Properties of determinants: Visual walkthrough

Everything here is about a $2 \times 2$ matrix because that is the smallest case you can draw. The exact same story runs in 3D (volumes) and higher; pictures just get harder.

Step 0 — What are we even looking at?

Before any symbol, here is the object.

A matrix is a box of numbers. Ours has two rows: $A = [a c b d]$ The top row $(a, b)$ and the bottom row $(c, d)$ are the only things we care about.
Read each row as an arrow (a vector) drawn from the origin: the orange arrow points to the spot $(a, b)$ , the teal arrow points to $(c, d)$ .
Those two arrows are two sides of a parallelogram (a leaning rectangle). The determinant is going to be the area of that parallelogram — with a plus or minus sign we'll explain.

WHAT the figure shows: the two rows as arrows and the parallelogram they enclose. WHY we draw it: from now on "determinant" means "signed area of this shape", and every property is a fact about how the shape moves. WHAT IT LOOKS LIKE: a leaning box; its area is the shaded region.

Step 1 — Where does $a d - b c$ come from? (the area, built from a rectangle)

WHAT we do: compute the parallelogram's area by boxing it inside a big rectangle of width $a$ and height $d$ , then subtracting the bits that stick out.

WHY this tool: we want a number for the area. The cleanest honest way is "big rectangle minus corner triangles" — no trigonometry needed, just areas of rectangles and triangles you already know.

Take the simple case where both arrows sit in the first quarter of the plane with the orange arrow lower and flatter. Enclose everything in an $a \times d$ rectangle.

$Area = whole rectangle a d - the two corner cut-offs b c$

$a d$ — the big rectangle: width $a$ (how far orange reaches sideways) times height $d$ (how far teal reaches up).
$b c$ — the total of the triangular/rectangular corners that are inside the rectangle but outside the parallelogram. They carve away exactly $b c$ .

So the area is $a d - b c$ — that is where the formula is born.

WHAT IT LOOKS LIKE: the shaded parallelogram is the big rectangle with two matching corner pieces (hatched) removed. The leftover shaded area is $a d - b c$ .

Step 2 — The sign: what does a negative area mean?

WHAT we do: ask what happens when we swap the two rows — put teal on top, orange on bottom.

WHY: the formula $a d - b c$ becomes $c b - d a = - (a d - b c)$ . The number flipped sign but the shape is identical. A shape can't have negative area — so the minus sign must mean something else.

The meaning: the sign tracks orientation — which way you turn to get from the first arrow to the second.

If you sweep counter-clockwise from row 1 to row 2 (the short way), the area is positive.
If you must sweep clockwise, the area is negative — the plane has been flipped over, like turning a page.

Swapping the rows is exactly turning the page: same shape, opposite orientation, opposite sign.

$det (swap rows) = - det (A) (this is P2)$

WHAT IT LOOKS LIKE: two copies of the same parallelogram. The left one has a small counter-clockwise arc (orange → teal) labelled $+$ ; the right, after the swap, has a clockwise arc labelled $-$ .

Step 3 — Equal rows: the parallelogram collapses (P3)

WHAT we do: make the two rows the same arrow: $(c, d) = (a, b)$ .

WHY: this is the most important degenerate case — and it drops straight out of Step 2.

Two identical arrows can't fence off any area — the "parallelogram" is squashed flat onto a single line. Flat shape ⇒ zero area.

You can also see it algebraically from the swap rule: $det = - det ⟹ 2 det = 0 ⟹ det = 0.$

Swapping two equal rows changes nothing, so the value equals itself.
But P2 says a swap negates the value.
The only number that equals its own negative is $0$ .

$det (two equal rows) = 0 (P3)$

WHAT IT LOOKS LIKE: both arrows lie on top of each other; the shaded region has been pinched down to a thin sliver of literally zero width.

Step 4 — Scaling one row stretches the area by the same factor (P4)

WHAT we do: multiply only the top row by a number $k$ : $(a, b) \to (k a, k b)$ .

WHY: we want to know what stretching one side of the parallelogram does. Intuition says "stretch one side $k$ times → area $k$ times", and the formula confirms it.

$det [k a c k b d] = (k a) d - (k b) c = k (a d - b c) = k det A$

$(k a) d$ and $(k b) c$ — every surviving term carries exactly one $k$ , because each product uses one entry from the top row.
So $k$ pulls out cleanly: the area is $k$ times bigger.

All cases of $k$ :

$k > 1$ : base longer → bigger area.
$0 < k < 1$ : base shorter → smaller area.
$k = 0$ : the top arrow vanishes to a point → flat shape → area $0$ (that's P7, a zero row).
$k < 0$ : the arrow flips to the opposite side → orientation flips → negative area, exactly matching the sign of $k$ .

WHAT IT LOOKS LIKE: the original parallelogram in solid outline, and a $k = 1.6$ stretched version in a lighter tint reaching $1.6 \times$ farther, covering $1.6 \times$ the area.

Step 5 — Sliding a row: the shear that saves the area (P6)

WHAT we do: replace the top row by $R_{1} + k R_{2}$ — add $k$ copies of the bottom arrow onto the top arrow. Nothing is added to the bottom row.

WHY: this is the single most useful operation for hand computation, and it's the one people find magical: the area does not change at all.

$det [a + k c c b + k d d] = (a + k c) d - (b + k d) c$

Multiply out and watch the extra pieces cancel: $= original a d - b c + = 0 k c d - k d c = a d - b c = det A$

$a d - b c$ — the original determinant, untouched.
$k c d - k d c$ — the new stuff created by the slide; it is $k (c d - d c) = 0$ because $c d = d c$ . It always cancels itself.

The geometry is the reason: sliding the top arrow parallel to the bottom arrow keeps the base and the perpendicular height identical. Area $=$ base $\times$ height, and neither changed. Think of a stack of cards pushed sideways — it leans, but its area is fixed.

$det (R_{1} \to R_{1} + k R_{2}) = det A (P6)$

WHAT IT LOOKS LIKE: the bottom arrow stays put; the top arrow slides along the dashed line parallel to the bottom arrow. The parallelogram leans into a new shape, but the shaded area is visibly the same (same base, same height marked with a right angle).

Step 6 — Cashing it in: shear to a triangle, read the diagonal (P8)

WHAT we do: use the "slide" of Step 5 to knock the bottom-left entry to $0$ , making the matrix triangular, then read the answer off the diagonal.

WHY: once a matrix is triangular (a zero below the diagonal), one arrow points straight along an axis. The parallelogram becomes an axis-aligned box tilted only one way, and its area is just the product of the two diagonal entries — no arithmetic circus.

Worked micro-example, all three tools at once:

\;\xrightarrow[\text{slide (P6)}]{R_1\to R_1-2R_2}\; \begin{vmatrix}0&-2\\1&3\end{vmatrix} \;\xrightarrow[\text{swap (P2)}]{R_1\leftrightarrow R_2}\; -\begin{vmatrix}1&3\\0&-2\end{vmatrix} =-(1\cdot -2)=2.$$ - **Slide (P6):** made a zero in the top-left, area unchanged. - **Swap (P2):** put the zero *below* the diagonal — remember the $-1$! - **Triangular (P8):** multiply the diagonal $1\cdot(-2)=-2$, apply the tracked sign $\Rightarrow 2$. Check directly: $2\cdot3-4\cdot1=6-4=2$. ✓ ![[deepdives/dd-maths-2.6.09-d2-s07.png]] **WHAT IT LOOKS LIKE:** three side-by-side frames — the original leaning box, the sheared box after the slide, and the final axis-aligned box whose area is literally width $\times$ height. --- ## The one-picture summary ![[deepdives/dd-maths-2.6.09-d2-s08.png]] Every property is one gesture on the same parallelogram: | Gesture | Picture | Effect on det | Rule | |---|---|---|---| | **Swap** rows | flip the page | $\times(-1)$ | P2 | | **Copy** a row | squash flat | $=0$ | P3 | | **Scale** a row by $k$ | stretch one side | $\times k$ | P4 | | **Zero** a row | shrink to a point | $=0$ | P7 | | **Slide** (add $kR_j$) | shear (cards leaning) | unchanged | P6 | | **Triangular** | axis-aligned box | product of diagonal | P8 | > [!mnemonic] One line to rule them all > **Swap Flips, Copy Kills, Scale Scales, Slide Saves.** > [!recall]- Feynman retelling — the whole walkthrough in plain words > Draw two arrows from a corner; they fence off a leaning box. The determinant is that box's area, > plus a sign that just remembers whether the arrows turn the "right" way (counter-clockwise) or the > mirror way (clockwise). Boxing the shape inside a big $a\times d$ rectangle and subtracting the > corner scraps gives $ad-bc$ — that's the whole formula. Now play: flip the two arrows and the box > turns over, so the sign flips (swap → $-$). Make both arrows identical and the box is squashed to > a line with no area (equal rows → $0$). Stretch one arrow $k$ times and the box is $k$ times > bigger (scale → $\times k$); stretch it $0$ times and it's a point (zero row → $0$). The sneaky > one: slide one arrow *along the direction of the other*. The box leans like a pushed deck of > cards, but its base and height never change — so the area is exactly the same (add a multiple → > saved). That last trick is what lets you shear any matrix into a triangle and just multiply the > diagonal. Nine "properties," one shape, four gestures. --- ## Recall check Which gesture leaves the area unchanged, and why? ::: Sliding one row by a multiple of another (P6); it's a shear that keeps base and perpendicular height fixed. Where does the minus sign in $ad-bc$ come from geometrically? ::: Orientation — clockwise sweep from row 1 to row 2 flips the page and negates the signed area. Why does scaling one row by $k$ give $k\det A$ but scaling the whole matrix gives $k^2\det A$? ::: One row scaled stretches one side ($\times k$); the whole matrix stretches both sides ($\times k\times k$). How do equal rows force $\det=0$ from the swap rule? ::: Swapping equal rows changes nothing yet must negate the value, and only $0$ equals its own negative. --- ## Connections - [[Properties of determinants]] — the parent: all nine rules stated and named. - [[Determinant — Definition and Expansion by Minors]] — the algebra behind the area picture. - [[Elementary Row Operations & Rank]] — swap/scale/slide are exactly the elementary operations. - [[Area of a Triangle using Determinants]] — half of this parallelogram. - [[Linear Transformations & Scaling of Area]] — the "area scaling factor" viewpoint made general. - [[Cofactors and Adjoint of a Matrix]] · [[Inverse of a Matrix via Adjoint]] · [[Cramer's Rule and Systems of Linear Equations]]