11
$\begingroup$

If $A$ is a $2\times2$ matrix, what is $\det(4A)$ in terms of $\det(A)$?

This seems trivial, but I'm not sure exactly what they are asking. I'm guessing I have some matrix $A = \begin{bmatrix}a&b\\c&d\end{bmatrix}$ where I know $\det(A) = ad - bc$. So if they want to know what is $\det(4A)$ wouldn't it just be $4A = 4\begin{bmatrix}a&b\\c&d\end{bmatrix} = \begin{bmatrix}4a&4b\\4c&4d\end{bmatrix} = \det(A) = 16ad - 16bc$?

$\endgroup$
1
  • 3
    $\begingroup$ This is not really an answer, but your confusion seems more based in the wording of the question than in properties of determinants. To generalize: write $\mathsf E$ in terms of $\mathsf V$ means the expected answer is an equation $\mathsf E = \text{RHS}$, where the right-hand side is an expression in only the variable(s) $\mathsf V$. $\endgroup$ – Lynn Jul 14 '16 at 10:25
36
$\begingroup$

Let's consider your equation $$4A = 4\begin{bmatrix}a&b\\c&d\end{bmatrix} = \begin{bmatrix}4a&4b\\4c&4d\end{bmatrix} = \det(A) = 16ad - 16bc$$

The first three objects in this are $2\times 2$ matrices while the last 2 are numbers. So clearly these can't all be equal. What you really want to say here is

$$\color{red}{\det(}4A\color{red}{)} = \color{red}{\det\left(\color{black}{4\begin{bmatrix}a&b\\c&d\end{bmatrix}}\right)} = \color{red}{\begin{vmatrix}\color{black}{4a}&\color{black}{4b}\\\color{black}{4c}&\color{black}{4d}\end{vmatrix}} = 16ad - 16bc$$

Then just complete the logic with a final $$=16\det(A)$$ and you're done.

$\endgroup$
1
  • 22
    $\begingroup$ And, in general, for $n\times n$ matrix $A$, $$\det(kA) = k^n\det A$$ This is because a determinant is a sum of $n$–term products of the matrix items, so multipying the matrix by a numer $k$ causes each of items be multiplied by $k$, hence each term of the sum gets a $k^n$ factor. $\endgroup$ – CiaPan Jul 13 '16 at 8:12
20
$\begingroup$

More generally, if $A$ is $n \times n$, $\det(cA) = c^n\det(A)$ for any scalar $c$. This is because the determinant is a multilinear function of its columns.

In fact, one can define the determinant as the unique function from $n \times n$ matrices to scalars that is $n$-linear alternating in the columns, and takes the value $1$ for the identity matrix.

$\endgroup$
2
  • 2
    $\begingroup$ @Yusha see Leibniz formula of determinant if you want the detail. $\endgroup$ – Henricus V. Jul 13 '16 at 6:43
  • 1
    $\begingroup$ @Yusha Do you know the "elementary row operation" that scales one single row of the matrix by a factor $c$? And know how the determinant behaves under this elementary row operation? Going from a matrix $A$ to the matrix $cA$ is exactly the same as performing such row scalings $n$ times. So you can see the formula $\det(cA) = c^n\det(A)$ in this perspective. There is an interesting special case, $$\det(-I) = (-1)^n,$$ where I have taken $c=-1$ and $A=I$, the identity matrix. This shows that reflection in the origin is a rotation (i.e. orientation-preserving) if and only if $n$ is even. $\endgroup$ – Jeppe Stig Nielsen Jul 13 '16 at 11:41
9
$\begingroup$

In particular, $\det(4A) = 4^2 \det(A)$.

Another way to think about it is to remember that the determinant of an $n \times n$ matrix gives the (signed) volume of the $n$-dimensional box enclosed by its column vectors. Increasing each side length by a factor of $k$ will increase the volume by a factor of $k^n$.

$\endgroup$
2
  • $\begingroup$ Could you please explain the line enclosed by its column vectors in your answer. $\endgroup$ – sameerkn Jul 13 '16 at 10:08
  • $\begingroup$ @sameerkn, see the pictures in [ askamathematician.com/2013/05/… ], for example. $\endgroup$ – Vectornaut Jul 13 '16 at 19:49
8
$\begingroup$

determinant is a multilinear form so for all matrix $A\in M_n(\mathbb{R})$ and for all $\alpha\in \mathbb{R}$ $$ \det(\alpha A)=\alpha^n \det(A) $$

$\endgroup$
5
$\begingroup$

It is indeed trivial. With $A$ as you have it, $\det(A)=ad-bc$.

Now, $\det(4A)=\det(\begin{bmatrix} 4a & 4b\\ 4c & 4d \end{bmatrix})$=$16ad-16bc=16(ad-bc)=16*\det(A)$.

Alternatively, just use the formula: if $A$ is a real valued $n \times n$ matrix, then $\det(cA) = c^n\det(A)$ $\forall$ $c\in \mathbb{R}$.

$\endgroup$
4
$\begingroup$

Seeing the determinant as the volume of the unit cube $Q$ under the map $A$, one sees that the composition with the multiplication $x\mapsto kx$ will dialate the image $AQ$ by a factor $k$.

But as the volume scales like $k^n$ in $n-$dimensional space, you get that $$\det(kA)=vol(kA(Q))=k^n vol(A(Q))=k^n\det(A).$$

$\endgroup$
2
$\begingroup$

Another view of the multilinear aspect of the determinant of a square matrix $A$ is the "recursive form" involving minors. If the matrix is $n\times n$, one can compute determinant with minors stemming from $(n-1)\times (n-1)$ matrices:

$$|A| = \sum (-1)^{i+j} a_{ij} M_{ij}\,.$$

The determinant of a $1\times 1$ matrix (a scalar) grows linearly with the factor $\lambda$ applied to the (unique) matrix element. Hence the determinant of a $2\times 2$ matrix grows twice faster, since you get one $\lambda$ for the $a_{ij}$ terms (for instance $a$ and $c$ for you), and one for the $M_{ij}$ terms, which derive from dimension $1\times 1$ matrices (for instance $b$ and $d$ for you), as you wrote with $\lambda=4$.

For a $3\times 3$ matrix, you get one $\lambda$ for the $a_{ij}$ terms, and $\lambda^2$ for the $M_{ij}$ terms, coming from dimension $2\times 2$ objects. More generally i, you get one $\lambda$ for the $a_{ij}$ terms, and $\lambda^{n-1}$ for the $M_{ij}$ terms, which are computed from dimension $(n-1)\times (n-1)$ matrices, hence $\lambda^n $ total.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.