Case Study 2 — The Qubit: When the Inner Product Becomes a Probability

Field: quantum information / physics. This is the chapter's anchor culmination (and the promise of Chapter 5, paid in full): the state of a quantum bit is a unit vector in the complex inner product space $\mathbb{C}^2$, and every number an experiment measures is a squared inner product. We make the geometry of §34.5 and §34.8 do real physics.

A bit that is a vector

A classical bit is $0$ or $1$. A qubit is richer: its state is a unit vector in the two-dimensional complex inner product space $\mathbb{C}^2$,

$$ |\psi\rangle = \alpha\,|0\rangle + \beta\,|1\rangle = \begin{bmatrix}\alpha\\\beta\end{bmatrix},\qquad \alpha,\beta\in\mathbb{C},\qquad |\alpha|^2+|\beta|^2 = 1, $$

where $|0\rangle=\mathbf{e}_1$ and $|1\rangle=\mathbf{e}_2$ are an orthonormal basis (physicists' "ket" notation). The complex inner product is the conjugating one from §34.5, written by physicists as a bracket: $\langle\psi|\phi\rangle = \overline{\alpha}\gamma+\overline{\beta}\delta$ for $|\phi\rangle=\gamma|0\rangle+\delta|1\rangle$. The conjugate is not optional — §34.5 showed that without it the state $|{+}i\rangle=\tfrac{1}{\sqrt2}(|0\rangle+i|1\rangle)$ would have "length" zero, an absurdity. With the conjugate, $\langle\psi|\psi\rangle = |\alpha|^2+|\beta|^2$ is real, non-negative, and the normalization condition says exactly $\lVert\psi\rVert=1$.

Why must the state be a unit vector? Because the geometry is about to become probability. This is the deepest single idea in quantum mechanics, and it is nothing but the inner-product geometry of this chapter.

Measurement is projection; probability is squared overlap

When you measure a qubit $|\psi\rangle$ in the basis $\{|0\rangle,|1\rangle\}$, you get outcome $0$ or $1$ — and which one is genuinely random. The Born rule gives the probabilities as squared overlaps:

$$ P(0) = |\langle 0|\psi\rangle|^2 = |\alpha|^2,\qquad P(1) = |\langle 1|\psi\rangle|^2 = |\beta|^2. $$

Read through this chapter's lens, this is a sentence you already know how to parse. The overlap $\langle 0|\psi\rangle$ is the projection of the state onto the $|0\rangle$ direction (Chapter 19's projection onto a unit vector, $\hat{\mathbf{u}}\cdot\mathbf{v}$, now over $\mathbb{C}$). Its squared magnitude is a squared length. So the probability of an outcome is the squared length of the projection of the state onto that outcome's direction. Projection, squared, reinterpreted as probability.

And the probabilities sum to one automatically — not by a separate axiom, but by the Pythagorean/Parseval identity of §34.7. Because $|0\rangle,|1\rangle$ are orthonormal and $|\psi\rangle$ is a unit vector,

$$ P(0)+P(1) = |\alpha|^2 + |\beta|^2 = \lVert\psi\rVert^2 = 1. $$

The law of total probability — "something is always observed" — is the Pythagorean theorem in an orthonormal basis. Geometry and physics are the same statement.

# A qubit lives in C^2; Born rule: P(outcome) = |<outcome|psi>|^2 = squared overlap.
import numpy as np
def braket(u, v):                    # <u|v> = conj(u) . v  (np.vdot conjugates the FIRST arg)
    return complex(np.vdot(u, v))

ket0, ket1 = np.array([1, 0]), np.array([0, 1])
psi = np.array([1, np.sqrt(3)]) / 2          # (|0> + sqrt(3)|1>)/2 : alpha=1/2, beta=sqrt3/2
print(round(braket(psi, psi).real, 6))       # 1.0   -> unit vector (normalized state)
print(round(abs(braket(ket0, psi))**2, 6))   # 0.25  -> P(0) = |1/2|^2
print(round(abs(braket(ket1, psi))**2, 6))   # 0.75  -> P(1) = |sqrt3/2|^2
print(round(abs(braket(ket0, psi))**2 + abs(braket(ket1, psi))**2, 6))  # 1.0 -> probabilities sum to 1

The outputs 1.0, 0.25, 0.75, 1.0 confirm it: the state $\tfrac12|0\rangle + \tfrac{\sqrt3}{2}|1\rangle$ is normalized, yields outcome $0$ with probability $\tfrac14$ and outcome $1$ with probability $\tfrac34$, and those probabilities sum to $1$. A measured probability is a squared inner product — full stop.

Distinguishability is orthogonality

Here is where the geometry pays a dividend that pure intuition would not give you. When can two quantum states be told apart with certainty by a single measurement? The answer is crisp: exactly when they are orthogonal, $\langle\psi|\phi\rangle=0$. Orthogonal states share no overlap, so there is a measurement that returns one answer for $|\psi\rangle$ and a different answer for $|\phi\rangle$, every time. States that are nearly parallel (overlap close to $1$ in magnitude) are nearly impossible to distinguish, and the probability of confusing them is governed by the squared overlap $|\langle\psi|\phi\rangle|^2$.

Consider the four states at the heart of quantum cryptography. The "computational" basis $|0\rangle, |1\rangle$ is orthonormal: $\langle 0|1\rangle = 0$, so these two are perfectly distinguishable. The "diagonal" basis $|+\rangle = \tfrac{1}{\sqrt2}(|0\rangle+|1\rangle)$ and $|-\rangle = \tfrac{1}{\sqrt2}(|0\rangle-|1\rangle)$ is also orthonormal: $\langle +|-\rangle = \tfrac12(1\cdot 1 + 1\cdot(-1)) = 0$. But a state from one basis is not orthogonal to a state from the other: $\langle 0|+\rangle = \tfrac{1}{\sqrt2}$, so $|0\rangle$ and $|+\rangle$ overlap, and a measurement in the wrong basis confuses them with probability $|\langle 0|+\rangle|^2 = \tfrac12$.

# Distinguishability = orthogonality. Overlaps decide everything.
import numpy as np
def braket(u, v): return complex(np.vdot(u, v))
ket0, ket1 = np.array([1, 0]), np.array([0, 1])
plus  = np.array([1,  1]) / np.sqrt(2)       # |+>
minus = np.array([1, -1]) / np.sqrt(2)       # |->
print(round(braket(ket0, ket1).real, 6))     # 0.0  -> |0>,|1> orthogonal: perfectly distinguishable
print(round(braket(plus, minus).real, 6))    # 0.0  -> |+>,|-> orthogonal: also distinguishable
print(round(abs(braket(ket0, plus))**2, 6))  # 0.5  -> |0>,|+> overlap: confused half the time

The outputs 0.0, 0.0, 0.5 are the geometric facts that the BB84 quantum key distribution protocol is built on. BB84 encodes bits in randomly chosen bases ($\{|0\rangle,|1\rangle\}$ or $\{|+\rangle,|-\rangle\}$); an eavesdropper who measures in the wrong basis gets a random answer (probability $\tfrac12$ of error, because the cross-basis overlap is $\tfrac{1}{\sqrt2}$) and unavoidably disturbs the state, revealing their presence. The security of one of the few provably secure cryptographic schemes rests on the non-orthogonality of $|0\rangle$ and $|+\rangle$ — a single inner product, $\langle 0|+\rangle=\tfrac{1}{\sqrt2}$.

Why the conjugate matters: a genuinely complex state

To see that the complex inner product (not just a real one) is doing essential work, take the state $|{+}i\rangle = \tfrac{1}{\sqrt2}(|0\rangle + i|1\rangle)$ — the one whose naive squared length was zero in §34.5. Under the correct conjugating inner product it is a perfectly good unit vector, and its physics comes out real precisely because of the conjugate.

# A genuinely complex state |+i> = (|0> + i|1>)/sqrt2: physics is real thanks to the conjugate.
import numpy as np
def braket(u, v): return complex(np.vdot(u, v))
plus   = np.array([1, 1])  / np.sqrt(2)
plus_i = np.array([1, 1j]) / np.sqrt(2)
print(round(braket(plus_i, plus_i).real, 6))       # 1.0   -> unit vector (conjugate makes it real & positive)
print(round((plus_i @ plus_i).real, 6))            # 0.0   -> WRONG: no-conjugate "norm" is nonsense
print(np.round(braket(plus, plus_i), 6))           # (0.5+0.5j)  -> a complex overlap
print(round(abs(braket(plus, plus_i))**2, 6))      # 0.5   -> but |overlap|^2 is real: a valid probability
print(np.round(braket(plus_i, plus), 6))           # (0.5-0.5j)  -> = conj(<+|+i>): conjugate symmetry

The outputs tell the whole story of §34.5. The conjugating inner product gives $\langle {+}i|{+}i\rangle = 1.0$ (real, positive — a valid state), while the naive plus_i @ plus_i returns the absurd $0.0$ for a nonzero vector. The overlap $\langle +|{+}i\rangle = 0.5+0.5i$ is genuinely complex, yet its squared magnitude $|\langle +|{+}i\rangle|^2 = 0.5$ is real — a legitimate probability. And $\langle {+}i|+\rangle = 0.5-0.5i$ is the complex conjugate of $\langle +|{+}i\rangle$, the conjugate-symmetry axiom in action. The conjugate we were forced to introduce in §34.5 is exactly what keeps measured probabilities real. Strip it out and quantum mechanics returns nonsense.

What this case study shows

The qubit is the chapter's thesis at its most dramatic. A quantum state is a unit vector; measurement is projection; probability is the squared overlap; total probability is the Pythagorean identity; distinguishability is orthogonality; and the complex conjugate of §34.5 is what makes every measured quantity real. None of this is metaphor — it is the literal mathematics, and it is the inner-product geometry of Part IV transplanted into $\mathbb{C}^2$. When Chapter 5 said "the state of a qubit is a vector" and promised the full story would wait for Hilbert space, this was the story: the geometry generalizes, and in its complex form it is the operating language of quantum information. The full physical development — many qubits, entanglement, the infinite-dimensional Hilbert spaces of wave mechanics — is the subject of Hilbert space in quantum mechanics, but the geometric heart is the one operation you now own: the inner product.