Concept of a function — input, output, mapping: Visual walkthrough

Every step below has three parts: WHAT we just did, WHY we did it, and PICTURE — the figure that carries the idea.

Step 1 — Two boxes and some dots

WHAT. Draw two collections of things. On the left, the things we are allowed to put in — call this box $A$ . On the right, the things that are allowed to come out — call this box $B$ . Inside each box we draw dots; each dot is one element (one "thing").

WHY. Before we talk about any rule, we must know where inputs live and where outputs live. A machine is useless if you don't know what you may feed it and what may fall out. Those two boxes are the entire stage on which a function performs.

PICTURE. Left box $A = {1, 2, 3}$ , right box $B = {4, 5, 6, 7}$ . Nothing connects them yet — this is the empty stage.

Figure — Concept of a function — input, output, mapping

Step 2 — Draw one arrow from every left dot

WHAT. From each dot in $A$ we draw exactly one arrow to some dot in $B$ . Here: $1 \to 4$ , $2 \to 5$ , $3 \to 6$ .

WHY. An arrow is the rule made visible. "Feed in $1$ , get out $4$ " is drawn literally as an arrow from the $1$ -dot to the $4$ -dot. The phrase "the rule assigns to each input an output" now means one concrete thing: from every left dot, one arrow leaves.

PICTURE. Three arrows, one leaving each of $1$ , $2$ , $3$ . Notice dot $7$ receives no arrow — that is completely fine; the codomain is allowed to have spare dots.

Step 3 — The forbidden picture: two arrows from one dot

WHAT. Now deliberately break it. Add a second arrow leaving dot $2$ : keep $2 \to 5$ but also draw $2 \to 7$ .

WHY. We want to see why the rule bans this. Ask the machine " $f (2) = ?$ ". Follow the arrows from $2$ — they split. One goes to $5$ , one goes to $7$ . The machine gives two answers at once. You cannot write down a single value; prediction collapses. That is the whole reason for "exactly one output".

PICTURE. Dot $2$ has a red fork — two arrows leaving it. This is the shape of "NOT a function".

Step 4 — The other broken picture: a left dot with no arrow

WHAT. Different breakage. Draw $1 \to 4$ and $3 \to 6$ , but leave dot $2$ with no arrow leaving it at all.

WHY. The rule said "to each element of $A$ ". Ask " $f (2) = ?$ " and there is nothing — the machine has no answer for an input it was supposed to accept. A vending machine that jams on button $2$ isn't a working machine. So every left dot must have an arrow, no exceptions.

PICTURE. Dot $2$ sits alone with a red " $\times$ " — no arrow. This too is "NOT a function", for the opposite reason to Step 3.

Recall The two-sided rule from Steps 3 and 4

From every left dot: at least one arrow (Step 4) and at most one arrow (Step 3) $=$ exactly one arrow. How many arrows leave each domain dot in a real function? ::: Exactly one — never zero, never two.

Step 5 — Many arrows arriving at one dot is allowed

WHAT. Draw $1 \to 4$ , $2 \to 4$ , $3 \to 6$ . Now dot $4$ on the right receives two arrows.

WHY. The ban is only on arrows leaving a dot, never on arrows arriving. Different inputs are free to share an output. Think of squaring: $3$ and $- 3$ both land on $9$ — two inputs, one output. Perfectly legal, because if you ask " $f (3) = ?$ " you still get one answer.

PICTURE. Two arrows funnel into dot $4$ (green — allowed), while every left dot still fires exactly one arrow.

Step 6 — Slide the boxes onto a graph

WHAT. Instead of two side-by-side boxes, lay the input box $A$ along a horizontal line (the $x$ -axis) and the output box $B$ along a vertical line (the $y$ -axis). Now the pair "input $x$ , output $y$ " is drawn as one point $(x, y)$ on the plane.

WHY. Real functions have infinitely many dots — we can't draw an arrow for each. A graph packs "input $\to$ output" into a single point, and infinitely many points into a smooth curve. The arrow picture and the graph picture are the same idea, just re-plotted.

PICTURE. The mapping $1 \to 4, 2 \to 5, 3 \to 6$ from Step 2 becomes the three points $(1, 4), (2, 5), (3, 6)$ — the arrows have folded into dots on a curve.

Step 7 — The Vertical Line Test drops out

WHAT. Stand at an input $x = c$ on the horizontal axis and shoot a vertical line straight up. Every point where this line meets the curve is an output the machine gives for that single input $c$ .

WHY. This is Steps 3 and 4 re-seen on the graph:

If the vertical line hits the curve twice (say at $(c, y_{1})$ and $(c, y_{2})$ ) — that is the fork of Step 3. Input $c$ gave two outputs. Not a function.
If it hits zero times where $c$ is supposed to be a valid input — that is the lonely dot of Step 4.
If it hits exactly once, everywhere — one arrow out of every input. Function. ✓

PICTURE. Left: a curve where a vertical line crosses once (green ✓). Right: the unit circle $x^{2} + y^{2} = 1$ , where the vertical line at $x = 0$ crosses at $y = + 1$ and $y = - 1$ (red fork ✗) — the picture-proof that the circle is not a function.

Step 8 — The degenerate cases, deliberately

WHAT & WHY, one line each — the cases the reader must never trip over:

Empty spare dots (Step 2, dot $7$ ): codomain may hold dots that no arrow reaches. Allowed — that just makes the range smaller than the codomain.
Every arrow to one dot (a constant machine, e.g. $f (x) = 5$ ): one arrow leaves every input, all landing on $5$ . Still exactly-one-out per input → a function, its whole range is ${5}$ .
Vertical line between points, hitting nothing: if $c$ is not in the domain (e.g. $x = 0$ for $f (x) = 1/ x$ ), a vertical line hitting the curve zero times is fine — we simply never fed in $c$ . The "no arrow" ban only applies to inputs we claimed to accept.
A single dot / single point ( $A = {2}$ , arrow $2 \to 7$ ): the tiniest possible function. One input, one arrow — the definition holds.

The one-picture summary

Everything on one canvas: out is strict (one arrow), in is free (many arrows), and the vertical line is just this rule seen sideways on a graph.

Recall Feynman retelling (say it to a 12-year-old)

Picture two boxes of dots. From every dot on the left you must draw exactly one arrow to the right — never zero (that dot would jam the machine), never two (the machine would give two answers and you couldn't trust it). Arrows are allowed to crash into the same right-dot, though — lots of inputs can share one output, no problem. Some right-dots may get no arrow at all; they're just spares.

Now tilt the picture: put inputs along the floor and outputs up the wall, so each "arrow" becomes a single point floating in the room. Stand on an input and look straight up: whatever the curve does above you is the output. If the curve is above you twice, that input has two answers — the machine cheated, so it's not a function. If the curve is above you exactly once everywhere your input is allowed, it's a genuine function. That "look straight up" move is the whole Vertical Line Test — and it's nothing more than the one-arrow-out rule wearing a graph costume.

Where this leads

Graphs of Functions — the Vertical Line Test in full.
Types of Functions — what happens when in is also strict (one-one).
Inverse Functions — flipping every arrow around, and when that's allowed.
Composition of Functions — feeding one machine's output into the next.
Real-world Applications — arrows that model temperature, money, motion.