Visual walkthrough — Abstract vector spaces — axioms, examples beyond ℝⁿ

Before we touch a symbol, let us name the pieces we are allowed to use. Think of them as the only tools in the box.

We will lean on exactly four axioms. Here they are, restated as "moves you are licensed to make":

---

Step 1 — Split the scalar zero in two

WHAT. We start from the object we want to understand, $0\cdot v$, and rewrite the number $0$ using a fact about ordinary numbers: $0=0+0$.

$$\underset{\text{number}}{\underbrace{0}}\cdot \underset{\text{tile}}{\underbrace{v}}=\underset{\text{same number, split}}{\underbrace{(0+0)}}\cdot v$$

WHY. We are not allowed to say what $0\cdot v$ "equals" yet — that is the whole mystery. But we are allowed to replace the number $0$ by anything equal to it, and $0+0=0$ is plain school arithmetic inside the field. This is the only crack we can wedge open. Splitting the scalar is what lets A8 get a grip.

PICTURE. The red tile is $v$. Scaling by $0$ (left) is the same as scaling by the split number $0+0$ (right) — a single knob turned to zero, versus two knobs each at zero that we'll separate next.

---

Step 2 — Distribute: one scaled tile becomes two

WHAT. Apply A8 to the right-hand side of Step 1.

$$(0+0)v\stackrel{\text{A8}}{=}\underset{\text{call this }w}{\underbrace{0\cdot v}}+\underset{\text{also }w}{\underbrace{0\cdot v}}$$

WHY. A8 says a scalar-sum in front of a tile splits into two scaled copies snapped together. This is the pivotal move: the mysterious object $0\cdot v$ now appears twice, added to itself. That self-addition is the trap we'll spring.

PICTURE. The single tile "$0\cdot v$" on the left unzips into two identical copies of itself on the right, joined by a snap "+". Give the shared name $w:=0\cdot v$ so we can talk cleanly.

---

Step 3 — Name it $w$ and read the equation we've earned

WHAT. Chaining Steps 1 and 2, and writing $w=0\cdot v$:

$$\underset{=0v}{\underbrace{w}}=\underset{\text{itself, snapped twice}}{\underbrace{w+w}}$$

WHY. This tiny line is the entire heart of the proof. In ordinary numbers, "a number that equals itself plus itself" must be zero ($w=2w\Rightarrow w=0$). We want to say the tile version, but "divide by 2" is not one of our licensed moves — there is no division of tiles. So we need the axioms to finish the job instead.

PICTURE. A balance scale: one $w$ on the left pan exactly balances two $w$'s on the right. Something must give — and the only way to restore balance without dividing is to undo a copy.

---

Step 4 — Snap the "undo" tile onto both sides

WHAT. Every tile has a partner (A4). Snap $(-w)$ onto both sides of $w=w+w$:

$$\underset{\text{left pan}}{\underbrace{w+(-w)}}=\underset{\text{right pan}}{\underbrace{(w+w)+(-w)}}$$

WHY. Since the two sides were equal, snapping the same tile onto each keeps them equal — that is just "do the same thing to both pans." We choose $(-w)$ specifically because A4 promises it collapses a $w$ to the empty tile. This is our substitute for the forbidden "divide."

PICTURE. The green undo-tile $(-w)$ lands on each pan. Watch the left pan first — it is about to vanish to empty.

---

Step 5 — Collapse the left, regroup the right

WHAT. Left side by A4 becomes $0$. Right side: regroup with A2, then collapse the inner pair with A4:

$$\underset{w+(-w)\text{(A4)}}{\underbrace{0}}=w+\underset{\text{regroup, A2}}{\underbrace{(w+(-w))}}=w+\underset{\text{A4 again}}{\underbrace{0}}$$

WHY. On the right we couldn't touch the middle $w$ directly, so we re-bracketed (A2) to pair the second $w$ with its own undo tile $(-w)$. That pair collapses to $0$, leaving just $w$ snapped to the empty tile.

PICTURE. Left pan is now empty ($0$). On the right, the brackets slide over so the neighbouring $w$ and $(-w)$ meet and annihilate to empty, leaving a lone $w$ beside an empty tile.

---

Step 6 — The empty tile does nothing → the result

WHAT. Use A3 ($w+0=w$) on the right:

$$0=w+0\stackrel{\text{A3}}{=}w=0\cdot v$$

WHY. The empty tile snapped onto $w$ leaves $w$ unchanged. Reading the chain end to end: $0=w=0\cdot v$. We have proved, using nothing but the axioms, that scaling any tile by the number zero yields the empty tile.

$$\boxed{0\cdot v=0}\text{for every }v\in V.$$

PICTURE. The empty tile absorbs into $w$; the balance settles reading "$0=0\cdot v$".

---

Step 7 — Bonus: why $(-1)\cdot v=-v$ falls out for free

WHAT. We now spend our new theorem to prove the second claim. Start from $v$ and snap on $(-1)v$:

$$v+(-1)v\stackrel{\text{A5}}{=}1\cdot v+(-1)\cdot v\stackrel{\text{A8}}{=}(1+(-1))v=0\cdot v\stackrel{\text{Step 6}}{=}0$$

Term by term: $1\cdot v=v$ turns the first tile into a scaled copy (A5); A8 merges the two scaled copies into one scalar $1+(-1)=0$ out front; that scalar is the number zero, so Step 6 finishes it to $0$.

WHY. If $v+(-1)v=0$, then $(-1)v$ is a tile that undoes $v$. And undo-tiles are unique (a standard one-line consequence of A2–A4). So $(-1)v$ must be the very tile we call $-v$.

$$\boxed{(-1)\cdot v=-v}$$

PICTURE. Tile $v$ (red) snapped to tile $(-1)v$ (green) annihilates to the empty tile — visibly the same relationship as $v+(-v)=0$.

---

The one-picture summary

The whole argument is one flow: split the scalar → distribute → get $w=w+w$ → undo one copy → land on $0$, then spend that result to crack $(-1)v=-v$.

Recall Feynman: the whole walkthrough in plain words

I want to know what happens when I scale a tile by the number zero, but I'm not allowed to assume anything — only snap, scale, undo, and the empty tile. So I use a trick: the number zero is secretly $0+0$. The distribute rule says scaling by $0+0$ is the same as taking two zero-scaled copies and snapping them together. So my mystery tile equals itself plus itself. Normally I'd "divide by two," but tiles don't divide! Instead I snap the tile's undo-partner onto both sides. The left side collapses to empty, and on the right the neighbouring pair also collapses, leaving just the mystery tile beside an empty tile — which is still the mystery tile. Reading it back: empty equals mystery tile. Done: scaling by zero always gives the empty tile. Then I get $(-1)v=-v$ almost for free, because $v$ plus $(-1)v$ merges (via distribute) into scaling by $1+(-1)=0$, which I just proved is empty — so $(-1)v$ is exactly the thing that undoes $v$, i.e. $-v$. No arrows, no coordinates — works for functions, matrices, everything obeying the rules.

Recall Quick self-check

Which axiom lets us turn $w=w+w$ into $0=w$? ::: A4 (additive inverse), snapped on both sides, plus A2 (regroup) and A3 (zero tile) to finish. Why can't we "divide by 2" to solve $w=w+w$? ::: Division of vectors is not a vector-space operation; only add/scale/undo are guaranteed. In the proof of $(-1)v=-v$, which earlier result is reused? ::: $0\cdot v=0$ from Step 6. Which two objects does Step 6 secretly connect? ::: The scalar $0$ (a number) and the vector $0$ (a tile in $V$).

---

Connections

• Abstract vector spaces — axioms, examples beyond ℝⁿ (parent)
• Fields and scalars
• Subspaces and the subspace test
• Linear independence, basis and dimension
• Linear maps and matrix representations
• Inner product spaces
• Function spaces and Fourier series

Proof Flow