Subgroups & Cyclic Groups
00 · Symbol Glossary
Read " is a subgroup of ." Means and is itself a group under 's operation. The symbol deliberately echoes for numbers — subgroups really do behave like an ordering on the "sub-structures" of .
The subgroup containing only the identity. Every group has this as its smallest possible subgroup — never empty, since is always present.
The set — every power (positive, negative, and zero) of a single element . Read "the subgroup generated by ."
Means every element of is some power of the single element . A group satisfying this for some is called cyclic.
The largest positive integer dividing both and . Central to determining generators of and orders of powers, via the identity developed in this chapter.
Recalled from Chapter 01: for some integer . Governs exactly which subgroups a cyclic group of order has — one for each positive divisor of , and no others.
The subgroup of consisting of even permutations (those expressible as a product of an even number of transpositions). . Previewed here as an example of a subgroup found by isolating a property closed under composition.
01 · What Is a Subgroup?
A subset can sit inside a group without inheriting group structure — for instance, has no identity for once you leave out. A subgroup is a subset that manages to be a self-contained group in its own right, using the ambient operation unchanged.
Let be a group. A subset is a subgroup of , written , if is itself a group under (restricted to ).
Checking all four group axioms (G1–G4) from scratch every time would be wasteful — associativity is automatically inherited from (it held for all triples in , so certainly for the triples that happen to live in ). This leaves only three things to check, exactly mirroring the subspace test from linear algebra (Chapter 01 of Linear Algebra) and the abstract pattern set by Theorem 1.3's partition argument.
A nonempty subset is a subgroup if and only if:
H1. is closed under : .
H2. is closed under inverses: .
() If , it satisfies its own G1 and G4, which are exactly H1 and H2.
() Suppose satisfies H1, H2, and is nonempty. Closure (G1) is H1 directly. Associativity (G2) holds for all elements of , hence for the subset . Identity (G3): since , pick any ; by H2, ; by H1, . Inverses (G4) is H2 directly. All four axioms hold, so is a group under , i.e. .
Notice the proof did not need to separately assume ; nonemptiness plus H1 and H2 forces automatically. This mirrors the subspace test's insistence on checking first, except here it's a consequence rather than a hypothesis — a small but genuine difference between the group and vector-space versions of this pattern.
: closed under the operation (), closed under inverses (), nonempty. Always a subgroup, called the trivial subgroup.
itself: trivially satisfies both conditions. Always a subgroup of itself.
Any subgroup strictly between these two extremes is called a proper nontrivial subgroup — the interesting case.
Let . Closure fails immediately: . Not even closed, so certainly not a subgroup — but notice this is a weaker failure than being closed-but-missing-inverses. Always check H1 before H2; if H1 fails, there's no need to check inverses at all.
02 · More Subgroup Examples
inside : closed (sum of two multiples of is a multiple of ), closed under inverses (negative of a multiple of is a multiple of ), nonempty. for every positive integer — infinitely many subgroups of a single infinite group, one per choice of (plus itself, the case ).
Let (the rotations alone, no reflections). Closed: a rotation composed with a rotation is a rotation. Inverses: the inverse of a rotation is a rotation (rotating the opposite way). , and exactly half of .
A transposition (2-cycle) like swaps exactly two elements. Every permutation can be written as a product of transpositions, and — although a given permutation has many such expressions — the parity (even or odd number of transpositions used) is always the same. Call a permutation even if it needs an even number of transpositions.
: composing two even permutations uses an even-plus-even (even) total count of transpositions, so is closed. The inverse of an even permutation is even (reversing a product of transpositions uses the same count). , with for — this specific subgroup will resurface heavily once Chapter 09's Sylow theory and Chapter 21's solvability-by-radicals both need it.
03 · A Shortcut for Finite Subgroups
For a finite group (or a finite candidate subset), checking inverses separately turns out to be unnecessary.
If is finite and is nonempty and closed under (H1 only), then automatically.
Let . Since is closed under , all powers lie in . Because is finite, this list cannot contain infinitely many distinct elements, so eventually two powers coincide: for some . Cancelling (valid by Theorem 2.1, since is a group) gives , so some positive power of equals — meaning has finite order, say , and by closure.
Now if , and trivially. If , then (since ), and by closure (it's a power of , and closure gives every power). Either way, . Since was arbitrary, H2 holds automatically, and Theorem 3.1 finishes the job.
Consider (the nonnegative integers) inside . Closed under addition: sum of two nonnegative integers is nonnegative. But is infinite, and indeed H2 fails: but . Theorem 3.2 does not apply here — finiteness is doing real work in the proof (guaranteeing powers eventually repeat), and dropping it breaks the conclusion.
04 · Cyclic Groups
For , define .
, and any subgroup with satisfies .
Subgroup: is nonempty ( is in it). Closure: . Inverses: . By Theorem 3.1, .
Smallest: suppose and . By closure, , and inductively for every . By H2, , and by the same induction for . Also . So every power of lies in , i.e. .
is cyclic if for some , called a generator of .
In additive notation, "" means ( times) . So — every element of is some multiple of . is cyclic, and is always a valid generator (though frequently not the only one).
. Every integer is a multiple of . is cyclic — the prototypical infinite cyclic group, in contrast to the finite cyclic groups .
Is cyclic? Its elements have orders (identity), (three transpositions), or (two 3-cycles) — checkable directly from Section 04 of Chapter 02. No element has order , so no single element's powers can produce all six elements. is not cyclic. Being generated by some collection of elements (which every finite group is, trivially, by all its elements) is a much weaker property than being generated by one.
(with both sides possibly infinite).
Case : suppose for contradiction for some . Then with , contradicting that has infinite order (no positive power equals ). So all powers () are distinct, and is infinite: .
Case finite: first, are distinct elements — if for , then with , contradicting that is the smallest positive power giving . Second, every power of equals one of these : for any integer , write with (division algorithm), so , one of the listed elements. Hence has exactly elements: .
05 · Subgroups of Cyclic Groups
Cyclic groups have an unusually clean subgroup structure — clean enough to fully classify without case-by-case checking.
If and , then is cyclic.
If , then , cyclic. Otherwise contains some ; replacing with if necessary (since is closed under inverses), assume contains for some positive . Let be the smallest positive integer with — this exists because the set of such positive integers is nonempty (it contains ) and every nonempty set of positive integers has a least element (well-ordering).
Claim: . Certainly , since is closed under the operation and inverses, so every power of lies in .
For the reverse inclusion, take any . By the division algorithm, with . Then . Both and (a power of ), so their product by closure. But , and was chosen as the smallest positive integer with — so must be (otherwise would be a smaller positive witness, contradicting minimality of ). Hence , and .
So and , giving — cyclic.
Setting recovers a classical fact: every subgroup of has the form for some (the subgroups from Section 02). The proof of Theorem 3.5 is literally the proof of that classical fact, written so it works for any cyclic group, not just — the payoff of working abstractly in the first place.
For each positive divisor of , has exactly one subgroup of order , namely . These are all the subgroups of .
Divisors of : — six divisors, hence exactly six subgroups.
| Divisor (order of subgroup) | Subgroup | |
|---|---|---|
| 1 | 12 | |
| 2 | 6 | |
| 3 | 4 | |
| 4 | 3 | |
| 6 | 2 | |
| 12 | 1 |
Compare row 3 to Section 04's step-by-step computation of — Theorem 3.6 predicted this exact subgroup from divisibility alone, no repeated computation needed.
has several elements of order (not just ): both and have order (check: , and no smaller positive multiple works). But — the same subgroup, generated two different ways. Theorem 3.6 counts subgroups, not generators; a single subgroup of order can have multiple valid generators, exactly as and both generate all of .
06 · Exercises
Apply Theorem 3.1: check nonemptiness, closure under matrix multiplication, and closure under matrix inverses.
Let (the special linear group ).
Nonempty: since .
Closure: if and , then , so .
Inverses: if , then , so .
Both conditions of Theorem 3.1 hold: .
Show that is a subgroup of .
Compute in first (the smallest with ), then apply Theorem 3.4.
, , , , . Smallest positive is , so .
By Theorem 3.4, , and .
Find in , and list its elements.
List the orders of all six elements of (identity, three transpositions, two 3-cycles) and compare to .
Orders in : has order . Each transposition (e.g. ) has order (). Each 3-cycle (e.g. ) has order .
No element has order . By Theorem 3.4, for every , so the largest cyclic subgroup any single element generates has order — never all of .
is not cyclic.
Using Theorem 3.4, explain why cannot be cyclic without checking every possible generator by brute force.
Apply Theorem 3.6: the divisors of 20 determine the subgroup orders directly.
Divisors of : — six divisors.
By Theorem 3.6, has exactly six subgroups, of orders respectively (namely , , , , , ).
How many subgroups does have? List their orders using Theorem 3.6, without computing each subgroup's elements directly.
Mimic the proof of Theorem 3.2 exactly: what does closure under guarantee about the powers of a single element of , and why must they eventually repeat?
Let . Closure gives . Since and is finite, is finite, so this list of powers cannot all be distinct — some with . Cancelling (Theorem 2.1, valid since is a group) gives with , so has some finite order , and .
If : , and . If : by closure (a power of ). Either way — exactly the argument of Theorem 3.2, since nothing in it depended on actually being all of a candidate subgroup versus a general nonempty closed subset.
Restate the proof of Theorem 3.2 in your own words: why does closure alone force inverses to exist, in a finite group?
07 · Chapter Summary
| Concept | Statement |
|---|---|
| Subgroup | that is itself a group under 's operation |
| Subgroup test | Nonempty + closed under + closed under inverses (Thm 3.1) |
| Finite subgroup test | Nonempty + closed under suffices, if finite (Thm 3.2) |
| ; the smallest subgroup containing (Thm 3.3) | |
| Cyclic group | for some generator |
| Order equality | $ |
| Subgroups of cyclic groups | Always cyclic (Thm 3.5) |
| Subgroups of | One per positive divisor of , namely (Thm 3.6) |
Next: Chapter 04 — Cosets & Lagrange's Theorem uses subgroups to partition an entire group into equal-sized pieces, producing the single most-used counting tool in finite group theory: the order of any subgroup must divide the order of the group.