A group can be complicated for two different reasons. It may have many elements, or it may resist quotient decompositions that preserve nonidentity information. The second kind of complexity is controlled by normal subgroups: whenever $N \trianglelefteq G$, the quotient $G/N$ records a piece of $G$ in which all elements of $N$ have been collapsed to the identity. If a nonidentity group has no [normal subgroup](/page/Normal%20Subgroup) available for this process, then quotient methods reach a hard endpoint. That endpoint is a simple group. The word simple is slightly misleading at first encounter. A simple group need not be easy to understand; $A_5$ already contains enough structure to encode the rotational symmetries of the icosahedron. The point is not computational simplicity. The point is indivisibility with respect to normal subgroups. A simple group is a group with no nonidentity quotient decomposition coming from an internal normal subgroup. This page develops the general definition, but most of its tests and examples are finite-group tests. Infinite simple groups exist and are important, especially in geometric group theory and Lie theory, but the first structural story is clearest in the finite setting, where orders, Sylow subgroups, and conjugacy-class counts turn normality into arithmetic. [example: Why Normality Is the Right Test] Let $G=S_3$, and let $H=\{e,(12)\}$. The subgroup $H$ has order $2$, since it contains exactly the identity and the transposition $(12)$. To test normality, conjugate its nonidentity element by $(123)$. Since $(123)^{-1}=(132)$, we compute \begin{align*} (123)(12)(123)^{-1}=(123)(12)(132)=(23). \end{align*} The result $(23)$ is not in $H$, so $H$ is not stable under conjugation by elements of $S_3$. Thus $H$ is not normal in $S_3$. Equivalently, the left and right cosets by $(123)$ do not match: \begin{align*} (123)H=\{(123),(123)(12)\}=\{(123),(13)\}. \end{align*} \begin{align*} H(123)=\{(123),(12)(123)\}=\{(123),(23)\}. \end{align*} Because these cosets are different, $H$ cannot be used as the subgroup collapsed in a [quotient group](/theorems/790) $S_3/H$. By contrast, $A_3=\{e,(123),(132)\}$ is normal in $S_3$. Indeed, for any $\sigma\in S_3$, \begin{align*} \sigma(123)\sigma^{-1}=(\sigma(1)\ \sigma(2)\ \sigma(3)), \end{align*} which is either $(123)$ or $(132)$, and the same statement holds for the inverse cycle $(132)$. Hence conjugation by every element of $S_3$ sends elements of $A_3$ back into $A_3$, so $A_3\trianglelefteq S_3$. The quotient has exactly two cosets, \begin{align*} A_3=\{e,(123),(132)\} \end{align*} and \begin{align*} (12)A_3=\{(12),(23),(13)\}, \end{align*} so $|S_3/A_3|=2$. This is why arbitrary subgroups do not measure quotient decomposability: quotient groups come from normal subgroups, because normality is precisely the condition that makes coset multiplication well-defined. [/example] The example also shows that the definition of simple group belongs under the broader definition-family of [group](/page/Group): it is not a new kind of algebraic operation, but a special condition on the normal subgroup structure of an existing group. ## Definition ### The Indivisible Case The obstruction to decomposing a group by quotients is the absence of proper nonidentity normal subgroups. The identity subgroup is always normal, and the whole group is always normal, so these two unavoidable cases must be excluded from the test. [definition: Simple Group] A group $G$ is simple if $G$ is nonidentity and the only normal subgroups of $G$ are $\{e\}$ and $G$. [/definition] This definition is deliberately about normal subgroups rather than all subgroups. A group may have many subgroups and still be simple. What it cannot have is a subgroup that is stable under every conjugation action of $G$ and therefore capable of serving as the kernel of a quotient map. The condition $G \ne \{e\}$ is included so that the identity group does not become an awkward degenerate simple group. In decomposition theory, simple groups are intended to be the indivisible non-zero pieces, much as prime numbers are the indivisible positive integers greater than $1$. ### Kernels and Quotients The most effective way to use simplicity is not usually to list normal subgroups directly. Kernels of homomorphisms are automatically normal, so every attempted homomorphic image of a simple group faces a two-way test: either nothing except the identity is killed, or the whole source is killed. [quotetheorem:10103] The theorem says that a homomorphism out of a simple group has only two possibilities: it is injective, or it collapses all of $G$ to the identity element of the target. Thus a simple group cannot map non-faithfully onto a proper quotient while retaining nonidentity information. The same idea can be phrased through quotient groups, but now the normal subgroup must be named explicitly: quotient groups of $G$ are precisely the groups $G/N$ with $N \trianglelefteq G$. Simplicity says that this construction has only the two unavoidable outcomes. [quotetheorem:10104] This makes simple groups the terminal objects for quotient-based decomposition. A non-simple group exposes a proper normal subgroup $N$, and the quotient $G/N$ is a nonidentity quotient image in which the elements of $N$ have been identified with the identity. ## Normal Subgroups as Quotient Data ### Minimal Normal Structure To see why simple groups matter, it helps to treat normal subgroups as algebraic information rather than as a special kind of subset. A normal subgroup records a way of identifying elements of $G$ while preserving the group law. If no such proper identification exists, the group is rigid from the quotient point of view. The definition of simplicity is often paired with the idea of a smallest non-zero normal subgroup. Larger groups may not be simple, but they can contain normal pieces that are already indivisible inside the ambient group. This motivates isolating the first possible layer of normal structure. [definition: Minimal Normal Subgroup] Let $G$ be a group. A subgroup $N \trianglelefteq G$ is a minimal normal subgroup of $G$ if $N \ne \{e\}$ and there is no normal subgroup $M \trianglelefteq G$ such that $\{e\} \subsetneq M \subsetneq N$. [/definition] A simple group is exactly a group that is minimal as a nonidentity normal subgroup of itself. This reformulation is useful when studying larger groups: the simple objects often appear as minimal normal pieces inside a group that is not itself simple. ### Composition Series Minimal normal subgroups show the first layer of normal structure, but a finite group may require several layers before it has been decomposed as far as possible. The next definition records a maximal normal-subgroup chain in which every successive quotient has reached the simple endpoint. [definition: Composition Series] Let $G$ be a finite group. A composition series for $G$ is a chain of subgroups \begin{align*} \{e\} = G_0 \trianglelefteq G_1 \trianglelefteq \cdots \trianglelefteq G_m = G \end{align*} such that $G_i \trianglelefteq G_{i+1}$ and $G_{i+1}/G_i$ is simple for each $0 \le i < m$. [/definition] The quotients $G_{i+1}/G_i$ are the pieces extracted from $G$ by repeated normal decomposition. To compare two decompositions, those quotient pieces need a name independent of the particular chain notation. [definition: Composition Factor] Let $G$ be a finite group with a composition series \begin{align*} \{e\} = G_0 \trianglelefteq G_1 \trianglelefteq \cdots \trianglelefteq G_m = G. \end{align*} The composition factors of this series are the simple quotient groups $G_{i+1}/G_i$ for $0 \le i < m$. [/definition] A finite group can have many different composition series. This creates a genuine ambiguity: if two different chains could produce unrelated quotient factors, then the phrase building block would depend on the chosen decomposition rather than on the group itself. The obstruction is that a composition series records choices of normal subgroups, while the intended factors should be invariants of $G$. [quotetheorem:2625] This theorem is the group-theoretic analogue of uniqueness of prime factorisation. It justifies the slogan that finite simple groups are the building blocks of finite groups, while also warning that building blocks do not determine the whole group by direct product alone. ## First Examples and Non-Examples The smallest simple groups are abelian. In an [abelian group](/page/Abelian%20Group) every subgroup is normal, so simplicity becomes a much stricter condition: there may be no proper nonidentity subgroup of any kind. [quotetheorem:10105] This theorem explains the first analogy between simple groups and prime numbers. The [cyclic group](/page/Cyclic%20Group) $C_p$ has no subgroup of order strictly between $1$ and $p$, while $C_n$ for composite $n$ has a proper nonidentity subgroup. [example: Cyclic Groups] Let $p$ be prime, and write $C_p=\langle x\rangle$ with $x^p=e$. If $K\le C_p$ and $K\ne \{e\}$, choose $x^r\in K$ with $x^r\ne e$. Then $p\nmid r$, so $\gcd(r,p)=1$. Hence there are integers $u,v$ with $ur+vp=1$, and therefore \begin{align*} x=x^{ur+vp}=(x^r)^u(x^p)^v=(x^r)^u e^v=(x^r)^u. \end{align*} Since $x^r\in K$, this gives $x\in K$, so $K=\langle x\rangle=C_p$. Thus the only subgroups of $C_p$ are $\{e\}$ and $C_p$. Because $C_p$ is abelian, every subgroup is normal, so $C_p$ is simple. If $n$ is composite, write $n=ab$ with $1<a<n$ and $b>1$. In $C_n=\langle x\rangle$, the element $x^a$ satisfies \begin{align*} (x^a)^b=x^{ab}=x^n=e. \end{align*} Also, if $(x^a)^k=e$, then $x^{ak}=e$, so $n\mid ak$. Since $n=ab$, this means $ab\mid ak$, hence $b\mid k$. Therefore $x^a$ has order $b$, and \begin{align*} \langle x^a\rangle=\{e,x^a,x^{2a},\ldots,x^{(b-1)a}\}. \end{align*} Because $b>1$, this subgroup is nonidentity. Because $b<n$, its order is smaller than $|C_n|$, so it is proper. Finally, $C_n$ is abelian, so for every $g\in C_n$ and every $h\in \langle x^a\rangle$, \begin{align*} ghg^{-1}=hgg^{-1}=h. \end{align*} Thus $\langle x^a\rangle\trianglelefteq C_n$ is a proper nonidentity normal subgroup, so $C_n$ is not simple. [/example] The first non-abelian examples require more care. A small group may fail to be simple for a reason that is not visible from its subgroup count alone. The normality condition is the decisive one. [example: The Symmetric Group $S_3$] The group $S_3$ is not simple. Its alternating subgroup is \begin{align*} A_3=\{e,(123),(132)\}. \end{align*} Thus $|A_3|=3$, so $A_3\ne \{e\}$. Also $(12)\in S_3$ but $(12)\notin A_3$, so $A_3\ne S_3$. To see that $A_3$ is normal, use the [sign homomorphism](/theorems/778) \begin{align*} \operatorname{sgn}: S_3 \to C_2. \end{align*} The even permutations in $S_3$ are exactly $e,(123),(132)$, so \begin{align*} \ker(\operatorname{sgn})=\{\sigma\in S_3:\operatorname{sgn}(\sigma)=1\}=A_3. \end{align*} If $\tau\in A_3$ and $\sigma\in S_3$, then $\operatorname{sgn}(\tau)=1$ and \begin{align*} \operatorname{sgn}(\sigma\tau\sigma^{-1})=\operatorname{sgn}(\sigma)\operatorname{sgn}(\tau)\operatorname{sgn}(\sigma^{-1})=\operatorname{sgn}(\sigma)\cdot 1\cdot \operatorname{sgn}(\sigma)^{-1}=1. \end{align*} Hence $\sigma\tau\sigma^{-1}\in A_3$ for every $\sigma\in S_3$ and every $\tau\in A_3$, so $A_3\trianglelefteq S_3$. Therefore $S_3$ has a proper nonidentity normal subgroup, namely $A_3$, and so $S_3$ is not simple. [/example] The next failure is subtler. A group may have no normal subgroup of prime index forced by an obvious sign map and still fail to be simple because conjugacy classes combine into a normal subgroup. [example: The Alternating Group $A_4$] The group $A_4$ is not simple. Let \begin{align*} V_4=\{e,(12)(34),(13)(24),(14)(23)\}. \end{align*} Write $a=(12)(34)$, $b=(13)(24)$, and $c=(14)(23)$. Each of $a,b,c$ has order $2$, because it is a product of disjoint transpositions: \begin{align*} a^2=b^2=c^2=e. \end{align*} The remaining products stay inside $V_4$: \begin{align*} ab=(12)(34)(13)(24)=(14)(23)=c. \end{align*} \begin{align*} ac=(12)(34)(14)(23)=(13)(24)=b. \end{align*} \begin{align*} bc=(13)(24)(14)(23)=(12)(34)=a. \end{align*} Since $a^{-1}=a$, $b^{-1}=b$, and $c^{-1}=c$, this proves that $V_4$ is a subgroup of $A_4$. Now let $\sigma\in A_4$. For distinct $i,j,k,\ell\in\{1,2,3,4\}$, conjugation sends the double transposition $(ij)(k\ell)$ to \begin{align*} \sigma(ij)(k\ell)\sigma^{-1}=(\sigma(i)\ \sigma(j))(\sigma(k)\ \sigma(\ell)). \end{align*} Indeed, the right-hand side sends $\sigma(i)$ to $\sigma(j)$, $\sigma(j)$ to $\sigma(i)$, $\sigma(k)$ to $\sigma(\ell)$, and $\sigma(\ell)$ to $\sigma(k)$, matching the action of $\sigma(ij)(k\ell)\sigma^{-1}$ on each element of $\{1,2,3,4\}$. Thus every conjugate of $a$, $b$, or $c$ is again one of $a,b,c$, and $\sigma e\sigma^{-1}=e$. Hence $\sigma V_4\sigma^{-1}=V_4$ for every $\sigma\in A_4$, so $V_4\trianglelefteq A_4$. Finally, $V_4\ne\{e\}$ because $(12)(34)\in V_4$, and $V_4\ne A_4$ because $(123)\in A_4$ but $(123)\notin V_4$. Therefore $A_4$ has a proper nonidentity normal subgroup, so $A_4$ is not simple. [/example] These examples show the common ways simplicity fails: abelian groups have too many normal subgroups unless they have prime order; symmetric groups usually have an alternating subgroup; and even alternating groups may contain hidden normal subgroups at small degree. ## Simplicity in Finite Groups ### Sylow Obstructions Finite groups give a sharp arithmetic test for possible simplicity. The order of a subgroup must divide the order of the group, but the existence of normal subgroups depends on conjugation. The tension between divisibility and conjugation is where many finite simplicity arguments begin. The first tool is [Cauchy's theorem](/theorems/797). It does not identify normal subgroups by itself, but it guarantees elements of prime order, and hence cyclic subgroups, inside finite groups. [quotetheorem:797] Cauchy's theorem finds prime-order elements, but a cyclic subgroup generated by such an element need not be normal. Sylow theory asks for maximal $p$-power subgroups and gives a uniqueness test; if uniqueness occurs, normality follows. For simple groups, that creates the following obstruction. [quotetheorem:10106] The theorem is often used contrapositive: if Sylow theory forces a unique Sylow subgroup, then the group is not simple unless it is itself a cyclic group of prime order. [example: Groups of Order $pq$] Let $p<q$ be primes, and suppose $|G|=pq$. By *[Sylow's Theorems](/theorems/847)*, a Sylow $q$-subgroup $P\le G$ exists, has order $q$, and the number $n_q$ of Sylow $q$-subgroups satisfies \begin{align*} n_q \mid p \quad\text{and}\quad n_q \equiv 1 \pmod q. \end{align*} Since $p$ is prime, the positive divisors of $p$ are $1$ and $p$, so \begin{align*} n_q\in\{1,p\}. \end{align*} If $n_q=p$, then $p\equiv 1\pmod q$, so $q\mid p-1$. But $p<q$ gives $0<p-1<q$, and no positive integer strictly between $0$ and $q$ is divisible by $q$. Hence $n_q\ne p$, so \begin{align*} n_q=1. \end{align*} Thus there is a unique Sylow $q$-subgroup $P$. For any $g\in G$, the conjugate $gPg^{-1}$ has order $|P|=q$, so it is also a Sylow $q$-subgroup. Uniqueness gives $gPg^{-1}=P$ for every $g\in G$, hence $P\trianglelefteq G$. Also $P\ne\{e\}$ because $|P|=q>1$, and $P\ne G$ because $|P|=q<pq=|G|$. Therefore $G$ has a proper nonidentity normal subgroup, so $G$ is not simple. This is the standard finite-group pattern: count Sylow subgroups, force uniqueness, and then use uniqueness to produce a normal subgroup. [/example] ### Centres Sylow theory detects normal subgroups through counting. A different source of normal subgroups comes from commutation: central elements commute with everything, so subgroups of the centre are stable under conjugation. For a non-abelian simple group, this leaves only one possible centre. [quotetheorem:10107] The theorem separates the abelian and non-abelian cases. Abelian simple groups have $Z(G)=G$ and prime order; non-abelian simple groups have identity centre. This is one of the first signs that non-abelian simple groups are rigid objects. ## Alternating Groups and the First Infinite Family The most important elementary source of non-abelian simple groups is the alternating group. The small cases $A_3$ and $A_4$ behave differently, so the theorem starts only at degree $5$. Before stating the theorem, recall that $A_n$ is generated by $3$-cycles for $n \ge 3$. This matters because normal subgroups are stable under conjugation, and conjugacy classes of $3$-cycles provide a way to spread one non-identity element through the group. [quotetheorem:849] This theorem gives the first infinite family of non-abelian finite simple groups. The case $A_5$ is the smallest non-abelian simple group, and it is the prototype for many computational tests of simplicity. [example: The Group $A_5$] The group $A_5$ has order \begin{align*} |A_5|=\frac{|S_5|}{2}=\frac{5!}{2}=\frac{120}{2}=60. \end{align*} Its nonidentity elements are exactly the even permutations in $S_5$ other than $e$. A $3$-cycle is even because $(abc)=(ac)(ab)$, so the number of $3$-cycles is \begin{align*} \binom{5}{3}\cdot 2=10\cdot 2=20. \end{align*} A product of two disjoint transpositions is also even. To count these, choose the four moved letters and then pair them in one of three ways, giving \begin{align*} \binom{5}{4}\cdot 3=5\cdot 3=15. \end{align*} A $5$-cycle is even because it is a product of $4$ transpositions, and the number of $5$-cycles is \begin{align*} (5-1)!=4!=24. \end{align*} Together with the identity, these account for \begin{align*} 1+20+15+24=60 \end{align*} elements, so these are all the elements of $A_5$. Inside $A_5$, the conjugacy classes have sizes \begin{align*} 1,\ 20,\ 15,\ 12,\ 12, \end{align*} where the two classes of size $12$ are the two $A_5$-conjugacy classes of $5$-cycles. By *Normal Subgroups as Unions of Conjugacy Classes*, any normal subgroup of $A_5$ must be a union of these classes and must contain the identity. Hence the order of a normal subgroup must be $1$ plus a sum of some of \begin{align*} 20,\ 15,\ 12,\ 12. \end{align*} The possible sums strictly between $1$ and $60$ are \begin{align*} 13,\ 16,\ 21,\ 25,\ 28,\ 33,\ 36,\ 40,\ 45,\ 48. \end{align*} None of these divides $60$. By [Lagrange's theorem](/theorems/782), no subset of any of these sizes can be a subgroup of $A_5$. Therefore the only normal subgroups of $A_5$ are $\{e\}$ and $A_5$, so $A_5$ is simple. Geometrically, $A_5$ is isomorphic to the rotational symmetry group of the icosahedron and the dodecahedron. Thus the smallest non-abelian simple group appears naturally as a symmetry group, not only as a permutation group. [/example] The restriction $n \ge 5$ is essential. The earlier examples showed $A_4$ is not simple, while $A_3 \cong C_3$ is simple but abelian. The theorem therefore marks the point where alternating groups become a stable non-abelian family. A useful warning is that $S_n$ itself is almost never simple. The sign homomorphism always produces $A_n$ as a normal subgroup for $n \ge 2$, and for $n \ge 3$ this subgroup is proper and nonidentity. [example: Why $S_n$ Is Not Simple] For $n \ge 3$, view $C_2$ multiplicatively as $\{1,-1\}$, and consider the sign homomorphism \begin{align*} \operatorname{sgn}: S_n \to C_2. \end{align*} By definition, $A_n$ is the set of even permutations, so \begin{align*} \ker(\operatorname{sgn})=\{\sigma\in S_n:\operatorname{sgn}(\sigma)=1\}=A_n. \end{align*} The subgroup $A_n$ is normal in $S_n$: if $\tau\in A_n$ and $\sigma\in S_n$, then $\operatorname{sgn}(\tau)=1$, and since $\operatorname{sgn}$ is a homomorphism, \begin{align*} \operatorname{sgn}(\sigma\tau\sigma^{-1})=\operatorname{sgn}(\sigma)\operatorname{sgn}(\tau)\operatorname{sgn}(\sigma^{-1})=\operatorname{sgn}(\sigma)\cdot 1\cdot \operatorname{sgn}(\sigma)^{-1}=1. \end{align*} Thus $\sigma\tau\sigma^{-1}\in A_n$ for every $\sigma\in S_n$ and every $\tau\in A_n$, so $A_n\trianglelefteq S_n$. It remains to check that this normal subgroup is neither $\{e\}$ nor all of $S_n$. Since $n\ge 3$, the cycle $(123)$ lies in $S_n$, and \begin{align*} (123)=(13)(12). \end{align*} It is therefore a product of two transpositions, so $(123)\in A_n$ and $A_n\ne\{e\}$. Also $(12)\in S_n$, and $(12)$ is a single transposition, so it is odd and hence $(12)\notin A_n$. Therefore $A_n\ne S_n$. Thus $S_n$ has a proper nonidentity normal subgroup, namely $A_n$, so $S_n$ is not simple. This example detects non-simplicity by producing a homomorphism whose kernel is a proper nonidentity subgroup. [/example] ## Actions, Kernels, and Classification Pressure ### Faithful Actions Group actions turn simplicity into a practical test. Whenever $G$ acts on a set $X$, the action determines a homomorphism from $G$ to a [symmetric group](/page/Symmetric%20Group). The kernel consists of the elements acting as the identity on every point of $X$. If $G$ is simple, that kernel is forced to be either all of $G$ or just $\{e\}$. To use this observation, we need language for actions whose kernel has disappeared. Such actions identify the acting group with an actual permutation group on the set. [definition: Faithful Group Action] Let $G$ be a group acting on a set $X$ by an action map $\alpha: G \times X \to X$, written $\alpha(g,x)=g \cdot x$. The action is faithful if the associated homomorphism \begin{align*} \varphi: G &\to S_X \end{align*} has kernel $\{e\}$, where $S_X$ denotes the full symmetric group on $X$. [/definition] For a simple group, every action with nonidentity image is close to faithful: if the action does not collapse all elements to the identity permutation, then the kernel cannot be all of $G$, so it must be $\{e\}$. The next theorem packages this as a reusable test. [quotetheorem:10108] This dichotomy is one reason simple groups are constrained. A non-collapsed action of a simple group embeds it into a symmetric group on the underlying set. If the set is small, this gives strong divisibility and order restrictions. [example: Acting on Cosets] Let $G$ be a simple group and let $H \subsetneq G$ be a proper subgroup of finite index $m$. The left action of $G$ on the coset space $G/H$ is given by \begin{align*} g\cdot aH=gaH. \end{align*} This is well-defined: if $aH=bH$, then $b^{-1}a\in H$, and \begin{align*} (gb)^{-1}(ga)=b^{-1}g^{-1}ga=b^{-1}a\in H, \end{align*} so $gaH=gbH$. Thus the action gives a homomorphism \begin{align*} \varphi:G\to S_{G/H}. \end{align*} After choosing a numbering of the $m$ cosets, we may identify $S_{G/H}$ with $S_m$. Let \begin{align*} K=\ker(\varphi)=\{g\in G:g\cdot aH=aH\text{ for every }aH\in G/H\}. \end{align*} The kernel of a [group homomorphism](/page/Group%20Homomorphism) is normal, so $K\trianglelefteq G$. Since $G$ is simple, either $K=\{e\}$ or $K=G$. If $K=G$, then every $g\in G$ fixes the coset $H$, so \begin{align*} gH=H. \end{align*} This implies $g\in H$ for every $g\in G$, hence $G\subseteq H$, contradicting $H\subsetneq G$. Therefore $K\ne G$, and simplicity forces \begin{align*} K=\{e\}. \end{align*} So $\varphi$ is injective. Hence $G$ is isomorphic to the subgroup $\varphi(G)\le S_m$. If $G$ is finite, then \begin{align*} |G|=|\varphi(G)|. \end{align*} By *[Lagrange's theorem](/theorems/841)*, $|\varphi(G)|$ divides $|S_m|=m!$, so \begin{align*} |G|\mid m!. \end{align*} Thus a proper subgroup of finite index $m$ forces a finite simple group to embed in $S_m$, giving the strong divisibility bound $|G|\mid m!$. [/example] ### Conjugacy Classes Coset actions turn subgroups into permutation representations. Conjugation gives another action, this time of $G$ on itself. Because normality is exactly stability under all conjugations, the orbit structure of this action provides a concrete way to search for normal subgroups. [definition: Conjugacy Class] Let $G$ be a group and let $g \in G$. The [conjugacy class](/page/Conjugacy%20Class) of $g$ in $G$ is \begin{align*} \operatorname{Cl}(g) = \{hgh^{-1} : h \in G\}. \end{align*} [/definition] Conjugacy classes are not subgroups in general, and an arbitrary union of conjugacy classes need not be closed under the group operation. The useful point is more precise: once a subgroup has been identified, normality is exactly the condition that it be assembled from whole conjugacy classes rather than pieces of them. [quotetheorem:5007] This theorem is especially useful in finite groups where conjugacy classes can be counted. The counting test always has two parts: first choose whole conjugacy classes including $\{e\}$, and then check whether the resulting set could be a subgroup. It is a key ingredient in elementary proofs that $A_5$ is simple and in many arguments showing that particular groups are not simple. [example: A Conjugacy-Class Test] Suppose the conjugacy class sizes of a finite group $G$ are exactly $1,12,12,15,20$, and the class of size $1$ is $\{e\}$. Then \begin{align*} |G|=1+12+12+15+20=60. \end{align*} Let $N\trianglelefteq G$. By *Normal Subgroups as Unions of Conjugacy Classes*, the subgroup $N$ is a union of whole conjugacy classes of $G$. Since $N$ is a subgroup, it contains $e$, so its order must be $1$ plus a sum of some of the four remaining class sizes $12,12,15,20$. The possible orders obtained this way are \begin{align*} 1,\ 13,\ 16,\ 21,\ 25,\ 28,\ 33,\ 36,\ 40,\ 45,\ 48,\ 60. \end{align*} Indeed, the proper nonidentity possibilities come from adding $1$ to one of the nonempty proper sums of $12,12,15,20$: \begin{align*} 12,\ 15,\ 20,\ 24,\ 27,\ 32,\ 35,\ 39,\ 44,\ 47. \end{align*} Thus a proper nonidentity normal subgroup would have order one of \begin{align*} 13,\ 16,\ 21,\ 25,\ 28,\ 33,\ 36,\ 40,\ 45,\ 48. \end{align*} The divisors of $60$ are \begin{align*} 1,\ 2,\ 3,\ 4,\ 5,\ 6,\ 10,\ 12,\ 15,\ 20,\ 30,\ 60. \end{align*} None of $13,16,21,25,28,33,36,40,45,48$ appears in this divisor list. By *Lagrange's theorem*, the order of any subgroup of $G$ must divide $|G|=60$, so none of those possible unions of conjugacy classes can be a subgroup. Therefore every normal subgroup is either $\{e\}$ or $G$, and $G$ is simple. This is the counting test used for $A_5$: once the conjugacy class sizes are known, normality reduces the search to whole-class unions, and divisibility eliminates every proper nonidentity candidate. [/example] ## Almost Simple Groups and Extensions Simple groups rarely appear in isolation. They often sit inside larger groups as a normal subgroup that controls most of the structure. When the surrounding group lies between a non-abelian simple group and its automorphism group, the larger group has one simple core plus possible outer symmetries. [definition: Almost Simple Group] A group $G$ is almost simple if there exists a non-abelian simple group $S$ such that \begin{align*} S \le G \le \operatorname{Aut}(S), \end{align*} where $S$ is identified with its inner automorphism group $\operatorname{Inn}(S)$ inside $\operatorname{Aut}(S)$. [/definition] The group $S$ is then the main simple core, while $G$ may include outer automorphisms of $S$. Almost simple groups are important because many natural symmetry groups are slightly larger than their simple normal subgroup. [example: $S_n$ as an Almost Simple Group] For $n \ge 5$, the subgroup $A_n$ is normal in $S_n$ because it is the kernel of the sign homomorphism $\operatorname{sgn}:S_n\to C_2$. Also $A_n$ is simple by *Simplicity of Alternating Groups*. We compute the conjugation action of $S_n$ on this normal subgroup. For $\sigma\in S_n$, define \begin{align*} \Phi(\sigma):A_n\to A_n,\qquad \Phi(\sigma)(\tau)=\sigma\tau\sigma^{-1}. \end{align*} This map has image in $A_n$ because $A_n\trianglelefteq S_n$. It is a homomorphism in the variable $\tau$, since for $\tau_1,\tau_2\in A_n$, \begin{align*} \Phi(\sigma)(\tau_1\tau_2)=\sigma\tau_1\tau_2\sigma^{-1}=(\sigma\tau_1\sigma^{-1})(\sigma\tau_2\sigma^{-1})=\Phi(\sigma)(\tau_1)\Phi(\sigma)(\tau_2). \end{align*} Its inverse is conjugation by $\sigma^{-1}$, because \begin{align*} \sigma^{-1}(\sigma\tau\sigma^{-1})\sigma=\tau. \end{align*} Thus $\Phi(\sigma)\in\operatorname{Aut}(A_n)$ for every $\sigma\in S_n$. The assignment $\Phi:S_n\to\operatorname{Aut}(A_n)$ is itself a group homomorphism. For $\sigma,\rho\in S_n$ and $\tau\in A_n$, \begin{align*} \Phi(\sigma\rho)(\tau)=\sigma\rho\tau\rho^{-1}\sigma^{-1}=\Phi(\sigma)(\rho\tau\rho^{-1})=(\Phi(\sigma)\circ\Phi(\rho))(\tau). \end{align*} Hence $\Phi(\sigma\rho)=\Phi(\sigma)\circ\Phi(\rho)$. We next check that $\Phi$ is injective. If $\sigma\in\ker(\Phi)$, then $\sigma\tau\sigma^{-1}=\tau$ for every $\tau\in A_n$. In particular, for every $3$-cycle $(ijk)$, \begin{align*} \sigma(ijk)\sigma^{-1}=(\sigma(i)\ \sigma(j)\ \sigma(k))=(ijk). \end{align*} Fix $i\in\{1,\ldots,n\}$. Since $n\ge 5$, choose distinct $j,k,\ell,m$ different from $i$. From the equality for $(ijk)$ we get $\sigma(i)\in\{i,j,k\}$, and from the equality for $(i\ell m)$ we get $\sigma(i)\in\{i,\ell,m\}$. Therefore \begin{align*} \sigma(i)\in \{i,j,k\}\cap\{i,\ell,m\}=\{i\}. \end{align*} So $\sigma(i)=i$. Since this holds for every $i$, $\sigma=e$, and therefore $\ker(\Phi)=\{e\}$. Thus $\Phi$ embeds $S_n$ into $\operatorname{Aut}(A_n)$. Under this embedding, every element of $A_n$ acts on $A_n$ by an inner automorphism, so \begin{align*} A_n\le \Phi(S_n)\le \operatorname{Aut}(A_n). \end{align*} For $n\ne 6$, the standard automorphism theorem for alternating groups says $\operatorname{Aut}(A_n)\cong S_n$, so this embedding identifies $S_n$ with $\operatorname{Aut}(A_n)$. For $n=6$, $\operatorname{Aut}(A_6)$ is larger, but the same injective conjugation map still gives \begin{align*} A_6\le \Phi(S_6)\le \operatorname{Aut}(A_6). \end{align*} Thus $S_n$ is not simple for $n\ge 5$, since $A_n$ is a proper nonidentity normal subgroup, but it is almost simple: its structure lies between the simple core $A_n$ and the full automorphism group of that core. [/example] Almost simplicity shows why the study of simple groups does not end once the simple objects have been listed. One must also understand automorphisms, extensions, and how simple groups act on other structures. ## Beyond and Connected Topics Simple groups are the endpoint of normal-subgroup decomposition, so the immediate next topic is composition theory. The Jordan--Holder theorem explains why composition factors are well-defined up to isomorphism and ordering. For a course-level development of these ideas, see [Cambridge IA Groups](/page/Cambridge%20IA%20Groups) and [Cambridge IB Groups, Rings and Modules](/page/Cambridge%20IB%20Groups%2C%20Rings%20and%20Modules). Finite simple groups lead into one of the largest classification theorems in mathematics. The classification of finite simple groups says that every finite simple group is cyclic of prime order, alternating of degree at least $5$, a group of Lie type, or one of the sporadic groups. This result is far beyond the elementary definition, but the definition of simple group is the entry point. Linear algebra enters through representations. A group homomorphism $G \to GL(V)$ has a kernel, so a simple group representation is either faithful or kills the whole group when the representation kernel is nonidentity. The background language of vector spaces and linear maps is developed in [Cambridge IB Linear Algebra](/page/Cambridge%20IB%20Linear%20Algebra). Lie theory gives infinite analogues and relatives of this story. Simple Lie algebras play a role parallel to simple groups, with ideals replacing normal subgroups. Many connected Lie groups are not simple as abstract groups until one accounts for their centres; for example, projective versions such as $PSL_n(k)$ are often the simple objects rather than $SL_n(k)$ itself. The page [Lie Algebras I: Foundations](/page/Lie%20Algebras%20I%3A%20Foundations) is the natural continuation for the infinitesimal version of simplicity. Infinite group theory also contains simple groups that are not Lie-theoretic. Some arise from actions on trees, homeomorphism groups, or finitely generated constructions designed so that every nonidentity normal subgroup spreads across the whole group. These examples are more advanced than the finite tests on this page, but they show that simplicity is a normal-subgroup condition, not a finiteness condition. Another direction is geometric [group action](/page/Group%20Action). Groups such as $A_5$ arise as rotation groups of polyhedra, and many simple groups are best understood through faithful actions on sets, vector spaces, buildings, or geometries. The action viewpoint turns the abstract absence of normal subgroups into concrete rigidity. ## References Androma, [Cambridge IA Groups](/page/Cambridge%20IA%20Groups). Androma, [Cambridge IB Groups, Rings and Modules](/page/Cambridge%20IB%20Groups%2C%20Rings%20and%20Modules). Androma, [Cambridge IB Linear Algebra](/page/Cambridge%20IB%20Linear%20Algebra). Androma, [Lie Algebras I: Foundations](/page/Lie%20Algebras%20I%3A%20Foundations). Dummit and Foote, *Abstract Algebra* (2004). Rotman, *An Introduction to the Theory of Groups* (1995). Wilson, *The Finite Simple Groups* (2009).