Symmetry often appears first as a group. A rotation group, a matrix group, or a group of transformations records finite motions and their composition. In analysis and geometry, however, many symmetries vary continuously. Near the identity transformation, such a group has first-order directions. Those infinitesimal directions form a [vector space](/page/Vector%20Space), and their commutator records the first non-commuting effect of moving in two directions. This infinitesimal algebraic object is a Lie algebra. A basic example comes from square matrices. Let $M_n(F)$ be the vector space of $n\times n$ matrices over a field $F$. Matrix multiplication is not commutative in general. The expression \begin{align*} [A,B]=AB-BA \end{align*} measures the failure of $A$ and $B$ to commute. This commutator is bilinear, alternates in its two variables, and satisfies a special three-term relation. Those three features are the axioms of a Lie algebra. The point of the theory is that the bracket is not a second multiplication of the same kind as a ring product. It is an operation designed to measure interaction between infinitesimal transformations. Many structural questions then become linear-algebraic while still remembering noncommutativity. ## Definition The ambient scalar field matters throughout the subject. In many analytic applications the field is $\mathbb R$ or $\mathbb C$. In algebraic applications one often works over an arbitrary field, with extra care in small characteristic. [definition: Lie Algebra] Let $F$ be a field. A Lie algebra over $F$ is an $F$-vector space $\mathfrak g$ equipped with a bilinear map \begin{align*} [\cdot,\cdot]\colon \mathfrak g\times \mathfrak g\to \mathfrak g \end{align*} such that, for all $x,y,z\in \mathfrak g$, \begin{align*} [x,x]=0. \end{align*} The Jacobi identity is \begin{align*} [x,[y,z]]+[y,[z,x]]+[z,[x,y]]=0. \end{align*} The map $[\cdot,\cdot]$ is called the Lie bracket. [/definition] The identity $[x,x]=0$ is the alternating law. When the field has characteristic different from $2$, it is equivalent to skew-symmetry: \begin{align*} [x,y]=-[y,x]. \end{align*} The three-term identity is the Jacobi identity. Since a Lie bracket is not an associative operation, Jacobi plays the structural role that makes iterated brackets coherent. Small characteristic changes some familiar shortcuts. The alternating law $[x,x]=0$ is part of the definition because skew-symmetry alone does not imply it in characteristic $2$. In characteristics $2$ and $3$, several classification theorems and representation-theoretic tests require extra hypotheses or modified statements, so results first learned over $\mathbb C$ should not be transferred to arbitrary fields without checking the characteristic. The notation $[x,y]$ is part of the meaning of the structure. It suggests a commutator, but the definition does not require an associative multiplication behind the scenes. Some Lie algebras arise from associative algebras, while others are introduced directly by generators, relations, or geometric constructions. ## First Examples ### Matrix Lie Algebras The matrix example gives a template for much of the subject. [example: General Linear Lie Algebra] The vector space $M_n(F)$ becomes a Lie algebra when equipped with the commutator bracket \begin{align*} [A,B]=AB-BA. \end{align*} This Lie algebra is denoted $\mathfrak{gl}_n(F)$. The bracket is bilinear because matrix multiplication is bilinear: for $\lambda,\mu\in F$ and $A,A',B\in M_n(F)$, \begin{align*} [\lambda A+\mu A',B]=(\lambda A+\mu A')B-B(\lambda A+\mu A')=\lambda(AB-BA)+\mu(A'B-BA')=\lambda[A,B]+\mu[A',B]. \end{align*} The same calculation in the second variable gives \begin{align*} [A,\lambda B+\mu B']=A(\lambda B+\mu B')-(\lambda B+\mu B')A=\lambda(AB-BA)+\mu(AB'-B'A)=\lambda[A,B]+\mu[A,B']. \end{align*} It is alternating, since for every $A\in M_n(F)$, \begin{align*} [A,A]=AA-AA=0. \end{align*} It remains to verify the Jacobi identity. For $A,B,C\in M_n(F)$, using the definition of the commutator and associativity of matrix multiplication, \begin{align*} [A,[B,C]]=A(BC-CB)-(BC-CB)A=ABC-ACB-BCA+CBA. \end{align*} Similarly, \begin{align*} [B,[C,A]]=B(CA-AC)-(CA-AC)B=BCA-BAC-CAB+ACB. \end{align*} And \begin{align*} [C,[A,B]]=C(AB-BA)-(AB-BA)C=CAB-CBA-ABC+BAC. \end{align*} Adding the three displayed expressions, the terms cancel in pairs: \begin{align*} (ABC-ABC)+(-ACB+ACB)+(-BCA+BCA)+(CBA-CBA)+(-BAC+BAC)+(-CAB+CAB)=0. \end{align*} Thus \begin{align*} [A,[B,C]]+[B,[C,A]]+[C,[A,B]]=0. \end{align*} So $M_n(F)$ with the commutator bracket is a Lie algebra, and the example shows that ordinary associative matrix multiplication produces a Lie bracket by measuring noncommutativity. [/example] The trace-zero matrices form a smaller Lie algebra that is central in representation theory. [example: Special Linear Lie Algebra] The set \begin{align*} \mathfrak{sl}_n(F)=\{A\in M_n(F):\operatorname{tr}(A)=0\} \end{align*} is a vector subspace of $M_n(F)$. Indeed, if $A,B\in \mathfrak{sl}_n(F)$ and $\lambda,\mu\in F$, then linearity of trace gives \begin{align*} \operatorname{tr}(\lambda A+\mu B)=\lambda\operatorname{tr}(A)+\mu\operatorname{tr}(B)=\lambda\cdot 0+\mu\cdot 0=0. \end{align*} So $\lambda A+\mu B\in \mathfrak{sl}_n(F)$. It remains to check that this subspace is closed under the commutator bracket inherited from $\mathfrak{gl}_n(F)$. Let $A=(a_{ij})$ and $B=(b_{ij})$ be elements of $\mathfrak{sl}_n(F)$. The diagonal entries of $AB$ are \begin{align*} (AB)_{ii}=\sum_{j=1}^n a_{ij}b_{ji}. \end{align*} Hence \begin{align*} \operatorname{tr}(AB)=\sum_{i=1}^n(AB)_{ii}=\sum_{i=1}^n\sum_{j=1}^n a_{ij}b_{ji}. \end{align*} Similarly, \begin{align*} \operatorname{tr}(BA)=\sum_{j=1}^n\sum_{i=1}^n b_{ji}a_{ij}. \end{align*} Since entries lie in the field $F$, $a_{ij}b_{ji}=b_{ji}a_{ij}$, and the two finite double sums have the same terms. Therefore \begin{align*} \operatorname{tr}(AB)=\operatorname{tr}(BA). \end{align*} Thus \begin{align*} \operatorname{tr}([A,B])=\operatorname{tr}(AB-BA)=\operatorname{tr}(AB)-\operatorname{tr}(BA)=0. \end{align*} So $[A,B]\in\mathfrak{sl}_n(F)$. The commutator bracket on $\mathfrak{sl}_n(F)$ is the restriction of the commutator bracket on $M_n(F)$, so bilinearity, alternation, and the Jacobi identity are inherited from the same matrix identities. Hence $\mathfrak{sl}_n(F)$ is a Lie algebra, called the special linear Lie algebra. [/example] There are also Lie algebras with no nonzero brackets. [example: Abelian Lie Algebra] Let $V$ be an $F$-vector space, and define a bracket on $V$ by \begin{align*} [v,w]=0 \end{align*} for all $v,w\in V$. We check that this is a Lie bracket. For bilinearity in the first variable, let $u,u',w\in V$ and $\lambda,\mu\in F$. By the definition of the bracket, \begin{align*} [\lambda u+\mu u',w]=0. \end{align*} Also, \begin{align*} \lambda[u,w]+\mu[u',w]=\lambda\cdot 0+\mu\cdot 0=0. \end{align*} Thus $[\lambda u+\mu u',w]=\lambda[u,w]+\mu[u',w]$. The same argument gives bilinearity in the second variable: for $u,w,w'\in V$, \begin{align*} [u,\lambda w+\mu w']=0=\lambda\cdot 0+\mu\cdot 0=\lambda[u,w]+\mu[u,w']. \end{align*} The bracket is alternating because, for every $v\in V$, \begin{align*} [v,v]=0. \end{align*} The Jacobi identity also holds. For $u,v,w\in V$, each inner bracket is zero, so \begin{align*} [u,[v,w]]=[u,0]=0. \end{align*} Similarly, \begin{align*} [v,[w,u]]=0. \end{align*} And \begin{align*} [w,[u,v]]=0. \end{align*} Therefore \begin{align*} [u,[v,w]]+[v,[w,u]]+[w,[u,v]]=0+0+0=0. \end{align*} Hence every $F$-vector space becomes a Lie algebra with the zero bracket. Such a Lie algebra is called abelian; it records a situation where all infinitesimal directions commute, so no nonzero bracket interaction remains. [/example] Abelian Lie algebras are important because they isolate what is genuinely caused by the bracket. They also occur as tangent Lie algebras of commutative Lie groups, such as vector groups and tori. ### Why the Jacobi Identity Is Needed Bilinearity and alternation alone do not give a workable infinitesimal symmetry theory. The missing condition is exactly the rule that makes brackets behave consistently under repeated commutation. [example: An Alternating Bracket That Is Not Lie] Let $F$ be a field, and let $V$ have basis $e_1,e_2,e_3$. Define a bilinear bracket on basis vectors by \begin{align*} [e_1,e_2]=e_1,\quad [e_2,e_3]=e_2,\quad [e_3,e_1]=e_3, \end{align*} together with $[e_i,e_i]=0$ and $[e_j,e_i]=-[e_i,e_j]$ for all $i,j$. This bracket is alternating: if $v=a_1e_1+a_2e_2+a_3e_3$, then bilinearity gives terms $a_i^2[e_i,e_i]=0$ and paired terms $a_ia_j([e_i,e_j]+[e_j,e_i])=a_ia_j([e_i,e_j]-[e_i,e_j])=0$, so $[v,v]=0$. Now compute the Jacobi expression on $e_1,e_2,e_3$. Since $[e_2,e_3]=e_2$, \begin{align*} [e_1,[e_2,e_3]]=[e_1,e_2]=e_1. \end{align*} Since $[e_3,e_1]=e_3$, \begin{align*} [e_2,[e_3,e_1]]=[e_2,e_3]=e_2. \end{align*} Since $[e_1,e_2]=e_1$, \begin{align*} [e_3,[e_1,e_2]]=[e_3,e_1]=e_3. \end{align*} Adding these three identities gives \begin{align*} [e_1,[e_2,e_3]]+[e_2,[e_3,e_1]]+[e_3,[e_1,e_2]]=e_1+e_2+e_3. \end{align*} Because $e_1,e_2,e_3$ are a basis, the vector $e_1+e_2+e_3$ is not zero. Thus the Jacobi identity fails for this bilinear alternating bracket, so it is not a Lie bracket. [/example] This failure is not cosmetic. Without Jacobi, the operation $y\mapsto [x,y]$ need not act like a derivation of the bracket, and the basic mechanism behind adjoint actions breaks down. ## Subalgebras and Ideals ### Bracket-Closed Subspaces The first structural notions imitate subgroups and normal subgroups, but they are linear. They let a large Lie algebra be studied through smaller bracket-closed pieces that still carry the same kind of algebraic structure. [definition: Lie Subalgebra] Let $\mathfrak g$ be a Lie algebra over $F$. A Lie subalgebra of $\mathfrak g$ is a vector subspace $\mathfrak h\subset \mathfrak g$ such that \begin{align*} [x,y]\in \mathfrak h \end{align*} for all $x,y\in\mathfrak h$. [/definition] A subalgebra is a smaller Lie algebra using the same bracket. For example, $\mathfrak{sl}_n(F)$ is a subalgebra of $\mathfrak{gl}_n(F)$. ### Ideals and Quotients Quotient constructions require a stronger condition, because brackets with elements outside the chosen subspace must not leave the subspace. The reason is the same obstruction that appears in quotient groups: if $a$ is going to become zero in the quotient, then every bracket involving $a$ must also become zero. Otherwise the bracket of two cosets would depend on the representatives chosen, so the quotient would remember an arbitrary choice rather than a well-defined infinitesimal symmetry. [definition: Ideal of a Lie Algebra] Let $\mathfrak g$ be a Lie algebra over a field $F$. An ideal of $\mathfrak g$ is an $F$-vector subspace $\mathfrak i\subset \mathfrak g$ such that \begin{align*} [x,a]\in \mathfrak i \end{align*} for every $x\in\mathfrak g$ and every $a\in\mathfrak i$. [/definition] Ideals are the subspaces by which one can form quotients. The theorem below is the formal check that the obstruction just described is the only obstruction: the bracket on cosets is forced to be $[x+\mathfrak i,y+\mathfrak i]=[x,y]+\mathfrak i$, and the ideal condition is precisely what makes this formula independent of the representatives $x$ and $y$. [quotetheorem:8146] This theorem is the first place where the ideal condition earns its name: it is exactly the condition that makes quotient Lie algebras behave like quotient groups and quotient rings. ## Homomorphisms Maps of Lie algebras must respect both the vector-space structure and the bracket. This condition is what makes Lie algebras into a category, and it is the right notion for comparing infinitesimal symmetry structures. [definition: Lie Algebra Homomorphism] Let $\mathfrak g$ and $\mathfrak h$ be Lie algebras over the same field $F$. A Lie algebra homomorphism is an $F$-[linear map](/page/Linear%20Map) $\phi\colon \mathfrak g\to \mathfrak h$ such that \begin{align*} \phi([x,y])=[\phi(x),\phi(y)] \end{align*} for all $x,y\in\mathfrak g$. [/definition] The next result is the Lie-algebra analogue of the familiar [first isomorphism theorem](/theorems/791). It answers two linked questions that appear whenever a homomorphism is used to simplify a Lie algebra: which part of the domain has been collapsed, and which part of the codomain has actually been reached? The answer explains why ideals, rather than arbitrary subalgebras, control quotients, while images remain bracket-closed subalgebras. [quotetheorem:3764] This result turns many questions about homomorphisms into questions about ideals and quotients. It is also the reason that kernels are not just incidental subspaces: they are precisely the subspaces that can be collapsed without losing the Lie-algebra structure. ## The Adjoint Representation A Lie algebra acts on itself by taking brackets. For each $x\in\mathfrak g$, define a linear map $\operatorname{ad}_x\colon \mathfrak g\to\mathfrak g$ by \begin{align*} \operatorname{ad}_x(y)=[x,y]. \end{align*} The Jacobi identity is often best read as saying that these maps differentiate the bracket. Moving the term $[x,[y,z]]$ to the left side of the Jacobi identity gives exactly the following structural rule. [quotetheorem:3762] To build a representation from the operators $\operatorname{ad}_x$, one more compatibility condition is needed: bracketing two elements of $\mathfrak g$ should match taking the commutator of the two corresponding operators. The next theorem proves exactly this condition, so it lets the internal bracket of $\mathfrak g$ be studied inside the matrix-style Lie algebra $\mathfrak{gl}(\mathfrak g)$. The assignment \begin{align*} \operatorname{ad}: \mathfrak g \to \mathfrak{gl}(\mathfrak g) \end{align*} is defined by \begin{align*} \operatorname{ad}(x)=\operatorname{ad}_x. \end{align*} For this assignment to deserve the name representation, it must preserve brackets. The obstruction is that there are two brackets in sight: the original bracket $[x,y]$ in $\mathfrak g$, and the commutator $[\operatorname{ad}_x,\operatorname{ad}_y]$ of endomorphisms of $\mathfrak g$. The Jacobi identity is exactly what makes these two operations agree under $\operatorname{ad}$, so the internal action by bracketing is compatible with the ambient linear-operator Lie algebra. [quotetheorem:3774] The adjoint homomorphism is useful because it turns the internal bracket of $\mathfrak g$ into ordinary commutators of linear maps. The next invariant needed for structure theory is the kernel of this always-available representation: it consists exactly of the elements that cannot be detected by bracketing with the rest of the algebra. That kernel deserves its own name. Which infinitesimal symmetries commute, under the bracket, with every other infinitesimal symmetry? Those elements form the center, and the definition below isolates them for later quotient and decomposition arguments. [definition: Center of a Lie Algebra] Let $\mathfrak g$ be a Lie algebra over a field $F$. The center of $\mathfrak g$ is the $F$-vector subspace \begin{align*} Z(\mathfrak g)=\{z\in\mathfrak g:[z,x]=0\text{ for all }x\in\mathfrak g\}. \end{align*} [/definition] The center measures the part of the Lie algebra that does not interact with the rest under the bracket. In an abelian Lie algebra the center is all of $\mathfrak g$. In many matrix Lie algebras the center is much smaller. ## Structure Through Commutators ### Derived Subalgebras and Abelianization Repeated brackets give a way to measure how far a Lie algebra is from being abelian. The first step is the derived subalgebra. [definition: Derived Subalgebra] Let $\mathfrak g$ be a Lie algebra over a field $F$. The derived subalgebra is the $F$-linear span \begin{align*} [\mathfrak g,\mathfrak g]=\operatorname{span}\{[x,y]:x,y\in\mathfrak g\}. \end{align*} [/definition] The derived subalgebra is the part generated by all commutators, so quotienting by it should erase exactly the non-abelian information. Before using that quotient, one needs the derived subalgebra to be an ideal rather than merely a subalgebra; otherwise the quotient bracket would not be available. [quotetheorem:3758] Once this ideal property is in place, the quotient by $[\mathfrak g,\mathfrak g]$ is available as a Lie algebra. The next question is whether this quotient is merely one convenient way to force commutators to vanish, or whether it is the canonical way. The abelianization theorem answers that question: it identifies $\mathfrak g/[\mathfrak g,\mathfrak g]$ as the largest abelian quotient through which every homomorphism from $\mathfrak g$ to an abelian Lie algebra must pass. [quotetheorem:8147] The abelianization only tests the first obstruction to being abelian. A Lie algebra may have nonzero commutators, but those commutators may themselves behave more simply than the original algebra. To detect this layered behavior, we need a recursive construction that asks the same question again inside the commutator algebra: after passing from $\mathfrak g$ to $[\mathfrak g,\mathfrak g]$, how much non-abelian structure is still left? The derived series is the bookkeeping device for that question. Instead of bracketing each stage with the original algebra, it repeatedly takes the commutator subalgebra of the current stage, so each step measures the internal noncommutativity that survived the previous step. [definition: Derived Series] Let $\mathfrak g$ be a Lie algebra over a field $F$. The derived series of $\mathfrak g$ is the sequence $(\mathfrak d_k)_{k\ge 0}$ defined by \begin{align*} \mathfrak d_0&=\mathfrak g, & \mathfrak d_{k+1}&=[\mathfrak d_k,\mathfrak d_k]. \end{align*} [/definition] The derived series records a weaker form of commutator decay than the lower central series below. It asks for repeated self-commutators to disappear, which captures Lie algebras built by successive abelian extensions. [definition: Solvable Lie Algebra] Let $\mathfrak g$ be a Lie algebra over a field $F$, and let $(\mathfrak d_k)_{k\ge 0}$ be its derived series. The Lie algebra $\mathfrak g$ is solvable if there exists $N\in\mathbb N$ such that \begin{align*} \mathfrak d_N=0. \end{align*} [/definition] Solvability and nilpotence are related but not identical. Solvability allows commutators to be taken inside the current stage, while nilpotence demands vanishing even when each stage is bracketed against all of $\mathfrak g$. ### Lower Central Series Another construction keeps bracketing with the whole algebra. This stronger commutator test is needed to define nilpotence, where every sufficiently long nested bracket with arbitrary outside entries must vanish. [definition: Lower Central Series] Let $\mathfrak g$ be a Lie algebra over a field $F$. The lower central series of $\mathfrak g$ is the sequence $(\mathfrak g_k)_{k\ge 1}$ defined by \begin{align*} \mathfrak g_1=\mathfrak g \end{align*} and, for every $k\ge 1$, by \begin{align*} \mathfrak g_{k+1}=[\mathfrak g,\mathfrak g_k]. \end{align*} Here $[\mathfrak g,\mathfrak g_k]$ means the $F$-linear span of all brackets $[x,y]$ with $x\in\mathfrak g$ and $y\in\mathfrak g_k$. [/definition] The point of the lower central series is to test whether repeated bracketing with arbitrary elements eventually kills all commutators. The next definition names the Lie algebras for which this stronger series reaches zero; this condition is central in the structure theory of highly constrained non-abelian algebras. [definition: Nilpotent Lie Algebra] Let $\mathfrak g$ be a Lie algebra over a field $F$. Define its lower central series by $\mathfrak g_1=\mathfrak g$ and $\mathfrak g_{k+1}=[\mathfrak g,\mathfrak g_k]$ for $k\ge 1$. The Lie algebra $\mathfrak g$ is nilpotent if there exists $N\in\mathbb N$ such that \begin{align*} \mathfrak g_N=0. \end{align*} [/definition] Nilpotence should be compared with solvability through the two series: the lower central series brackets with all of $\mathfrak g$, while the derived series only brackets the current term with itself. This comparison matters because the two definitions look similar but measure different strengths of commutator decay. The next theorem supplies the basic calibration: the stronger lower-central vanishing forces derived-series vanishing, so nilpotent Lie algebras automatically sit inside the larger solvable class. [quotetheorem:3788] This implication is one-way: solvable Lie algebras need not be nilpotent. These conditions organize many classification results, especially for finite-dimensional Lie algebras over algebraically closed fields of characteristic $0$. ## Representations Lie algebras are often studied through their actions on vector spaces. A representation turns an abstract bracket into commutators of concrete linear operators, so it exposes the algebra through linear algebra. [definition: Representation of a Lie Algebra] Let $\mathfrak g$ be a Lie algebra over $F$, and let $V$ be an $F$-vector space. A representation of $\mathfrak g$ on $V$ is a Lie algebra homomorphism \begin{align*} \rho\colon \mathfrak g\to \mathfrak{gl}(V)=\operatorname{End}_F(V), \end{align*} where $\operatorname{End}_F(V)$ is the $F$-vector space of $F$-linear maps $V\to V$, equipped with the commutator bracket. [/definition] Equivalently, each $x\in\mathfrak g$ acts as a linear operator $\rho(x)$ on $V$, and the action satisfies \begin{align*} \rho([x,y])=\rho(x)\rho(y)-\rho(y)\rho(x). \end{align*} Representations translate bracket identities into identities among linear maps. This is why linear algebra remains central even when the Lie algebra itself is not a matrix algebra at the start. The adjoint representation is the representation of $\mathfrak g$ on its own underlying vector space. Other representations reveal how $\mathfrak g$ can act on geometric, analytic, or algebraic objects. ## How Lie Algebras Enter Analysis ### Infinitesimal Generators In analysis, Lie algebras often appear near one-parameter families of transformations. The first infinitesimal object is the generator of a flow or semigroup: it differentiates motion in one time direction. This is not yet the Lie bracket. The bracket enters when two infinitesimal motions are compared by a commutator, either as commutators of operators in an associative algebra or as Lie brackets of vector fields. [definition: Infinitesimal Generator] Let $X$ be a real [Banach space](/page/Banach%20Space). Let either $T:\mathbb R\to \mathcal{L}(X)$ be a strongly continuous one-parameter group of bounded linear operators or $T:[0,\infty)\to \mathcal{L}(X)$ be a strongly continuous one-parameter semigroup of bounded linear operators, with $T(0)=I$ and $T(s+t)=T(s)T(t)$ whenever the parameters lie in the chosen domain. The infinitesimal generator of $T$ is the operator \begin{align*} A:D(A)\subset X\to X \end{align*} whose domain is \begin{align*} D(A)=\left\{x\in X:\lim_{t\to 0}\frac{T(t)x-x}{t}\text{ exists in }X\right\} \end{align*} for a group, with the corresponding one-sided limit $t\downarrow 0$ for a semigroup, and whose value on $x\in D(A)$ is \begin{align*} Ax=\lim_{t\to 0}\frac{T(t)x-x}{t} \end{align*} for a group, with the corresponding one-sided limit for a semigroup. [/definition] This operator may be unbounded and need not be defined on all of $X$. If the family is differentiable in operator norm, then the special bounded case occurs and $A\in\mathcal{L}(X)$. The full functional-analytic theory studies domains, closedness, resolvents, and generation theorems; this page only uses the generator as one route from continuous motion to infinitesimal algebra. The bracket is a second layer of structure. For bounded operators on the same Banach space, the products $AB$ and $BA$ are bounded operators on all of $X$, so the commutator $[A,B]=AB-BA$ is again a bounded operator. For unbounded generators, the same formula becomes domain-sensitive: $AB$ and $BA$ may have different domains or may fail to share a useful common domain, so the commutator is not automatically an operator on all of $X$. For smooth vector fields on a manifold, the Lie bracket is the vector field measuring the infinitesimal commutator of their flows. Thus generators explain where infinitesimal directions come from, while Lie brackets explain how two such directions fail to commute. ### From Lie Groups to Lie Algebras For a Lie group $G$, the tangent space at the identity element carries a natural Lie bracket. This Lie algebra is denoted $\operatorname{Lie}(G)$. The passage from $G$ to $\operatorname{Lie}(G)$ turns local questions about smooth group actions into questions about a bracketed vector space. This connection explains the term "Lie algebra." The algebra is named for Sophus Lie and was developed to study continuous transformation groups and differential equations. The modern definition is abstract enough to support representation theory, algebraic geometry, differential geometry, and operator theory. ## A Small Computation ### Matrix Units Let $E_{ij}$ denote the matrix with $1$ in the $(i,j)$-entry and $0$ elsewhere. In $\mathfrak{gl}_n(F)$, multiplication of matrix units gives \begin{align*} E_{ij}E_{kl}=\delta_{jk}E_{il}. \end{align*} Therefore \begin{align*} [E_{ij},E_{kl}]=\delta_{jk}E_{il}-\delta_{li}E_{kj}. \end{align*} This formula shows how the bracket moves indices. It also gives a concrete way to compute inside $\mathfrak{sl}_n(F)$ and its subalgebras. ### The $\mathfrak{sl}_2$ Relations For example, suppose $F$ has characteristic not equal to $2$. In $\mathfrak{sl}_2(F)$ with basis \begin{align*} e=E_{12},\qquad f=E_{21},\qquad h=E_{11}-E_{22}, \end{align*} the brackets are \begin{align*} [h,e]=2e. \end{align*} Also, \begin{align*} [h,f]=-2f. \end{align*} Finally, \begin{align*} [e,f]=h. \end{align*} Over fields of characteristic $0$, these three relations are a compact model for many phenomena in semisimple Lie theory. ## Beyond and Connected Topics After the basic definitions, the subject splits into several linked directions. One direction studies ideals, quotients, solvable Lie algebras, nilpotent Lie algebras, simple Lie algebras, and the radical, which separates the solvable part of a finite-dimensional Lie algebra from its semisimple quotient. Another direction studies representations and modules, where abstract brackets become concrete commutators of linear maps. A third direction connects Lie algebras to Lie groups, differential operators, and flows. The finite-dimensional complex semisimple theory has a deep classification by root systems and Dynkin diagrams. Universal enveloping algebras form another bridge: they package Lie brackets into associative algebras so that representation theory can use tools from modules over rings. That theory rests on the elementary ideas introduced here: brackets, ideals, adjoint actions, and representations. Even in analytic settings where infinite-dimensional Lie algebras occur, these same local notions remain the starting vocabulary. For broader companion treatments, [Lie Algebras I: Foundations](/page/Lie%20Algebras%20I%3A%20Foundations) develops the entry-level algebraic framework, while [Lie Algebras II: Structure and Classification](/page/Lie%20Algebras%20II%3A%20Structure%20and%20Classification) continues toward the finite-dimensional structure theory. The central mental picture is this: a Lie algebra is a vector space whose bracket records infinitesimal noncommutativity. The bracket is linear enough to calculate with, but rigid enough to preserve the structure of symmetry. ## References Androma, [Lie Algebras I: Foundations](/page/Lie%20Algebras%20I%3A%20Foundations). Androma, [Lie Algebras II: Structure and Classification](/page/Lie%20Algebras%20II%3A%20Structure%20and%20Classification). Androma, [Cambridge IB Linear Algebra](/page/Cambridge%20IB%20Linear%20Algebra). Androma, [Cambridge III Commutative Algebra](/page/Cambridge%20III%20Commutative%20Algebra). Humphreys, *Introduction to Lie Algebras and Representation Theory* (1972). Hall, *Lie Groups, Lie Algebras, and Representations* (2015). Serre, *Complex Semisimple Lie Algebras* (1987).