2.6.6 · D5Matrices & Determinants — Introduction
Question bank — Transpose — definition, properties
This page is a rapid-fire trap detector for the Transpose — definition, properties. Each line is a claim or question; read it, decide your answer, THEN reveal. The point is not arithmetic — it is catching the conceptual slips that this topic invites. Reasoning lives on the answer side, never a bare "yes/no".
Before we start, one anchor we use everywhere: the transpose is the flip across the main diagonal (the diagonal running from top-left downward). The picture below is the whole idea — look at the red diagonal fold line, and notice that entries on it stay put while off-diagonal partners swap across it.
True or false — justify
The transpose of an matrix is again
False — flipping rows and columns swaps the shape, so is . It is only when (a square matrix).
If is symmetric then must be square
True — forces and to have the same shape, but is while is , so .
Every diagonal matrix is symmetric
True — off-diagonal entries are all zero, so whenever , which is exactly the symmetry condition.
A skew-symmetric matrix can have a nonzero number on its main diagonal
False — forces , so every diagonal entry must be zero.
holds for every matrix, even non-square ones
True — flipping across the diagonal twice returns each entry to its start: , shape and all.
False — the order reverses: . The wrong version usually isn't even a legal product because the inner dimensions no longer match.
requires and to be the same shape
True — you can only add matrices of equal shape, and transposing each preserves the (now flipped) shared shape, so the sum is defined on both sides.
If then is orthogonal
True — that is precisely the definition of an orthogonal matrix: its transpose equals its inverse, so .
For any square matrix,
False — this holds only for orthogonal matrices. Transpose is a structural flip; inverse is what undoes multiplication. They coincide rarely.
is always symmetric (for square )
True — , so it equals its own transpose.
is always skew-symmetric (for square )
True — , which is exactly the skew-symmetric condition.
If and is , then can be any real number
False — a matrix is its own transpose, so forces ; only the zero scalar qualifies.
Spot the error
" so we must transpose the scalar too."
A scalar is not a matrix that gets flipped; . The number has no rows or columns to swap — "" is meaningless.
"."
The full reversal is . Transpose reverses the whole chain like reading it backwards, not term by term in place.
" is symmetric, is symmetric, so is symmetric."
Not generally — , which equals only if and commute. Symmetry of the product is a stronger, extra condition (see Matrix Operations).
"To transpose I just rewrite the same row."
A row vector transposes to a column vector; the numbers stack vertically. The shape genuinely changes from to .
", so the transpose of the inverse is ."
The correct identity is . The reader confused "transpose" with "inverse" mid-line — two different operations (see Matrix Inverse).
"Since transpose flips the diagonal, the diagonal entries all move to new spots."
The diagonal entries sit exactly on the fold line, so they stay put. Only off-diagonal entries swap partners.
"Every matrix equals a symmetric matrix plus a skew-symmetric one, so every matrix is symmetric."
The decomposition is true, but the second piece is skew-symmetric, not symmetric — so itself need not be symmetric at all.
Why questions
Why does the order reverse in ?
Because entry ; rewriting each factor via transpose, and , gives — the shared summed index only lines up when comes first.
Why isn't — show it on a tiny example?
Take : then , whereas — clearly different, so only the reversed order works.
Why must a symmetric matrix be square, not just "shaped nicely"?
The equation compares two matrices entry-for-entry, and that is only possible if they have identical dimensions, which forces 's row count to equal its column count.
Why does transposing an equation like prove ?
Applying transpose to both sides and using the reversal rule gives , which says is exactly the matrix that undoes — the definition of .
Why do scalars and sums "ignore" the transpose flip, but products don't?
Scaling and addition act on each entry alone: and have nothing to reroute when you swap and . A product entry mixes a whole row of with a column of , so flipping indices reshuffles which row meets which column — and that is what forces the order to reverse.
Why is always symmetric, for any (even non-square) ?
Compute ; it equals its own transpose. This is why appears so often in Inner Product Spaces and least-squares work.
Why is the transpose relevant to dot products at all?
For column vectors , the dot product is — the transpose turns a column into a row so that matrix multiplication produces the single sum-of-products number that defines the inner product.
Why does hold for every square matrix?
The determinant is built from sums of products picking one entry per row and per column; transposing just relabels every "row , column " as "row , column ", which permutes the same terms without changing their signs. Minimal check: for , and , identical. See Determinants.
Why can a symmetric matrix's eigen-story be "nicer" than a general matrix?
Because is a strong constraint; it guarantees real eigenvalues and orthogonal eigenvectors, a fact that leans directly on Symmetric Matrices and feeds into Eigenvalues and Eigenvectors.
Edge cases
What is the transpose of a matrix ?
It is itself — a single number sits on the diagonal with nothing to swap, so every matrix is (trivially) symmetric.
What is the transpose of an empty matrix (no rows, columns)?
It is the matrix — the shape-swap rule still applies, turning "no rows, columns" into " rows, no columns". No entries exist to move, so the operation is well-defined but vacuous.
Is the zero matrix symmetric, skew-symmetric, or both?
Both — every entry is , so (symmetric) and (skew) hold simultaneously; the square zero matrix is the only matrix that is both.
What happens to the identity matrix under transpose?
— the identity is symmetric, since its only nonzero entries are 's on the diagonal, which stay fixed under the flip.
Does still hold if and are (non-square)?
Yes — the sum rule never needs squareness, only that and share a shape; each side is a valid matrix and they agree entry by entry.
For a square matrix, do and always have the same determinant?
Yes — always, because transposing only permutes the same signed products in the determinant expansion; it is a clean invariant even when isn't symmetric (see Determinants).
If is with , is ever equal to ?
No — a non-square matrix has no ordinary inverse at all, so the question "" cannot even be posed; orthogonality is a square-matrix notion.
Can a non-square matrix be symmetric?
No — symmetry requires , and has swapped dimensions, so a matrix can never equal its transpose.
The one takeaway
If you remember a single rule from this whole page, make it this: