---
schema_version: 1
id: PHYS-2026-07-21-01
date: 2026-07-21
updated_at: 2026-07-21
title: Self-adjointness as boundary data
summary: Why choosing the domain of a quantum Hamiltonian is part of specifying the observable.
language: en
entry_kind: daily
status: published
level: graduate-advanced
user_difficulty: unrated
domains:
  - quantum-theory
  - mathematical-physics
estimated_minutes: 25
---

## Key claim

A differential expression is not yet an observable. The fracture appears when the same formal Hamiltonian acquires different spectra because its domain changes.

## Theme

Self-adjointness is boundary data in disguise.

## Guiding question

For a particle on the interval $[0,L]$, suppose we write

$$
H=-\frac{\hbar^2}{2m}\frac{\dd^2}{\dd x^2}.
$$

What physical information is missing from this line?

## Setup

Integration by parts exposes the boundary form

$$
\bra{\phi}H\ket{\psi}-\bra{H\phi}\ket{\psi}
=-\frac{\hbar^2}{2m}
\left[\phi^*(x)\psi'(x)-\phi'^*(x)\psi(x)\right]_{0}^{L}.
$$

Symmetry requires this expression to vanish on the chosen domain. Self-adjointness further requires that the adjoint have exactly the same domain.

## Analysis

### Derivation

Build a domain by imposing Dirichlet data, $\psi(0)=\psi(L)=0$. The boundary form vanishes, and the familiar discrete spectrum follows.

### Scope and assumptions

Why privilege Dirichlet data? Periodic data,

$$
\psi(L)=e^{\ii\theta}\psi(0),\qquad
\psi'(L)=e^{\ii\theta}\psi'(0),
$$

also makes the boundary form vanish and defines a different self-adjoint operator.

### Physical interpretation

The formal differential rule describes local evolution. The domain tells the wavefunction how the ends of configuration space are physically identified.

## False claim to diagnose

> If two Hamiltonians have the same differential expression, they represent the same observable.

Locate the hidden assumption that makes this statement false.

## What follows — and what does not

The Builder treats the boundary condition as a construction choice. The Skeptic reveals a family of equally consistent choices. The Translator identifies the choice as global physical structure rather than a mathematical afterthought.

## Exercise

Let $H=-\dd^2/\dd x^2$ on $[0,L]$. Show that Robin conditions

$$
\psi'(0)=\alpha\psi(0),\qquad
\psi'(L)=\beta\psi(L),
$$

with real $\alpha,\beta$, make the boundary form vanish for every pair of functions satisfying the same conditions.

<details>
<summary>Hint 1</summary>

Evaluate the boundary form separately at $0$ and $L$.

</details>

<details>
<summary>Hint 2</summary>

Use the reality of $\alpha$ and $\beta$ when complex-conjugating the condition on $\phi$.

</details>

**Oral check 1.** Why would complex $\alpha$ generally spoil symmetry?

**Oral check 2.** Does vanishing of the boundary form alone prove self-adjointness, or only symmetry?

<details class="solution" open>
<summary>Solution</summary>

At $x=0$,

$$
\phi^*(0)\psi'(0)-\phi'^*(0)\psi(0)
=\alpha\phi^*(0)\psi(0)-\alpha\phi^*(0)\psi(0)=0.
$$

The same cancellation holds at $L$ with $\beta$. Thus the boundary form vanishes. Establishing self-adjointness also requires verifying that the adjoint domain introduces no additional boundary freedom.

</details>

## Check your understanding

In one sentence: what extra datum turns a formal Hamiltonian into a quantum observable?

## Connections and next step

- New pressure point: operator domain versus differential expression.
- Retrieve later: deficiency indices and the $U(2)$ family of interval extensions.
- Unresolved: which extensions can be generated as limits of regular boundary potentials?