LINEAR-ALGEBRA · Interactive Practice | Unit 25 · Video 1

When Transpose Isn't Enough: The Conjugate Transpose

IKey Formulas

Concept	Real Case	Complex Case
Length squared	$x^T x$	$z^H z = \\|z_1\\|^2 + \\|z_2\\|^2 + \cdots + \\|z_n\\|^2$
Inner product	$y^T x$	$\langle x, y \rangle = y^H x$
Symmetric / Hermitian	$A^T = A$	$A^H = A$
Orthogonal / Unitary	$Q^T Q = I$	$Q^H Q = I$ , so $Q^{-1} = Q^H$

The conjugate transpose $A^H$ is: transpose, then conjugate every entry. The rule is simple — whenever you transpose, also conjugate.

IIVisualization 1 — Why Transpose Alone Fails

For a complex vector, plain $z^T z$ can be zero or negative — only $z^H z$ gives a real, non-negative length.

IIIVisualization 2 — Building a Hermitian Matrix

$A$ is Hermitian when $A^H = A$ : that forces a real diagonal and conjugate-paired off-diagonals.

A[0,1] real 3 A[0,1] imag 1 A[0,0] imag 0 A[1,0]

IVVisualization 3 — Verifying a Unitary Matrix

$Q$ is unitary when $Q^H Q = I$ — its columns orthonormal under the complex inner product.

Scale s 1.0 Product

VQuiz Questions

Question 1

Let $z = \begin{bmatrix} 2 \\ 1 + i \end{bmatrix}$ . What is $z^H z$ (the length squared)?

✅ Correct! Each $|z_i|^2$ is real and non-negative, and $|2|^2 + |1+i|^2 = 4 + 2 = 6$ .

❌ Not quite. That's what you'd get from $z^T z$ without conjugating — exactly the kind of nonsense answer that motivates using $H$ .

❌ Not quite. Zero would mean $z$ is the zero vector, but $z \ne 0$ . With the conjugate, the length squared is positive.

❌ Not quite. Compute $|z_1|^2 + |z_2|^2$ where $|a+bi|^2 = a^2 + b^2$ .

Show solution

Solution:

The conjugate transpose of $z$ is $z^H = \begin{bmatrix} 2 & 1 - i \end{bmatrix}$ (transpose, then conjugate each entry).

$z^H z = 2 \cdot 2 + (1 - i)(1 + i) = 4 + (1 - i^2) = 4 + (1 + 1) = 6$

Equivalently, $z^H z = |2|^2 + |1 + i|^2 = 4 + 2 = 6$ . The result is real and positive, as guaranteed.

Question 2

If $z$ is the column vector $\begin{bmatrix} 1 \\ 2i \end{bmatrix}$ , what is $z^H$ ?

✅ Correct! Transpose flips column to row, then conjugation flips $2i$ to $-2i$ .

❌ Not quite. You transposed but forgot to conjugate. $H$ requires both steps.

❌ Not quite. You conjugated but forgot to transpose. $H$ swaps shape and conjugates.

❌ Not quite. $H$ never negates the real part — only flips the sign of imaginary parts.

Show solution

Solution:

Apply $H$ in two steps:

Transpose: the column $\begin{bmatrix} 1 \\ 2i \end{bmatrix}$ becomes the row $\begin{bmatrix} 1 & 2i \end{bmatrix}$ .
Conjugate every entry: $1$ stays $1$ , and $2i$ becomes $-2i$ .

So $z^H = \begin{bmatrix} 1 & -2i \end{bmatrix}$ , a row vector.

Question 3

Consider the matrix $A = \begin{bmatrix} 4 & 2 + 3i \\ 2 - 3i & 7 \end{bmatrix}.$

Is $A$ Hermitian?

✅ Correct! The diagonal is real and the off-diagonal entries $2+3i$ and $2-3i$ form a conjugate pair, so $A^H = A$ .

❌ Not quite. Check both conditions: diagonal entries $4, 7$ are real, and off-diagonal $2+3i$ and $2-3i$ are conjugates — Hermitian is satisfied.

❌ Not quite. Try again — the hints above can help.

Show solution

Solution:

Check the structural conditions for $A^H = A$ :

Diagonal entries real? $4$ and $7$ — yes, both real.
Off-diagonal conjugate pair? $A_{12} = 2 + 3i$ and $A_{21} = 2 - 3i$ . We need $A_{21} = \overline{A_{12}} = \overline{2 + 3i} = 2 - 3i$ . Yes — they are conjugates.

Both conditions hold, so $A^H = A$ and $A$ is Hermitian.

Question 4

A matrix $Q$ is unitary, meaning $Q^H Q = I$ . Which statement about $Q^{-1}$ is correct?

✅ Correct! From $Q^H Q = I$ we read off $Q^{-1} = Q^H$ . Inverting a unitary matrix is as cheap as conjugate-transposing it.

❌ Not quite. That's the rule for real orthogonal matrices. For complex unitary matrices, conjugation is essential — $Q^T Q \ne I$ in general.

❌ Not quite. Only special matrices (involutions) satisfy $Q^2 = I$ . The general unitary rule is $Q^{-1} = Q^H$ .

❌ Not quite. Unitary matrices are always invertible — the formula $Q^H Q = I$ tells us $Q^{-1} = Q^H$ .

Show solution

Solution:

By definition, $Q^H Q = I$ . Multiplying both sides by $Q^{-1}$ on the right:

$Q^H = Q^{-1}.$

So the inverse of a unitary matrix is just its conjugate transpose — extremely cheap to compute. This is the complex generalization of "the inverse of an orthogonal matrix is its transpose."

Note: $Q^T$ alone is not the inverse for complex $Q$ — conjugation is essential. For instance, with $Q = \frac{1}{\sqrt{2}}\begin{bmatrix} 1 & i \\ i & 1 \end{bmatrix}$ , you can check that $Q^T Q \ne I$ but $Q^H Q = I$ .

Solved: 0 / 4