From counting to structure
Let be distinct primes and . Sylow's theorems tell us how many subgroups of each prime order can occur. We will use the unique subgroup of order to put every element into a normal form, then see exactly where noncommutativity can enter. The preceding note supplies Sylow's theorems, Cauchy's theorem and Lagrange's theorem. Throughout, exponents of are taken modulo , and exponents of modulo .
The normal subgroup and a normal form
Theorem 2.2.1
(Normal form for a group of order )
Let be primes and . Then has exactly one subgroup of order , and it is normal. If is any subgroup of order , every element of is uniquely with and , and there is a residue with
Proof
(Sylow counting, then conjugation inside )
Let be the number of Sylow -subgroups. It divides and is congruent to modulo . Since , the only possibility is . Thus has order and is normal. Choose a Sylow -subgroup . Prime-order groups are cyclic, and by Lagrange's theorem.
The products , for and , are distinct: if , then belongs to , forcing and in these ranges. They therefore exhaust . This proof supplies both existence and uniqueness; it does not assume that is normal.
Since is normal, conjugation by is an automorphism of . It sends a generator to a generator, so for a nonzero residue modulo . The relation follows by conjugating by repeatedly: , and gives .
Corollary 2.2.2
(When every group of order is cyclic)
Let be primes with . Then every group of order is cyclic.
Proof
(The twisting residue must be one)
The multiplicative order of in divides , hence is either or . If , Lagrange's theorem in the group of order rules out order , so . Then and commute. If , then lies in the trivial intersection, so both and divide . Hence has order and is cyclic.
Proof X-ray
Four facts, each used once
The argument chains four separate facts. Sylow counting uses : the only divisor of that is modulo is , so is normal, while nothing is claimed about . Normality of is what makes conjugation by an automorphism of , producing the single number that records the whole interaction between the two subgroups. The relation turns into , and Lagrange's theorem is then applied not in but in , whose order is where the hypothesis enters. Finally, commuting elements of coprime orders and have product of order ; the trivial intersection is the step that rules out a smaller order.
The hypothesis on cannot be removed.
Counterexample mode
Order $pq$ does not force a cyclic group
Claim: every group of order with is cyclic. Take , and , of order . With right-to-left composition, while , so is not abelian, let alone cyclic. The proof breaks exactly at the step : here , , and , so , a residue of order in . This is possible because divides . The corrected statement is the corollary, which assumes .
What changes when divides ?
The cyclic group of order still exists. We now construct a noncommutative example and verify that its multiplication is a group operation. By Cauchy's theorem applied to the multiplicative group of order , there is an of order .
Definition 2.2.3
(The pair group with twisting residue )
Let be a residue modulo with . On pairs in , define
The first coordinate is calculated modulo and the second modulo .
Proposition 2.2.4
(A noncommutative group of order )
If , the rule above defines a group of order in which and satisfy and . It is noncommutative exactly when .
Proof
(Checking the group axioms for the pair rule)
The rule is well defined: replacing by leaves unchanged, because . Associativity follows because both ways of multiplying three pairs give first coordinate and second coordinate . The identity is and the inverse of is . These checks use that is invertible modulo , which follows from .
Powers of are , so has order ; powers of are , so has order . Also while , so exactly when . Multiplying on the right by gives .
The construction is called a semidirect product: the -part acts on the normal -part by multiplication by . Conversely, the normal form above shows that any group of order is represented by this same multiplication for its own , because . We do not need a classification of which choices of give isomorphic groups here.
Examples
Worked example 2.2.5
(Order fifteen)
For order , take , . Since , the corollary shows that every group of order is cyclic. The Sylow counts agree: divides and is modulo , so ; divides and is modulo , so as well. This conclusion applies to every such group, not just to a selected construction.
Worked example 2.2.6
(Order twenty-one)
For order , take , and . Indeed are the first three positive powers of modulo , so has order . The pair construction has elements and is noncommutative: but . It resolves the ambiguity left by Sylow counting: divides and is modulo , hence is or . In this noncommutative example it must be . If it were , both and would be normal; their commutators would lie in both subgroups and hence be trivial, making their generators commute, a contradiction. The cyclic group of order instead has .
Worked example 2.2.7
(The smallest prime: order )
Let and be an odd prime, so always. The condition means , so or , because is prime. The residue gives the cyclic group. For the relations become
which are the relations of the rotation and a reflection of a regular -gon. Since every group of order is described by the pair rule for its own , and the rule depends only on , a noncommutative group of order multiplies exactly like . For this recovers , where as in the counterexample above.
Common mistakes
Common mistake
(A unique Sylow -subgroup says nothing about )
The count makes normal; it does not make the Sylow -subgroup normal. In the noncommutative group of order , , so none of the seven subgroups of order three is normal. Normality of must be proved or disproved separately, as in the last exercise.
Common mistake
(The twisting residue must satisfy )
The residue cannot be an arbitrary unit modulo . For , and , the rule is not well defined: and name the same element of , yet while . The products and would be and . Only residues with , here , give a group.
Exercises
Quick check
(Prove that every group of order is cyclic. Identify the values of and and state the divisibility test you use.)
Solution · Answer
Here and ; since , the corollary forces a generator of order .
Quick check
(In the order- pair group above, calculate and the inverse of .)
Solution · Answer
The product is . Since modulo , the inverse is .
Quick check
(In the same group, explain why is normal, but is not. Do not infer non-normality merely from the fact that the whole group is noncommutative.)
Solution · Answer
Conjugation by and by preserves , and they generate the group; hence is normal. If were also normal, commutators between and would belong to their trivial intersection. This would force and to commute, contrary to the explicit calculation above.
The argument assumes two distinct primes with ; it does not classify groups of order .