Chapter 10: Symmetry and Conservation Laws — Why Quantum Mechanics Loves Group Theory

Symmetry: The Most Powerful Tool in Physics

"It is only slightly overstating the case to say that physics is the study of symmetry." — Philip W. Anderson

There is a hierarchy of ideas in physics. At the top, above the equations of motion, above the Hamiltonian, above even the choice of dynamical variables, sits symmetry. If you know the symmetries of a system, you know an extraordinary amount about its physics — often before you solve a single equation. You know which quantities are conserved. You know which matrix elements vanish. You know which states are degenerate. You know how to classify the spectrum. And, in quantum mechanics, this power reaches its most elegant form.

Consider what you already know, even if you have not framed it in these terms. In Chapter 5, the hydrogen atom's energy levels turned out to depend only on the principal quantum number $n$, not on $l$ or $m$. Why? Rotational symmetry. In Chapter 4, the QHO eigenstates were alternately even and odd functions. Why? Parity symmetry. In Chapter 7, you learned that stationary states have time-independent probability distributions. Why? Time-translation symmetry. Every clean, elegant result you have encountered so far has symmetry hiding behind it.

This chapter makes these hidden connections explicit. It connects three ideas that turn out to be the same idea viewed from three angles:

1. Symmetry: A transformation that leaves the physics unchanged.
2. Conservation law: A quantity whose expectation value does not change with time.
3. Commutation with the Hamiltonian: An operator $\hat{G}$ satisfying $[\hat{H}, \hat{G}] = 0$.

The unification of these three perspectives is not merely aesthetically satisfying — it is the most powerful problem-solving tool in theoretical physics. Before you compute a single integral, symmetry tells you which integrals are zero, which energy levels are degenerate, and which quantum numbers are "good." Mastering symmetry does not replace computation; it tells you which computations are worth doing.

The theorem that unifies these — the quantum Noether's theorem — is, without exaggeration, the most powerful single result in theoretical physics. Emmy Noether proved the classical version in 1918. The quantum version follows from the operator formalism you learned in Chapter 8. By the end of this chapter, you will be able to derive it yourself, apply it to every fundamental symmetry in physics, and use it to solve problems that would be intractable by brute force.

🔗 Connection (Ch 8) — In Chapter 8, you learned that unitary operators $\hat{U}$ satisfy $\hat{U}^\dagger\hat{U} = \hat{I}$ and preserve inner products: $\langle\phi'|\psi'\rangle = \langle\phi|\psi\rangle$. That was abstract algebra. Here is the physical reason unitary operators matter: symmetry transformations in quantum mechanics are unitary operators. Every symmetry you have ever encountered — translation, rotation, reflection, time reversal — is implemented by a specific operator acting on the Hilbert space. The machinery of Chapter 8 was built for exactly this purpose.

🔗 Connection (Ch 5) — In Chapter 5, you encountered the angular momentum quantum numbers $l$ and $m$ in the hydrogen atom solution. At the time, these arose from the separation of variables in spherical coordinates. But where do they really come from? The answer is rotational symmetry. The quantum numbers $l$ and $m$ are labels for irreducible representations of the rotation group $SO(3)$. This chapter explains what that sentence means.

Let us make the general idea precise.

10.1 What Is a Symmetry?

The classical definition

In classical mechanics, a symmetry is a transformation of the dynamical variables that leaves the equations of motion unchanged. If you translate a free particle's position by a constant vector $\mathbf{a}$ — replacing $\mathbf{x}$ with $\mathbf{x} + \mathbf{a}$ — Newton's second law $\mathbf{F} = m\ddot{\mathbf{x}}$ is unchanged (because $\mathbf{F} = 0$ for a free particle, and the acceleration of $\mathbf{x} + \mathbf{a}$ equals the acceleration of $\mathbf{x}$). Translation is a symmetry of the free particle.

More generally, if the Lagrangian $L(\mathbf{q}, \dot{\mathbf{q}})$ is invariant under a transformation $\mathbf{q} \to \mathbf{q}'(\mathbf{q})$, then the Euler-Lagrange equations are invariant, and — by Noether's theorem — there is a conserved quantity associated with that transformation.

The quantum definition

In quantum mechanics, the "equations of motion" are replaced by the Schrodinger equation:

$$i\hbar \frac{\partial}{\partial t}|\psi(t)\rangle = \hat{H}|\psi(t)\rangle$$

A symmetry is an operation $\hat{U}$ on the Hilbert space that satisfies two requirements:

1. It preserves all physical predictions. Since all predictions in quantum mechanics come from inner products $\langle\phi|\psi\rangle$ (probabilities, expectation values, transition amplitudes), the transformation must preserve these: $\langle\phi'|\psi'\rangle = \langle\phi|\psi\rangle$ for all states.

2. It is compatible with time evolution. The transformed state $|\psi'(t)\rangle = \hat{U}|\psi(t)\rangle$ must also satisfy the Schrodinger equation — possibly with a transformed Hamiltonian $\hat{H}'$.

The first requirement, by Wigner's theorem (1931), implies that $\hat{U}$ must be either unitary ($\hat{U}^\dagger\hat{U} = \hat{I}$, linear) or antiunitary ($\hat{U}^\dagger\hat{U} = \hat{I}$, antilinear). Most symmetries are unitary. The sole important exception is time reversal, which is antiunitary — we will treat it in Section 10.7.

💡 Key Insight — Wigner's theorem is the foundational result of quantum symmetry analysis. It says: the only way to map rays in Hilbert space to rays in Hilbert space while preserving transition probabilities is via a unitary or antiunitary operator. This is not a postulate — it is a mathematical theorem. The structure of Hilbert space itself constrains what "symmetry" can mean.

If $\hat{U}$ is a symmetry of the system — meaning the physics is the same before and after the transformation — then the second requirement forces the Hamiltonian to commute with $\hat{U}$:

$$\hat{U}\hat{H}\hat{U}^\dagger = \hat{H} \qquad \Longleftrightarrow \qquad [\hat{H}, \hat{U}] = 0$$

This is the quantum statement that "$\hat{U}$ is a symmetry." The Hamiltonian is invariant under $\hat{U}$.

✅ Checkpoint — Before proceeding, verify that you can derive the equivalence above. Start from $i\hbar\frac{\partial}{\partial t}|\psi'\rangle = \hat{H}|\psi'\rangle$ with $|\psi'\rangle = \hat{U}|\psi\rangle$, and use the fact that $\hat{U}$ is time-independent. You should arrive at $\hat{U}\hat{H} = \hat{H}\hat{U}$.

Three examples to build intuition

Example 1: Free particle and translation. The free-particle Hamiltonian $\hat{H} = \hat{p}^2/2m$ is invariant under spatial translations $\hat{T}(\mathbf{a})$. We will construct $\hat{T}(\mathbf{a})$ explicitly in Section 10.4 and verify that $[\hat{H}, \hat{T}(\mathbf{a})] = 0$.

Example 2: Central potential and rotation. The hydrogen atom Hamiltonian depends only on $r = |\mathbf{r}|$, not on the direction of $\mathbf{r}$. It is invariant under all rotations $\hat{R}(\hat{n}, \theta)$. This is why angular momentum is conserved, and it is the origin of the quantum numbers $l$ and $m$ you met in Chapter 5.

Example 3: QHO and parity. The harmonic oscillator potential $V(x) = \frac{1}{2}m\omega^2 x^2$ is even: $V(-x) = V(x)$. The parity operator $\hat{\Pi}$ that sends $x \to -x$ commutes with $\hat{H}$. This is why QHO eigenstates are either even or odd functions — they are eigenstates of $\hat{\Pi}$ with eigenvalues $\pm 1$.

The power of symmetry: what it tells you before you compute

Let us pause and appreciate what symmetry analysis gives you for free. Consider a particle in an unknown potential $V(r)$ that depends only on $r$ (a central potential). You have never seen this potential before. You know nothing about its specific form. Yet, before solving a single equation, symmetry tells you:

1. Angular momentum is conserved: $[\hat{H}, \hat{L}^2] = [\hat{H}, \hat{L}_z] = 0$.
2. Energy eigenstates can be labeled by $l$ and $m$: $|E, l, m\rangle$.
3. $(2l+1)$-fold degeneracy: For each $l$, the $2l+1$ states with $m = -l, \ldots, l$ all have the same energy.
4. Selection rules: The matrix element $\langle E', l', m'|\hat{r}|E, l, m\rangle$ vanishes unless $\Delta l = \pm 1$ (parity) and $\Delta m = 0, \pm 1$ (angular momentum addition rules).
5. Radial-angular separation: The angular part of every eigenstate is a spherical harmonic $Y_l^m(\theta, \phi)$. Only the radial part depends on the specific potential.

All of this follows from $V = V(r)$ alone — rotational symmetry. You still need to solve the radial equation to find the specific energy levels, but the structure of the problem is completely determined by symmetry. This is the principle at work throughout modern physics, from atomic spectroscopy to particle physics: symmetry constrains; dynamics fills in the details.

10.2 Symmetry Transformations as Unitary Operators

Continuous vs. discrete symmetries

Symmetries come in two fundamental types:

• Continuous symmetries can be performed "a little bit at a time." Translations, rotations, and time evolution are continuous — you can translate by an infinitesimally small amount $\epsilon$, rotate by an infinitesimally small angle $d\theta$.

• Discrete symmetries are all-or-nothing. Parity (spatial reflection) either flips the coordinates or it does not. There is no "half a parity transformation."

This distinction matters because continuous symmetries have a beautiful mathematical structure: they form Lie groups, and the corresponding unitary operators are generated by Hermitian operators (the generators) via exponentiation.

The generator of a continuous symmetry

Consider a continuous family of unitary transformations $\hat{U}(\epsilon)$ parameterized by a real number $\epsilon$, with $\hat{U}(0) = \hat{I}$ (the identity). For small $\epsilon$, we can expand:

$$\hat{U}(\epsilon) = \hat{I} - \frac{i\epsilon}{\hbar}\hat{G} + O(\epsilon^2)$$

The operator $\hat{G}$ is called the generator of the transformation. The factor of $-i/\hbar$ is a convention that ensures $\hat{G}$ is Hermitian when $\hat{U}$ is unitary. Let us verify this.

Proof that $\hat{G}$ is Hermitian: Unitarity requires $\hat{U}^\dagger(\epsilon)\hat{U}(\epsilon) = \hat{I}$. Substituting the expansion:

$$\left(\hat{I} + \frac{i\epsilon}{\hbar}\hat{G}^\dagger\right)\left(\hat{I} - \frac{i\epsilon}{\hbar}\hat{G}\right) = \hat{I}$$

Expanding to first order in $\epsilon$:

$$\hat{I} + \frac{i\epsilon}{\hbar}(\hat{G}^\dagger - \hat{G}) + O(\epsilon^2) = \hat{I}$$

Therefore $\hat{G}^\dagger = \hat{G}$. The generator is Hermitian. $\square$

Since $\hat{G}$ is Hermitian, it is an observable — it corresponds to a measurable physical quantity. This is the first hint that symmetries and conserved quantities are related: the generator of a symmetry transformation is itself a physical observable.

Exponentiation: from infinitesimal to finite

To build the finite transformation $\hat{U}(\alpha)$ from its infinitesimal version, we compose $N$ infinitesimal transformations with $\epsilon = \alpha/N$ and take $N \to \infty$:

$$\hat{U}(\alpha) = \lim_{N\to\infty}\left(\hat{I} - \frac{i\alpha}{N\hbar}\hat{G}\right)^N = e^{-i\alpha\hat{G}/\hbar}$$

This is the operator exponential. You encountered it in Chapter 7, where the time evolution operator $\hat{U}(t) = e^{-i\hat{H}t/\hbar}$ is the exponential of the Hamiltonian. That was a specific case of the general principle: every continuous symmetry is generated by exponentiation of a Hermitian operator.

💡 Key Insight — The relationship between Lie groups and Lie algebras, expressed in the equation $\hat{U}(\alpha) = e^{-i\alpha\hat{G}/\hbar}$, is the mathematical backbone of modern physics. The group elements $\hat{U}(\alpha)$ perform the symmetry transformations. The algebra elements $\hat{G}$ generate those transformations and are themselves the physical observables (momentum, angular momentum, energy, etc.). Group theory is not an optional add-on to quantum mechanics — it is the language in which the theory is written.

🔵 Historical Note — The connection between Lie groups and quantum mechanics was first made explicit by Hermann Weyl in his 1928 book Gruppentheorie und Quantenmechanik. At the time, many physicists resisted the use of group theory, calling it the "Gruppenpest" (group plague). By 1950, it had become indispensable. By 2000, it was unthinkable to do theoretical physics without it. Eugene Wigner, who received the Nobel Prize in 1963 partly for his application of group theory to physics, once remarked that the "unreasonable effectiveness of group theory" was a miracle on par with the "unreasonable effectiveness of mathematics" itself.

The fundamental table

Here is the master table connecting symmetries, generators, and conserved quantities. We will derive each row in the sections that follow.

🔄 Retrieval Practice (Ch 8) — The time evolution operator $\hat{U}(t) = e^{-i\hat{H}t/\hbar}$ was introduced in Chapter 8 as the solution to the Schrodinger equation for time-independent Hamiltonians. Without looking back, can you show that $\hat{U}(t)$ is unitary? (Hint: use $\hat{H}^\dagger = \hat{H}$.) Can you verify that $\hat{U}(t_1)\hat{U}(t_2) = \hat{U}(t_1 + t_2)$? These are the defining properties of a one-parameter unitary group — exactly the structure we are now generalizing.

10.3 Continuous Symmetries and Conservation Laws: The Quantum Noether Theorem

The classical Noether theorem (brief review)

In classical Lagrangian mechanics, Noether's theorem states: if the Lagrangian is invariant under a continuous transformation parameterized by $\epsilon$, then there exists a conserved quantity — a function of the coordinates and momenta that does not change along any trajectory.

The proof uses the Euler-Lagrange equations and the chain rule. It is one of the most beautiful results in classical physics, but we will not reproduce it here (see Further Reading). Instead, we derive its quantum counterpart directly.

The quantum Noether theorem

Theorem. Let $\hat{G}$ be the Hermitian generator of a continuous unitary transformation $\hat{U}(\alpha) = e^{-i\alpha\hat{G}/\hbar}$. Then the following three statements are equivalent:

(i) The transformation is a symmetry: $\hat{U}(\alpha)\hat{H}\hat{U}^\dagger(\alpha) = \hat{H}$ for all $\alpha$.

(ii) The generator commutes with the Hamiltonian: $[\hat{H}, \hat{G}] = 0$.

(iii) The expectation value of $\hat{G}$ is conserved in time: $\frac{d}{dt}\langle\hat{G}\rangle = 0$ for any state $|\psi(t)\rangle$.

Proof:

(i) $\Leftrightarrow$ (ii): Statement (i) says $[\hat{H}, \hat{U}(\alpha)] = 0$ for all $\alpha$. Differentiate with respect to $\alpha$ at $\alpha = 0$:

$$0 = \frac{d}{d\alpha}\bigg|_{\alpha=0} [\hat{H}, \hat{U}(\alpha)] = \left[\hat{H}, \frac{d\hat{U}}{d\alpha}\bigg|_{\alpha=0}\right] = \left[\hat{H}, -\frac{i}{\hbar}\hat{G}\right] = -\frac{i}{\hbar}[\hat{H}, \hat{G}]$$

Therefore $[\hat{H}, \hat{G}] = 0$. Conversely, if $[\hat{H}, \hat{G}] = 0$, then $[\hat{H}, \hat{G}^n] = 0$ for all $n$ (by induction), so $[\hat{H}, e^{-i\alpha\hat{G}/\hbar}] = 0$. $\checkmark$

(ii) $\Leftrightarrow$ (iii): The Ehrenfest theorem (which you derived in Chapter 7) gives:

$$\frac{d}{dt}\langle\hat{G}\rangle = \frac{1}{i\hbar}\langle[\hat{G}, \hat{H}]\rangle + \left\langle\frac{\partial\hat{G}}{\partial t}\right\rangle$$

If $\hat{G}$ has no explicit time dependence (which is the case for the generators we consider — $\hat{p}$, $\hat{L}$, etc.), the second term vanishes, and:

$$\frac{d}{dt}\langle\hat{G}\rangle = \frac{1}{i\hbar}\langle[\hat{G}, \hat{H}]\rangle$$

This is zero for all states if and only if $[\hat{G}, \hat{H}] = 0$. $\square$

This is the quantum Noether theorem. It is simpler than its classical counterpart because quantum mechanics is already formulated in terms of operators and commutators. The connection between symmetry and conservation is not an additional theorem bolted onto the formalism — it is built into the operator algebra from the start.

Let us also note the beauty of the logical structure. The Ehrenfest theorem from Chapter 7 gave us a general equation for the time evolution of expectation values. The algebraic structure of Chapter 8 (Hermitian operators, unitary transformations, commutators) provided the language. And now Noether's theorem emerges naturally from combining these pieces. We did not need any new postulates — only the recognition that the machinery we already built has profound implications.

Worked Example: Energy conservation from time-translation symmetry. The time-evolution operator is $\hat{U}(t) = e^{-i\hat{H}t/\hbar}$. This is a continuous unitary transformation with generator $\hat{G} = \hat{H}$. Does the Hamiltonian commute with itself? Trivially: $[\hat{H}, \hat{H}] = 0$. Therefore energy is conserved: $\frac{d}{dt}\langle\hat{H}\rangle = 0$. This is almost embarrassingly simple, yet it captures the most fundamental conservation law in physics. (The caveat: this holds for time-independent Hamiltonians. If $\hat{H}$ depends explicitly on $t$, time-translation symmetry is broken, and energy is not conserved — think of a particle in a time-varying external field.)

💡 Key Insight — The quantum Noether theorem is actually stronger than its classical analogue. In classical mechanics, Noether's theorem tells you that a certain function of the phase-space variables is constant along trajectories. In quantum mechanics, statement (ii) says much more: $[\hat{H}, \hat{G}] = 0$ means that $\hat{H}$ and $\hat{G}$ can be simultaneously diagonalized (by the spectral theorem from Chapter 9). This means you can label energy eigenstates by the eigenvalues of $\hat{G}$. Symmetry does not just conserve a quantity — it organizes the entire spectrum.

🔵 Historical Note — Amalie Emmy Noether (1882--1935) was one of the greatest mathematicians of the twentieth century. Her 1918 theorem connecting symmetries and conservation laws was originally proved in the context of general relativity — David Hilbert and Felix Klein asked her to resolve a puzzle about energy conservation in Einstein's theory. The result she produced transcended the original question and became the foundation of modern theoretical physics. Noether faced extraordinary obstacles as a woman in early-twentieth-century German academia. She was not permitted to hold a formal position at Gottingen until 1919, lecturing for years under Hilbert's name. Einstein called her "the most significant creative mathematical genius thus far produced since the higher education of women began." Her theorem is the subject of Case Study 1.

Consequences for the energy spectrum

If $[\hat{H}, \hat{G}] = 0$, then:

1. Simultaneous eigenstates exist. There is a basis $\{|E_n, g_k\rangle\}$ where each state is an eigenstate of both $\hat{H}$ and $\hat{G}$:

$$\hat{H}|E_n, g_k\rangle = E_n|E_n, g_k\rangle, \qquad \hat{G}|E_n, g_k\rangle = g_k|E_n, g_k\rangle$$

2. Degeneracy is explained by symmetry. If $\hat{U}(\alpha)$ is a symmetry and $|E\rangle$ is an energy eigenstate, then $\hat{U}(\alpha)|E\rangle$ is also an energy eigenstate with the same energy. Different states related by the symmetry transformation form a multiplet — a set of degenerate states that transform into each other. The dimension of the multiplet is the dimension of the irreducible representation of the symmetry group.

3. Selection rules follow. Matrix elements $\langle E', g'|\hat{O}|E, g\rangle$ of an operator $\hat{O}$ vanish unless certain relationships between the quantum numbers hold. These selection rules are derived entirely from symmetry, without solving the Schrodinger equation.

⚠️ Common Misconception — Students sometimes think that $[\hat{H}, \hat{G}] = 0$ means $\hat{G}$ and $\hat{H}$ are the same operator, or that they have the same eigenvalues. They are not and they do not. It means they share a common eigenbasis — but the eigenvalues are completely independent. For example, $[\hat{H}, \hat{L}^2] = 0$ for the hydrogen atom, but $\hat{H}$ has eigenvalues $-13.6\,\text{eV}/n^2$ and $\hat{L}^2$ has eigenvalues $\hbar^2 l(l+1)$. They share eigenstates $|n, l, m\rangle$ but have completely different spectra.

10.4 Translation Symmetry and Momentum Conservation

Constructing the translation operator

The spatial translation operator $\hat{T}(\mathbf{a})$ shifts every position by a constant vector $\mathbf{a}$. In position representation, its action on a wave function is:

$$\hat{T}(\mathbf{a})\psi(\mathbf{x}) = \psi(\mathbf{x} - \mathbf{a})$$

Note the minus sign: $\hat{T}(\mathbf{a})$ takes the state whose probability is peaked at $\mathbf{x} = \mathbf{x}_0$ and moves it to $\mathbf{x} = \mathbf{x}_0 + \mathbf{a}$. To see why, note that the translated wave function evaluates at $\mathbf{x}$ what the original evaluated at $\mathbf{x} - \mathbf{a}$.

In one dimension, for an infinitesimal translation by $\epsilon$:

$$\hat{T}(\epsilon)\psi(x) = \psi(x - \epsilon) = \psi(x) - \epsilon\frac{d\psi}{dx} + O(\epsilon^2)$$

In Dirac notation:

$$\hat{T}(\epsilon) = \hat{I} - \epsilon\frac{d}{dx} + O(\epsilon^2) = \hat{I} - \frac{i\epsilon}{\hbar}\hat{p} + O(\epsilon^2)$$

where we used $\hat{p} = -i\hbar\frac{d}{dx}$. Comparing with $\hat{U}(\epsilon) = \hat{I} - \frac{i\epsilon}{\hbar}\hat{G}$, we identify:

$$\boxed{\text{The generator of spatial translations is the momentum operator } \hat{p}.}$$

The finite translation operator is therefore:

$$\hat{T}(a) = e^{-i\hat{p}a/\hbar}$$

And in three dimensions:

$$\hat{T}(\mathbf{a}) = e^{-i\hat{\mathbf{p}}\cdot\mathbf{a}/\hbar}$$

The momentum eigenstates as irreducible representations

Here is a perspective that connects directly to the language of group theory. The translation operators $\hat{T}(a)$ form a group (the translation group). An eigenstate of $\hat{T}(a)$ satisfies:

$$\hat{T}(a)|p\rangle = e^{-ipa/\hbar}|p\rangle$$

The eigenvalue $e^{-ipa/\hbar}$ is a one-dimensional representation of the translation group (a homomorphism from the group to the complex numbers). Each value of $p$ gives a different representation. The momentum eigenstate $|p\rangle$ "carries" the representation labeled by $p$.

This is the pattern we will see repeatedly: quantum numbers label representations of symmetry groups. The quantum number $p$ labels representations of the translation group. The quantum numbers $l$ and $m$ label representations of the rotation group. The quantum number $\pi = \pm 1$ labels representations of the parity group $\mathbb{Z}_2$. Understanding this pattern is the key to understanding why quantum mechanics is organized the way it is.

Verifying the group structure

Translation operators form an abelian group — the order does not matter:

$$\hat{T}(\mathbf{a})\hat{T}(\mathbf{b}) = e^{-i\hat{\mathbf{p}}\cdot\mathbf{a}/\hbar}e^{-i\hat{\mathbf{p}}\cdot\mathbf{b}/\hbar} = e^{-i\hat{\mathbf{p}}\cdot(\mathbf{a}+\mathbf{b})/\hbar} = \hat{T}(\mathbf{a} + \mathbf{b})$$

The second equality holds because $\hat{p}_x$, $\hat{p}_y$, and $\hat{p}_z$ all commute with each other (this is the Baker-Campbell-Hausdorff theorem in the trivial case where $[\hat{A}, \hat{B}] = 0$, so $e^{\hat{A}}e^{\hat{B}} = e^{\hat{A}+\hat{B}}$).

🔄 Retrieval Practice (Ch 5) — In Chapter 5, you solved the Schrodinger equation for the hydrogen atom using spherical coordinates, finding the quantum numbers $n$, $l$, $m$. The energy depended only on $n$: $E_n = -13.6\,\text{eV}/n^2$. For a given $n$, all states with different $l$ and $m$ values had the same energy. This $n^2$-fold degeneracy arises because the hydrogen atom has more symmetry than just rotational invariance — it has an additional "hidden" symmetry (the Laplace-Runge-Lenz vector), which we will not prove here but which is one of the most beautiful results in mathematical physics.

The conservation law

If the Hamiltonian is translationally invariant — $[\hat{H}, \hat{T}(\mathbf{a})] = 0$ for all $\mathbf{a}$ — then:

$$[\hat{H}, \hat{p}_x] = [\hat{H}, \hat{p}_y] = [\hat{H}, \hat{p}_z] = 0$$

By the quantum Noether theorem, momentum is conserved: $\frac{d}{dt}\langle\hat{\mathbf{p}}\rangle = 0$.

When is the Hamiltonian translationally invariant? For a single particle, $\hat{H} = \hat{p}^2/2m + V(\hat{x})$. The kinetic energy is always translation-invariant (it depends on $\hat{p}$, not $\hat{x}$). The potential energy is translation-invariant only if $V(x)$ is constant — i.e., for a free particle or (in 3D) for a uniform potential. If $V(x)$ varies with position, translation is broken and momentum is not conserved.

Worked Example. Consider a particle in a uniform gravitational field: $\hat{H} = \hat{p}_x^2/2m + \hat{p}_y^2/2m + \hat{p}_z^2/2m + mg\hat{z}$.

• Is momentum in the $x$-direction conserved? Compute $[\hat{H}, \hat{p}_x] = [mg\hat{z}, \hat{p}_x] = 0$ (since $[\hat{z}, \hat{p}_x] = 0$). Yes.
• Is momentum in the $z$-direction conserved? Compute $[\hat{H}, \hat{p}_z] = [mg\hat{z}, \hat{p}_z] = mg \cdot i\hbar \neq 0$. No.

Gravity breaks translation symmetry in $z$ but preserves it in $x$ and $y$, so $p_x$ and $p_y$ are conserved but $p_z$ is not.

✅ Checkpoint — Verify the commutator $[mg\hat{z}, \hat{p}_z] = img\hbar$ using the canonical commutation relation $[\hat{x}_i, \hat{p}_j] = i\hbar\delta_{ij}$. Then explain physically why $p_z$ is not conserved.

The Ehrenfest-Noether connection

The result $\frac{d}{dt}\langle\hat{p}\rangle = -\langle V'(\hat{x})\rangle$ from the Ehrenfest theorem (Chapter 7) is a special case of the quantum Noether theorem. Setting $\hat{G} = \hat{p}$:

$$\frac{d}{dt}\langle\hat{p}\rangle = \frac{1}{i\hbar}\langle[\hat{p}, \hat{H}]\rangle = \frac{1}{i\hbar}\langle[\hat{p}, V(\hat{x})]\rangle$$

Using $[\hat{p}, f(\hat{x})] = -i\hbar f'(\hat{x})$ (which follows from the canonical commutation relation):

$$\frac{d}{dt}\langle\hat{p}\rangle = \frac{1}{i\hbar}\langle -i\hbar V'(\hat{x})\rangle = -\langle V'(\hat{x})\rangle$$

This is exactly Ehrenfest's result — and it is the quantum version of Newton's second law $\dot{p} = -\frac{dV}{dx} = F$. When $V$ is constant (no force), momentum is conserved. The chain of reasoning is: no force → translation symmetry → $[\hat{H}, \hat{p}] = 0$ → momentum conservation. Newton's second law, Noether's theorem, and Ehrenfest's theorem are all saying the same thing in different languages.

10.5 Rotation Symmetry and Angular Momentum Conservation

Constructing the rotation operator

A rotation about the $z$-axis by an infinitesimal angle $d\phi$ transforms the position coordinates:

$$x \to x - y\,d\phi, \qquad y \to y + x\,d\phi, \qquad z \to z$$

The effect on a wave function is:

$$\hat{R}_z(d\phi)\psi(x, y, z) = \psi(x + y\,d\phi, y - x\,d\phi, z)$$

Taylor expanding to first order:

$$\hat{R}_z(d\phi)\psi = \psi + d\phi\left(y\frac{\partial\psi}{\partial x} - x\frac{\partial\psi}{\partial y}\right) = \left(\hat{I} + d\phi\left(y\frac{\partial}{\partial x} - x\frac{\partial}{\partial y}\right)\right)\psi$$

Recalling that $\hat{L}_z = -i\hbar\left(x\frac{\partial}{\partial y} - y\frac{\partial}{\partial x}\right)$, this becomes:

$$\hat{R}_z(d\phi) = \hat{I} - \frac{id\phi}{\hbar}\hat{L}_z$$

Comparing with the general formula, we identify:

$$\boxed{\text{The generator of rotations about the } z\text{-axis is the angular momentum operator } \hat{L}_z.}$$

The finite rotation by angle $\phi$ about $\hat{z}$ is:

$$\hat{R}_z(\phi) = e^{-i\hat{L}_z\phi/\hbar}$$

More generally, a rotation by angle $\theta$ about an arbitrary axis $\hat{n}$:

$$\hat{R}(\hat{n}, \theta) = e^{-i\hat{\mathbf{J}}\cdot\hat{n}\theta/\hbar}$$

where $\hat{\mathbf{J}} = \hat{\mathbf{L}} + \hat{\mathbf{S}}$ is the total angular momentum, including both orbital and spin contributions. (The spin contribution has no classical analogue — it will be treated fully in Chapter 13.)

Non-commutativity of rotations

Unlike translations, rotations about different axes do not commute. Rotate a book 90 degrees about $x$, then 90 degrees about $z$ — the result is different from doing $z$ first, then $x$. This is reflected in the commutation relations of the generators:

$$[\hat{J}_i, \hat{J}_j] = i\hbar\epsilon_{ijk}\hat{J}_k$$

where $\epsilon_{ijk}$ is the Levi-Civita symbol. This is the angular momentum algebra (also called the $\mathfrak{su}(2)$ Lie algebra), and it will be the central subject of Chapter 12.

🔴 Warning — The non-commutativity of rotations has a mathematical consequence that tripped up even experienced physicists: the Baker-Campbell-Hausdorff formula for rotations is not trivial. You cannot simply write $e^{i\hat{A}}e^{i\hat{B}} = e^{i(\hat{A}+\hat{B})}$ when $[\hat{A}, \hat{B}] \neq 0$. The correct formula involves an infinite series of nested commutators. For rotations, this subtlety is the origin of the Thomas precession and the spin-orbit coupling correction to the hydrogen spectrum (Chapter 18).

The conservation law

If the Hamiltonian is rotationally invariant — $[\hat{H}, \hat{R}(\hat{n}, \theta)] = 0$ for all $\hat{n}$ and $\theta$ — then:

$$[\hat{H}, \hat{J}_x] = [\hat{H}, \hat{J}_y] = [\hat{H}, \hat{J}_z] = 0$$

and equivalently $[\hat{H}, \hat{\mathbf{J}}^2] = 0$.

Angular momentum is conserved: $\frac{d}{dt}\langle\hat{\mathbf{J}}\rangle = 0$.

When is the Hamiltonian rotationally invariant? For a particle in a potential, $\hat{H} = \hat{p}^2/2m + V(\hat{r})$. The kinetic energy is rotationally invariant (it depends on $|\hat{\mathbf{p}}|^2$, which is a scalar). The potential is rotationally invariant if it depends only on $r = |\mathbf{r}|$ — a central potential. This includes the Coulomb potential ($V = -e^2/r$), the harmonic oscillator ($V = \frac{1}{2}m\omega^2 r^2$), and all other spherically symmetric potentials.

Degeneracy from rotational symmetry

For a rotationally invariant Hamiltonian, the energy eigenstates can be labeled by the angular momentum quantum numbers $l$ (for $\hat{L}^2$) and $m$ (for $\hat{L}_z$):

$$\hat{H}|n, l, m\rangle = E_{nl}|n, l, m\rangle$$

The crucial point: $E_{nl}$ does not depend on $m$. For a given $l$, the states $|n, l, m\rangle$ with $m = -l, -l+1, \ldots, l-1, l$ all have the same energy. This $(2l+1)$-fold degeneracy is a direct consequence of rotational symmetry: the rotation operator $\hat{R}(\hat{n}, \theta)$ maps one $m$ value to another without changing the energy.

Worked Example. The hydrogen atom with $n = 2$ has states:

• $|2, 0, 0\rangle$ (the $2s$ state): $l = 0$, one $m$ value → 1 state
• $|2, 1, m\rangle$ (the $2p$ states): $l = 1$, $m \in \{-1, 0, 1\}$ → 3 states

Rotational symmetry guarantees that the three $2p$ states are degenerate. (In hydrogen, the $2s$ and $2p$ states happen to be degenerate too, at $E_2 = -3.4\,\text{eV}$. That additional degeneracy requires the hidden $SO(4)$ symmetry — beyond our scope here.)

🧪 Experiment — The Zeeman effect provides direct experimental verification of the symmetry-degeneracy connection. An external magnetic field $\mathbf{B} = B\hat{z}$ breaks rotational symmetry by picking out a preferred direction. The Hamiltonian acquires an additional term $\hat{H}' = -\hat{\mu} \cdot \mathbf{B}$, which is proportional to $\hat{L}_z$. This lifts the $m$-degeneracy: states with different $m$ acquire different energies, and spectral lines split into multiplets. Remove the field, restore the symmetry, and the degeneracy returns.

✅ Checkpoint — For a particle in a 2D isotropic harmonic oscillator $V = \frac{1}{2}m\omega^2(x^2 + y^2)$, what symmetry is present? What is the conserved quantity? What degeneracy do you predict for the energy levels? (Verify by recalling the energy formula $E_{n_x, n_y} = (n_x + n_y + 1)\hbar\omega$ from Chapter 4.)

Symmetry and the classification of atomic spectra

The power of rotational symmetry extends far beyond the hydrogen atom. Consider any atom — helium, carbon, iron, uranium — in the absence of external fields. The Hamiltonian includes the kinetic energies of all electrons, the Coulomb attraction to the nucleus, and the electron-electron repulsion:

$$\hat{H} = \sum_i \frac{\hat{p}_i^2}{2m_e} - \sum_i \frac{Ze^2}{r_i} + \sum_{i < j} \frac{e^2}{|\mathbf{r}_i - \mathbf{r}_j|}$$

Every term is rotationally invariant (they all depend on distances $r_i$ and $|\mathbf{r}_i - \mathbf{r}_j|$, which are scalars). Therefore the total angular momentum $\hat{\mathbf{J}} = \sum_i (\hat{\mathbf{L}}_i + \hat{\mathbf{S}}_i)$ commutes with $\hat{H}$, and the energy levels can be labeled by $J$ and $M_J$.

This is the foundation of term symbol notation in atomic spectroscopy. When a spectroscopist writes ${}^3P_2$ for a state of carbon, the subscript "2" means $J = 2$, and the superscript "3" means $2S+1 = 3$ (spin triplet). The letter "P" means $L = 1$. These quantum numbers are representation labels of the rotation group — exactly the machinery we have developed.

An external magnetic field breaks rotational symmetry to rotation about a single axis, reducing the symmetry from $SO(3)$ to $SO(2)$ (the "cylindrical" subgroup). This lifts the $(2J+1)$-fold degeneracy in $M_J$, producing the Zeeman effect that you will study in Chapter 18. An external electric field (Stark effect) breaks different symmetries and produces a different splitting pattern. In every case, the degeneracy structure is dictated by which symmetries survive and which are broken.

10.6 Discrete Symmetries: Parity

The parity operator

The parity operator $\hat{\Pi}$ reverses all spatial coordinates:

$$\hat{\Pi}|\mathbf{r}\rangle = |-\mathbf{r}\rangle$$

In position representation:

$$\hat{\Pi}\psi(\mathbf{r}) = \psi(-\mathbf{r})$$

In one dimension: $\hat{\Pi}\psi(x) = \psi(-x)$.

Properties of the parity operator

$\hat{\Pi}$ is Hermitian and unitary. Since $\hat{\Pi}^2 = \hat{I}$ (reflecting twice returns to the original), the eigenvalues satisfy $\pi^2 = 1$, giving:

$$\pi = \pm 1$$

States with eigenvalue $+1$ are even (symmetric under reflection): $\psi(-x) = +\psi(x)$.

States with eigenvalue $-1$ are odd (antisymmetric under reflection): $\psi(-x) = -\psi(x)$.

These are the only possibilities. There is no "intermediate parity."

$\hat{\Pi}$ is not generated by exponentiation of a Hermitian operator. Unlike translation or rotation, parity is a discrete symmetry — there is no continuous parameter and no generator. You cannot "continuously deform" the identity into a reflection.

Action of parity on operators

How does parity affect the position and momentum operators? Under reflection, position reverses sign and momentum reverses sign:

$$\hat{\Pi}\hat{x}\hat{\Pi}^\dagger = -\hat{x}, \qquad \hat{\Pi}\hat{p}\hat{\Pi}^\dagger = -\hat{p}$$

We say that $\hat{x}$ and $\hat{p}$ are odd under parity (they pick up a minus sign). An operator $\hat{A}$ is:

• Even (parity-even) if $\hat{\Pi}\hat{A}\hat{\Pi}^\dagger = +\hat{A}$
• Odd (parity-odd) if $\hat{\Pi}\hat{A}\hat{\Pi}^\dagger = -\hat{A}$

Some important examples:

⚠️ Common Misconception — Angular momentum is even under parity, not odd. This is because $\hat{\mathbf{L}} = \hat{\mathbf{r}} \times \hat{\mathbf{p}}$, and both $\hat{\mathbf{r}}$ and $\hat{\mathbf{p}}$ are odd, so their cross product is even. Vectors that are even under parity are called pseudovectors or axial vectors. Magnetic fields are pseudovectors (this is why magnetic dipole transitions have different selection rules from electric dipole transitions).

Parity conservation and selection rules

If $[\hat{H}, \hat{\Pi}] = 0$ — that is, the potential is symmetric, $V(-\mathbf{r}) = V(\mathbf{r})$ — then parity is conserved. The consequences are:

1. Energy eigenstates have definite parity. Each non-degenerate eigenstate of $\hat{H}$ is simultaneously an eigenstate of $\hat{\Pi}$, with eigenvalue $+1$ or $-1$.

2. Selection rules for matrix elements. Consider the matrix element $\langle n'|\hat{x}|n\rangle$ between two energy eigenstates with definite parities $\pi_{n'}$ and $\pi_n$. Since $\hat{x}$ is parity-odd:

$$\langle n'|\hat{x}|n\rangle = \langle n'|\hat{\Pi}^\dagger(\hat{\Pi}\hat{x}\hat{\Pi}^\dagger)\hat{\Pi}|n\rangle = \langle n'|\hat{\Pi}^\dagger(-\hat{x})\hat{\Pi}|n\rangle = -\pi_{n'}\pi_n\langle n'|\hat{x}|n\rangle$$

Therefore:

$$\langle n'|\hat{x}|n\rangle(1 + \pi_{n'}\pi_n) = 0$$

If $\pi_{n'} = \pi_n$ (same parity), then $(1 + 1) = 2 \neq 0$, so $\langle n'|\hat{x}|n\rangle = 0$. The matrix element vanishes unless $\pi_{n'}\pi_n = -1$ — the states must have opposite parity.

This is the parity selection rule for electric dipole transitions: $\Delta l = \pm 1$ (since the parity of a hydrogen eigenstate is $(-1)^l$, and a transition requires opposite parities).

Worked Example: QHO parity. The QHO potential $V(x) = \frac{1}{2}m\omega^2 x^2$ is even, so $[\hat{H}, \hat{\Pi}] = 0$. The energy eigenstates $|n\rangle$ have parity $(-1)^n$: even for $n = 0, 2, 4, \ldots$ and odd for $n = 1, 3, 5, \ldots$. Consequence: $\langle m|\hat{x}|n\rangle = 0$ unless $m$ and $n$ have opposite parity. You can verify this using the ladder operator expression $\hat{x} = \sqrt{\frac{\hbar}{2m\omega}}(\hat{a} + \hat{a}^\dagger)$, which changes $n$ by exactly 1.

Let us push this further. Consider the matrix element $\langle m|\hat{x}^3|n\rangle$. Since $\hat{x}^3$ is parity-odd (the cube of a parity-odd operator), the same selection rule applies: the matrix element vanishes unless $m$ and $n$ have opposite parity. This means, for instance, that if we add a perturbation $\hat{H}' = \lambda\hat{x}^3$ to the QHO Hamiltonian, the first-order energy correction $E_n^{(1)} = \langle n|\hat{H}'|n\rangle = \lambda\langle n|\hat{x}^3|n\rangle$ is exactly zero for every eigenstate — not approximately zero, not numerically small, but identically zero by symmetry. You can compute the integral $\int_{-\infty}^{\infty} \psi_n^*(x) x^3 \psi_n(x) dx$ and confirm this, but the parity argument gives the result instantly.

This is the practical power of selection rules: they eliminate entire families of calculations. In atomic physics, where matrix elements of the dipole operator $\hat{d} = -e\hat{\mathbf{r}}$ determine transition rates, the parity selection rule $\Delta l = \pm 1$ immediately tells you that roughly half of all conceivable transitions are forbidden. The remaining allowed transitions still require computation, but the workload is cut in half — or more — by symmetry.

Worked Example: Why atoms do not have permanent electric dipole moments. A permanent electric dipole moment would mean $\langle\hat{d}\rangle = \langle\psi|\hat{d}|\psi\rangle \neq 0$ for an energy eigenstate. But $\hat{d} = -e\hat{\mathbf{r}}$ is parity-odd. If the Hamiltonian is parity-symmetric (no external electric field), each non-degenerate energy eigenstate has definite parity, and $\langle n|\hat{d}|n\rangle = 0$ by the parity selection rule. Atoms in energy eigenstates cannot have permanent electric dipole moments. (In a degenerate subspace, linear combinations can have non-zero dipole moments — this is the mechanism behind the linear Stark effect in hydrogen, which exploits the $l$-degeneracy.)

🔄 Retrieval Practice (Ch 5) — In Chapter 5, you found that the hydrogen atom wave functions $\psi_{nlm}(r, \theta, \phi)$ include the spherical harmonics $Y_l^m(\theta, \phi)$. What is the parity of $Y_l^m$? (Recall: under $\mathbf{r} \to -\mathbf{r}$ in spherical coordinates, $r \to r$, $\theta \to \pi - \theta$, $\phi \to \phi + \pi$. The result is $Y_l^m(\pi - \theta, \phi + \pi) = (-1)^l Y_l^m(\theta, \phi)$.) Verify that the parity selection rule $\Delta l = \pm 1$ follows.

10.7 Discrete Symmetries: Time Reversal

The physical idea

Time reversal, denoted $\hat{\Theta}$, is the operation of "running the movie backward." In classical mechanics, if you reverse the velocities of all particles in a system (while keeping positions fixed), the system retraces its trajectory. In quantum mechanics, the situation is more subtle.

What should time reversal do to a quantum state? Consider a free particle with wave function $\psi(x, t) = Ae^{i(kx - \omega t)}$, representing a particle moving to the right with momentum $p = \hbar k > 0$. Running the movie backward should give a particle moving to the left with momentum $-p$. The time-reversed wave function should be $\psi_T(x, t) = Ae^{i(-kx - \omega t)}$.

Notice: the time-reversed wave function is the complex conjugate of the original (up to a time-dependent phase). This observation is the key to understanding why time reversal is antiunitary.

Why time reversal must be antiunitary

An antilinear operator $\hat{\Theta}$ satisfies:

$$\hat{\Theta}(\alpha|\psi\rangle + \beta|\phi\rangle) = \alpha^*\hat{\Theta}|\psi\rangle + \beta^*\hat{\Theta}|\phi\rangle$$

It conjugates all complex coefficients. An antiunitary operator is antilinear and preserves inner products in the following sense:

$$\langle\hat{\Theta}\phi|\hat{\Theta}\psi\rangle = \langle\phi|\psi\rangle^* = \langle\psi|\phi\rangle$$

Why must time reversal be antiunitary rather than unitary? The argument (due to Wigner) proceeds by reductio:

Suppose $\hat{\Theta}$ were unitary (linear). Then it would commute with $\hat{H}$, and $\hat{\Theta}|E\rangle$ would be an eigenstate with the same energy. Consider the time evolution: $|\psi(t)\rangle = e^{-i\hat{H}t/\hbar}|\psi(0)\rangle$. Applying $\hat{\Theta}$:

$$\hat{\Theta}|\psi(t)\rangle = \hat{\Theta}e^{-i\hat{H}t/\hbar}|\psi(0)\rangle = e^{-i\hat{H}t/\hbar}\hat{\Theta}|\psi(0)\rangle$$

But this gives forward time evolution of the time-reversed state — the opposite of what we want. We need the time-reversed state to evolve backward in time. The fix is antilinearity: if $\hat{\Theta}$ is antilinear, then $\hat{\Theta}(i) = -i$, and:

$$\hat{\Theta}e^{-i\hat{H}t/\hbar}|\psi(0)\rangle = e^{+i\hat{H}t/\hbar}\hat{\Theta}|\psi(0)\rangle$$

This gives backward time evolution, which is exactly what we want. $\square$

Action of $\hat{\Theta}$ on operators

Time reversal transforms operators as follows:

$$\hat{\Theta}\hat{x}\hat{\Theta}^{-1} = +\hat{x} \qquad \text{(position is even)}$$

$$\hat{\Theta}\hat{p}\hat{\Theta}^{-1} = -\hat{p} \qquad \text{(momentum is odd)}$$

$$\hat{\Theta}\hat{L}\hat{\Theta}^{-1} = -\hat{L} \qquad \text{(angular momentum is odd)}$$

$$\hat{\Theta}\hat{S}\hat{\Theta}^{-1} = -\hat{S} \qquad \text{(spin is odd)}$$

$$\hat{\Theta}i\hat{\Theta}^{-1} = -i \qquad \text{(antilinearity)}$$

The reversal of momentum and angular momentum is physically natural: running the movie backward reverses all velocities and angular velocities.

The spin-1/2 time reversal operator

For a spin-1/2 particle, the time reversal operator in the $|\uparrow\rangle$, $|\downarrow\rangle$ basis is:

$$\hat{\Theta} = -i\hat{\sigma}_y \hat{K}$$

where $\hat{K}$ denotes complex conjugation. The factor $-i\hat{\sigma}_y$ rotates the spinor, and $\hat{K}$ conjugates the coefficients. Let us verify that this has the correct properties.

Acting on spin-up: $\hat{\Theta}|\uparrow\rangle = -i\hat{\sigma}_y|\uparrow\rangle = -i \begin{pmatrix} 0 & -i \\ i & 0 \end{pmatrix} \begin{pmatrix} 1 \\ 0 \end{pmatrix} = -i \begin{pmatrix} 0 \\ i \end{pmatrix} = \begin{pmatrix} 0 \\ 1 \end{pmatrix} = |\downarrow\rangle$

Acting on spin-down: $\hat{\Theta}|\downarrow\rangle = -i\hat{\sigma}_y|\downarrow\rangle = -i \begin{pmatrix} 0 & -i \\ i & 0 \end{pmatrix} \begin{pmatrix} 0 \\ 1 \end{pmatrix} = -i \begin{pmatrix} -i \\ 0 \end{pmatrix} = \begin{pmatrix} -1 \\ 0 \end{pmatrix} = -|\uparrow\rangle$

Time reversal flips the spin — as it should, since angular momentum reverses under time reversal. But notice the minus sign in $\hat{\Theta}|\downarrow\rangle = -|\uparrow\rangle$. This asymmetry has deep consequences.

Kramers' degeneracy

For systems with an odd number of spin-1/2 particles (e.g., a single electron), time reversal leads to a remarkable result. The time-reversed state $\hat{\Theta}|\psi\rangle$ is orthogonal to $|\psi\rangle$ and has the same energy. This means every energy level is at least two-fold degenerate. This is Kramers' theorem (1930).

The proof uses a key property: for a spin-1/2 particle, $\hat{\Theta}^2 = -\hat{I}$ (not $+\hat{I}$). Let us verify this:

$$\hat{\Theta}^2|\uparrow\rangle = \hat{\Theta}|\downarrow\rangle = -|\uparrow\rangle$$

$$\hat{\Theta}^2|\downarrow\rangle = \hat{\Theta}(-|\uparrow\rangle) = -\hat{\Theta}|\uparrow\rangle^* = -|\downarrow\rangle$$

Wait — we must be careful with antilinearity. Since $\hat{\Theta}$ is antilinear: $\hat{\Theta}(-|\uparrow\rangle) = (-1)^*\hat{\Theta}|\uparrow\rangle = -\hat{\Theta}|\uparrow\rangle = -|\downarrow\rangle$. So indeed $\hat{\Theta}^2 = -\hat{I}$ on both basis states.

Now, using $\hat{\Theta}^2 = -\hat{I}$ and the antiunitary inner product rule:

$$\langle\psi|\hat{\Theta}\psi\rangle = \langle\hat{\Theta}^2\psi|\hat{\Theta}\psi\rangle = \langle(-\psi)|\hat{\Theta}\psi\rangle = -\langle\psi|\hat{\Theta}\psi\rangle$$

which implies $\langle\psi|\hat{\Theta}\psi\rangle = 0$. The state and its time-reverse are orthogonal and degenerate. $\square$

The logic of the first equality above uses the antiunitary property: $\langle\hat{\Theta}\alpha|\hat{\Theta}\beta\rangle = \langle\alpha|\beta\rangle^*$, setting $|\alpha\rangle = \hat{\Theta}|\psi\rangle$ and $|\beta\rangle = |\psi\rangle$.

Kramers' degeneracy is broken only by a magnetic field (which breaks time-reversal symmetry). It is essential in condensed matter physics: it protects the band structure of certain materials, leading to topological insulators (Chapter 36).

💡 Key Insight — The distinction between $\hat{\Theta}^2 = +\hat{I}$ (integer spin) and $\hat{\Theta}^2 = -\hat{I}$ (half-integer spin) is intimately connected to the $2\pi$ rotation behavior of spinors vs. vectors. A spin-1/2 particle acquires a minus sign under a $2\pi$ rotation ($e^{-i \cdot 2\pi \cdot \hat{S}_z/\hbar} = -\hat{I}$ for spin-1/2). This minus sign resurfaces in the time-reversal operator and produces Kramers' degeneracy. The deep connection between rotation and time reversal is one of the most beautiful structures in physics.

⚖️ Interpretation — Time reversal in quantum mechanics raises interpretive questions. If the fundamental laws are time-reversal symmetric (as they are in non-relativistic QM with real potentials), why does the macroscopic world have an arrow of time? The answer involves the second law of thermodynamics and the special boundary conditions of the universe — territory beyond our scope, but worth contemplating. In the Standard Model of particle physics, CP violation (a subtle asymmetry between matter and antimatter) means that T symmetry is not exact, even at the fundamental level.

10.8 Crystal Symmetries and Bloch's Theorem (Preview of Chapter 26)

From continuous to discrete translation symmetry

A perfect crystal has a special kind of translational symmetry: the potential is not invariant under arbitrary translations, but only under translations by lattice vectors:

$$V(\mathbf{r} + \mathbf{R}) = V(\mathbf{r}) \quad \text{for all lattice vectors } \mathbf{R} = n_1\mathbf{a}_1 + n_2\mathbf{a}_2 + n_3\mathbf{a}_3$$

where $\mathbf{a}_1, \mathbf{a}_2, \mathbf{a}_3$ are the primitive lattice vectors and $n_1, n_2, n_3$ are integers.

The translation operators $\hat{T}(\mathbf{R})$ for lattice vectors commute with the Hamiltonian:

$$[\hat{H}, \hat{T}(\mathbf{R})] = 0$$

Because the lattice translation operators commute with each other and with $\hat{H}$, the energy eigenstates can be chosen to be simultaneous eigenstates of all $\hat{T}(\mathbf{R})$.

Bloch's theorem

Since $\hat{T}(\mathbf{R})$ is unitary, its eigenvalue is a complex number of modulus 1. Writing it as $e^{i\mathbf{k}\cdot\mathbf{R}}$ for some wave vector $\mathbf{k}$:

$$\hat{T}(\mathbf{R})\psi_{\mathbf{k}}(\mathbf{r}) = e^{i\mathbf{k}\cdot\mathbf{R}}\psi_{\mathbf{k}}(\mathbf{r})$$

This means:

$$\psi_{\mathbf{k}}(\mathbf{r} + \mathbf{R}) = e^{i\mathbf{k}\cdot\mathbf{R}}\psi_{\mathbf{k}}(\mathbf{r})$$

Writing $\psi_{\mathbf{k}}(\mathbf{r}) = e^{i\mathbf{k}\cdot\mathbf{r}}u_{\mathbf{k}}(\mathbf{r})$, where $u_{\mathbf{k}}(\mathbf{r})$ has the periodicity of the lattice ($u_{\mathbf{k}}(\mathbf{r} + \mathbf{R}) = u_{\mathbf{k}}(\mathbf{r})$), we arrive at Bloch's theorem:

$$\boxed{\psi_{n\mathbf{k}}(\mathbf{r}) = e^{i\mathbf{k}\cdot\mathbf{r}}u_{n\mathbf{k}}(\mathbf{r})}$$

Every energy eigenstate in a periodic potential is a Bloch wave: a plane wave modulated by a function with the periodicity of the lattice. The quantum number $\mathbf{k}$ (the crystal momentum) labels the irreducible representations of the discrete translation group.

Energy bands

For each value of $\mathbf{k}$, there are infinitely many energy eigenvalues $E_n(\mathbf{k})$, indexed by the band index $n$. As $\mathbf{k}$ varies continuously through the Brillouin zone (the fundamental domain of the reciprocal lattice), the energy levels trace out continuous curves — the energy bands.

The band structure $E_n(\mathbf{k})$ is the central object of condensed matter physics. It determines whether a material is a metal, semiconductor, or insulator. The entire theory of electronic properties of solids rests on Bloch's theorem, which rests on the discrete translation symmetry of the crystal lattice. This is symmetry as problem-solving strategy at its most powerful.

A simple example: the Kronig-Penney model

To see Bloch's theorem in action, consider the simplest periodic potential: a one-dimensional array of delta-function barriers separated by distance $a$:

$$V(x) = \alpha \sum_{n=-\infty}^{\infty} \delta(x - na)$$

Between the barriers, the Schrodinger equation is the free-particle equation, and the solutions are plane waves. The delta-function barriers impose matching conditions on the wave function and its derivative at each lattice site. By Bloch's theorem, we need only solve in one unit cell ($0 < x < a$) and then require $\psi(x + a) = e^{ika}\psi(x)$.

The result (which you will derive in Chapter 26) is a transcendental equation relating the energy $E$ to the crystal momentum $k$:

$$\cos(ka) = \cos(qa) + \frac{m\alpha}{\hbar^2 q}\sin(qa)$$

where $q = \sqrt{2mE}/\hbar$. The crucial feature: this equation has solutions only for certain ranges of $E$. These ranges are the energy bands — continuous intervals of allowed energies. Between the bands are band gaps — intervals where no solutions exist. Electrons with energies in the gap cannot propagate through the crystal.

This band-gap structure is the reason silicon is a semiconductor, diamond is an insulator, and aluminum is a metal. It all follows from the discrete translation symmetry of the lattice — Bloch's theorem in its most elementary application.

Point group symmetries

Beyond translation, crystals have additional symmetries: rotations, reflections, and improper rotations that map the lattice onto itself. These form the point group of the crystal. The combination of point group and translation group gives the space group.

In three dimensions, there are exactly 32 crystallographic point groups and 230 space groups. The representation theory of these groups determines:

• Which energy bands must cross or touch at special points in the Brillouin zone.
• Which electronic states are degenerate at high-symmetry points.
• Selection rules for optical transitions in solids.

This is a vast subject — the full development awaits Chapter 26. The point here is that all of it begins with the idea you learned in Section 10.1: identify the symmetry, find the representations, and the physics follows.

🔗 Connection (Ch 26) — Chapter 26 will develop band theory in full, starting from the tight-binding model and working up to realistic band structures. You will compute energy bands numerically and understand why silicon is a semiconductor, copper is a metal, and diamond is an insulator. The foundation for all of that is the symmetry argument you have just seen.

📊 By the Numbers — The International Tables for Crystallography classify all possible crystal symmetries into exactly 230 space groups in three dimensions (combining point group operations with lattice translations). This complete enumeration, achieved by Fedorov, Schoenflies, and Barlow independently in 1891, means that every crystal in the universe belongs to one of exactly 230 symmetry types. The classification of possible band structures follows from the representation theory of these groups.

10.9 Summary and Project Checkpoint

What we have established

This chapter developed the deepest organizing principle in physics: symmetry determines structure. Here is the logical chain:

1. A symmetry is a transformation that leaves the physics unchanged, implemented by a unitary (or antiunitary) operator $\hat{U}$.

2. For continuous symmetries, $\hat{U}(\alpha) = e^{-i\alpha\hat{G}/\hbar}$, where the generator $\hat{G}$ is Hermitian and thus an observable.

3. The quantum Noether theorem establishes the equivalence: symmetry $\Leftrightarrow$ $[\hat{H}, \hat{G}] = 0$ $\Leftrightarrow$ conservation of $\hat{G}$.

4. Translation symmetry is generated by $\hat{p}$ → conservation of linear momentum.

5. Rotation symmetry is generated by $\hat{J}$ → conservation of angular momentum. The non-commutativity of rotations produces the angular momentum algebra $[\hat{J}_i, \hat{J}_j] = i\hbar\epsilon_{ijk}\hat{J}_k$, which is the foundation of Chapter 12.

6. Parity ($\hat{\Pi}$) is a discrete symmetry with eigenvalues $\pm 1$. It gives selection rules for matrix elements and explains the even/odd structure of the QHO eigenstates.

7. Time reversal ($\hat{\Theta}$) is antiunitary. For half-integer spin, $\hat{\Theta}^2 = -\hat{I}$, leading to Kramers' degeneracy.

8. Crystal symmetry (discrete translation invariance) produces Bloch's theorem and energy bands — the foundation of solid-state physics.

💡 Key Insight — Symmetry is not just a tool for solving problems more efficiently (though it is that). It is the organizing principle that determines which problems exist and what form their solutions take. The quantum numbers that label states ($n, l, m, m_s$, crystal momentum $\mathbf{k}$, band index $n$, parity $\pm 1$) are all labels for irreducible representations of the symmetries of the Hamiltonian. If you understand the symmetry group, you understand the qualitative structure of the spectrum before you compute anything.

The deep structure: quantum numbers are representation labels

Throughout this chapter, we have seen the same pattern: a symmetry of the Hamiltonian produces quantum numbers that label the energy eigenstates. Let us make this explicit.

A representation of a symmetry group $G$ is a map from group elements to matrices: $g \mapsto D(g)$, such that $D(g_1)D(g_2) = D(g_1 g_2)$. An irreducible representation cannot be decomposed into smaller representations.

The energy eigenspace for a given energy $E$ carries a representation of the symmetry group. If this representation is irreducible with dimension $d$, then the energy level is $d$-fold degenerate. The labels that distinguish the degenerate states within a multiplet are the representation labels.

This is the conceptual core of the chapter: quantum mechanics and group theory are different languages for the same physics. The quantum numbers are not arbitrary labels — they are mathematically forced by the symmetry structure. Any quantum system with the same symmetry group will have the same pattern of quantum numbers and degeneracies, regardless of the specific Hamiltonian. This universality is why symmetry is the deepest organizing principle in physics.

The symmetry toolbox

Here is a practical summary of the symmetry problem-solving strategy:

Step 1: Identify the symmetries. What transformations leave $\hat{H}$ invariant? Check: translations (in which directions?), rotations (full $SO(3)$ or a subgroup?), parity, time reversal, and any discrete symmetries specific to the system (crystal point group, permutation symmetry, etc.).

Step 2: Find the conserved quantities. For each symmetry, identify the generator (for continuous symmetries) or the quantum number (for discrete symmetries). These are the "good quantum numbers" that label the energy eigenstates.

Step 3: Classify the spectrum. Use the representation theory of the symmetry group to determine the degeneracy pattern. States that transform into each other under the symmetry group must have the same energy.

Step 4: Derive selection rules. Determine which matrix elements $\langle n'|\hat{O}|n\rangle$ vanish by symmetry. This eliminates entire classes of calculations.

Step 5: Solve the reduced problem. After symmetry has done its work, the remaining problem is smaller and more tractable. Solve it by the methods of Chapters 3--5 or the approximation methods of Chapters 17--22.

Progressive Project: Quantum Toolkit v1.0 — Symmetry Module

Your toolkit should now include a symmetry.py module with the following capabilities:

1. Symmetry transformation generators. Functions that build the matrix representations of translation, rotation, and parity operators in a given basis.

2. Conservation law verifier. A check_conservation(H, G) function that numerically verifies $[\hat{H}, \hat{G}] = 0$ — i.e., that $\hat{G}$ commutes with the Hamiltonian.

3. Parity operator. A function that constructs the parity operator in the position basis (for a discretized 1D system) and verifies its eigenvalues are $\pm 1$.

4. Group representation tools. Functions that build matrix representations for discrete symmetry groups (e.g., $C_2$, $C_3$, $C_{4v}$ — at least the cyclic groups).

See code/project-checkpoint.py for the implementation and test cases. After completing this checkpoint, your toolkit covers:

Looking ahead

Chapter 11 (Tensor Products and Composite Systems) will combine the symmetry tools of this chapter with the linear algebra of Chapter 8 to handle multi-particle quantum mechanics. The tensor product structure of composite Hilbert spaces is where entanglement — the most characteristically quantum phenomenon — makes its appearance.

Chapter 12 (Angular Momentum Algebra) will take the rotation generators $\hat{J}_x, \hat{J}_y, \hat{J}_z$ and their commutation relations and develop the complete theory of angular momentum. Every result in Chapter 12 is a consequence of rotational symmetry. The quantum numbers $j$ and $m$ that you will compute there are labels for irreducible representations of $SU(2)$ — the language of this chapter made precise.

🔗 Connection (Ch 32) — The adiabatic theorem (Chapter 32) adds a new dimension to symmetry: what happens when the Hamiltonian changes slowly in time, preserving symmetry at each instant but allowing the parameters to vary? The answer involves Berry's phase — a geometric phase determined by the topology of the parameter space. Berry's phase is a direct descendant of the symmetry ideas in this chapter, elevated from algebra to geometry.

Glossary of New Terms

1. Symmetry transformation — An operation on the Hilbert space that preserves all physical predictions (transition probabilities).

2. Wigner's theorem — The mathematical result that any symmetry transformation must be implemented by a unitary or antiunitary operator.

3. Generator — The Hermitian operator $\hat{G}$ from which a continuous unitary transformation is built via exponentiation: $\hat{U}(\alpha) = e^{-i\alpha\hat{G}/\hbar}$.

4. Lie group — A continuous group of symmetry transformations that can be parameterized smoothly (e.g., rotations, translations).

5. Lie algebra — The algebraic structure formed by the generators of a Lie group, defined by their commutation relations.

6. Noether's theorem (quantum) — The equivalence of three statements: (i) $\hat{U}$ is a symmetry, (ii) $[\hat{H}, \hat{G}] = 0$, (iii) $\langle\hat{G}\rangle$ is conserved.

7. Translation operator — $\hat{T}(\mathbf{a}) = e^{-i\hat{\mathbf{p}}\cdot\mathbf{a}/\hbar}$, generated by momentum.

8. Rotation operator — $\hat{R}(\hat{n}, \theta) = e^{-i\hat{\mathbf{J}}\cdot\hat{n}\theta/\hbar}$, generated by angular momentum.

9. Multiplet — A set of degenerate states related by a symmetry transformation, forming an irreducible representation.

10. Selection rule — A constraint on matrix elements derived from symmetry, specifying which transitions are forbidden.

11. Parity operator ($\hat{\Pi}$) — The operator that reverses all spatial coordinates: $\hat{\Pi}|\mathbf{r}\rangle = |-\mathbf{r}\rangle$. Eigenvalues $\pm 1$.

12. Pseudovector (axial vector) — A vector quantity that is even under parity (e.g., angular momentum, magnetic field).

13. Time reversal operator ($\hat{\Theta}$) — The antiunitary operator implementing time reversal. Reverses momentum and angular momentum, conjugates complex coefficients.

14. Antiunitary operator — An operator that is antilinear ($\hat{\Theta}(\alpha|\psi\rangle) = \alpha^*\hat{\Theta}|\psi\rangle$) and preserves inner products up to conjugation.

15. Kramers' degeneracy — For half-integer spin systems with time-reversal symmetry, every energy level is at least two-fold degenerate.

16. Bloch's theorem — In a periodic potential, energy eigenstates have the form $\psi_{n\mathbf{k}}(\mathbf{r}) = e^{i\mathbf{k}\cdot\mathbf{r}}u_{n\mathbf{k}}(\mathbf{r})$ where $u$ has the lattice periodicity.

17. Crystal momentum ($\hbar\mathbf{k}$) — The quantum number labeling Bloch states; analogous to ordinary momentum but defined modulo a reciprocal lattice vector.

18. Brillouin zone — The fundamental domain of crystal momentum space; the set of independent $\mathbf{k}$ values.