Symmetries and Conservation in Quantum Dynamics

Symmetry-first reasoning is the default of modern physics. The Standard Model is defined by its gauge symmetry group, with conserved charges as the generators — this article's theorem, iterated locally; particle multiplets (the pion triplet, quark flavor families) are the forced degeneracies of Derivation B at accelerator scale. In condensed matter the last two decades reorganized entire phase diagrams by symmetry: topological insulators are classified by which discrete symmetries (time reversal, particle-hole) protect their edge states, and "symmetry-protected" is precisely "degeneracy forced by a non-commuting pair, robust until the symmetry breaks". The laboratory face is symmetry breaking as a precision instrument: atomic parity-violation experiments measure the weak interaction's parity breaking inside a single atom to parts in $10^{11}$, and optical-clock comparisons watch nominally conserved quantities for drifts that would betray new symmetry-violating physics — the warning box's "measured drift indicts the Hamiltonian" run as a discovery program.

Slide the entire experiment three meters to the left. If the world is translation-invariant, no measurement can notice — and that unnoticeability is not a platitude but an operator equation: the sliding unitary commutes with $\hat{H}$. Now recall what the unitary chapter taught: continuous unitary families have hermitian generators, $\hat{U}(\varepsilon)=e^{-i\varepsilon\hat{G}/\hbar}$. The generator of sliding is momentum (the translation family of Canonical Commutation Relation). So "sliding is unnoticeable" and "momentum commutes with $\hat{H}$" are the same sentence — and by the previous article, that sentence means momentum's entire distribution is frozen. Symmetry has paid out a conservation law, without a single dynamical calculation.

The second dividend has no classical parallel worth the name. Suppose two symmetries hold at once but refuse to commute with each other — rotations about different axes, say. Apply one to an energy eigenstate: the result has the same energy (symmetry commutes with $\hat{H}$). If it were always the same state up to phase, both symmetries would act as mere phases and would effectively commute. Since they don't, somewhere the transformed state is a genuinely new state at the old energy: the level is forced to hold more than one state. Non-commuting symmetries sculpt degenerate multiplets — and the size and structure of the multiplets is pure representation theory, fixed before any Schrödinger equation is solved. The machine-checked 2D oscillator below shows level $n$ holding exactly $n{+}1$ states, with $\hat{L}_z$ ticking through them.

Two layers of definition, one operational and one dynamical. Operationally, a symmetry transformation is a map of states preserving all measurable structure — every $|\langle\varphi|\psi\rangle|^2$. Wigner's theorem (stated and owned by Unitary Operators and Transformations) says such maps come in exactly two kinds: unitary or antiunitary. This article works the unitary branch; the antiunitary branch has one physical citizen — time reversal — and is deferred by name to its own article. Dynamically, a symmetry of the system must also respect the Hamiltonian:

where: the equivalence is one rearrangement (multiply by $\hat{U}_s$ from the left). Reading (1) forward: transformed states obey the same dynamics, since $\hat{U}_s$ commutes with every power of $\hat{H}$ and hence with $\hat{U}(t)=e^{-i\hat{H}t/\hbar}$ — symmetry operations and time evolution can be performed in either order. Reading it as spectroscopy: $\hat{U}_s$ maps each energy eigenspace into itself's energy — the degeneracy engine below. For a continuous symmetry the criterion compresses onto the generator. The unitary chapter's generator doctrine gives every strongly continuous family the form $\hat{U}(\varepsilon)=e^{-i\varepsilon\hat{G}/\hbar}$ with $\hat{G}$ hermitian; the theorem is:

where: the left side is infinitely many conditions, the right side one commutator — the theorem's practical force is exactly this compression. And the right side is the previous article's criterion verbatim: the generator is a constant of the motion, full frozen-census theory included. Equation (2) is quantum mechanics' Noether theorem — leaner than the classical one (no Lagrangians, no boundary terms), because in quantum mechanics both symmetries and observables are already operators on the same stage.

The catalogue then writes itself, one commutator per row. Translation → momentum: $\hat{T}(a)=e^{-ia\hat{p}/\hbar}$ (the series proof that it shifts wavefunctions is Canonical Commutation Relation's); $[\hat{p},\hat{H}]=0$ iff $V'=0$ — free particles and translation-invariant systems conserve momentum, and the verifier's two-particle check $[\hat{P}_{\rm tot},V(x_1-x_2)]=0$ shows the interacting version: internal forces respect total momentum. Time translation → energy: the family is $\hat{U}(t)$ itself and the generator is $\hat{H}$ — the self-referential row, conserved for free when static. Rotation → angular momentum: stated here, owned by the next chapter, where $[\hat{L}_i,\hat{L}_j]=i\hbar\epsilon_{ijk}\hat{L}_k$ makes the symmetries mutually non-commuting — the degeneracy engine's showcase. Global phase → identity: $e^{i\theta}$ costs nothing and conserves the norm — the symmetry reading of Conservation of Normalization's theorem.

Two structural notes complete the theory. First, the symmetries of a given $\hat{H}$ form a group (products and inverses of symmetries are symmetries — check (1) twice), and their conserved generators close under commutation into the algebra the previous article's Jacobi harvest built; non-abelian group $\leftrightarrow$ non-commuting constants $\leftrightarrow$ forced multiplets is one triangle of equivalences. The classification of which continuous groups exist at all — the shelf quantum mechanics shops from — was completed by Cartan before quantum mechanics needed it (portrait below). Second, the active–passive card: transforming the state ($|\psi\rangle\to\hat{U}_s|\psi\rangle$, apparatus fixed) and transforming the apparatus (operators $\hat{A}\to\hat{U}_s^\dagger\hat{A}\hat{U}_s$, states fixed) produce identical physics when (1) holds — a symmetry is exactly a transformation for which the active and passive readings cannot be told apart. (Transforming both is the picture-change of Equivalence of the Dynamical Pictures — bookkeeping, no physics, no condition on $\hat{H}$ needed.) Discrete symmetries join through criterion (1) directly, no generator attached. Parity is the workhorse: unitary, hermitian, its own inverse ($\hat{\Pi}^2=\hat{1}$ — all three machine-checked on a grid representation), conserved whenever $V(-x)=V(x)$ (Parity's theorem; the frozen-census demo ran in the previous article). Time reversal is the deliberate exception to everything above: Wigner's theorem's antiunitary branch, complex-conjugating, with its own strange arithmetic ($\hat{\Theta}^2=-1$ for half-integer spin and the Kramers degeneracy it forces) — all owned by Time Reversal in Quantum Dynamics at the chapter's end. And one line of honesty, bought expensively in 1956–57: parity conservation in the weak interaction was assumed for thirty years and is false — Lee and Yang questioned it, Wu's cobalt-60 experiment answered, and "symmetries are hypotheses subject to measurement" became a permanent lesson of the field.

Example 1 — reading a quantum dot's spectrum by symmetry alone. A 2D isotropic harmonic trap (a gate-defined quantum dot) has $\hbar\omega=3$ meV. Predict the structure of its fourth level before solving anything.

Step 1. Symmetry census: the trap is rotation-invariant in the plane, so $[\hat{L}_z,\hat{H}]=0$ — and reflections about in-plane axes are also symmetries, which fail to commute with rotations. Degeneracy is forced (Derivation B, Step 3).

Step 2. The multiplet size: representation theory of the 2D oscillator gives level $n$ a degeneracy $g=n+1$. For the fourth level, $n=3$: $g=$ 4 — machine-checked by exact diagonalization in a two-mode Fock basis.

Step 3. The labels: $\hat{L}_z$ ticks through the multiplet with eigenvalues $m\hbar$, $m=-3,-1,+1,+3$ (steps of 2 — a parity constraint within the level; verified numerically). The conserved label sorts the degenerate states into an address book, the CSCO story with dynamics attached.

Step 4. The energy: $E_3=(3+1)\hbar\omega=$ 12 meV above the trap bottom. Everything in Steps 1–3 came from symmetry; the only "solving" was the one number $\hbar\omega$. When a real dot's levels split this multiplet, the splitting pattern reads out which symmetry the device broke — anisotropy splits differently than a magnetic field does.

Example 2 — degeneracy you have already met: the free particle's ±k pair. An electron plane wave has de Broglie wavelength $\lambda=1$ nm. Diagnose its degeneracy by symmetry.

Step 1. The energy: $E=h^2/2m_e\lambda^2=$ 1.50 eV — shared exactly by the two states $e^{\pm ikx}$, movers to the right and left.

Step 2. The symmetry diagnosis: the free Hamiltonian admits both translations $\hat{T}(a)$ and parity $\hat{\Pi}$, and they do not commute — $\hat{\Pi}\hat{T}(a)=\hat{T}(-a)\hat{\Pi}$ (flipping space reverses the direction of a shift). Non-commuting symmetry pair ⇒ forced degeneracy: parity maps $e^{ikx}$ into the distinct same-energy state $e^{-ikx}$.

Step 3. The two labeling schemes: diagonalize $\{\hat{p}\}$ (travelling waves, momentum labels $\pm\hbar k$) or $\{\hat{H},\hat{\Pi}\}$ (standing waves $\cos kx$, $\sin kx$, parity labels $\pm$) — the two CSCOs of Simultaneous Diagonalization's free-particle exhibit, now explained: two incompatible symmetries offer two incompatible address books for the same forced multiplet.

Step 4. The takeaway: whenever an energy level repeats, hunt the non-commuting symmetry pair. Bound states in 1D never show this (no degeneracy — Non-Degeneracy of 1D Bound States), precisely because their boundary conditions break the translation half of the pair.

References. Sakurai & Napolitano, Modern Quantum Mechanics, 3rd ed., Ch. 3–4 · Shankar, Principles of Quantum Mechanics, 2nd ed., Ch. 11 · Cohen-Tannoudji, Diu & Laloë, Quantum Mechanics, Vol. II · Weyl, The Theory of Groups and Quantum Mechanics (1928; Dover) · Gross, "The role of symmetry in fundamental physics", PNAS 93, 14256 (1996).

Provenance. All commutation relations and generator forms are standard (Sakurai 3rd ed. Ch. 3–4; Shankar Ch. 11; Cohen-Tannoudji Vol. 2). A verifier should: (1) confirm $\hat{p}=-i\hbar\partial_x$ generates translation via Taylor expansion of $\psi(x-a)$; (2) re-derive $\mathrm{d}\langle p\rangle/\mathrm{d}t=-\langle\partial_x V\rangle$ from the Heisenberg/Ehrenfest equation; (3) check $[J_z,J_\pm]=\pm\hbar J_\pm$ and the $(2j+1)$ dimension count; (4) verify the hydrogen $SO(4)$ bound-state representations have dimension $n^2$. No nonstandard numerical values are claimed. For this edition the generator theorem was machine-executed both directions (invariance over an $\varepsilon$-grid to $10^{-11}$; first-order extraction numeric and symbolic), plus: $[\hat{T}(a),\hat{H}_{\rm free}]=0$ exact; $[\hat{p},V(x)]=-i\hbar V'$ and the two-particle $[\hat{P}_{\rm tot},V(x_1{-}x_2)]=0$ symbolic; the degeneracy-forcing 4×4 exhibit; the 2D oscillator ($[\hat{L}_z,\hat{H}]=0$, $g(3)=4$, $\hat{L}_z$ eigenvalues $\{-3,-1,1,3\}\hbar$); parity's grid credentials $\hat{\Pi}^2=\hat{1}=\hat{\Pi}^\dagger\hat{\Pi}$ and $[\hat{H},\hat{\Pi}]=0$ for symmetric quartic; and the worked numbers (12 meV; 1.50 eV). Items (1)–(2) are owned upstream (Canonical Commutation Relation; Ehrenfest's Theorem) and were consumed by citation; items (3)–(4) are deliberately NOT executed here — the angular-momentum algebra belongs to the ANGM chapter and the SO(4) count to the hydrogen chapter, both teased by name only. Per the chapter ownership map: Wigner's theorem statement and the unitary/generator doctrine are Unitary Operators and Transformations'; conservation theory is Constants of the Motion's (consumed); time reversal and antiunitarity → Time Reversal in Quantum Dynamics BY NAME; field-theoretic Noether → the relativistic track.

Builds on: Constants of the Motion and Conservation Laws · Unitary Operators and Transformations · Canonical Commutation Relation · The Commutator and Compatible Observables · Parity

Leads to: Angular momentum and rotations (ANGM chapter) · Central potentials and hydrogen's SO(4) · Time Reversal in Quantum Dynamics · The Quantum Propagator