Scalar multiplication: Visual walkthrough — Maths deep dive

We will build up a few symbols first, so nothing surprises you:

Step 1 — Start with one number and one arrow

WHAT. Forget matrices for a moment. Take a single number $a$ . Scaling it by $k$ means the product $k \cdot a$ — that's just ordinary multiplication.

WHY. A matrix is made of single numbers. If we can't agree what "scale one number" means, we can't scale a whole grid. So we pin down the simplest case first.

PICTURE. Draw the number $a$ as an arrow of length $a$ on a line. Multiplying by $k = 2$ makes the arrow twice as long. Multiplying by $k = \frac{1}{2}$ makes it half as long. That stretch is all scaling ever does to a single value.

the stretch factor k \cdot the original number a = stretched number k a

Step 2 — What happens to a negative number, or zero?

WHAT. We must check the awkward inputs before we trust the idea. If $a$ is negative, its arrow points left. If $k$ is negative, it flips the arrow to the other side of $0$ and then stretches. If $k = 0$ , the arrow collapses to a single point at the origin.

WHY. A rule you only tested on nice positive numbers is a rule that will break the first time real data has a minus sign. We cover every sign now so no scenario surprises us later.

PICTURE. Three arrows from the origin: $k = 2$ (longer, same side), $k = - 1$ (same length, flipped side — a reflection), $k = 0$ (shrunk to nothing).

Here $∣ k ∣$ (read "the size of $k$ ") is just $k$ with any minus sign removed — it measures length, which is never negative.

Step 3 — Lay the numbers out in a grid

WHAT. Now put four numbers in a $2 \times 2$ grid so we can see the whole matrix at once:

A = [a_{11} a_{21} a_{12} a_{22}]

WHY. The matrix is our container. Before we scale it, we must be crystal clear that each entry lives in a fixed cell, and each cell is just one of the numbers from Step 1.

PICTURE. A grid of four cells, each labelled by its address $a_{ij}$ . The accent colour marks one cell, $a_{21}$ (row 2, column 1), so you can watch that exact cell in the next steps.

one number a_{ij} \in the grid holding all of them A

Step 4 — Scale one cell, ignore the rest

WHAT. Take just the highlighted cell $a_{21}$ and scale it by $k$ using Step 1. It becomes $k \cdot a_{21}$ . Nothing else moves yet.

WHY. Here is the key insight the parent note stated but did not show: the cells do not talk to each other. The value in one cell has no effect on how another cell is scaled. So scaling the grid is nothing more than repeating the single-number rule, cell by cell.

PICTURE. The same grid; the red cell's number is replaced by $k a_{21}$ while the three black cells sit untouched — grey, waiting their turn.

(k A)_{21} = stretch k \cdot this cell only a_{21}

The symbol $(k A)_{21}$ means "the entry in row 2, column 1 of the result $k A$ ."

Step 5 — Do it to every cell at once

WHAT. Repeat Step 4 for all four cells simultaneously. Each $a_{ij}$ becomes $k a_{ij}$ , right where it sits.

WHY. Because the cells are independent (Step 4), "scale the matrix" can only mean "scale each cell" — there is nothing else to scale. This is the whole definition, now forced on us rather than declared.

PICTURE. All four cells turn red at once, each showing $k a_{ij}$ . The grid's shape — 2 rows, 2 columns — is identical; only the contents changed.

The general statement (any size $m \times n$ ) is the exact same idea copied into more cells:

(k A)_{ij} = k \cdot a_{ij} for every i, j

Step 6 — The shape cannot change (a degenerate check)

WHAT. Ask: could scaling ever add a row, delete a column, or move a cell? No. Multiplication by $k$ only changes the number inside each cell.

WHY. This kills a common fear (see Mistake 3 in the parent): scaling is a contents operation, not a structure operation. An $m \times n$ matrix stays $m \times n$ . Even the extreme case $k = 0$ keeps the shape — you get a full $m \times n$ grid of zeros (the zero matrix $O$ ), not an empty matrix.

PICTURE. Left: $A$ with its outline. Right: $k A$ with the same outline but scaled numbers. A red bracket shows the outline is byte-for-byte identical.

A is m \times n ⟹ k A is m \times n

$k = 1$ : every cell unchanged, so $1 \cdot A = A$ .
$k = 0$ : every cell becomes $0$ , so $0 \cdot A = O$ (same shape, all zeros).

Step 7 — Watch a real example transform

WHAT. Let $A = [25 - 3 0]$ and scale by $k = 4$ .

WHY. To prove the picture and the arithmetic agree, and to include a negative entry ( $- 3$ ) and a zero entry ( $0$ ) so every kind of cell shows up.

PICTURE. Each cell of $A$ has an arrow of its own length; every arrow grows to $4 \times$ its length, and the $- 3$ arrow (pointing the negative way) grows while keeping its direction. The zero cell stays a dot.

4 [25 - 3 0] = [4 \cdot 2 4 \cdot 5 4 \cdot (- 3) 4 \cdot 0] = [820 - 12 0]

Notice: $8$ grew, $- 12$ grew and stayed negative, $20$ grew, $0$ stayed $0$ . Every case from Step 2 appears in one matrix.

Recall Check yourself

Compute $\frac{1}{3} [612 - 9 15]$ . ::: $[24 - 3 5]$ — each cell multiplied by $\frac{1}{3}$ ; a scalar less than 1 shrinks. What is $0 \cdot [7129]$ , and what shape is it? ::: The zero matrix $[0000]$ , still $2 \times 2$ .

The one-picture summary

Everything above compresses into a single flow: one number scales (Step 1–2), the grid is just many such numbers side by side (Step 3), and because cells never interact, scaling the grid is scaling each cell in place while the outer shape holds fixed (Steps 4–6).

This is also exactly the seed of a linear transformation (uniform stretching of space), and it plays cleanly with addition via the distributive laws collected in properties of matrix operations. It is not the row-by-column combining of matrix multiplication — that is a different, harder operation.

Recall Feynman retelling — the whole walkthrough in plain words

Imagine a tic-tac-toe board with a number written in every square. Someone hands you a "stretch factor," say $4$ . Here's the trick I want you to see, not just believe: the number in each square is a little arrow, and each square is completely alone — the top-left square has no idea what the bottom-right square contains. So all you can do is walk to each square and stretch its arrow by $4$ . Positive numbers get longer. Negative numbers get longer but keep pointing the negative way. A zero just stays a dot no matter what. When you finish, the board has the exact same number of rows and columns as before — you never added or removed a square, you only rewrote what was inside. If your stretch factor were $1$ , nothing would change; if it were $0$ , every square would become $0$ but the board would still be there, just full of zeros. That's the entire story of $k A$ : same board, every arrow stretched by $k$ .