Part I · Chapter 7 · Final chapter of Part I

Postulates & Measurement

Read ~22 min Prerequisites Ch. 1–6 Tools repeated-measurement histogram, code cell Live

Nothing in this chapter is new. Every piece has already been built, one chapter at a time: vectors and norms (Ch. 1–2), unitary evolution (Ch. 4), tensor products (Ch. 5), projectors (Ch. 2) and eigenvalues (Ch. 1) feeding into Hermitian observables (Ch. 4 and Ch. 6). This chapter's only job is to state, plainly, the four rules that say how those pieces actually govern a physical system — the postulates of quantum mechanics.

Postulate 1 — State

The state of an isolated physical system is a unit vector |ψ⟩ in a complex vector space. "Unit" means ⟨ψ|ψ⟩ = 1 (Ch. 2) — exactly the |α|²+|β|²=1 constraint you first met in Ch. 3, now stated as the general rule it always was.

Postulate 2 — Evolution

A closed system's state evolves over time according to |ψ′⟩ = U|ψ⟩ for some unitary U (Ch. 4). Every gate in Ch. 3 and every circuit in the Composer was already an example of this postulate in action — you were applying it before you had a name for it.

Postulate 3 — Measurement (the Born rule)

An observable is represented by a Hermitian operator M (Ch. 4). Measuring it can only ever produce one of M's eigenvalues (Ch. 1) — that's the actual mechanism behind "measurement outcomes are real numbers," since Hermitian operators guarantee real eigenvalues. If |φᵢ⟩ is the eigenvector for eigenvalue λᵢ, the probability of getting that outcome is:

P(λᵢ) = |⟨φᵢ|ψ⟩|²

— the squared magnitude of an inner product, exactly the projector machinery from Ch. 2. After the measurement, the state collapses to |φᵢ⟩: the outcome you got is now, with certainty, what the system is in.

Postulate 4 — Composite systems

The state space of several systems together is the tensor product of their individual state spaces (Ch. 5). This is the entire reason a 2-qubit system needs 4 amplitudes rather than 4 — it's where every two-qubit state in the Composer and Ch. 3's circuit widgets actually came from.

Watching the Born rule converge

P(λᵢ) = |⟨φᵢ|ψ⟩|² is a single-shot probability — it tells you the odds for one measurement. It doesn't say what one measurement will give you (that's genuinely random), but it makes a sharp prediction about what happens over many measurements: the fraction of each outcome should converge to that exact probability as the count grows. That convergence is something you can actually watch happen.

Fig. 7.1 — set a state, pick a basis, run shotswatch convergence
Measure in:
|0⟩
|1⟩
The blue bar is the empirical fraction so far; the amber line is the exact Born-rule probability |⟨φᵢ|ψ⟩|², computed live as a real inner product — the same calculation Ch. 2's projector section walked through by hand. Switching basis with shots already run resets the count, since you're now measuring a genuinely different observable.

Why collapse matters for what you can know next

The collapse part of Postulate 3 is easy to skim past, but it's the reason measurement order matters at all. Set θ = 90°, φ = 0° (state |+⟩) and run some shots in the Z basis — you'll see roughly 50/50, as expected. But once a single measurement actually happens and collapses the state to, say, |0⟩, a second Z-basis measurement immediately after will give |0⟩ again with certainty — the randomness was only ever about the first measurement of a given observable on a given state, not a property of the universe re-rolling dice forever.

Fig. 7.2 — code, runs in your browser
Run this a few times — the empirical P(0) jitters around 0.5 but never strays far, the same convergence Fig. 7.1 shows visually.

End of Part I

Everything from here on — density matrices, entanglement, error correction, algorithms — is built entirely out of these four postulates and the linear algebra underneath them. There's no new foundational machinery coming; just new things to build with what you already have.

Check your understanding

State the four postulates in one phrase each.

State: a unit vector. Evolution: unitary transformation. Measurement: Hermitian operator, outcome is an eigenvalue with probability |⟨eigenvector|ψ⟩|², then collapse. Composite systems: tensor product of individual spaces.

Check your understanding

Why must a measurement outcome's probability use the inner product, rather than just reading off an amplitude directly?

Because the relevant probability is for a specific basis vector, found via the projection of ψ onto that basis — exactly what an inner product computes. P(λᵢ) = |⟨φᵢ|ψ⟩|² is literally "how much of ψ points along φᵢ," squared.

Check your understanding

For |+⟩, what's P(0) in the Z basis, and what's P(+) in the X basis?

P(0) = 0.5 in the Z basis (|+⟩ is an equal superposition of |0⟩ and |1⟩), but P(+) = 1.0 in the X basis, since |+⟩ is one of the X-basis vectors — measuring an observable in its own eigenbasis always gives a certain outcome.

Check your understanding

After a measurement collapses |ψ⟩ to |0⟩, what happens if you immediately measure the same observable again?

You get |0⟩ again, with certainty (probability 1). The randomness is only in the first measurement of a given state — collapse means the state genuinely is |0⟩ afterward, not just "probably."

Check your understanding

Which earlier chapter's idea does each postulate depend on: (a) Evolution, (b) Composite systems?

(a) Ch. 4 — unitary operators. (b) Ch. 5 — the tensor product.

Check your understanding

Why does the Born rule specifically require the observable to be Hermitian?

Because the outcome must be one of the observable's eigenvalues, and that eigenvalue has to be a real number for the outcome to mean anything physically. Hermitian operators are exactly the ones guaranteed to have real eigenvalues (Ch. 4).

Review queue

Scroll back to a chapter's quiz cards to review them once they're due — a full cross-chapter queue view is a planned upgrade.