A module is meant to be a linear world: we can add elements, multiply them by scalars from a ring, and study maps that respect both operations. But linear worlds often contain redundant directions. A system of equations may identify two solutions if they differ by a homogeneous solution; a homomorphism may forget exactly the elements in its kernel; a representation may have a stable subspace that we want to collapse. The quotient module is the construction that turns such identifications into a new module rather than an informal rule. The guiding question is this: if $N$ is a submodule of an $R$-module $M$, can we treat all elements of $N$ as zero while keeping the remaining module structure intact? The answer is yes, and the quotient module $M/N$ is the most efficient object that does exactly that. [example: Collapsing Even Integers] Let $R=\mathbb{Z}$, let $M=\mathbb{Z}$ as a $\mathbb{Z}$-module, and let $N=2\mathbb{Z}$. For integers $a,b\in\mathbb{Z}$, the cosets $a+2\mathbb{Z}$ and $b+2\mathbb{Z}$ are equal exactly when $a-b\in 2\mathbb{Z}$, so the quotient identifies precisely the integers with the same parity. Every integer is either even or odd. If $m$ is even, then $m=2q$ for some $q\in\mathbb{Z}$, so $m\in 2\mathbb{Z}$ and \begin{align*} m+2\mathbb{Z}=0+2\mathbb{Z}. \end{align*} If $m$ is odd, then $m=2q+1$ for some $q\in\mathbb{Z}$, so $m-1=2q\in 2\mathbb{Z}$ and \begin{align*} m+2\mathbb{Z}=1+2\mathbb{Z}. \end{align*} Thus the only cosets are $0+2\mathbb{Z}$ and $1+2\mathbb{Z}$. Addition is computed by adding representatives and then passing to the resulting coset: \begin{align*} (1+2\mathbb{Z})+(1+2\mathbb{Z})=(1+1)+2\mathbb{Z}. \end{align*} Since $1+1=2$ and $2-0=2\in 2\mathbb{Z}$, this becomes \begin{align*} 2+2\mathbb{Z}=0+2\mathbb{Z}. \end{align*} Scalar multiplication is computed in the same way: \begin{align*} n(1+2\mathbb{Z})=n\cdot 1+2\mathbb{Z}=n+2\mathbb{Z}. \end{align*} Therefore $n(1+2\mathbb{Z})=0+2\mathbb{Z}$ when $n$ is even, and $n(1+2\mathbb{Z})=1+2\mathbb{Z}$ when $n$ is odd. The quotient module $\mathbb{Z}/2\mathbb{Z}$ keeps exactly the parity of an integer and forgets everything divisible by $2$. [/example] This example is small, but it shows the essential feature: the quotient does not choose preferred representatives. It makes the entire submodule invisible and remembers only cosets. The construction is useful exactly because all computations are invariant under changing representatives by elements of the submodule. ## Definition The parent concept is a [module](/page/Module): an [abelian group](/page/Abelian%20Group) equipped with scalar multiplication by a ring. A quotient module asks for one extra ingredient, a submodule $N \subset M$, because only submodules are stable enough to be collapsed without breaking addition and scalar multiplication. The definition answers the practical question of how to turn the instruction "treat every element of $N$ as zero" into an actual module. [definition: Quotient Module] Let $R$ be a ring, let $M$ be a left $R$-module, and let $N \subset M$ be an $R$-submodule. The quotient module $M/N$ is the set of additive cosets \begin{align*} M/N = \{m + N : m \in M\}, \end{align*} with addition map $+: (M/N)\times(M/N)\to M/N$ defined by \begin{align*} (m + N) + (m' + N) = (m + m') + N, \end{align*} and scalar multiplication map $R\times(M/N)\to M/N$ defined by \begin{align*} r(m + N) = rm + N, \end{align*} for all $m,m' \in M$ and $r \in R$. [/definition] The definition is deliberately short because all the work is hidden in the word submodule. If $N$ were only a subset, addition or scalar multiplication could depend on the chosen representative. The condition $N \subset M$ as a submodule is precisely what makes the two displayed operations meaningful. The formulas in the definition still require a structural check. A coset has many names, and a formula using a representative is legitimate only if every name gives the same result. The obstruction is exactly the ambiguity created by adding elements of $N$, so the quotient construction must show that addition and scalar multiplication ignore that ambiguity. The following result is only about the ordinary module structure on $M/N$: it verifies that these two operations are well-defined and satisfy the module axioms. [quotetheorem:8502] This theorem is the reason the notation $M/N$ is not just set-theoretic bookkeeping. It says that the quotient inherits exactly enough linear structure from $M$ to remain inside the same category of modules. It does not by itself prove that extra structure descends. For example, if $M$ carries operators or a [Lie algebra](/page/Lie%20Algebra) action, one must separately check that the relevant submodule is stable under those operations and that the induced operations satisfy the required identities on cosets. ## Quotient Maps and Kernels Once a quotient exists, we need a canonical map that performs the collapse. This map is the device that lets us compare $M$ with $M/N$, transport homomorphisms through the quotient, and name precisely which elements have disappeared. [definition: Quotient Map] Let $R$ be a ring, let $M$ be a left $R$-module, and let $N \subset M$ be an $R$-submodule. The quotient map is the $R$-[linear map](/page/Linear%20Map) $\pi: M \to M/N$ defined by \begin{align*} \pi(m) = m+N. \end{align*} [/definition] The quotient map should erase exactly $N$, not more and not less. Verifying this gives the basic diagnostic for quotient maps and prepares the link with kernels of general homomorphisms. [quotetheorem:7860] The quotient module is therefore a controlled way of creating a map with a prescribed kernel. This is the bridge from quotients to homomorphism theorems. [example: Quotient of a Free Module by One Relation] Let $R$ be a ring, let $a,b\in R$, and let $R^2$ denote the free left $R$-module with standard basis $e_1=(1,0)$ and $e_2=(0,1)$. Let $N=R(a,b)$, meaning the submodule generated by the single vector $(a,b)$: explicitly, $N=\{r(a,b):r\in R\}$. We compute the quotient $R^2/N$ in terms of the images $e_1+N$ and $e_2+N$. For every $(x,y)\in R^2$, \begin{align*} (x,y)=x(1,0)+y(0,1)=xe_1+ye_2. \end{align*} Passing to cosets gives \begin{align*} (x,y)+N=x(e_1+N)+y(e_2+N). \end{align*} Thus $e_1+N$ and $e_2+N$ generate $R^2/N$. The generator of $N$ is \begin{align*} (a,b)=a(1,0)+b(0,1)=ae_1+be_2. \end{align*} Since $(a,b)\in N$, its coset is the zero coset: \begin{align*} (a,b)+N=0+N. \end{align*} Therefore \begin{align*} a(e_1+N)+b(e_2+N)=(ae_1+be_2)+N=(a,b)+N=0+N. \end{align*} Conversely, every element of $N$ has the form $r(a,b)$ for some $r\in R$, and \begin{align*} r(a,b)=r(ae_1+be_2)=(ra)e_1+(rb)e_2. \end{align*} In the quotient, this becomes \begin{align*} r(a,b)+N=r\bigl(a(e_1+N)+b(e_2+N)\bigr)=0+N. \end{align*} So the single relation $a(e_1+N)+b(e_2+N)=0+N$ accounts exactly for the multiples of $(a,b)$ that are collapsed. When $R=\mathbb{Z}$ and $(a,b)=(2,3)$, the submodule is $N=\mathbb{Z}(2,3)$, and the defining relation is \begin{align*} 2(e_1+N)+3(e_2+N)=0+N. \end{align*} Thus $\mathbb{Z}^2/N$ is generated by the two cosets $e_1+N$ and $e_2+N$, with the relation that $2e_1+3e_2$ becomes zero. The quotient turns the chosen generator of the relation submodule into a linear dependence among the images of the free generators. [/example] ## Cosets and Well-Defined Operations The main danger in quotient constructions is representative dependence. A coset $m+N$ may also be written $(m+n)+N$ for any $n \in N$, so an operation on cosets is legitimate only if changing representatives does not change the answer. The right language for this is congruence: two elements of $M$ should count as the same after quotienting precisely when their difference lies in the part we plan to erase. [definition: Congruence Modulo a Submodule] Let $R$ be a ring, let $M$ be a left $R$-module, and let $N \subset M$ be an $R$-submodule. For $m,m' \in M$, write \begin{align*} m \equiv m' \pmod{N} \end{align*} if $m - m' \in N$. [/definition] This relation translates the quotient into a familiar language, but the construction still has a possible ambiguity: the same coset can be named by many different elements of $M$. To compute in $M/N$, we need an intrinsic equality test that does not depend on which names were chosen. The obstruction is exactly the difference of the two representatives, and equality should hold precisely when that difference lies in the submodule being collapsed. [remark: Coset Equality Criterion] With $R$, $M$, and $N$ as above, two additive cosets in $M/N$ are equal exactly when their representatives differ by an element of $N$: \begin{align*} m+N=m'+N \quad\Longleftrightarrow\quad m-m'\in N. \end{align*} Equivalently, equality in the quotient is the same relation as congruence modulo $N$. [/remark] Once equality in the quotient is understood, computations become systematic. You compute upstairs in $M$, then pass to the coset, while remembering that any element of $N$ may be discarded. [example: Representative Independence] Let $M=\mathbb{Z}^2$ and let $N=\mathbb{Z}(1,1)$. The elements $(3,0)$ and $(2,-1)$ determine the same element of $M/N$ because their difference lies in the submodule being collapsed: \begin{align*} (3,0)-(2,-1)=(3-2,0-(-1))=(1,1)=1(1,1)\in N. \end{align*} Equivalently, $(3,0)=(2,-1)+(1,1)$ with $(1,1)\in N$, so adding $N$ to both representatives gives the same coset: \begin{align*} (3,0)+N=((2,-1)+(1,1))+N=(2,-1)+N. \end{align*} Scalar multiplication is also independent of this choice of representative. Multiplying both representatives by $5$ gives \begin{align*} 5(3,0)=(15,0). \end{align*} and \begin{align*} 5(2,-1)=(10,-5). \end{align*} Their difference is \begin{align*} (15,0)-(10,-5)=(5,5)=5(1,1)\in N. \end{align*} Hence \begin{align*} 5((3,0)+N)=(15,0)+N=(10,-5)+N=5((2,-1)+N). \end{align*} The quotient forgets motion in the diagonal direction $\mathbb{Z}(1,1)$; in particular, both representatives have the same value $x-y=3$, since $3-0=3$ and $2-(-1)=3$. [/example] The preceding example also hints at what fails without the submodule condition. A subset closed under neither addition nor scalar multiplication cannot be collapsed while preserving the module laws. [example: A Non-Submodule Cannot Be Quotiented] Let $R=\mathbb{Z}$, let $M=\mathbb{Z}$, and let $S=\{0,2\}$. The first obstruction is that the translated sets $a+S=\{a+s:s\in S\}$ do not behave like cosets of a subgroup. For example, \begin{align*} 0+S=\{0+0,0+2\}=\{0,2\}. \end{align*} Also, \begin{align*} 2+S=\{2+0,2+2\}=\{2,4\}. \end{align*} These two sets overlap because $2\in\{0,2\}$ and $2\in\{2,4\}$, but they are not equal because $0\in\{0,2\}$ while $0\notin\{2,4\}$. Thus the proposed translates do not partition $\mathbb{Z}$, so they cannot serve as the elements of a quotient module. The representative problem appears immediately in addition. Since $0\in 0+S$, using the representative $0$ for each summand would give \begin{align*} (0+S)+(0+S)=(0+0)+S=0+S=\{0,2\}. \end{align*} But $2\in 0+S$ as well, so using the representative $2$ for each summand would give \begin{align*} (0+S)+(0+S)=(2+2)+S=4+S. \end{align*} Here \begin{align*} 4+S=\{4+0,4+2\}=\{4,6\}. \end{align*} The two answers are different because $0\in\{0,2\}$ but $0\notin\{4,6\}$. This dependence on representatives occurs because $S$ is not closed under addition: $2\in S$ and $2\in S$, but $2+2=4\notin S$. The quotient construction therefore breaks at the level of addition, before scalar multiplication can define a module structure. [/example] ## Universal Property and Factorisation ### Maps that Kill a Submodule The quotient module is best understood not as a set of cosets but as the object through which maps factor. If a homomorphism $f: M \to P$ kills every element of $N$, then $f$ cannot distinguish any two elements that differ by an element of $N$. The quotient $M/N$ is the module obtained by making exactly those identifications and no additional ones. The universal property turns this observation into a construction principle. It explains how to define maps out of $M/N$ without choosing representatives and gives uniqueness, so there is only one map on the quotient compatible with the original map on $M$. [quotetheorem:8338] This theorem is the practical test for constructing maps out of a quotient. Instead of defining a map on cosets and separately checking representatives, define a map upstairs on $M$ and verify that it vanishes on $N$. [example: Defining a Map out of a Quotient] Let $R=\mathbb{Z}$, let $M=\mathbb{Z}^2$, and let $N=\mathbb{Z}(2,3)$. Define $f:\mathbb{Z}^2\to \mathbb{Z}/5\mathbb{Z}$ by \begin{align*} f(x,y)=x+y+5\mathbb{Z}. \end{align*} This is a $\mathbb{Z}$-[module homomorphism](/page/Module%20Homomorphism): for $(x,y),(x',y')\in \mathbb{Z}^2$, \begin{align*} f((x,y)+(x',y'))=f(x+x',y+y')=(x+x')+(y+y')+5\mathbb{Z}. \end{align*} The right-hand side equals \begin{align*} (x+y)+(x'+y')+5\mathbb{Z}=f(x,y)+f(x',y'). \end{align*} For $n\in \mathbb{Z}$, \begin{align*} f(n(x,y))=f(nx,ny)=nx+ny+5\mathbb{Z}. \end{align*} Since $nx+ny=n(x+y)$, this is \begin{align*} n(x+y)+5\mathbb{Z}=n\,f(x,y). \end{align*} Now \begin{align*} f(2,3)=2+3+5\mathbb{Z}=5+5\mathbb{Z}. \end{align*} Because $5-0=5\in 5\mathbb{Z}$, we have \begin{align*} 5+5\mathbb{Z}=0+5\mathbb{Z}. \end{align*} Every element of $N$ has the form $m(2,3)$ with $m\in \mathbb{Z}$, and by $\mathbb{Z}$-linearity, \begin{align*} f(m(2,3))=m f(2,3)=m(0+5\mathbb{Z})=0+5\mathbb{Z}. \end{align*} Thus $N\subset \ker(f)$. By *Universal Property of the Quotient Module*, there is a unique homomorphism $\bar{f}:\mathbb{Z}^2/N\to \mathbb{Z}/5\mathbb{Z}$ satisfying $\bar{f}((x,y)+N)=f(x,y)$. Therefore \begin{align*} \bar{f}((x,y)+N)=x+y+5\mathbb{Z}. \end{align*} The formula does not depend on the representative: if $(x,y)+N=(x',y')+N$, then $(x,y)-(x',y')\in N$, so $f((x,y)-(x',y'))=0+5\mathbb{Z}$, and hence \begin{align*} f(x,y)-f(x',y')=0+5\mathbb{Z}. \end{align*} Equivalently, $f(x,y)=f(x',y')$ in $\mathbb{Z}/5\mathbb{Z}$. The quotient map is therefore exactly the operation of remembering $x+y$ modulo $5$ after collapsing all integer multiples of $(2,3)$. [/example] ### Kernels as Lost Information The universal property also explains why quotient modules and kernels are paired throughout algebra. Every homomorphism forgets exactly the elements in its kernel. The [first isomorphism theorem](/theorems/791) says that after collapsing this lost information, the remaining quotient is the same module as the image. [quotetheorem:862] This result says that a homomorphism is the same as two operations performed in sequence: collapse the kernel, then identify the quotient with the image. Quotients measure precisely the information retained by the map. ## Relations and Presentations Quotients become especially powerful when modules are described by generators and relations. A free module gives independent formal generators; a submodule of relations declares which linear combinations should become zero. The quotient is the module obtained after enforcing those relations. We need a name for this way of describing modules because it is the computational language of modules over rings. A presentation records both the free module of generators and the submodule of relations whose elements are to be killed. [definition: Module Presentation] Let $R$ be a ring and let $M$ be a left $R$-module. A presentation of $M$ consists of a free left $R$-module $F$ with chosen generators and a submodule $K\subset F$ of relations such that \begin{align*} M \cong F/K. \end{align*} When the chosen generators are indexed by a set $I$, one often writes the free module as $R^{(I)}$, meaning the [direct sum](/page/Direct%20Sum) of one copy of $R$ for each element of $I$. If the relations are generated by a set indexed by $J$, this presentation can later be encoded by a map from $R^{(J)}$ to $R^{(I)}$ whose image is $K$. [/definition] The elements indexed by $I$ are generators, and if $\varphi:F\to M$ is the chosen surjection from a free module $F\cong R^{(I)}$, then the kernel $A=\ker(\varphi)$ is the submodule of relations. After this definition, the quotient module appears as the last step of every presentation: the presented module is $R^{(I)}/A$. The next theorem justifies that this is not a special situation; every module can be reached by starting from a free module and quotienting by relations. [quotetheorem:9941] The theorem gives a constructive viewpoint: to build a module, start with too many independent generators and then quotient by the relations you want to impose. [example: Integer Matrix Presentation] Let $R=\mathbb{Z}$ and let $F=\mathbb{Z}^2$ with basis $e_1=(1,0)$ and $e_2=(0,1)$. The relations $2e_1=0$ and $3e_2=0$ are encoded by the homomorphism $A:\mathbb{Z}^2\to\mathbb{Z}^2$ with $A=\operatorname{diag}(2,3)$, meaning \begin{align*} A(m,n)=(2m,3n). \end{align*} In particular, \begin{align*} A(1,0)=(2,0)=2e_1. \end{align*} and \begin{align*} A(0,1)=(0,3)=3e_2. \end{align*} Thus \begin{align*} \operatorname{im}(A)=\{(2m,3n):m,n\in\mathbb{Z}\}. \end{align*} We show that the presented module $\mathbb{Z}^2/\operatorname{im}(A)$ is naturally isomorphic to $\mathbb{Z}/2\mathbb{Z}\oplus\mathbb{Z}/3\mathbb{Z}$. Define \begin{align*} \Phi\bigl((x,y)+\operatorname{im}(A)\bigr)=(x+2\mathbb{Z},y+3\mathbb{Z}). \end{align*} If $(x,y)+\operatorname{im}(A)=(x',y')+\operatorname{im}(A)$, then \begin{align*} (x,y)-(x',y')=(x-x',y-y')\in\operatorname{im}(A). \end{align*} So there are $m,n\in\mathbb{Z}$ with \begin{align*} (x-x',y-y')=(2m,3n). \end{align*} Hence $x-x'\in 2\mathbb{Z}$ and $y-y'\in 3\mathbb{Z}$, which gives \begin{align*} x+2\mathbb{Z}=x'+2\mathbb{Z}. \end{align*} and \begin{align*} y+3\mathbb{Z}=y'+3\mathbb{Z}. \end{align*} Therefore $\Phi$ is well-defined. The map $\Phi$ is surjective because, for any $a,b\in\mathbb{Z}$, \begin{align*} \Phi\bigl((a,b)+\operatorname{im}(A)\bigr)=(a+2\mathbb{Z},b+3\mathbb{Z}). \end{align*} Its kernel consists exactly of the zero coset. Indeed, if \begin{align*} \Phi\bigl((x,y)+\operatorname{im}(A)\bigr)=(0+2\mathbb{Z},0+3\mathbb{Z}), \end{align*} then $x\in 2\mathbb{Z}$ and $y\in 3\mathbb{Z}$, so $x=2m$ and $y=3n$ for some $m,n\in\mathbb{Z}$. Therefore \begin{align*} (x,y)=(2m,3n)=A(m,n)\in\operatorname{im}(A). \end{align*} Thus \begin{align*} (x,y)+\operatorname{im}(A)=0+\operatorname{im}(A). \end{align*} So $\Phi$ is an isomorphism: \begin{align*} \mathbb{Z}^2/\operatorname{im}(A)\cong \mathbb{Z}/2\mathbb{Z}\oplus\mathbb{Z}/3\mathbb{Z}. \end{align*} The first column of $A$ contributes the relation $2e_1=0$, the second column contributes the relation $3e_2=0$, and the quotient remembers exactly the residue class of the first coordinate modulo $2$ and the second coordinate modulo $3$. [/example] Matrix presentations are useful because row and column operations can simplify the relation submodule without changing the isomorphism type of the quotient. The cyclic case is the one-generator version of the same idea. [example: Cyclic Modules as Quotients] Let $R$ be a commutative ring and let $I \trianglelefteq R$ be an ideal. In the quotient module $R/I$, every element has the form $r+I$ for some $r\in R$. Since scalar multiplication is defined by multiplying representatives, we have \begin{align*} r(1+I)=r\cdot 1+I. \end{align*} Because $r\cdot 1=r$, this gives \begin{align*} r(1+I)=r+I. \end{align*} Thus every coset $r+I$ is an $R$-multiple of $1+I$, so $1+I$ generates $R/I$ as a left $R$-module. We now compute the annihilator of this generator. If $r\in I$, then $r-0=r\in I$, so \begin{align*} r+I=0+I. \end{align*} Using $r(1+I)=r+I$, this gives \begin{align*} r(1+I)=0+I. \end{align*} Conversely, if $r(1+I)=0+I$, then $r+I=0+I$. Equality of these cosets means that $r-0\in I$, hence $r\in I$. Therefore \begin{align*} r(1+I)=0+I \iff r\in I. \end{align*} So $R/I$ is the cyclic module generated by $1+I$, and the scalars that kill this generator are exactly the elements of the ideal $I$. [/example] The cyclic example leaves a classification question: if a module is built from finitely many generators, can its relations be simplified into a standard list of quotient pieces? Over a [principal ideal domain](/page/Principal%20Ideal%20Domain) the answer is yes: the structure theorem for finitely generated modules says that the torsion part breaks into quotient modules of the form $R/(a)$. That classification is deeper than the basic quotient construction, so here it serves mainly as a signpost. It shows that even when the main problem is classification rather than construction, quotient modules remain the standard blocks from which finitely generated modules are assembled. ## Subquotients and Isomorphism Theorems ### Quotienting in Stages Quotients rarely appear alone. Algebra often compares several submodules at once: a submodule inside another submodule, a sum $A+B$, or an intersection $A \cap B$. The isomorphism theorems explain how quotienting interacts with these operations. The first recurring construction is a quotient of a quotient. If $N \subset P \subset M$, then collapsing $N$ first and then collapsing $P/N$ should give the same result as collapsing $P$ all at once. The theorem makes this staged collapse independent of representatives. [remark: Quotienting in Stages] Let $M$ be an $R$-module, and let $N \subset P \subset M$ be submodules. Then $P/N$ is a submodule of $M/N$, and the staged quotient is described by a natural map from $(M/N)/(P/N)$ to $M/P$. On representatives, this map sends $(m+N)+(P/N)$ to $m+P$. This is the module version of the third isomorphism theorem. [/remark] The theorem says that quotienting is compatible with doing collapses in stages. It prevents quotient notation from becoming ambiguous when a chain of submodules is present. ### Images of Submodules in Quotients Another common comparison starts with a submodule $A \subset M$ and asks what part of $A$ remains visible after quotienting $M$ by another submodule $B$. The answer should depend only on the overlap $A \cap B$, because that is precisely the part of $A$ killed by the quotient map $M \to M/B$. [remark: Images of Submodules in Quotients] Let $A$ and $B$ be submodules of an $R$-module $M$. The image of $A$ under the quotient map $M\to M/B$ is $(A+B)/B$, while the elements of $A$ killed by that map are exactly $A\cap B$. The resulting correspondence is a natural map from $A/(A\cap B)$ to $(A+B)/B$. On representatives, this map sends $a+(A\cap B)$ to $a+B$. This is the module version of the second isomorphism theorem. [/remark] This result is often the cleanest way to compute a quotient. Instead of describing the image of $A$ inside $M/B$ directly, it identifies that image as a quotient of $A$ by the overlap $A \cap B$. [example: A Subquotient in $\mathbb{Z}^2$] Let $M=\mathbb{Z}^2$, let $A=\mathbb{Z}(1,0)$, and let $B=\mathbb{Z}(1,1)$. We compute the part of the horizontal axis that is killed when passing to the quotient by the diagonal submodule $B$. An element of $A$ has the form $n(1,0)=(n,0)$ with $n\in\mathbb{Z}$, and an element of $B$ has the form $m(1,1)=(m,m)$ with $m\in\mathbb{Z}$. If $(n,0)\in A\cap B$, then there is some $m\in\mathbb{Z}$ such that \begin{align*} (n,0)=(m,m). \end{align*} Equality in $\mathbb{Z}^2$ means equality of coordinates, so \begin{align*} n=m. \end{align*} and \begin{align*} 0=m. \end{align*} Hence $m=0$ and then $n=0$, so \begin{align*} (n,0)=(0,0). \end{align*} Conversely, $(0,0)=0(1,0)\in A$ and $(0,0)=0(1,1)\in B$, so $(0,0)\in A\cap B$. Therefore \begin{align*} A\cap B=\{(0,0)\}=0. \end{align*} By *Second Isomorphism Theorem for Modules*, \begin{align*} A/(A\cap B)\cong (A+B)/B. \end{align*} Using $A\cap B=0$, this becomes \begin{align*} A/(A\cap B)=A/0. \end{align*} The map $A\to A/0$ sends $(n,0)$ to $(n,0)+0$, and no nonzero element is killed because \begin{align*} (n,0)+0=(0,0)+0 \end{align*} holds exactly when \begin{align*} (n,0)-(0,0)=(n,0)\in 0. \end{align*} This forces $n=0$. Thus $A/0\cong A$, and $A\cong \mathbb{Z}$ by the map $n(1,0)\mapsto n$. Consequently, \begin{align*} (A+B)/B\cong \mathbb{Z}. \end{align*} The image of the horizontal axis in $\mathbb{Z}^2/B$ is therefore still a copy of $\mathbb{Z}$: quotienting by the diagonal direction kills no nonzero horizontal vector. [/example] ## Exact Sequences and Homological Viewpoint ### Quotients as Cokernels Quotient modules are the last term in every short exact sequence. This matters because exact sequences turn algebraic constructions into a language of kernels, images, extensions, and obstructions. A quotient $M/N$ is not just a module; it is the cokernel of the inclusion $N \hookrightarrow M$. The next definition names this terminal role. It is useful because many constructions in homological algebra are defined by universal properties of kernels and cokernels rather than by elements, so quotient modules become part of a larger categorical language. [definition: Cokernel of a Module Homomorphism] Let $R$ be a ring, let $f: M \to P$ be an $R$-linear map of left $R$-modules. The cokernel of $f$ is the quotient module \begin{align*} \operatorname{coker}(f) = P/\operatorname{im}(f). \end{align*} [/definition] When $f$ is an inclusion $N \hookrightarrow M$, the cokernel is exactly $M/N$. To use this observation systematically, we need a compact way to record an inclusion, a projection, and the assertion that the projection has exactly that inclusion as its kernel. [definition: Short Exact Sequence of Modules] Let $R$ be a ring. A short exact sequence of left $R$-modules is a sequence of $R$-linear maps \begin{align*} 0 \to A \xrightarrow{i} B \xrightarrow{p} C \to 0 \end{align*} such that $i$ is injective, $p$ is surjective, and $\operatorname{im}(i)=\ker(p)$. [/definition] Short exact sequences are the right language for saying that $C$ is obtained from $B$ by collapsing the submodule $i(A)$. The point that needs checking is not just that $p$ kills $i(A)$, but that it kills exactly those elements and loses no further information. At the module level, this means that $p$ factors through the quotient $B/i(A)$ and gives an isomorphism $B/i(A)\cong C$. Thus every quotient map gives a short exact sequence \begin{align*} 0 \to N \to M \to M/N \to 0, \end{align*} and every short exact sequence of modules can be read as a quotient description of its last term. This viewpoint is the doorway into homological algebra. Many questions about quotients become questions about whether a short exact sequence splits, how functors behave on it, or what extension class it represents. ### Splitting and Complements A quotient map may or may not admit a linear choice of representatives. When such a choice exists, the quotient sits inside the original module as a complementary summand. When it does not, the quotient still exists but carries extension information that no choice of submodule can remove. [definition: Split Short Exact Sequence] Let $R$ be a ring. A short exact sequence \begin{align*} 0 \to A \xrightarrow{i} B \xrightarrow{p} C \to 0 \end{align*} of left $R$-modules is split if there exists an $R$-linear map $s: C \to B$ such that \begin{align*} p \circ s = \operatorname{id}_C. \end{align*} [/definition] A split quotient behaves like a direct summand: collapsing $A$ loses a component that can be independently recovered inside $B$. The question is whether this recovery can be realised by an actual submodule of representatives inside the original module. That requires more than surjectivity of the quotient map: it requires a complementary submodule meeting the collapsed part only in zero and together spanning the whole module. [quotetheorem:4529] The condition is strong. Vector spaces always have complements, but modules over general rings may not. Quotients therefore carry extension information that cannot always be replaced by a chosen submodule of representatives. [example: A Quotient with No Complement] Consider the $\mathbb{Z}$-module $M=\mathbb{Z}$ and the submodule $N=2\mathbb{Z}$. The quotient has exactly two cosets: if $m=2q$, then $m+2\mathbb{Z}=0+2\mathbb{Z}$, and if $m=2q+1$, then $m-1=2q\in 2\mathbb{Z}$, so $m+2\mathbb{Z}=1+2\mathbb{Z}$. Thus $\mathbb{Z}/2\mathbb{Z}$ records only the parity of an integer. Suppose, for contradiction, that $2\mathbb{Z}$ has a complement $P\subset \mathbb{Z}$, so \begin{align*} \mathbb{Z}=2\mathbb{Z}\oplus P. \end{align*} The quotient map $\pi:\mathbb{Z}\to \mathbb{Z}/2\mathbb{Z}$, $\pi(n)=n+2\mathbb{Z}$, restricts to a map $\pi|_P:P\to \mathbb{Z}/2\mathbb{Z}$. This restriction is injective: if $p\in P$ and $\pi(p)=0+2\mathbb{Z}$, then $p\in 2\mathbb{Z}$, so $p\in P\cap 2\mathbb{Z}$. Since the sum is direct, $P\cap 2\mathbb{Z}=\{0\}$, hence $p=0$. The restriction is also surjective. Since $\mathbb{Z}=2\mathbb{Z}+P$, we can write \begin{align*} 1=2q+p \end{align*} for some $q\in\mathbb{Z}$ and $p\in P$. Passing to cosets gives \begin{align*} 1+2\mathbb{Z}=(2q+p)+2\mathbb{Z}=p+2\mathbb{Z}. \end{align*} Thus $\pi|_P$ is an isomorphism $P\cong \mathbb{Z}/2\mathbb{Z}$. Let $p\in P$ be the element with $p+2\mathbb{Z}=1+2\mathbb{Z}$. Since $2(1+2\mathbb{Z})=2+2\mathbb{Z}=0+2\mathbb{Z}$, the isomorphism forces \begin{align*} 2p=0. \end{align*} But $p$ is an integer, and $2p=0$ in $\mathbb{Z}$ implies $p=0$. Then $p+2\mathbb{Z}=0+2\mathbb{Z}$, contradicting $p+2\mathbb{Z}=1+2\mathbb{Z}$. Therefore no such complement $P$ exists. Consequently the short exact sequence \begin{align*} 0 \to 2\mathbb{Z} \to \mathbb{Z} \to \mathbb{Z}/2\mathbb{Z} \to 0 \end{align*} does not split. The quotient exists, but it cannot be realised as a complementary submodule inside $\mathbb{Z}$. [/example] ## Quotients in Representation Theory Modules appear not only over rings but also as representations of algebras. For a Lie algebra $\mathfrak{g}$, a $\mathfrak{g}$-module is a [vector space](/page/Vector%20Space) on which elements of $\mathfrak{g}$ act linearly and compatibly with the Lie bracket. Quotients let us collapse invariant subspaces while preserving the action. To form a quotient representation, the subspace being collapsed must be stable under the action. Otherwise the formula $x\cdot(v+W)=x\cdot v+W$ would depend on the chosen representative, because changing $v$ by an element of $W$ might change $x\cdot v$ by something outside $W$. [definition: Quotient $\mathfrak{g}$-Module] Let $\mathfrak{g}$ be a Lie algebra over a field $k$, let $V$ be a $\mathfrak{g}$-module with representation $\rho: \mathfrak{g} \to \mathfrak{gl}(V)$, and let $W \subset V$ be a $\mathfrak{g}$-submodule. The quotient $\mathfrak{g}$-module $V/W$ is the vector space quotient equipped with the representation $\bar{\rho}: \mathfrak{g} \to \mathfrak{gl}(V/W)$, where \begin{align*} \bar{\rho}(x)(v+W)=\rho(x)(v)+W \end{align*} for all $x \in \mathfrak{g}$ and $v \in V$. [/definition] This is the same quotient principle in representation-theoretic clothing. The submodule condition means $\rho(x)(W) \subset W$ for every $x \in \mathfrak{g}$. That condition makes the formula on cosets independent of representatives and lets the representation identity descend. [remark: Quotient Representations] If $v+W=v'+W$, then $v-v'\in W$. Since $W$ is stable under the action, $\rho(x)(v-v')\in W$, so \begin{align*} \rho(x)(v)+W=\rho(x)(v')+W. \end{align*} Thus each $\bar{\rho}(x)$ is well-defined on $V/W$. The identity \begin{align*} \bar{\rho}([x,y])=[\bar{\rho}(x),\bar{\rho}(y)] \end{align*} then follows by applying the corresponding identity in $V$ and passing to cosets. [/remark] Quotient representations are central in composition series: if a representation has a nonzero proper submodule, quotienting by it produces a smaller representation and exposes another layer. [example: Quotient of a Two-Dimensional Lie Algebra Module] Let $\mathfrak{g}=kx$ be the one-dimensional Lie algebra over $k$ with $[x,x]=0$, and let $V=k^2$ have basis $e_1,e_2$. Define $\rho(x)$ on basis vectors by \begin{align*} \rho(x)(e_1)=0, \qquad \rho(x)(e_2)=e_1. \end{align*} This determines a linear map $\rho(x):V\to V$. Since $[x,x]=0$, the representation identity on the basis element $x$ reads \begin{align*} \rho([x,x])=\rho(0)=0. \end{align*} Also, \begin{align*} [\rho(x),\rho(x)]=\rho(x)\rho(x)-\rho(x)\rho(x)=0. \end{align*} Thus $\rho([x,x])=[\rho(x),\rho(x)]$, and bilinearity gives a representation of the one-dimensional Lie algebra $\mathfrak{g}$. Let $W=ke_1$. To check that $W$ is a $\mathfrak{g}$-submodule, take $w=ae_1\in W$ and $cx\in \mathfrak{g}$. By linearity of the action, \begin{align*} \rho(cx)(w)=c\rho(x)(ae_1)=ca\rho(x)(e_1)=ca\cdot 0=0. \end{align*} Since $0\in W$, the subspace $W$ is stable under the action. Every vector $v\in V$ has the form $v=ae_1+be_2$ with $a,b\in k$. Because $ae_1\in W$, its coset is zero in $V/W$: \begin{align*} ae_1+W=0+W. \end{align*} Hence \begin{align*} v+W=(ae_1+be_2)+W=be_2+W=b(e_2+W). \end{align*} Thus $V/W$ is generated by $e_2+W$. This generator is nonzero: if $e_2+W=0+W$, then $e_2\in W$, so $e_2=ae_1$ for some $a\in k$, contradicting the [linear independence](/page/Linear%20Independence) of $e_1,e_2$. Therefore $V/W$ is one-dimensional. The induced quotient action is computed on the generator by applying $\rho(x)$ in $V$ and then passing to the coset: \begin{align*} \bar{\rho}(x)(e_2+W)=\rho(x)(e_2)+W=e_1+W. \end{align*} Since $e_1\in W$, we have \begin{align*} e_1+W=0+W. \end{align*} Therefore \begin{align*} \bar{\rho}(x)(e_2+W)=0+W. \end{align*} For any $b\in k$, \begin{align*} \bar{\rho}(x)(b(e_2+W))=b\bar{\rho}(x)(e_2+W)=b(0+W)=0+W. \end{align*} So $\bar{\rho}(x)$ is the zero operator on $V/W$, and then $\bar{\rho}(cx)=c\bar{\rho}(x)=0$ for every $cx\in\mathfrak{g}$. The quotient action is therefore the zero action, while the original action on $V$ is not zero because $\rho(x)(e_2)=e_1\ne 0$. [/example] ## Beyond and Connected Topics Quotient modules sit at the intersection of elementary algebra, commutative algebra, representation theory, and homological algebra. The same construction reappears whenever mathematics replaces objects by equivalence classes while preserving linear structure. In [Cambridge IB Groups, Rings and Modules](/page/Cambridge%20IB%20Groups%2C%20Rings%20and%20Modules), quotient modules belong with quotient groups and quotient rings as part of the general language of algebraic objects modulo stable subobjects. The module version is often cleaner than the group version because every submodule is automatically normal in the additive group. In [Cambridge III Commutative Algebra](/page/Cambridge%20III%20Commutative%20Algebra), quotient modules become local objects. Modules such as $M/\mathfrak{p}M$, $M/IM$, and localised quotients measure fibres, supports, and infinitesimal structure over rings. Exactness is the operational reason this works so often: localization preserves quotients, and tensoring converts suitable presentations into quotients such as $(M/N)\otimes_R S \cong (M\otimes_R S)/\operatorname{im}(N\otimes_R S)$ when the induced map records the relations after base change. Ideals and quotient rings are special cases of the same mechanism. In [Homological Algebra I: Complexes and Resolutions](/page/Homological%20Algebra%20I%3A%20Complexes%20and%20Resolutions), quotient modules are encoded as cokernels. Exactness, derived functors, projective resolutions, $\operatorname{Tor}_n^R(M,N)$, and $\operatorname{Ext}_R^n(M,N)$ all depend on understanding how kernels and quotients interact. In [Lie Algebras I: Foundations](/page/Lie%20Algebras%20I%3A%20Foundations), quotient modules appear as quotient representations. Submodules, simple modules, composition factors, and Jordan-Hölder theory all rely on repeatedly passing from a module to a quotient by an invariant subspace. ## References Androma, [Cambridge IB Groups, Rings and Modules](/page/Cambridge%20IB%20Groups%2C%20Rings%20and%20Modules). Androma, [Cambridge III Commutative Algebra](/page/Cambridge%20III%20Commutative%20Algebra). Androma, [Homological Algebra I: Complexes and Resolutions](/page/Homological%20Algebra%20I%3A%20Complexes%20and%20Resolutions). Androma, [Lie Algebras I: Foundations](/page/Lie%20Algebras%20I%3A%20Foundations). Atiyah and Macdonald, *Introduction to Commutative Algebra* (1969). Dummit and Foote, *Abstract Algebra* (2004). Lang, *Algebra* (2002).