Motivation
An abelian group has no nontrivial commutators. A nonabelian group may still become abelian after a normal subgroup is collapsed, and that subgroup may itself simplify in the same way. Solvability asks whether finitely many such layers account for the whole group. It is a structural condition, not a statement that a particular multiplication table is easy to compute.
The commutator subgroup from chapter 1 provides a canonical next layer. Repeating it avoids arbitrary choices of normal subgroups and makes the behavior of maps transparent. We will prove the closure properties before using them, then calculate examples where the layers terminate and a finite simple group where they cannot terminate.
Groups need not be finite in the closure theorems. Finiteness is imposed explicitly for the -group and conjugacy-counting arguments. We use the quotient, action, class-equation, and Sylow results already proved in the earlier group chapters.
Definitions
Definition 6.1.1
(Derived series and solvability)
Write and let be the subgroup generated by all commutators. Define and . This is the derived series. A group is solvable if for some finite . The smallest such is its derived length; the trivial group has length zero, and a nontrivial abelian group has length one.
A subgroup is characteristic if every automorphism preserves it. An automorphism sends to the commutator of the images, so is characteristic. Inductively every derived term is characteristic in : automorphisms restrict to the previous term and preserve its commutators. Characteristic subgroups are normal because inner conjugations are automorphisms. Thus every derived quotient used below is a legitimate quotient group.
A subnormal series with abelian factors is a finite descending chain with each normal in and abelian. It does not require each member to be normal in all of . The derived series has the stronger characteristic property, but the existence criterion will need only this stated condition.
For permutations, define as the determinant of its permutation matrix, using the elementary multiplicative determinant identity. It is a homomorphism to , and a transposition has sign minus one. The alternating group is its kernel. For the sign map is onto, so has order and is normal in .
Theorems
Theorem 6.1.2
(Maps, subgroups, quotients, and extensions)
Every homomorphism sends into the corresponding derived term of its target; a surjective homomorphism maps it onto that term. Subgroups and quotients of solvable groups are solvable. If has derived length at most and has derived length at most , then has derived length at most .
Theorem 6.1.3
(Abelian layers and finite -groups)
A group is solvable exactly when it has a finite subnormal series with abelian factors. Every finite -group is solvable. A nonabelian simple group is not solvable, where simple means that its only normal subgroups are the trivial subgroup and the whole group.
Concept lensStructural
Abelian layers refined to prime-order layers
For finite groups there is a sharper equivalent description: is solvable exactly when it has a subnormal series whose factors are cyclic of prime order. Such factors are abelian, so one direction is immediate. Conversely, if a factor is abelian of non-prime order greater than one, Cauchy's theorem gives a proper nontrivial subgroup, normal because the factor is abelian. By the correspondence theorem its preimage satisfies . Each insertion splits a factor order, so refinement stops. For the factor orders become .
Proof sketch or proof idea
Proof
(Commutators under maps, extensions and -groups)
For a homomorphism , the identity proves the first-derived inclusion, then induction proves all of them. If is onto, every target commutator has a source commutator preimage; finite products and inverses give equality of the generated subgroups. Restricting the same observation to an inclusion gives , so subgroups inherit termination. Surjectivity of a quotient map gives the quotient conclusion.
For the extension assertion, let . Since the quotient's th term is trivial, is trivial, hence . Taking further derived steps gives . This argument separates what vanishes in the quotient from what remains in its kernel; it never assumes the extension splits or is a direct product.
If the derived series terminates, its successive factors are abelian by the characterization abelian. Conversely, suppose a subnormal series has abelian factors. Then . Starting at , induction gives , because taking commutators respects subgroup inclusion. Thus the th term is trivial. This proves the equivalence without strengthening subnormality to normality in the entire group.
For a finite -group, induct on its order. The trivial case is immediate. A nontrivial group has a nontrivial center by the class equation. If the center is the whole group, the group is abelian. Otherwise is a smaller -group, so it is solvable by induction, while is abelian. The extension theorem proves solvability of .
Finally, if is nonabelian and simple, is a nontrivial normal subgroup, hence equals . Every derived term is then , so no term can become trivial. Nonabelianness is essential: a cyclic group of prime order is simple and solvable.
Proof X-ray
Vanishing in a quotient leaves a kernel to resolve
After derived steps, the quotient calculation proves only containment in . The next steps use solvability of that kernel to finish. Knowing that a quotient is abelian therefore does not immediately make the original group abelian, or even prove solvability unless the kernel is controlled. The two contributions to the length bound correspond to these two separate obligations.
Why three-cycles generate the alternating group
Proof
(Pairing transpositions into three-cycles)
Every permutation is a product of transpositions: decompose it into disjoint cycles and write . Its sign is therefore minus one to the number of transpositions. An even permutation has a product with an even number, which can be paired. Identical adjacent factors cancel. Two transpositions sharing one letter multiply to a three-cycle, with the orientation determined by their order. For four distinct letters,
Checking the images of the four letters verifies this identity. Thus each pair is a product of three-cycles, and every even permutation is generated by three-cycles. Conversely each three-cycle is a product of two transpositions and is even. This supplies the generation fact used in the examples, rather than inferring it merely from signs.
Worked examples
Worked example 6.1.4
( is solvable but not abelian)
The sign map gives . The commutator is a nonidentity three-cycle and generates , so . Since is cyclic, its commutator subgroup is trivial. Thus the derived series is and the exact derived length is two. The two abelian factors do not assert that itself is abelian or that it is their direct product.
Worked example 6.1.5
(The three layers of )
Again . The same three-cycle commutator belongs to , and normality gives all its conjugates, hence all three-cycles. The generation result proves .
Let . Direct multiplication shows it is an abelian subgroup of order four: each nonidentity element squares to the identity, and two distinct such elements multiply to the third. Conjugation merely relabels the double transpositions, so . The quotient has order three and is cyclic, giving .
For the reverse inclusion, . Conjugating by and its square yields the other two nonidentity elements of . Since is normal in , it contains all of . Thus , and . The exact derived series is , of length three, with factor orders two, three, and four. Each equality used both containments; displaying a possible normal chain alone would give only an upper bound on length.
Worked example 6.1.6
( is simple, so its derived series cannot shrink)
The even cycle types on five letters are the identity, three-cycles, double transpositions, and five-cycles. There are three-cycles, double transpositions, and five-cycles. These counts plus the identity total sixty, the order of .
A three-cycle has centralizer in of order six: three powers on its support and either permutation of the other two letters. Exactly the three powers are even, so its centralizer has order three and its conjugacy class has size twenty. A double transposition has a unique fixed point; a commuting permutation fixes that point and can exchange its two pairs and independently flip each pair, giving eight choices. Half are even, since multiplication by the commuting odd transposition pairs the two parities for the representative . The centralizer has order four and the class size is fifteen.
A permutation commuting with a five-cycle is determined by the image of one letter and must be a power of that cycle. Its centralizer has order five, and all its elements are even. Thus an class of five-cycles has size twelve, and the twenty-four five-cycles form two classes. The nonidentity class sizes are therefore .
A normal subgroup must be a union of complete conjugacy classes containing the identity, and its order must divide sixty. The possible proper nontrivial union sizes are , none of which divides sixty. Only orders one and sixty remain. Hence is simple. It is nonabelian, for the two three-cycles used in the preceding commutator calculation already lie in it and do not commute. The theorem now gives , so it is not solvable. Class counting supplied the simplicity proof; neither its name nor its order alone would do so.
Counterexample mode
Solvable generators do not suffice
The group is generated by transpositions, each lying in a cyclic subgroup of order two and therefore solvable. Nevertheless contains the nonsolvable subgroup , so it is not solvable. The repair is the extension theorem's actual hypothesis: a normal solvable kernel and a solvable quotient, not merely a collection of solvable generating subgroups.
Reading a proposed solvability argument
First identify the normal subgroup that makes the next quotient meaningful. Then verify the quotient is abelian or already known solvable, and account separately for the kernel. If a chain is given, distinguish the upper bound on derived length supplied by its factors from an exact length calculation, which requires proving the actual derived subgroups.
For finite groups, numerical constraints often help choose a candidate kernel, but they are not themselves a series. A unique Sylow subgroup is normal, which can make the extension theorem applicable. A nonunique Sylow subgroup may still be solvable without serving as a normal kernel. The order-twelve exercise shows how a counting argument can produce the normal subgroup needed for the structural conclusion.
Common mistakes
Common mistake
(Solvable is not abelian or split)
Solvable does not mean abelian, and a sequence of abelian factors does not automatically give a direct-product decomposition. Splitting is an additional condition. Normality also cannot be dropped simply because the intended quotient has a plausible size.
Common mistake
(One commutator does not identify the derived subgroup)
Derived subgroups are generated by commutators; one nontrivial commutator proves nonabelianness but does not by itself identify the whole derived subgroup. The calculation used conjugation and a separate quotient upper bound to complete both inclusions. Likewise, ruling out a few subgroup orders is not a simplicity proof unless all possible normal-subgroup class unions have been covered.
Summary
Derived series measure successive nonabelian layers. Homomorphisms, inclusions and normal extensions control those layers, and the center proves finite -groups solvable. and have explicit terminating series, while the full class calculation proves a nonabelian simple obstruction. These arguments distinguish normal-series existence, exact derived length, and simplicity.
Exercises
Quick check
(A group is called perfect if . Prove that a solvable perfect group is trivial.)
Solution · Answer
If , induction gives for every . Solvability makes one term trivial, so itself is trivial. A nontrivial abelian group is not perfect, because its derived subgroup is already trivial.
Quick check
(For the dihedral group of order with , find its derived subgroup and exact derived length.)
Solution · Answer
The relation gives , so . The subgroup is normal, since conjugation by fixes it and by inverts its powers. In , the images satisfy , so the defining relation makes the two generating images commute. The quotient is abelian, giving and hence equality. This derived subgroup is cyclic, so its own derived subgroup is trivial. The original group is nonabelian for , since would force ; the exact length is therefore two.
Quick check
(Prove that every group of order twelve is solvable, using the Sylow theorems rather than a classification list.)
Solution · Answer
The number of Sylow three-subgroups is one or four. If one, the normal subgroup has order three, and its quotient has order four and is abelian by the earlier theorem. The extension is solvable. If four, the distinct three-subgroups intersect only at the identity and contribute eight nonidentity elements of order three. Only four elements remain, including the identity. Any Sylow two-subgroup has four elements and none of its nonidentity elements has order three, so it is exactly this remaining set. It is consequently unique and normal. Its order-four group is abelian and its quotient has order three, again giving a solvable extension.
Quick check
(Over any field , prove that the group of invertible upper triangular two-by-two matrices is solvable with derived length at most two.)
Solution · Answer
Send to . This is a surjective homomorphism to an abelian group. Its kernel consists of and is isomorphic to the additive group of , hence abelian. The extension theorem gives length at most two. Over the two-element field the whole group is already abelian; the asserted bound does not require equality for every field.