Reading Griffiths, Chapter 3

Chapter 3 of Griffiths is where quantum mechanics stops feeling like a bag of tricks and starts feeling like a place you can think inside. The topic is formalism — Hilbert space, operators, eigenvalues — and for a while it felt like symbol-pushing with no destination. Then something clicked.

What Hilbert space actually is

A Hilbert space is a complete inner product space. That definition is precise but initially opaque. What helped me was thinking of it as a generalization of vectors in $\mathbb{R}^3$ to functions.

The inner product of two wavefunctions $\psi$ and $\phi$ is:

$$\langle \psi | \phi \rangle = \int_{-\infty}^{\infty} \psi^*(x)\, \phi(x)\, dx$$

The analogy to the dot product is exact. “Orthogonal states” means $\langle \psi | \phi \rangle = 0$ — they carry no overlap, no shared information.

Hermitian operators and why they matter

Observable quantities (position, momentum, energy) correspond to Hermitian operators $\hat{Q}$, satisfying $\langle f | \hat{Q} g \rangle = \langle \hat{Q} f | g \rangle$. Griffiths shows this forces two things:

1. All eigenvalues are real — so measurements return real numbers.
2. Eigenfunctions belonging to different eigenvalues are orthogonal.

The inline version: a Hermitian operator $\hat{Q}$ has $\hat{Q}^\dagger = \hat{Q}$, where $\dagger$ is the conjugate transpose.

The eigenvalue equation in code

When I wanted to build intuition for discrete spectra, I sketched out a finite-difference eigenvalue solver:

import numpy as np

def particle_in_box_energies(n_levels=5, L=1.0):
    """Exact analytic energies E_n = n² π² ℏ² / (2mL²), in natural units."""
    return [(n**2 * np.pi**2) / (2 * L**2) for n in range(1, n_levels + 1)]

energies = particle_in_box_energies()
for n, E in enumerate(energies, 1):
    print(f"E_{n} = {E:.4f}")

The output matches what the formalism predicts exactly. Seeing the eigenvalues come out of a matrix made “eigenvalue equation” concrete in a way the derivation alone didn’t.

What finally clicked

The thing that made Hilbert space feel like a place rather than an abstraction: every state $|\psi\rangle$ is a superposition of the eigenstates of whatever observable you’re measuring. The act of measurement collapses it to one of them. The probabilities are the squared magnitudes of the expansion coefficients.

That’s not a postulate bolted on — it follows from the inner product structure.¹

Footnotes

1. Griffiths section 3.3 walks through this carefully. The completeness of the eigenbasis is the key step I kept glossing over. ↩