Internal Symmetries and Conserved Charge

Energy and momentum come from sliding the whole experiment in time and space. This article's conservation laws come from doing nothing to spacetime at all — from re-labelling the fields in place. An internal symmetry rotates the components or the phase of a field multiplet identically at every point, $\delta x^\mu = 0$ , and Noether's machine, fed this input, returns the second great family of conserved quantities: charges. The prototype is the complex scalar's phase rotation $\phi \to e^{i\alpha}\phi$ , whose conserved $Q$ is the classical ancestor of electric charge, baryon number, lepton number — every additive quantum number on the particle-data sheets is an internal-symmetry charge. Three payoffs make this the most modern-facing article of the chapter. First, the charge has a sign structure the energy never had: decompose a free complex field into its two frequency branches and $Q$ counts one branch positively and the other negatively — the classical shadow of particles and antiparticles, visible before quantum mechanics says a word. Second, non-abelian symmetries (rotating a multiplet in ways that do not commute) produce families of charges that fail to Poisson-commute among themselves, closing instead on the group's Lie algebra — the structure the Poisson-brackets article derived and isospin exemplifies. Third, the demand that the symmetry hold locally — a different rotation at every point — fails for the plain Lagrangian in a precisely diagnosable way, and the unique repair is to introduce a new field that couples to the Noether current. That repair is the gauge principle: electromagnetism, and behind it the entire Standard Model, is the price of localizing an internal symmetry. This article ends with that door open.

Key takeaways

An internal symmetry mixes field components or phases with $\delta x^\mu = 0$ : the compasses rotate, spacetime does not move.
Noether output: one conserved current per group parameter, $j^{\mu a} = -i\,\frac{\partial\mathcal{L}}{\partial(\partial_\mu\phi_a)}(T^a)_{ac}\phi_c + \text{c.c.}$ , with $\partial_\mu j^{\mu a} = 0$ on-shell and charges $Q^a = \int\mathrm{d}^3x\,j^{0a}$ .
Internal currents have no orbital piece: since coordinates do not transform, there is no $K^\mu$ , no stress-energy contribution — pure field-space response.
The $U(1)$ prototype: $j^\mu = i(\phi^*\partial^\mu\phi - \phi\,\partial^\mu\phi^*)$ ; its charge counts positive-frequency content minus negative-frequency content — particles minus antiparticles, classically foreshadowed.
A real field has identically zero internal $U(1)$ current: no phase, no charge.
Non-abelian charges close on the Lie algebra under Poisson brackets, $\{Q^a, Q^b\} = f^{abc}Q^c$ , and generate the multiplet rotations — the bracket article's machinery applied.
Gauging ( $\alpha \to \alpha(x)$ ) breaks the naive symmetry through the derivative term; restoring it forces a gauge field coupled to exactly the Noether current — the gauge principle.
Key idea: re-labelling fields in place is a symmetry too, and its Noether tally — charge — is the additive quantum number of physics.

The core insight

For a symmetry that acts only on field indices, $\delta\phi_a = i\alpha^b(T^b)_{ac}\phi_c$ with $\delta x^\mu = 0$ , Noether's theorem gives a current $j^{\mu a} = -i\,\frac{\partial\mathcal{L}}{\partial(\partial_\mu\phi_a)}(T^a)_{ac}\phi_c + \text{c.c.}$ with $\partial_\mu j^{\mu a} = 0$ on-shell, and a conserved charge $Q^a = \int\mathrm{d}^3x\,j^{0a}$ .

Where this lives today

Internal-symmetry charges are the accounting system of particle physics: every interaction vertex in the Standard Model conserves electric charge, color, and (in the classical Lagrangian) baryon and lepton number, and detector reconstruction leans on those ledgers event by event — a collision at the LHC is "balanced" using exactly the additive quantum numbers this article derives. The gauge principle scaled from this article's $U(1)$ to $SU(3)\times SU(2)\times U(1)$ is the Standard Model's definition; the 2012 Higgs discovery completed the roster of its predicted fields, and the current precision program (Higgs couplings, electroweak fits) is stress-testing the local-symmetry construction at the per-mille level. The signed ledger runs the other way too: the observed baryon asymmetry of the universe — why matter outnumbers antimatter — is a statement that $Q_{\text{baryon}}$ today is nonzero, and Sakharov's conditions for generating it (baryon-number violation, C and CP violation, departure from equilibrium) are a checklist of how the classical conservation laws of this article must fail in the early universe; leptogenesis scenarios and the LHCb/Belle-II CP-violation program live on that checklist. In condensed matter, the same mathematics wears different clothes: superfluidity and superconductivity are spontaneously broken internal $U(1)$ 's (particle number), the stiffness $q^2v^2$ of the worked example is measured as superfluid density, and the gauge-field absorption of the phase twist is the Meissner effect — the worked example's Step 4 is, almost verbatim, why magnetic fields are expelled from superconductors and how gauge bosons acquire mass in the Higgs mechanism. Within the chapter: the charge algebra behind (4) is the Poisson-brackets article's; the energy tally that pairs with the charge ledger is the stress-energy article's; and the capstone will quantize the $a$ 's and $b$ 's into the particles and antiparticles their signs already announce.

The physical picture

Attach to every point of space an internal compass — an arrow living not in space but in field space. For a complex scalar the compass is the phase of $\phi$ ; for a two-component multiplet it is an orientation in an abstract two-dimensional field plane. A global internal symmetry rotates every compass by the same angle, everywhere, at once. Nothing measurable moves: energies, distances, field magnitudes, the dynamics — all depend only on invariant combinations like $|\phi|^2$ and $|\partial_\mu\phi|^2$ , which do not feel the rotation. That indifference is a flat direction of the action, and Noether converts it into a conserved tally of "which way the compasses point and how fast they turn": the charge.

Figure

Internal symmetry — every compass rotates together, spacetime holds still, and the tally of phase flow is the charge

The picture sharpens into the article's three themes. Abelian: one compass angle, one charge — and because the phase can circulate in two senses, the charge comes signed, which is where antiparticles will enter. Non-abelian: a compass that can be rotated in several non-commuting ways (orientations in a 3D internal space, say) — the order of rotations matters, and the charges inherit exactly that non-commutativity in their bracket algebra. Local: try to rotate each compass by a different angle and the derivative terms of the Lagrangian object immediately — neighbouring compasses no longer agree, gradients register the mismatch, and the action shifts. The failure is not a dead end but a construction principle: add a bookkeeping field that absorbs point-to-point mismatches and invariance is restored. Physics has a name for the bookkeeping field — the photon.

The internal Noether current

The transformation, in full generality: a multiplet $\phi_a$ rotated by generator matrices $T^b$ with constant parameters $\alpha^b$ ,

\delta\phi_a = i\,\alpha^b\,(T^b)_{ac}\,\phi_c, \qquad \delta x^\mu = 0 \qquad (1)

where: the matrices satisfy the group's algebra $[T^a, T^b] = if^{abc}T^c$ ( $f^{abc}$ the structure constants; all zero for the abelian $U(1)$ where $T = 1$ ). The absence of any coordinate change is the family's defining mark and its great simplification: the variation commutes with $\partial_\mu$ , the Lagrangian density is strictly invariant (not merely up to a divergence), and the Noether machinery runs with $K^\mu = 0$ .

Derivation AThe Noether current of a global internal symmetry

We specialize the master identity of the Noether article to internal transformations and obtain the general multiplet current — the formula every charge in particle physics instantiates. The strategy is the standard three moves (chain rule, Euler-Lagrange substitution, reverse product rule), but the point of running them again in this restricted setting is to see what disappears: with $\delta x^\mu = 0$ there is no transport term, no $K^\mu$ , and the current is nothing but canonical momenta contracted with the generator's action. Watch for where each independent parameter peels off its own conservation law.

Step 1. Vary the density. Since coordinates are untouched, $\delta(\partial_\mu\phi_a) = \partial_\mu(\delta\phi_a)$ , and

\delta\mathcal{L} = \frac{\partial\mathcal{L}}{\partial\phi_a}\,\delta\phi_a + \frac{\partial\mathcal{L}}{\partial(\partial_\mu\phi_a)}\,\partial_\mu(\delta\phi_a)

Step 2. Go on-shell: substitute the Euler-Lagrange equations $\frac{\partial\mathcal{L}}{\partial\phi_a} = \partial_\mu\big(\frac{\partial\mathcal{L}}{\partial(\partial_\mu\phi_a)}\big)$ into the first term, and run the product rule in reverse:

\delta\mathcal{L} = \partial_\mu\!\left(\frac{\partial\mathcal{L}}{\partial(\partial_\mu\phi_a)}\,\delta\phi_a\right)

Step 3. Invoke the symmetry: $\delta\mathcal{L} = 0$ (strict invariance — the internal rotation changes no invariant the density is built from). Substituting (1):

0 = \alpha^b\,\partial_\mu\!\left(\frac{\partial\mathcal{L}}{\partial(\partial_\mu\phi_a)}\;i(T^b)_{ac}\,\phi_c\right)

Each of the parameters $\alpha^b$ is independent, so each bracket is separately conserved — one current per generator.

Step 4. Name the current, fixing the sign so that the prototype's charge comes out positive for positive-frequency (particle-like) content, and adding the conjugate-field contribution for complex multiplets:

j^{\mu b} = -i\,\frac{\partial\mathcal{L}}{\partial(\partial_\mu\phi_a)}\,(T^b)_{ac}\,\phi_c + \text{c.c.}, \qquad \partial_\mu j^{\mu b} = 0 \qquad (2)

Why there is no orbital piece: compare the translation current of the Noether article, which carried both a field-response term and the dragged density $K^\mu = -a^\mu\mathcal{L}$ . Here the second ingredient never existed — the transformation moves nothing through spacetime. An internal current is pure response; if you find yourself adding a $\partial_\nu\phi\,\delta x^\nu$ term to one, you have mixed the families. *Verified: reduction of the master identity per Goldstein §13.7; the c.c. structure checked against the complex-scalar prototype below; parameter-peeling logic identical to the translation case of the Noether article.*

For the complex scalar prototype $\mathcal{L} = \partial_\mu\phi^*\partial^\mu\phi - m^2\phi^*\phi$ under $\phi \to e^{i\alpha}\phi$ , formula (2) evaluates (as the Noether article computed step by step) to

j^\mu = i\,(\phi^*\partial^\mu\phi - \phi\,\partial^\mu\phi^*), \qquad Q = \int\mathrm{d}^3x\; i\,(\phi^*\dot\phi - \phi\,\dot\phi^*) \qquad (3)

where: the relative minus sign between the two terms is the structural content — it makes $j^\mu$ vanish identically for a real field (no phase, no compass, no charge) and it is what gives $Q$ its sign structure below. Note the family resemblance to the quantum-mechanical probability current: the Schrödinger current is this object's nonrelativistic limit, a connection the classical-limit article closes.

What the charge counts: the two-frequency ledger

Here is the article's most consequential computation: evaluate $Q$ for a general free-field configuration and discover that it is not a positive count of field intensity. It is a difference of two counts.

Go deeper · Derivation B · 2 minThe charge counts positive-frequency minus negative-frequency content

Derivation BThe charge counts positive-frequency minus negative-frequency content

We evaluate the $U(1)$ charge (3) for a general free complex scalar in a periodic box and show it separates into mode contributions with signs: each positive-frequency (particle-like) mode contributes positively, each negative-frequency (antiparticle-like) mode negatively. This is the classical origin of the particle/antiparticle distinction and the reason $Q$ is a charge rather than a probability. Strategy: expand the general solution in the two frequency branches the Klein-Gordon equation admits, insert into (3), and let mode orthogonality collapse the integral. Watch the relative sign — it comes from the $\pm i\omega$ in the time derivatives, i.e. from the dynamics, not from a convention.

Step 1. General free solution in a box of volume $V$ (allowed wavevectors $\mathbf{k}$ , $\omega_k = \sqrt{k^2 + m^2}$ ). Because the Klein-Gordon equation is second order, each mode carries two branches, and a complex field keeps independent coefficients on each:

\phi(\mathbf{x}, t) = \frac{1}{\sqrt{V}}\sum_{\mathbf{k}}\Big[ a_{\mathbf{k}}\,e^{-i(\omega_k t - \mathbf{k}\cdot\mathbf{x})} + b^*_{\mathbf{k}}\,e^{+i(\omega_k t - \mathbf{k}\cdot\mathbf{x})} \Big]

where $a_{\mathbf{k}}$ weights the positive-frequency branch and $b^*_{\mathbf{k}}$ the negative-frequency one (starred by convention, so $b$ will label the conjugate field's positive-frequency content).

Step 2. Time derivative: $\dot\phi = \frac{1}{\sqrt{V}}\sum_{\mathbf{k}}(-i\omega_k)\big[a_{\mathbf{k}}e^{-i(\cdot)} - b^*_{\mathbf{k}}e^{+i(\cdot)}\big]$ — the two branches acquire opposite signs, the crucial asymmetry.

Step 3. Insert into $Q = \int\mathrm{d}^3x\; i(\phi^*\dot\phi - \phi\dot\phi^*)$ and integrate. Mode orthogonality $\int e^{i(\mathbf{k}-\mathbf{k}')\cdot\mathbf{x}}\mathrm{d}^3x = V\delta_{\mathbf{k}\mathbf{k}'}$ kills all cross- $\mathbf{k}$ terms; the surviving same- $\mathbf{k}$ cross terms between branches oscillate at $e^{\pm 2i\omega_k t}$ and cancel between the two halves of (3). The diagonal terms give

Q = \sum_{\mathbf{k}} 2\omega_k\Big( |a_{\mathbf{k}}|^2 - |b_{\mathbf{k}}|^2 \Big)

Step 4. Read the ledger. The charge is a signed tally: positive-frequency amplitude counts $+$ , negative-frequency counts $-$ , mode by mode. (With the conventional relativistic normalization $a \to a/\sqrt{2\omega}$ , the tally is literally $\sum(|a|^2 - |b|^2)$ — "number of particles minus number of antiparticles," a sentence that becomes exact upon quantization when $|a|^2 \to \hat n$ .) Meanwhile the energy of the same configuration, from the stress-energy article, is $\sum 2\omega_k^2(|a|^2 + |b|^2)$ with a plus — both branches cost energy, but they carry opposite charge. A field can be electrically neutral yet full of energy ( $|a| = |b|$ ); it cannot be charged for free. Verified: mode algebra recomputed; cross-term cancellation checked explicitly (the $e^{\pm2i\omega t}$ pieces cancel between $\phi^*\dot\phi$ and $-\phi\dot\phi^*$ ); matches the standard classical precursor of the QFT charge operator $\hat Q = \sum(\hat n_a - \hat n_b)$ .

Figure

The signed ledger — energy adds both frequency branches, charge subtracts them: the classical shadow of antiparticles

The non-abelian generalization needs one more fact, and the Poisson-brackets article has already proved it: the charges $Q^a = -i\int\mathrm{d}^3x\,\pi_b(T^a)_{bc}\phi_c$ generate the multiplet rotations through the bracket, $\{Q^a, \phi_d\} = -i(T^a)_{dc}\phi_c$ , and close on the algebra,

\{Q^a, Q^b\} = f^{abc}\,Q^c \qquad (4)

where: the structure constants on the right are the same ones defining the generator matrices — the conserved charges are a copy of the symmetry's Lie algebra sitting inside the observables. For $SU(2)$ isospin this is the statement that the three isospin charges bracket like angular momenta; only one of them (conventionally $Q^3$ ) can be diagonalized at a time, which is why particle tables list " $I_3$ " and not three independent isospin numbers. The derivation is the bracket article's closure computation verbatim; we do not repeat it.

Gauging: the door out of the chapter

Everything so far used a global rotation — the same $\alpha$ at every point of the universe, which should strike you as a strangely rigid demand. What happens if we ask for invariance under $\alpha = \alpha(x)$ , a different compass rotation at each point?

Go deeper · Derivation C · 2 minLocalizing the symmetry forces a gauge field coupled to the Noether current

Derivation CLocalizing the symmetry forces a gauge field coupled to the Noether current

We show that the complex scalar's Lagrangian fails to be invariant under local phase rotations, that the failure is proportional to $\partial_\mu\alpha$ contracted with the Noether current, and that invariance is restored — uniquely, at lowest order — by introducing a vector field $A_\mu$ with a shift transformation law and coupling it to that same current. This is the gauge principle: the derivation pattern that generates electromagnetism here and, run on non-abelian groups, the full Standard Model interaction structure. The strategy is direct computation of the offending terms, then their systematic cancellation. Watch which object multiplies $\partial_\mu\alpha$ : the theorem's fingerprints are on it.

Step 1. Transform with a spacetime-dependent phase: $\phi \to e^{i\alpha(x)}\phi$ . The potential term $m^2\phi^*\phi$ is still invariant (no derivatives), but the kinetic term picks up mismatch terms because the derivative now hits the phase:

\partial_\mu\phi \;\to\; e^{i\alpha}\big(\partial_\mu\phi + i(\partial_\mu\alpha)\,\phi\big)

Step 2. Collect the change of the density to first order in $\alpha$ :

\delta\mathcal{L} = i(\partial_\mu\alpha)\big(\phi\,\partial^\mu\phi^* - \phi^*\partial^\mu\phi\big) + O(\alpha^2)\cdot|\phi|^2 = -(\partial_\mu\alpha)\,j^\mu + (\partial_\mu\alpha)(\partial^\mu\alpha)\,|\phi|^2

Why the current appears: the derivative of the parameter couples to exactly the object that measured phase circulation — the Noether current (3). This is general: localizing any global symmetry produces a violation $-(\partial_\mu\alpha^a)\,j^{\mu a}$ . The current tells you what the repair must couple to.

Step 3. Introduce the repair. Add a vector field $A_\mu$ transforming as $A_\mu \to A_\mu + \frac{1}{e}\partial_\mu\alpha$ (a shift that can absorb the mismatch), and replace every derivative by the covariant derivative:

D_\mu\phi \equiv (\partial_\mu - ieA_\mu)\,\phi \quad\Longrightarrow\quad D_\mu\phi \to e^{i\alpha}\,D_\mu\phi

The covariant derivative of the field transforms exactly like the field itself — check: the $i(\partial_\mu\alpha)\phi$ from the derivative cancels against $-ie\cdot\frac{1}{e}(\partial_\mu\alpha)\phi$ from the shift. Hence $\mathcal{L} = D_\mu\phi^*\,D^\mu\phi - m^2\phi^*\phi$ is invariant under local rotations.

Step 4. Read off the physics. Expanding the covariantized kinetic term:

D_\mu\phi^*D^\mu\phi = \partial_\mu\phi^*\partial^\mu\phi + e\,A_\mu\,j^\mu + e^2 A_\mu A^\mu\,|\phi|^2

The new field couples linearly to the Noether current $j^\mu$ — the conserved charge current has become the source the gauge field responds to — with coupling strength $e$ : the charge. Giving $A_\mu$ its own dynamics ( $-\tfrac14 F_{\mu\nu}F^{\mu\nu}$ , the unique gauge-invariant choice at two derivatives) completes scalar electrodynamics. Nothing about this construction was optional once local invariance was demanded. Verified: cancellation in Step 3 checked term by term; the expansion in Step 4 recomputed; matches the standard minimal-coupling construction (any field-theory text's scalar QED opening; classical content only used here).

The conceptual inversion deserves a sentence of its own: in this article's first half, symmetry produced a conserved current; in the second half, the current produced an interaction. Global symmetry ⇒ charge; local symmetry ⇒ force coupled to that charge. That two-step is the organizing logic of modern particle physics, and it is entirely visible in classical field theory.

Limiting cases

Case	Result	Meaning
real scalar field	$j^\mu \equiv 0$	no phase to rotate: chargeless field
abelian $U(1)$	$f^{abc} = 0$ : single commuting charge	electric-charge prototype; all charges simultaneously sharp
non-abelian $SU(N)$	$\{Q^a, Q^b\} = f^{abc}Q^c \neq 0$	only a maximal commuting set diagonal (e.g. $I_3$ )
$\lvert a_{\mathbf{k}}\rvert = \lvert b_{\mathbf{k}}\rvert$	$Q = 0$ , $E > 0$	neutral but energetic: charge is a difference, energy a sum
static field configuration	$\dot\phi = 0 \Rightarrow j^0 = 0$	a frozen complex field carries no charge density
global limit of gauged theory ( $\partial\alpha \to 0$ )	coupling $\to$ inert bookkeeping	the global charge survives gauging as the physical invariant
approximate symmetry (e.g. isospin)	$\partial_\mu j^{\mu a} =$ breaking term	charges drift at a computable rate (percent-level for isospin)
nonrelativistic limit	$j^\mu \to$ Schrödinger probability current	charge density $\to \rho \propto \|\psi\|^2 > 0$ : one branch survives

Common trapNo orbital piece, and never forget the conjugate field

Two mechanical errors account for most wrong internal currents. First: adding a spacetime piece. Because $\delta x^\mu = 0$ , an internal current contains no $\frac{\partial\mathcal{L}}{\partial(\partial_\mu\phi)}\partial_\nu\phi\,\delta x^\nu$ term, no $K^\mu$ , no stress-energy contamination — it is canonical momenta contracted with the generator, full stop. If your "internal" current contains explicit derivatives of the coordinates or pieces of $T^{\mu\nu}$ , you have crossed the family boundary (the mixed objects that do legitimately combine both families, like the angular-momentum current's orbital-plus-spin split, are the spin-current article's subject — and there the spacetime piece belongs by construction). Second: dropping the conjugate. A complex field is two real degrees of freedom, and both transform: $\Delta\phi = i\phi$ , $\Delta\phi^* = -i\phi^*$ . The current (3) is the sum of both contributions; keep only the $\phi$ term and you get half the current — worse, a non-conserved half, since the cancellation in $\partial_\mu j^\mu$ needs both mass terms. The same omission in the charge-algebra computation silently breaks closure. Rule of thumb: in any complex-field Noether computation, count your fields as $\{\phi, \phi^*\}$ from the first line, and let the c.c. structure of (2) do its job.

Worked examples

1 · The signed ledger with numbers (confidence builder)

A complex scalar in a box of volume $V = 1.00$ carries one particle-branch mode at rest and one antiparticle-branch mode: amplitudes $a = 0.30$ at $\mathbf{k}_1 = 0$ and $b = 0.20$ at $k_2 = 0.75$ , with $m = 1.00$ . Compute the charge $Q$ and the sign of each contribution.

Step 1. Mode frequencies: $\omega_1 = \sqrt{0 + 1} = \mathbf{1.000}$ ; $\omega_2 = \sqrt{0.5625 + 1} = \mathbf{1.250}$ .

Step 2. Apply the ledger from the derivation, $Q = 2\omega_1|a|^2 - 2\omega_2|b|^2$ (single modes, $V = 1$ ):

Q = 2(1.000)(0.090) - 2(1.250)(0.040) = 0.1800 - 0.1000 = \mathbf{+0.0800}

Step 3. Audit the signs. The positive-frequency mode contributes $+0.1800$ ; the negative-frequency mode $-0.1000$ : net positive — this configuration is "particle-rich." The energy of the same field, by contrast, adds both contributions with positive sign (as the stress-energy article's $T^{00}$ confirms): charge is the difference of two occupancies, energy their weighted sum. Set $b \to 0.2683$ and $Q = 0$ exactly while the field remains loud everywhere — a classically neutral, energetic state, the ancestor of a particle-antiparticle gas.

2 · The price of a local phase twist, and the gauge field that waives it (exam level)

A complex scalar sits at uniform magnitude $|\phi| = v = 0.50$ with a linearly winding phase: $\phi(x) = v\,e^{iqx}$ , $q = 2\pi/L$ , $L = 10.0$ , $m$ -term irrelevant (compare configurations at fixed $|\phi|$ ). (i) Compute the gradient-energy density and the total over one period — the cost of refusing to keep the compasses aligned. (ii) Show a gauge field $A_x$ waives the cost, and find its value.

Step 1. The twist wavenumber: $q = 2\pi/10.0 = 0.6283$ . Gradient term: $\partial_x\phi = iq\phi$ , so $|\partial_x\phi|^2 = q^2v^2 = (0.3948)(0.25) = \mathbf{0.0987}$ per unit volume — pure phase-stiffness cost; the magnitude never moved.

Step 2. Total over one period (unit cross-section): $E_{\text{twist}} = q^2v^2 L = 0.0987 \times 10.0 = \mathbf{0.987}$ . Compare the untwisted configuration ( $q = 0$ ): zero. A global rotation is free; a local one is taxed at $q^2v^2$ per unit volume — quadratic in the misalignment rate, exactly the $(\partial\alpha)^2|\phi|^2$ term of the gauging derivation with $\alpha = qx$ .

Step 4. The gauge field absorbs the twist: choose $A_x = q/e$ (with $e = 1$ : $A_x = \mathbf{0.6283}$ , constant) and the covariant gradient energy is $(q - q)^2v^2 = \mathbf{0}$ — the twisted configuration becomes exactly degenerate with the untwisted one, as local gauge invariance demands (the two are now gauge-equivalent descriptions of one physical state). The lesson in numbers: the global theory charges $0.987$ for the twist; the gauged theory recognizes the twist as unphysical bookkeeping and charges nothing. What the gauged theory does pay for is $F_{\mu\nu} \neq 0$ — curvature of $A$ , a genuinely misaligned connection — which a constant $A_x$ does not have. That distinction, twist versus curvature, is the entire geometry of gauge theory in one worked example.

Origins · 1 minHermann Weyl and how this was discovered

Hermann Weyl

1885–1955 · Göttingen, Zürich, Princeton

Weyl invented the gauge principle — twice, the second time correctly. In 1918, aiming to unify gravity and electromagnetism, he proposed that the scale ("gauge," as in rail gauge) of lengths might vary from point to point, with a compensating field restoring consistency; Einstein shot the idea down within weeks (atomic spectra would depend on a clock's history), but Weyl's mathematics — a local symmetry demanding a compensating vector field — was exactly right with the wrong group. After quantum mechanics revealed the complex phase of the wavefunction, Weyl returned in 1929 with the fix: the local symmetry is a phase rotation, the compensating field is the electromagnetic potential, and "gauge invariance" — the name survived the failed original — became the construction principle of this article's final derivation. His 1929 paper also introduced the two-component spinors (Weyl spinors) that describe neutrinos and organized the handedness structure the Standard Model is built on. A geometer-philosopher of the Göttingen school and Hilbert's successor, Weyl wrote the first mathematically serious book on general relativity (Raum, Zeit, Materie, 1918), founded the representation theory of compact Lie groups — the mathematics of this article's non-abelian charges — and, forced from Germany in 1933, spent his last decades at the Institute for Advanced Study. Yang and Mills's 1954 non-abelian gauge theory is his 1929 move run on isospin; asked late in life what he considered his greatest work, Weyl chose the gauge idea.

History

1918

— Weyl's first gauge theory: local scale invariance with a compensating field — right structure, wrong symmetry.

1926

— Fock and London note that quantum mechanics couples to electromagnetism through the phase:

p \to p - eA

is a phase bookkeeping rule.

1929

— Weyl's second gauge theory: local phase rotation of the matter field forces the electromagnetic potential — the gauge principle in its modern form.

1932

— Heisenberg introduces isospin: proton and neutron as one nucleon doublet — the first non-abelian internal multiplet.

1954

— Yang and Mills gauge isospin: non-abelian gauge fields, the classical template of the strong and weak interactions.

Going further

References. Goldstein, Poole & Safko, Classical Mechanics, 3rd ed., §13.7 · Landau & Lifshitz, The Classical Theory of Fields, §32 · Peskin & Schroeder, An Introduction to QFT, §2.2 and §15.1 (gauge principle context) · Weyl, "Elektron und Gravitation" (1929) — the original gauge-principle paper

Provenance. A verifier should: (1) re-derive the general internal Noether current from $\delta\mathcal{L}=0$ with $\delta x^\mu=0$ (Goldstein §13.7); (2) recompute $j^\mu=i(\phi^*\partial^\mu\phi-\phi\partial^\mu\phi^*)$ and confirm $\partial_\mu j^\mu=0$ using $\Box\phi+m^2\phi=0$ (Landau–Lifshitz Vol. 2 §32); (3) verify the Poisson-bracket charge algebra $\{Q^a,Q^b\}=f^{abc}Q^c$ from equal-time canonical brackets. All equations are standard textbook results; no numerical constants are asserted. Additionally for this upgrade: the two-frequency charge ledger $Q = \sum 2\omega_k(|a_{\mathbf{k}}|^2 - |b_{\mathbf{k}}|^2)$ recomputed with explicit cross-term cancellation; gauging derivation follows the standard minimal-coupling construction; worked-example numbers recomputed directly ( $\omega_1 = 1.000$ , $\omega_2 = 1.250$ , $Q = +0.0800$ , neutrality at $b = 0.2683$ ; twist cost $q^2v^2 = 0.0987$ , $E = 0.987$ , absorbed by $A_x = q/e = 0.6283$ ). Scope note: the $U(1)$ current's step-by-step Noether derivation is owned by the Noether article; the charge-algebra closure by the Poisson-brackets article; the energy tally by the stress-energy article; full gauge-field dynamics ( $F_{\mu\nu}$ , Maxwell action) belongs to the EM track.

Builds on: Noether's Theorem for Continuous Symmetries · Conserved Currents and Charges · The Lagrangian Density and Field Action · Covariant Euler-Lagrange Equations for Fields Leads to: Poisson Brackets for Fields · Lorentz Covariance of Classical Field Theory · The Canonical Stress-Energy Tensor

The charge density j⁰ is not positive-definite. Doesn't that make it useless as a density?

It makes it useless as a probability density — and that failure is historically momentous — but perfect as a charge density. Born's interpretation of $|\psi|^2$ requires a positive quantity, and early attempts to read the Klein-Gordon $j^0 = i(\phi^*\dot\phi - \phi\dot\phi^*)$ as a relativistic probability collapsed precisely because both signs occur (this is one of the reasons Dirac hunted a first-order equation). The resolution came from reinterpreting the object: $j^0$ is charge density, negative values signal antiparticle-branch content, and nothing is paradoxical about negative charge density — your table salt has some wherever the chlorine ions are. The two-frequency derivation makes the structure exact: positive- and negative-frequency parts contribute with opposite signs, so a region can be charge-negative while energy-positive. The nonrelativistic limit explains why Schrödinger never noticed: sending $c \to \infty$ freezes out the negative-frequency branch (the classical-limit article's business), one sign survives, and $j^0$ degenerates into the positive $|\psi|^2$ of quantum mechanics. So the non-positivity is not a bug in the formalism; it is the formalism announcing antiparticles seventy years early.

If gauging makes the phase unphysical, what happens to the conserved charge — is it gauged away too?

No — the charge survives as the physical, gauge-invariant observable, and understanding why sharpens both concepts. Gauge transformations are local: they can twist the phase differently from point to point, and all such twists are declared redundant descriptions. But the global rotation — the same $\alpha$ everywhere — sits inside the gauge group as its rigid subgroup, and quotienting by local transformations does not remove it: configurations still organize into classes labeled by the total charge, which no gauge transformation changes. Operationally the charge even gains a new, more robust definition after gauging: Gauss's law ties $Q$ to the flux of the gauge field through a distant surface ( $Q = \oint\mathbf{E}\cdot\mathrm{d}\mathbf{S}$ in electromagnetism), so the charge of a region can be read at its boundary without touching the matter inside — a formulation manifestly independent of any phase convention. This is also why gauge "symmetries" are best called redundancies while their global remnants are true symmetries with true Noether charges: the electron's charge is physics; the electron field's phase is bookkeeping.

Isospin is only approximate. How useful is a conservation law that is a few percent wrong?

Enormously — approximate symmetries with computable breaking are among the most predictive structures in physics, and isospin is the founding example. Treating the proton and neutron as an $SU(2)$ doublet organizes nuclear states into multiplets that would be degenerate at exact symmetry; the observed splittings (a few MeV on a GeV scale — the percent level) measure the breaking, which comes from the up-down quark mass difference and electromagnetism. The Noether framework handles this gracefully via the off-shell identity of the Noether article: $\partial_\mu j^{\mu a}$ equals the breaking term, so charges drift at a rate set by small parameters rather than being simply "not conserved." The predictive harvest is concrete: isospin relates cross-sections of different charge channels (famously, ratios like $\sigma(\pi^+ p)/\sigma(\pi^- p)$ at the $\Delta$ resonance), fixes Clebsch-Gordan-weighted branching ratios, and underwrote the eightfold-way generalization to $SU(3)$ flavor that predicted the $\Omega^-$ baryon — mass and all — before its 1964 discovery. The working principle: an exact symmetry classifies; a slightly broken one classifies and predicts the pattern of its own violations.

Where does the electromagnetic field's own charge bookkeeping fit — does the photon carry the charge it mediates?

For the abelian $U(1)$ of electromagnetism, no: the gauge field transforms by a pure shift ( $A_\mu \to A_\mu + \partial_\mu\alpha/e$ ), carries no phase of its own, and hence no charge — the photon is neutral, and the matter current alone satisfies the ordinary continuity equation. This is the quiet structural fact that makes electromagnetism the easy gauge theory. Non-abelian gauge fields are not so innocent: transforming in the adjoint representation, they carry the very charge they mediate (gluons are colored, W bosons carry weak isospin and electric charge), so charge continually flows between matter and gauge field, and only the total obeys a plain conservation law — the covariant-conservation subtlety flagged in the conserved-currents article. The observable consequences are dramatic: charged mediators self-interact, which bends the running of couplings (asymptotic freedom of QCD — the strong coupling weakens at short distance because gluon self-interaction anti-screens) and confines color so thoroughly that no isolated colored charge has ever been measured. One classical-field-theory sentence captures the divide: abelian charge is a spectator ledger; non-abelian charge is a participant.

What is the relation between the internal charge and the number of particles when the field is quantized?

Upon canonical quantization (the capstone's substitution applied to this chapter's brackets), the mode amplitudes of the two-frequency derivation become operators: $a_{\mathbf{k}} \to \hat a_{\mathbf{k}}$ annihilating particles, $b_{\mathbf{k}} \to \hat b_{\mathbf{k}}$ annihilating antiparticles, and the classical ledger $Q = \sum 2\omega_k(|a|^2 - |b|^2)$ becomes (with relativistic normalization) the operator $\hat Q = \sum_{\mathbf{k}}(\hat n^{(a)}_{\mathbf{k}} - \hat n^{(b)}_{\mathbf{k}})$ — literally particle count minus antiparticle count, with integer eigenvalues. Two refinements arrive with the hats. First, charge becomes quantized in units of the coupling — a fact the classical theory permits but does not explain (and which deeper structures like compactness of the gauge group or magnetic monopoles would enforce). Second, the conservation law upgrades from a number to a selection rule: $[\hat Q, \hat H] = 0$ (the quantum image of $\{Q, H\} = 0$ ) means transitions can only connect states of equal charge — pair production creates particle and antiparticle together because $\hat Q$ must not budge. Everything structural — the sign ledger, the algebra (4), the generator property — passes through quantization intact; that fidelity is the whole premise of building QFT on this chapter's classical scaffolding.

Can an internal symmetry mix fields of different masses?

Only at the price of being broken — and the pattern of breaking is itself informative. The Noether derivation required $\delta\mathcal{L} = 0$ ; a mass term $\tfrac12 m_a^2\phi_a^2$ (or $m^2_{ab}\phi_a\phi_b$ generally) is invariant under a rotation $\phi_a \to R_{ab}\phi_b$ only if the mass matrix commutes with the rotation, i.e. if the mixed fields are degenerate. Mix a light field with a heavy one and the symmetry survives in the kinetic terms but is explicitly broken by the mass difference: currents leak at a rate proportional to $m_1^2 - m_2^2$ . Nature exploits both sides of this rule. Where masses are nearly equal, approximate multiplets appear (proton-neutron isospin; the near-degenerate kaon pairs whose mixing gives the celebrated CP-violation phenomenology). Where the underlying symmetry is exact but the vacuum pays the mass difference — spontaneous breaking — the Goldstone structure of the Noether article's FAQ takes over instead: currents stay exactly conserved and massless modes appear. Reading a spectrum for its near-degeneracies, and attributing them to internal symmetries slightly broken, remains one of the standard discovery tools of both particle and condensed-matter physics — band structures, hadron multiplets, and molecular spectra all yield to it.