lagrangian density is invariant under a continuous transformation of the field, there is a conserved current. For a spacetime translation $x^{\mu }\to x^{\mu }+a^{\mu }$, the conserved current is the canonical stress-energy tensor

$$T^{\mu }{}_{\nu }=\frac{\partial \mathcal{L}}{\partial ({\partial }_{\mu }\varphi )}{\partial }_{\nu }\varphi -{\delta }_{\nu }^{\mu }\mathcal{L},$$

satisfying ${\partial }_{\mu }{T}^{\mu \nu }=0$ on shell. The component $T^{00}$ is the energy density, $T^{0i}$ is the momentum density (energy flux), and $T^{ij}$ is the stress tensor. The full derivation is the subject of Noether's theorem 09.03.01.

Key theorem with proof Intermediate+

Theorem (The wave equation is the Euler-Lagrange equation for the elastic-string Lagrangian density). The one-dimensional wave equation $\mu \ddot{u}=\kappa u^{\prime \prime }$ is the Euler-Lagrange equation for the Lagrangian density $\mathcal{L}=\frac{1}{2}\mu {\dot{u}}^2-\frac{1}{2}\kappa (u^{\prime }{)}^2$.

Proof. The action is

$$S[u]={\int }_{t_0}^{t_1}{\int }_0^{\ell }\mathcal{L}(u,\dot{u},u^{\prime },x,t)dxdt.$$

Vary $u\to u+\delta u$ with $\delta u$ vanishing on the boundary of the space-time domain:

$$\delta S=\int \int \left[\frac{\partial \mathcal{L}}{\partial \dot{u}}\delta \dot{u}+\frac{\partial \mathcal{L}}{\partial u^{\prime }}\delta u^{\prime }+\frac{\partial \mathcal{L}}{\partial u}\delta u\right]dxdt.$$

Integrate by parts in $t$ for the first term and $x$ for the second:

$$\delta S=\int \int \left[-\frac{\partial }{\partial t}\frac{\partial \mathcal{L}}{\partial \dot{u}}-\frac{\partial }{\partial x}\frac{\partial \mathcal{L}}{\partial u^{\prime }}+\frac{\partial \mathcal{L}}{\partial u}\right]\delta udxdt.$$

The boundary terms vanish because $\delta u=0$ on the boundary. Since $\delta u$ is arbitrary in the interior:

$$\frac{\partial }{\partial t}\frac{\partial \mathcal{L}}{\partial \dot{u}}+\frac{\partial }{\partial x}\frac{\partial \mathcal{L}}{\partial u^{\prime }}-\frac{\partial \mathcal{L}}{\partial u}=0.$$

Substituting $\mathcal{L}=\frac{1}{2}\mu {\dot{u}}^2-\frac{1}{2}\kappa (u^{\prime }{)}^2$: $\partial \mathcal{L}/\partial \dot{u}=\mu \dot{u}$, $\partial \mathcal{L}/\partial u^{\prime }=-\kappa u^{\prime }$, $\partial \mathcal{L}/\partial u=0$. Therefore $\mu \ddot{u}-\kappa u^{\prime \prime }=0$. $\square$

Corollary (Normal modes). For fixed boundary conditions $u(0,t)=u(\ell ,t)=0$, the normal-mode solutions are $u_n(x,t)=\sin (n\pi x/\ell )\cos ({\omega }_{n}t)$ with frequencies ${\omega }_n=n\pi c/\ell$ where $c=\sqrt{\kappa /\mu }$. The general solution is a superposition $u(x,t)={\sum }_{n=1}^{\infty }{A}_n\sin (n\pi x/\ell )\cos ({\omega }_{n}t+{\delta }_n)$, determined by the initial displacement and velocity.

Corollary (D'Alembert solution). The general solution of the wave equation on the entire line is $u(x,t)=f(x-ct)+g(x+ct)$ for arbitrary twice-differentiable functions $f$ and $g$, representing right-moving and left-moving waves.

Exercises Intermediate+

Lean formalization Intermediate+

The Euler-Lagrange equations for fields require a calculus of variations on infinite-dimensional function spaces (the space of field configurations). Mathlib has finite-dimensional calculus-of-variations machinery but not the field-theoretic extension. The continuum limit (passage from discrete to continuous) would require assembling Mathlib's measure-theory integration with its finite-dimensional Lagrangian mechanics. The stress-energy tensor derivation requires tangent-space computations on the space of field configurations. The Klein-Gordon equation as an Euler-Lagrange PDE would require Mathlib's PDE and distribution-theory capabilities, which are in early stages. This unit ships without a Lean module.

Advanced results Master

The stress-energy tensor from translation invariance

For a field theory with Lagrangian density $\mathcal{L}(\varphi ,{\partial }_{\mu }\varphi )$ in $(n+1)$-dimensional spacetime, translation invariance in each spacetime direction produces a conserved current via Noether's theorem 09.03.01. The full set of $(n+1{)}^2$ components assembles into the canonical stress-energy tensor:

$$T^{\mu }{}_{\nu }=\frac{\partial \mathcal{L}}{\partial ({\partial }_{\mu }\varphi )}{\partial }_{\nu }\varphi -{\delta }_{\nu }^{\mu }\mathcal{L},$$

satisfying ${\partial }_{\mu }{T}^{\mu \nu }=0$ on solutions of the Euler-Lagrange equations. The component $T^{00}$ is the energy density, $T^{0i}$ is the momentum density (or energy flux), and $T^{ij}$ is the stress tensor (force per unit area in the $i$-direction on a surface with normal in the $j$-direction).

For the elastic string: $T^{00}=\frac{1}{2}\mu {\dot{u}}^2+\frac{1}{2}\kappa (u^{\prime }{)}^2$ (energy per unit length), $T^{01}=-\kappa \dot{u}{u}^{\prime }$ (power flux), and $T^{11}=\frac{1}{2}\mu {\dot{u}}^2+\frac{1}{2}\kappa (u^{\prime }{)}^2$ (tension). The conservation law ${\partial }_0{T}^{00}+{\partial }_1{T}^{01}=0$ states that the rate of change of energy in a segment equals the net energy flux through its boundaries.

The canonical stress-energy tensor is not always symmetric. For field theories that are also Lorentz invariant, the Belinfante-Rosenfeld procedure produces a symmetric, gauge-invariant improved tensor ${\Theta }^{\mu \nu }$ that also satisfies ${\partial }_{\mu }{\Theta }^{\mu \nu }=0$ and serves as the source of gravity in general relativity. The symmetric tensor is the one that couples to the metric in the Einstein field equations.

Three-dimensional elastic media

In three dimensions the displacement field $\mathbf{u}(\mathbf{x},t)$ is a vector field. The strain tensor is ${ϵ}_{ij}=\frac{1}{2}({\partial }_i{u}_j+{\partial }_j{u}_i)$ and for an isotropic elastic solid with Lame parameters $\lambda$ and $\mu$, the Lagrangian density is

$$\mathcal{L}=\frac{1}{2}\rho |\dot{\mathbf{u}}{|}^2-\mu {ϵ}_{ij}{ϵ}_{ij}-\frac{1}{2}\lambda ({ϵ}_{kk}{)}^2.$$

The Euler-Lagrange equations give the Navier equations of elasticity:

$$\rho \ddot{\mathbf{u}}=\mu {\nabla }^2\mathbf{u}+(\lambda +\mu )\nabla (\nabla \cdot \mathbf{u}).$$

These support two types of waves: longitudinal (P-waves, speed $\sqrt{(\lambda +2\mu )/\rho }$) and transverse (S-waves, speed $\sqrt{\mu /\rho }$). Seismology uses this framework to model earthquake wave propagation through the Earth's interior. The P-wave is irrotational ($\nabla \times \mathbf{u}=0$) and the S-wave is incompressible ($\nabla \cdot \mathbf{u}=0$).

Gauge fields and field Lagrangians

The simplest gauge field theory -- the free electromagnetic field -- has Lagrangian density

$${\mathcal{L}}_{\mathrm{EM}}=-\frac{1}{4}{F}_{\mu \nu }{F}^{\mu \nu },$$

where $F_{\mu \nu }={\partial }_{\mu }{A}_{\nu }-{\partial }_{\nu }{A}_{\mu }$ is the electromagnetic field tensor. The Euler-Lagrange equations for $A_{\mu }$ give Maxwell's equations in vacuum: ${\partial }_{\mu }{F}^{\mu \nu }=0$. This is the prototype for all gauge field theories in modern physics, and its Lagrangian structure is a direct descendant of the elastic-string Lagrangian studied here.

The electromagnetic Lagrangian is singular (the Hessian ${\partial }^2\mathcal{L}/\partial {\dot{A}}_{\mu }\partial {\dot{A}}_{\nu }$ is degenerate), reflecting gauge invariance $A_{\mu }\to A_{\mu }+{\partial }_{\mu }\Lambda$. The Dirac-Bergmann constraint analysis produces the primary constraint ${\pi }^0=0$ and the secondary Gauss-law constraint $\nabla \cdot \mathbf{E}=\rho$. This connects to the singular-Lagrangian theory of 09.04.01.

Canonical quantization of fields: a preview

The passage from classical to quantum field theory proceeds by canonical quantization. Given a classical field Lagrangian density $\mathcal{L}(\varphi ,{\partial }_{\mu }\varphi )$, define the conjugate momentum density

$$\pi (x,t):=\frac{\partial \mathcal{L}}{\partial \dot{\varphi }(x,t)}.$$

For the Klein-Gordon field $\mathcal{L}=\frac{1}{2}{\dot{\varphi }}^2-\frac{1}{2}|\nabla \varphi {|}^2-\frac{1}{2}{m}^2{\varphi }^2$, the conjugate momentum is $\pi =\dot{\varphi }$. The Hamiltonian density is obtained by the field-theoretic Legendre transform 09.04.01:

$$\mathcal{H}=\pi \dot{\varphi }-\mathcal{L}=\frac{1}{2}{\pi }^2+\frac{1}{2}|\nabla \varphi {|}^2+\frac{1}{2}{m}^2{\varphi }^2.$$

The classical Poisson bracket of the field and its conjugate momentum is $\{\varphi (x,t),\pi (y,t)\}=\delta (x-y)$ (the Dirac delta replaces the Kronecker delta of discrete mechanics). Canonical quantization promotes $\varphi$ and $\pi$ to operators satisfying the equal-time commutation relations

$$[\hat{\varphi }(x,t),\hat{\pi }(y,t)]=i\hslash \delta (x-y).$$

This is the field-theoretic generalisation of $[\hat{q},\hat{p}]=i\hslash$. The Hamiltonian operator $\hat{H}=\int \hat{\mathcal{H}}{d}^{3}x$ generates time evolution. The Fock-space representation decomposes the field into creation and annihilation operators:

$$\hat{\varphi }(x)=\int \frac{d^{3}p}{(2\pi {)}^3}\frac{1}{\sqrt{2{\omega }_p}}\left({\hat{a}}_p{e}^{ip\cdot x}+{\hat{a}}_p^{†}{e}^{-ip\cdot x}\right),$$

where ${\omega }_p=\sqrt{|\mathbf{p}{|}^2+m^2}$. The operators ${\hat{a}}_p^{†}$ and ${\hat{a}}_p$ create and destroy quanta (particles) of momentum $\mathbf{p}$, satisfying $[{\hat{a}}_p,{\hat{a}}_{p^{\prime }}^{†}]=(2\pi {)}^3\delta (p-p^{\prime })$. The vacuum $|0⟩$ is annihilated by all ${\hat{a}}_p$, and the one-particle state $|\mathbf{p}⟩={\hat{a}}_p^{†}|0⟩$ has energy ${\omega }_p=\sqrt{|\mathbf{p}{|}^2+m^2}$ -- exactly the relativistic energy-momentum relation. The classical field $\varphi$ becomes a quantum field whose excitations are particles -- this is the foundational insight of QFT 12.03.01.

The Hamiltonian in terms of creation and annihilation operators is

$$\hat{H}=\int \frac{d^{3}p}{(2\pi {)}^3}{\omega }_p{\hat{a}}_p^{†}{\hat{a}}_p+E_{\mathrm{vac}},$$

where $E_{\mathrm{vac}}=\int d^{3}p{\omega }_p/2$ is the (divergent) zero-point energy of the vacuum. Renormalisation -- the systematic subtraction of this and other divergences -- is the technical core of perturbative QFT. The normal-ordered Hamiltonian $:\hat{H}:$ discards $E_{\mathrm{vac}}$ and counts only excitations above the vacuum. The entire machinery of Feynman diagrams, scattering amplitudes, and cross-sections in particle physics is built on this foundation.

For the massless case $m=0$, the Klein-Gordon field reduces to the elastic-string field studied in this unit. The quanta are massless scalar particles analogous to (but distinct from) photons. The connection between the classical wave equation and the quantum description of massless particles is the simplest example of wave-particle duality in a field-theoretic setting.

Proof: conservation of the stress-energy tensor

Proposition (Stress-energy conservation). For any Lagrangian density $\mathcal{L}(\varphi ,{\partial }_{\mu }\varphi )$ that does not depend explicitly on $x^{\mu }$, the canonical stress-energy tensor $T^{\mu }{}_{\nu }=(\partial \mathcal{L}/\partial ({\partial }_{\mu }\varphi )){\partial }_{\nu }\varphi -{\delta }_{\nu }^{\mu }\mathcal{L}$ satisfies ${\partial }_{\mu }{T}^{\mu \nu }=0$ on solutions of the Euler-Lagrange equations.

Proof. Differentiate:

$${\partial }_{\mu }{T}^{\mu }{}_{\nu }={\partial }_{\mu }\left[\frac{\partial \mathcal{L}}{\partial ({\partial }_{\mu }\varphi )}\right]{\partial }_{\nu }\varphi +\frac{\partial \mathcal{L}}{\partial ({\partial }_{\mu }\varphi )}{\partial }_{\mu }{\partial }_{\nu }\varphi -{\partial }_{\nu }\mathcal{L}.$$

Expand ${\partial }_{\nu }\mathcal{L}$ by the chain rule:

$${\partial }_{\nu }\mathcal{L}=\frac{\partial \mathcal{L}}{\partial \varphi }{\partial }_{\nu }\varphi +\frac{\partial \mathcal{L}}{\partial ({\partial }_{\mu }\varphi )}{\partial }_{\nu }{\partial }_{\mu }\varphi .$$

The second terms match. The remaining terms are

$${\partial }_{\mu }{T}^{\mu }{}_{\nu }=\left[{\partial }_{\mu }\frac{\partial \mathcal{L}}{\partial ({\partial }_{\mu }\varphi )}-\frac{\partial \mathcal{L}}{\partial \varphi }\right]{\partial }_{\nu }\varphi .$$

On shell, the bracket vanishes by the Euler-Lagrange equations. $\square$

This is a special case of Noether's theorem 09.03.01: the symmetry is spacetime translation $x^{\mu }\to x^{\mu }+a^{\mu }$, and the conserved current for each choice of $\nu$ is $T^{\mu }{}_{\nu }$. The four conservation laws ${\partial }_{\mu }{T}^{\mu 0}=0$ (energy) and ${\partial }_{\mu }{T}^{\mu i}=0$ (momentum) encode the local balance of energy and momentum in the field.

The stress-energy tensor as the source of gravity

In general relativity, the symmetric stress-energy tensor ${\Theta }^{\mu \nu }$ appears on the right-hand side of the Einstein field equations:

$$G_{\mu \nu }=\frac{8\pi G}{c^4}{\Theta }_{\mu \nu }.$$

The Einstein-Hilbert action $S_{\mathrm{EH}}=\int R\sqrt{-g}{d}^{4}x$ generates the left-hand side through the Euler-Lagrange equations for the metric tensor $g_{\mu \nu }$ (treated as a field). The matter Lagrangian density ${\mathcal{L}}_m$ generates the right-hand side: ${\Theta }_{\mu \nu }=-2\delta {\mathcal{L}}_m/\delta g^{\mu \nu }$. The elastic-string Lagrangian of this unit, embedded in a curved spacetime, would contribute to ${\Theta }_{\mu \nu }$ through this variational formula. The variational principle for the metric is the ultimate generalisation of the Euler-Lagrange equations for fields developed here.

Multi-component fields and internal symmetries

The scalar field $\varphi$ studied so far carries no internal index. A multi-component field ${\varphi }^a$ (with $a=1,\dots ,N$) has a Lagrangian density that may be invariant under internal rotations ${\varphi }^a\to R^a{}_b{\varphi }^b$. For the $O(N)$-invariant Lagrangian

$$\mathcal{L}=\frac{1}{2}{\partial }_{\mu }{\varphi }^a{\partial }^{\mu }{\varphi }^a-\frac{1}{2}{m}^2{\varphi }^a{\varphi }^a-\frac{\lambda }{4}({\varphi }^a{\varphi }^a{)}^2,$$

Noether's theorem produces a conserved current $J_{ab}^{\mu }=(\partial \mathcal{L}/\partial ({\partial }_{\mu }{\varphi }^a)){\varphi }^b-(\partial \mathcal{L}/\partial ({\partial }_{\mu }{\varphi }^b)){\varphi }^a$ for each generator of $O(N)$. The case $N=4$ with spontaneous symmetry breaking ($m^2<0$) produces three Goldstone bosons and one massive Higgs mode -- the mathematical template for the Higgs mechanism of the Standard Model. The field-theoretic Lagrangian formalism is the language in which the entire particle-physics Standard Model is written.

Synthesis. The continuum limit of discrete Lagrangian systems provides the foundational reason that field theories inherit the variational structure of particle mechanics; the central insight is that the Lagrangian density replaces the particle Lagrangian, the Euler-Lagrange PDE replaces the Euler-Lagrange ODE, and the sum over particles becomes an integral over space. This is exactly the structure that quantum field theory, continuum mechanics, and general relativity all share. The wave equation derived here is the simplest example of a field equation obtained from a variational principle. The Klein-Gordon equation introduces a mass gap and connects directly to the scalar field of QFT. The stress-energy tensor, derived from translation invariance, is the bridge to gravity and to the conservation laws that constrain all field theories. Putting these together, the continuum limit provides the passage from particle mechanics to field physics -- the single most important conceptual leap in theoretical physics from the 18th century to the present.

Connections Master

• 09.02.01 The action principle for fields is the same $\delta S=0$ condition as for particles, but the action is now a functional of the field configuration $u(x,t)$ over a spacetime region rather than a path $q(t)$ in time.

• 09.02.02 The Euler-Lagfield equations generalise the finite-dimensional Euler-Lagrange equations by replacing ordinary derivatives with partial derivatives and sums with integrals. The derivation follows the same integration-by-parts argument.

• 09.03.01 Noether's theorem for fields produces the stress-energy tensor as the conserved current associated with spacetime translation invariance. This unit's field Lagrangian provides the concrete setting in which that conservation law is derived. The stress-energy tensor computed here is the Noether current.

• 09.04.01 The Legendre transform for fields (from Lagrangian density to Hamiltonian density) follows the same pattern as the Legendre transform for particles: define the conjugate momentum density $\pi =\partial \mathcal{L}/\partial \dot{\varphi }$ and form $\mathcal{H}=\pi \dot{\varphi }-\mathcal{L}$.

• 09.07.01 Continuum mechanics extends the one-dimensional elastic rod to three-dimensional elastic media with tensor-valued strain and stress. The Lagrangian density framework developed here is the starting point.

• 08.01.01 The Klein-Gordon equation ${\partial }_{\mu }{\partial }^{\mu }\varphi +m^2\varphi =0$ is the relativistic wave equation for a free scalar particle. The QFT units develop the same equation from the perspective of quantum fields and Fock space.

• 12.03.01 Quantum field theory begins with the same field Lagrangian densities studied here, promoting the fields to operator-valued distributions. The Klein-Gordon Lagrangian of Exercise 6 is the free scalar field of QFT. The canonical quantization procedure outlined in the Master section is the bridge.

Historical & philosophical context Master

The passage from discrete to continuous mechanics was a central theme of 18th-century mathematical physics. The debate over the nature of the vibrating string -- whether it should be modelled as a continuum or as a discrete chain of infinitesimal masses -- consumed the attention of d'Alembert, Euler, Daniel Bernoulli, and Lagrange from the 1740s through the 1760s.

D'Alembert derived the wave equation in 1747 and found its general solution $u=f(x-ct)+g(x+ct)$, one of the first PDE solutions ever found. Euler argued that arbitrary initial conditions should be admissible, while d'Alembert insisted on analyticity. Bernoulli proposed the trigonometric (Fourier) series solution in 1753, though its generality was not accepted until Fourier's 1807 memoir -- and rigorously justified only by Dirichlet in 1829.

Lagrange's Mechanique analytique (1788) contained the variational foundation for discrete mechanics, but the field-theoretic extension came later. Hamilton's principle (1834) was generalised to continuous systems by several authors in the late 19th century, but the systematic Lagrangian-density formulation was achieved by Hilbert in 1915 as part of his axiomatic approach to general relativity. Hilbert's derivation of the Einstein field equations from the Einstein-Hilbert action $\int R\sqrt{-g}{d}^{4}x$ demonstrated that the Euler-Lagfield equations applied to the metric tensor $g_{\mu \nu }$ produce the geometric theory of gravity. The same formalism was adopted by Dirac, Heisenberg, and Pauli for quantum electrodynamics in the 1920s and 1930s.

The canonical quantization programme was initiated by Dirac (1927) and developed systematically by Heisenberg and Pauli (1929-1930). The key step -- promoting the Poisson bracket $\{\varphi ,\pi \}=\delta$ to the commutator $[\hat{\varphi },\hat{\pi }]=i\hslash \delta$ -- is a direct generalisation of the particle quantization $[\hat{q},\hat{p}]=i\hslash$ that Schrodinger and Heisenberg had introduced in 1925-1926. The passage from the classical elastic string to the quantum field is thus the culmination of a line of thought that begins with the vibrating-string controversy of the 1740s and ends with the Standard Model of particle physics.

A philosophical point: the continuum limit is not merely a mathematical convenience. The physical question "is matter fundamentally discrete or continuous?" is left open by classical mechanics. The continuum limit produces the same predictions as the discrete model when the wavelength greatly exceeds the lattice spacing. Only at atomic scales does the discrete structure become detectable. The Lagrangian-density formalism is agnostic about this question; it applies equally well to fundamental fields (electromagnetism) and to emergent continuum descriptions of discrete systems (elastic solids). This is one of the deepest insights of classical field theory: the variational principle does not care whether the field is fundamental or emergent.

Bibliography Master

• d'Alembert, J. le R., "Recherches sur la courbe que forme une corde tendue mise en vibration," Hist. Acad. Berlin (1747), 214-219.
• Lagrange, J. L., Mechanique analytique (1788), Seconde Partie, §III.
• Hamilton, W. R., "On a general method in dynamics," Phil. Trans. Roy. Soc. 124 (1834), 247-308.
• Hilbert, D., "Die Grundlagen der Physik," Nachr. Ges. Wiss. Gottingen, Math.-Phys. Kl. (1915), 395-407.
• Dirac, P. A. M., "The quantum theory of the emission and absorption of radiation," Proc. Roy. Soc. A 114 (1927), 243-265.
• Heisenberg, W. & Pauli, W., "Zur Quantendynamik der Wellenfelder," Z. Phys. 56 (1929), 1-61; 59 (1930), 168-190.
• Goldstein, H., Poole, C. P. & Safko, J., Classical Mechanics, 3rd ed. (Pearson, 2002), Ch. 13.
• Taylor, J. R., Classical Mechanics (University Science Books, 2005), Ch. 13.
• Landau, L. D. & Lifshitz, E. M., Mechanics, 3rd ed. (Course of Theoretical Physics Vol. 1, Pergamon, 1976), §46-47.
• Landau, L. D. & Lifshitz, E. M., Theory of Elasticity, 3rd ed. (Course of Theoretical Physics Vol. 7, Pergamon, 1986), §1-5.
• Srednicki, M., Quantum Field Theory (Cambridge University Press, 2007), Ch. 3 (canonical quantization of the free scalar field).
• Peskin, M. E. & Schroeder, D. V., An Introduction to Quantum Field Theory (Westview, 1995), Ch. 2 (the Klein-Gordon field).