Before you can compose functions, you must be fluent in the little pieces the parent note throws around: arrows, letters like f, the mysterious circle ∘, "domain", "range", nested parentheses. We build them one at a time, and each one earns its place before it is used again.
Read f(x) out loud as "f of x". It does not mean "f times x". The brackets are not multiplication — they are a feeding slot: whatever sits inside them is the thing being dropped into the machine.
Figure s01 — a rounded box labelled f (the machine) with a magenta arrow feeding the input x=3 into its slot on the left, and an orange arrow carrying the output f(3)=10 out on the right; the rule "square it, add 1" is printed inside the box.
The namef (or g, or h) labels which machine we are using.
The input is whatever is inside the brackets.
The output is what comes out — the value f(x) stands for.
The parent writes f:A→B and, using a second machine named g (Section 1 — any letter names a machine), g:B→C. Let us earn every piece.
The → arrow is chosen because a function has a direction: things flow in one way. You feed A, you never feed B. This one-way arrow is exactly why order will matter later.
Figure s02 — three white circles labelled A, B, C left to right; a violet arrow marked f runs from A to B, an orange arrow marked g runs from B to C, showing the machines lined up head-to-tail so that B is both the out-pile of f and the in-pile of g.
Why does the topic need this? Because composition demands that the out-pile of the first machine fits into the in-slot of the second. The parent's rule "the outputs of f must be a subset of the inputs g accepts" is just: the arrows must line up head-to-tail, like A→B→C.
So we now have three named piles: the domain (A, inputs), the codomain (B, declared outputs), and the range (the part of B genuinely hit). See Domain and range of function for the full story. Here is the picture that matters for composition:
f(x)=xrefuses negative inputs — you cannot square-root −4 and stay in real numbers. Its domain is x≥0.
g(x)=x−4accepts anything — its domain is all real numbers.
The parent constantly writes things like g(f(x)). This is a nesting, and there is one iron rule.
This is the same rule you already use in arithmetic: in g(f(5)) you compute f(5) before g, exactly as in 2×(3+4) you compute 3+4 before multiplying. Innermost bracket wins.
Figure s03 — the expression g(f(x)) printed large; a magenta box rings the inner f(x) labelled "compute this inner value first", and a violet downward arrow leads to the caption "drop that value into g's slot", with the arithmetic analogy 2×(3+4) underneath.
Why the topic needs this: the entire meaning of g(f(x)) depends on reading it inside-out. Read it outside-in and you would apply g first — which is the wrong machine and the wrong answer.
The circle is just a compact label for "the assembly-line machine". Everything the parent proves — non-commutativity, associativity, the identity — is a fact about this combined machine.
Why does the topic need a do-nothing machine? Because composition needs a neutral element — something that, when bolted on, changes nothing. It plays the role that 0 plays for adding and 1 plays for multiplying: f∘I=f and I∘f=f (where I is taken on whichever set makes the arrows line up). It also secretly powers Inverse functions, where a function and its inverse compose back toI.
The map below reads top-to-bottom (graph TD = "graph, top-down"): an arrow X→Y means "you need X before Y makes sense." Trace it: the function machine (top) is the seed of everything. It grows two shoots — the arrow notation (which sets A and B) and the rule about nested brackets. The arrow notation feeds domain and range; nested brackets and domain/range together feed the circle symbol. The machine also spawns the identity. Finally, the circle, the identity, and domain/range all pour into full composition (the box near the bottom), which then flows onward to inverses and the chain rule.
Each box is a symbol or idea you now own. The bottom box — actual composition — is what the parent note runs with. From there the path continues to Inverse functions, Bijective functions, and Derivatives - chain rule, all of which lean on this same machinery.