7.1 Rings, Ideals and Quotient Rings
Two operations at once: rings and subrings, units, zero divisors, nilpotent and idempotent elements, integral domains and fields, the characteristic, ideals and quotient rings, ring homomorphisms and the isomorphism theorems — with $\mathbb{Z}_n$, $\mathbb{Z}[i]$ and matrix rings as the running examples.
A group has one operation. Most of the number systems we actually compute in have two: we add and multiply integers, residues modulo $n$, polynomials, matrices, Gaussian integers, continuous functions. A ring is the structure that keeps both operations and the distributive law that ties them together. Once the axioms are fixed, the arithmetic of $\mathbb{Z}$ splits into many separate phenomena that the integers hide: elements with inverses (units) and elements that multiply to zero without being zero (zero divisors); elements some power of which vanishes (nilpotents) and elements equal to their own square (idempotents). In $\mathbb{Z}_{12}$ all four kinds appear at once, and learning to sort the elements of a ring into these classes is the first skill this section teaches.
The second half of the section is the ring version of Chapter 6. Normal subgroups were exactly the subgroups that can be "divided out"; for rings the corresponding objects are ideals, and the quotient $R/I$ is again a ring. Ring homomorphisms, their kernels and the isomorphism theorems then let us identify quotients such as $\mathbb{Z}[x]/\langle x^2+1\rangle\cong\mathbb{Z}[i]$ or $\mathbb{Z}[i]/\langle2+i\rangle\cong\mathbb{Z}_5$ without computing a single multiplication table. CSIR NET tests this material with short computations (how many nilpotents in $\mathbb{Z}_{n}$, how many ring homomorphisms $\mathbb{Z}_m\to\mathbb{Z}_n$, what $\mathbb{Z}[i]/\langle\alpha\rangle$ is) and with "which of the following is an integral domain / an ideal / a field" questions whose false options are standard counterexamples. Everything here is used again in §7.2 (prime and maximal ideals) and in Chapter 8 (fields as quotients $F[x]/\langle f\rangle$).
Rings: Axioms and First Examples
A ring $(R,+,\cdot)$ is a set $R$ with two binary operations such that
- $(R,+)$ is an abelian group, with identity $0$ and the inverse of $a$ written $-a$;
- multiplication is associative: $(ab)c=a(bc)$;
- both distributive laws hold: $a(b+c)=ab+ac$ and $(a+b)c=ac+bc$.
$R$ is commutative if $ab=ba$ for all $a,b$. $R$ is a ring with identity (unity) if there is $1\in R$ with $1a=a1=a$ for all $a$. Unless we say otherwise, a ring need not be commutative and need not have an identity; from §7.2 on, "ring" will mean commutative ring with $1\ne0$.
The examples to keep in mind, and the features that make each one useful:
Let $R$ be a ring and $a,b\in R$. Then
- $0a=a0=0$;
- $(-a)b=a(-b)=-(ab)$ and $(-a)(-b)=ab$;
- $(ma)b=a(mb)=m(ab)$ for every integer $m$, where $ma$ means $a+\dots+a$ ($m$ terms) for $m\gt0$;
- if $R$ has an identity, it is unique, $(-1)a=-a$, and $1=0$ happens only in the zero ring $R=\{0\}$.
Proof. (1) $0a=(0+0)a=0a+0a$; adding $-(0a)$ to both sides gives $0=0a$. Similarly $a0=0$. (2) $ab+(-a)b=(a+(-a))b=0b=0$, so $(-a)b$ is the additive inverse of $ab$; the same for $a(-b)$. Then $(-a)(-b)=-(a(-b))=-(-(ab))=ab$. (3) follows from the distributive law by induction on $m\gt0$, and from (2) for $m\lt0$. (4) If $1$ and $1'$ are identities then $1=1\cdot1'=1'$. $(-1)a=-(1a)=-a$ by (2). If $1=0$ then $a=1a=0a=0$ for every $a$. ∎
The rule "minus times minus is plus" is therefore not a convention: it is forced by the distributive law. Part (4) is why we insist on $1\ne0$ for integral domains and fields — the zero ring is a ring with identity, but a useless one.
Proof. If $S$ is a subring, it is closed under subtraction and multiplication. Conversely, closure under subtraction makes $(S,+)$ a subgroup of $(R,+)$ by the one-step subgroup test (§6.1), and it is abelian because $(R,+)$ is. Closure under multiplication makes $\cdot$ an operation on $S$, and associativity and the distributive laws are inherited from $R$, since they hold for all elements of $R$. ∎
Examples: $n\mathbb{Z}$ is a subring of $\mathbb{Z}$; $\mathbb{Z}[i]$ is a subring of $\mathbb{C}$ because $(a+bi)(c+di)=(ac-bd)+(ad+bc)i$; the upper triangular matrices form a subring of $M_2(\mathbb{R})$; the constant polynomials form a subring of $R[x]$. The set $\{(a,0):a\in\mathbb{Z}\}$ is a subring of $\mathbb{Z}\times\mathbb{Z}$ with its own identity $(1,0)$, different from the identity $(1,1)$ of $\mathbb{Z}\times\mathbb{Z}$ — so it is a subring that has an identity but is not a unital subring. Similarly $\left\{\begin{pmatrix}a&0\\0&0\end{pmatrix}\right\}$ is a copy of $\mathbb{R}$ inside $M_2(\mathbb{R})$ with identity $E_{11}\ne I$.
Let $R$ be a ring with identity $1\ne0$.
- $u\in R$ is a unit if there is $v\in R$ with $uv=vu=1$. The units form a group under multiplication, the unit group $R^\times$. For $R=\mathbb{Z}_n$ this is the group $U(n)$ of Chapter 6.
- A nonzero $a\in R$ is a zero divisor if there is a nonzero $b\in R$ with $ab=0$ or $ba=0$.
- $a$ is nilpotent if $a^k=0$ for some $k\ge1$; the least such $k$ is the index of nilpotency.
- $e$ is idempotent if $e^2=e$. Every ring has the trivial idempotents $0$ and $1$.
(The definitions of zero divisor, nilpotent and idempotent make sense without an identity too.)
Two facts hold in every ring with identity. A unit is never a zero divisor: if $uv=1$ and $ub=0$ then $b=(vu)b=v(ub)=0$. And a nonzero nilpotent is always a zero divisor: if $a^k=0$ with $k$ minimal, then $k\ge2$ and $a\cdot a^{k-1}=0$ with $a^{k-1}\ne0$. A nontrivial idempotent $e$ is also a zero divisor, because $e(1-e)=e-e^2=0$ and $1-e\ne0$.
- In $\mathbb{Z}_n$ ($n\ge2$), $\bar a$ is a unit if and only if $\gcd(a,n)=1$, and every nonzero non-unit is a zero divisor. Hence $\mathbb{Z}_n$ has $\varphi(n)$ units and $n-1-\varphi(n)$ zero divisors.
- $\bar a$ is nilpotent in $\mathbb{Z}_n$ if and only if every prime divisor of $n$ divides $a$, i.e. $\operatorname{rad}(n)\mid a$, where $\operatorname{rad}(n)=p_1p_2\cdots p_k$ is the product of the distinct primes dividing $n$. So $\mathbb{Z}_n$ has exactly $n/\operatorname{rad}(n)$ nilpotent elements (counting $0$).
- More generally, in a finite commutative ring $R$ with identity, every nonzero element is either a unit or a zero divisor.
Proof. (1) If $\gcd(a,n)=1$, Bézout's identity (§5.2) gives $ax+ny=1$, so $\bar a\,\bar x=\bar1$. If $d=\gcd(a,n)\gt1$ and $\bar a\ne0$, then $b=n/d$ satisfies $0\lt b\lt n$ and $ab=(a/d)\,n\equiv0$, so $\bar a$ is a zero divisor, hence not a unit. The count follows from the definition of $\varphi$ (§5.3).
(2) If $\operatorname{rad}(n)\mid a$, write $n=p_1^{e_1}\cdots p_k^{e_k}$ and $e=\max e_i$; then $p_i^{e}\mid a^{e}$ for each $i$, so $n\mid a^{e}$ and $\bar a^{\,e}=0$. Conversely, if $n\mid a^m$ and $p\mid n$ is prime, then $p\mid a^m$, so $p\mid a$ by Euclid's lemma. The multiples of $r=\operatorname{rad}(n)$ in $\{0,\dots,n-1\}$ are $0,r,2r,\dots$, and there are $n/r$ of them.
(3) Let $a\ne0$ be a non-zero-divisor. The map $\lambda_a:R\to R$, $x\mapsto ax$, is injective, since $ax=ay$ gives $a(x-y)=0$ and so $x=y$. An injective map of a finite set to itself is surjective, so $ax=1$ for some $x$, and $a$ is a unit. ∎
Part (3) fails for infinite rings: $2$ is neither a unit nor a zero divisor in $\mathbb{Z}$. Idempotents of $\mathbb{Z}_n$ are counted in Theorem 7.18 below with the Chinese remainder theorem: there are $2^k$ of them, where $k$ is the number of distinct primes dividing $n$.
Integral Domains, Fields and the Characteristic
- An integral domain is a commutative ring with identity $1\ne0$ and no zero divisors: $ab=0\Rightarrow a=0$ or $b=0$.
- A division ring (skew field) is a ring with identity $1\ne0$ in which every nonzero element is a unit.
- A field is a commutative division ring.
So $\mathbb{Z}$ and $\mathbb{Z}[i]$ are integral domains but not fields, $\mathbb{Q},\mathbb{R},\mathbb{C},\mathbb{Z}_p$ are fields, and $\mathbb{H}$ is a division ring that is not a field. Every subring of a field that contains $1$ is an integral domain (a zero divisor in the subring would be one in the field). Every field is an integral domain, because a unit is not a zero divisor.
Proof. If $R$ is a domain and $ab=ac$, then $a(b-c)=0$ with $a\ne0$, so $b-c=0$. Conversely, if cancellation holds and $ab=0=a0$ with $a\ne0$, cancelling $a$ gives $b=0$. ∎
Proof. In a finite domain no nonzero element is a zero divisor, so by Theorem 7.3(3) every nonzero element is a unit. For $\mathbb{Z}_n$: if $n=p$ is prime, every $a\not\equiv0$ has $\gcd(a,p)=1$ and is a unit. If $n=rs$ with $1\lt r,s\lt n$, then $\bar r\,\bar s=0$ with $\bar r,\bar s\ne0$, so $\mathbb{Z}_n$ is not a domain. ∎
An infinite domain need not be a field ($\mathbb{Z}$), and finiteness is genuinely needed: the proof uses that an injective self-map of a finite set is onto, which fails for $x\mapsto2x$ on $\mathbb{Z}$.
- $N(\alpha\beta)=N(\alpha)N(\beta)$ for all $\alpha,\beta\in\mathbb{Z}[\sqrt d]$.
- $\alpha$ is a unit of $\mathbb{Z}[\sqrt d]$ if and only if $N(\alpha)=\pm1$.
- The units of $\mathbb{Z}[i]$ are $\pm1,\pm i$; for $d\le-2$ the only units of $\mathbb{Z}[\sqrt d]$ are $\pm1$. For $d\ge2$ there may be infinitely many units: in $\mathbb{Z}[\sqrt2]$ all the powers $(1+\sqrt2)^n$, $n\in\mathbb{Z}$, are distinct units.
Proof. (1) Conjugation $a+b\sqrt d\mapsto a-b\sqrt d$ respects products: writing $\alpha=a+b\sqrt d$, $\beta=c+e\sqrt d$, we have $\alpha\beta=(ac+dbe)+(ae+bc)\sqrt d$, whose conjugate $(ac+dbe)-(ae+bc)\sqrt d$ equals $\bar\alpha\bar\beta$ by direct expansion. Hence $N(\alpha\beta)=\alpha\beta\,\overline{\alpha\beta}=\alpha\bar\alpha\,\beta\bar\beta$.
(2) If $\alpha\beta=1$ then $N(\alpha)N(\beta)=N(1)=1$ in $\mathbb{Z}$, so $N(\alpha)=\pm1$. Conversely, if $N(\alpha)=\alpha\bar\alpha=\pm1$, then $\alpha\cdot(\pm\bar\alpha)=1$ and $\pm\bar\alpha\in\mathbb{Z}[\sqrt d]$.
(3) For $d=-1$: $a^2+b^2=1$ forces $(a,b)=(\pm1,0)$ or $(0,\pm1)$. For $d\le-2$: $a^2+|d|b^2=1$ forces $b=0$, $a=\pm1$. For $d=2$: $N(1+\sqrt2)=1-2=-1$, so $1+\sqrt2$ is a unit, and since $1+\sqrt2\gt1$ its powers are strictly increasing real numbers, hence distinct. ∎
The norm is the main computational tool for $\mathbb{Z}[\sqrt d]$: it turns questions about divisibility of algebraic integers into questions about divisibility of ordinary integers. We use it heavily in §7.2.
Proof. The determinant is multiplicative over any commutative ring ($\det(AB)=\det A\det B$ is a polynomial identity in the entries, true over $\mathbb{Z}[\text{entries}]$ and hence in every commutative ring), and $\det I=1$. If $AB=I$ then $\det A\det B=1$, so $\det A\in R^\times$. Conversely, the adjugate matrix $\operatorname{adj}A$ (transposed cofactors, whose entries are polynomials in the entries of $A$ with integer coefficients) satisfies $A\operatorname{adj}A=\operatorname{adj}A\,A=(\det A)I$ over any commutative ring. If $\det A=u$ is a unit, $B=u^{-1}\operatorname{adj}A$ satisfies $AB=BA=I$. ∎
For example $\begin{pmatrix}2&1\\1&1\end{pmatrix}$ has determinant $1$ and is invertible over $\mathbb{Z}$, while $\begin{pmatrix}2&0\\0&1\end{pmatrix}$ is invertible over $\mathbb{Q}$ but not over $\mathbb{Z}$, and not over $\mathbb{Z}_4$ (its determinant $2$ is a zero divisor there). Over a field, a matrix is invertible iff $\det A\ne0$, and a nonzero singular matrix $A$ is a zero divisor: if $Av=0$ with $v\ne0$, the matrix $B$ whose columns all equal $v$ satisfies $AB=0$. So in $M_n(F)$ every nonzero element is a unit or a zero divisor — the same dichotomy as in finite commutative rings, although $M_n(F)$ is infinite when $F$ is.
- If $R$ has an identity, $\operatorname{char}R$ is the additive order of $1$ (and $0$ if that order is infinite).
- The characteristic of an integral domain is $0$ or a prime. In particular every field has characteristic $0$ or a prime $p$, and a finite field has prime characteristic.
- $\operatorname{char}(R\times S)=\operatorname{lcm}(\operatorname{char}R,\operatorname{char}S)$ when both are positive, and $0$ if either is $0$.
- Freshman's dream. If $R$ is commutative of prime characteristic $p$, then $(a+b)^p=a^p+b^p$ for all $a,b\in R$, so $a\mapsto a^p$ (the Frobenius map) respects addition and multiplication.
Proof. (1) If $n\cdot1=0$ then $na=(n\cdot1)a=0$ for every $a$ (Theorem 7.1(3)); conversely $na=0$ for all $a$ includes $a=1$. So the least $n$ that kills every element is the least $n$ that kills $1$.
(2) Let $R$ be a domain with $\operatorname{char}R=n\gt0$. Since $1\ne0$, $n\ge2$. If $n=rs$ with $1\lt r,s\lt n$, then $0=n\cdot1=(r\cdot1)(s\cdot1)$, so $r\cdot1=0$ or $s\cdot1=0$, contradicting the minimality of $n$. A finite ring cannot have characteristic $0$, since the multiples $1,2\cdot1,3\cdot1,\dots$ must repeat.
(3) $m(a,b)=(ma,mb)=0$ for all $(a,b)$ iff $\operatorname{char}R\mid m$ and $\operatorname{char}S\mid m$ (when the characteristics are positive, the set of such $m$ is a subgroup $c\mathbb{Z}$ of $\mathbb{Z}$, so "kills everything" is the same as "is a multiple of the characteristic"). The least such $m\gt0$ is the lcm; if one ring has characteristic $0$ there is no such $m$.
(4) By the binomial theorem (valid in any commutative ring), $(a+b)^p=\sum_{k=0}^{p}\binom pk a^kb^{p-k}$. For $0\lt k\lt p$ the prime $p$ divides $\binom pk=\dfrac{p!}{k!(p-k)!}$, because $p$ divides the numerator and not the denominator. These terms are integer multiples of $p\cdot(\text{something})$, hence $0$. Multiplicativity $(ab)^p=a^pb^p$ is commutativity. ∎
Characteristic and size are independent: $\mathbb{Z}_p[x]$ and the field $\mathbb{Z}_p(x)$ of rational functions are infinite of characteristic $p$, while $\mathbb{Z}_2\times\mathbb{Z}_3$ is finite of characteristic $6$ (not a domain, which is consistent with (2)).
Ideals and Quotient Rings
For groups, the kernels of homomorphisms are exactly the normal subgroups, and these are what we can divide by. For rings, the kernel $K$ of a homomorphism $\phi$ has an extra property: if $\phi(a)=0$ then $\phi(ra)=\phi(r)\cdot0=0$ for every $r\in R$, not just for $r\in K$. A kernel absorbs multiplication by arbitrary ring elements. This absorption property defines ideals.
Let $R$ be a ring. An additive subgroup $I$ of $R$ is a left ideal if $ra\in I$ for all $r\in R$, $a\in I$; a right ideal if $ar\in I$ for all $r\in R$, $a\in I$; and an ideal (two-sided ideal), written $I\trianglelefteq R$, if it is both. In a commutative ring the three notions coincide. $\{0\}$ and $R$ are ideals of every ring; an ideal $I\ne R$ is proper.
If $R$ is commutative with identity and $a_1,\dots,a_k\in R$, the ideal generated by them is $\langle a_1,\dots,a_k\rangle=\{r_1a_1+\dots+r_ka_k:r_i\in R\}$. An ideal of the form $\langle a\rangle=aR$ is principal.
- A nonempty subset $I\subseteq R$ is an ideal iff $a-b\in I$, $ra\in I$ and $ar\in I$ for all $a,b\in I$, $r\in R$. Every ideal is a subring.
- If $R$ is commutative with identity, $\langle a_1,\dots,a_k\rangle$ is an ideal containing $a_1,\dots,a_k$, and it is the smallest such ideal: it is contained in every ideal that contains $a_1,\dots,a_k$.
Proof. (1) The first condition makes $I$ an additive subgroup (§6.1, one-step test), and the other two are the absorption conditions. Absorption with $r\in I$ gives closure under multiplication, so $I$ is a subring by Theorem 7.2. (2) $\sum r_ia_i-\sum s_ia_i=\sum(r_i-s_i)a_i$ and $r\sum r_ia_i=\sum(rr_i)a_i$, so the set is an ideal; it contains $a_j$ (take $r_j=1$, the other $r_i=0$). Any ideal $J\ni a_1,\dots,a_k$ contains each $r_ia_i$ by absorption and hence each finite sum. ∎
Without an identity the formula changes: in $2\mathbb{Z}$ the ideal generated by $2$ is $2\mathbb{Z}$, but $\{r\cdot2:r\in2\mathbb{Z}\}=4\mathbb{Z}$ misses $2$ itself; the correct description is $\{ra+na:r\in R,\ n\in\mathbb{Z}\}$.
- Every ideal of $\mathbb{Z}$ is principal: it equals $m\mathbb{Z}$ for a unique $m\ge0$.
- The ideals of $\mathbb{Z}_n$ are the principal ideals $\langle\bar d\rangle$ for the positive divisors $d$ of $n$; they are distinct, $\langle\bar d\rangle$ has $n/d$ elements, and $\mathbb{Z}_n$ has exactly $\tau(n)$ ideals. Moreover $\langle\bar d\rangle\subseteq\langle\bar e\rangle$ iff $e\mid d$.
Proof. In $\mathbb{Z}$ and in $\mathbb{Z}_n$ every element $r$ of the ring is an integer multiple of $1$, so $ra=a+\dots+a$ (or its negative) lies in every additive subgroup containing $a$. Hence the ideals are exactly the additive subgroups. (1) The subgroups of the cyclic group $\mathbb{Z}$ are $m\mathbb{Z}$, $m\ge0$ (§6.1). (2) The subgroups of the cyclic group $\mathbb{Z}_n$ are in bijection with the divisors of $n$: for each $d\mid n$ there is exactly one, $\langle\bar d\rangle$, of order $n/d$ (§6.1). The inclusion statement is the same lattice of subgroups. ∎
In most rings ideals are much rarer than additive subgroups: $\mathbb{Z}$ is an additive subgroup and a subring of $\mathbb{Q}$, but not an ideal, since $\tfrac12\cdot1\notin\mathbb{Z}$. The next theorem shows that fields have no interesting ideals at all.
- Let $R$ have identity. If an ideal (even a left ideal) $I$ contains a unit, then $I=R$.
- A commutative ring $R$ with $1\ne0$ is a field if and only if its only ideals are $\{0\}$ and $R$.
- For a field $F$ and $n\ge2$, the only two-sided ideals of $M_n(F)$ are $\{0\}$ and $M_n(F)$, although $M_n(F)$ is not a division ring. (A ring with this property is called simple.)
Proof. (1) If $u\in I$ is a unit then $1=u^{-1}u\in I$, and then $r=r\cdot1\in I$ for every $r$. (2) If $R$ is a field, every nonzero ideal contains a unit and equals $R$ by (1). Conversely, suppose $R$ has only the ideals $0$ and $R$, and let $a\ne0$. The ideal $\langle a\rangle$ is nonzero, so $\langle a\rangle=R\ni1$, i.e. $ab=1$ for some $b$: $a$ is a unit. (3) Let $J\ne0$ be a two-sided ideal and $A=(a_{st})\in J$ with $a_{kl}\ne0$. With the matrix units $E_{ij}$ (a $1$ in place $(i,j)$, zeros elsewhere), $E_{ik}AE_{lj}=a_{kl}E_{ij}$, so $E_{ij}=a_{kl}^{-1}(E_{ik}AE_{lj})\in J$ for all $i,j$. Then $I=E_{11}+\dots+E_{nn}\in J$ and $J=M_n(F)$ by (1). On the other hand $E_{11}$ is a nonzero non-unit, so $M_n(F)$ is not a division ring. ∎
So (2) is genuinely about commutative rings: $M_2(\mathbb{R})$ has no proper nonzero two-sided ideals, but it has many one-sided ones, for instance the matrices whose second column is zero (a left ideal).
Let $I,J$ be ideals of a ring $R$. Then $I+J=\{a+b:a\in I,b\in J\}$, $I\cap J$ and the product $IJ=\{\sum_{t=1}^{k}a_tb_t:k\ge1,\ a_t\in I,\ b_t\in J\}$ are ideals, and $IJ\subseteq I\cap J\subseteq I\subseteq I+J$.
- In $\mathbb{Z}$: $m\mathbb{Z}+n\mathbb{Z}=\gcd(m,n)\mathbb{Z}$, $\ m\mathbb{Z}\cap n\mathbb{Z}=\operatorname{lcm}(m,n)\mathbb{Z}$, $\ (m\mathbb{Z})(n\mathbb{Z})=mn\mathbb{Z}$.
- If $R$ is commutative with identity and $I+J=R$ (the ideals are comaximal), then $IJ=I\cap J$.
- $I\cup J$ is an ideal only when $I\subseteq J$ or $J\subseteq I$.
Proof. Closure of $I+J$ under subtraction is clear, and $r(a+b)=ra+rb\in I+J$; similarly on the right. $I\cap J$ is an intersection of additive subgroups closed under absorption. A product $r\sum a_tb_t=\sum(ra_t)b_t$ is again of the required form, and so is $(\sum a_tb_t)r=\sum a_t(b_tr)$. Each $a_tb_t$ lies in $I$ (absorption by $I$) and in $J$ (absorption by $J$), so $IJ\subseteq I\cap J$.
(1) $m\mathbb{Z}+n\mathbb{Z}$ is an ideal, hence $d\mathbb{Z}$ for some $d\ge0$ (Theorem 7.10); $d$ divides $m$ and $n$, and $d=mx+ny$ is divisible by every common divisor, so $d=\gcd(m,n)$ — this is Bézout's identity in ideal language. $m\mathbb{Z}\cap n\mathbb{Z}$ is the set of common multiples, i.e. the multiples of the lcm. Every product of a multiple of $m$ and a multiple of $n$ is a multiple of $mn$, and $mn$ itself is such a product.
(2) Write $1=i+j$ with $i\in I$, $j\in J$. For $x\in I\cap J$, $x=x\cdot1=xi+xj$; here $xi=ix\in IJ$ (as $i\in I$, $x\in J$) and $xj\in IJ$ (as $x\in I$, $j\in J$). So $I\cap J\subseteq IJ$.
(3) If $a\in I\setminus J$ and $b\in J\setminus I$, then $a+b\notin I$ (else $b=(a+b)-a\in I$) and $a+b\notin J$ (else $a\in J$), so $I\cup J$ is not closed under addition. ∎
Proof. Addition of cosets is well defined because $(R,+)$ is abelian, so $I$ is a normal subgroup and $R/I$ is the quotient group (§6.2). For multiplication, let $a'=a+x$ and $b'=b+y$ with $x,y\in I$. Then $a'b'=ab+ay+xb+xy$, and $ay\in I$ (left absorption), $xb\in I$ (right absorption), $xy\in I$; so $a'b'+I=ab+I$. Associativity and distributivity in $R/I$ follow from those in $R$, e.g. $(\bar a\bar b)\bar c=\overline{(ab)c}=\overline{a(bc)}=\bar a(\bar b\bar c)$; commutativity and the identity are inherited the same way. The statements about $\pi$ are the definitions of the operations. Conversely, if coset multiplication is well defined, then for $x\in I$ and $r\in R$ the cosets $x+I=0+I$ give $(x+I)(r+I)=(0+I)(r+I)$, i.e. $xr+I=0+I$, so $xr\in I$; similarly $rx\in I$. ∎
The model is $\mathbb{Z}/n\mathbb{Z}=\mathbb{Z}_n$. Computing in $R/I$ means computing in $R$ and then using the relations "every element of $I$ is $0$". In $\mathbb{R}[x]/\langle x^2+1\rangle$, for instance, every polynomial is congruent to its remainder $a+bx$ on division by $x^2+1$, and $x^2\equiv-1$: the quotient is a copy of $\mathbb{C}$ with $\bar x$ playing the role of $i$ (Example 6).
- $\mathbb{Z}[i]/\langle\alpha\rangle$ has exactly $N$ elements.
- If $\gcd(a,b)=1$, then $\mathbb{Z}\to\mathbb{Z}[i]/\langle\alpha\rangle$, $m\mapsto m+\langle\alpha\rangle$, is onto with kernel $N\mathbb{Z}$, so $\mathbb{Z}[i]/\langle\alpha\rangle\cong\mathbb{Z}_N$.
Proof. Identify $x+yi$ with $(x,y)\in\mathbb{Z}^2$. The ideal $\langle\alpha\rangle=\{\alpha(u+vi)\}$ consists of the integer combinations $u(a,b)+v(-b,a)$ of $\alpha$ and $i\alpha$. (1) Let $g=\gcd(a,b)$ and choose integers $s,t$ with $sa-tb=g$. The vectors $$R_1=s(a,b)+t(-b,a)=(g,\ sb+ta),\qquad R_2=\tfrac bg(a,b)+\tfrac ag(-b,a)=\left(0,\ \tfrac Ng\right)$$ lie in $\langle\alpha\rangle$, and the change-of-basis matrix $\begin{pmatrix}s&t\\b/g&a/g\end{pmatrix}$ has determinant $(sa-tb)/g=1$, so it is invertible over $\mathbb{Z}$ and $R_1,R_2$ also generate $\langle\alpha\rangle$. Given $(x,y)$, subtracting a multiple of $R_1$ brings $x$ into $\{0,\dots,g-1\}$, and then subtracting a multiple of $R_2$ brings $y$ into $\{0,\dots,N/g-1\}$. These $g\cdot N/g=N$ representatives are pairwise incongruent: a difference $kR_1+lR_2$ with first coordinate $kg\in(-g,g)$ has $k=0$, and then $lN/g\in(-N/g,N/g)$ forces $l=0$.
(2) Since $\gcd(b,N)=\gcd(b,a^2)=1$, choose $c$ with $bc\equiv1\pmod N$. In the quotient, $N=\alpha\bar\alpha\equiv0$, hence $bc\equiv1$, and $bi\equiv-a$ (because $a+bi\equiv0$). Multiplying by $c$: $i\equiv-ac$. So every $x+yi\equiv x-yac$ is congruent to an integer — the map from $\mathbb{Z}$ is onto. Its kernel is $\mathbb{Z}\cap\langle\alpha\rangle$. It contains $N$. If $m=(a+bi)(u+vi)=(au-bv)+(av+bu)i$ is an integer, then $av=-bu$; as $\gcd(a,b)=1$, $u=ak$, $v=-bk$ for some $k$, and $m=a^2k+b^2k=kN$. So the kernel is $N\mathbb{Z}$, and the first isomorphism theorem (Theorem 7.16 below) gives $\mathbb{Z}[i]/\langle\alpha\rangle\cong\mathbb{Z}_N$. ∎
When $g\gt1$ the quotient is not $\mathbb{Z}_N$: its additive group is $\mathbb{Z}_g\times\mathbb{Z}_{N/g}$, which is not cyclic because $g\mid N/g$. For instance $\mathbb{Z}[i]/\langle3\rangle$ has $9$ elements $x+yi$, $x,y\in\mathbb{Z}_3$ — it is the field of order $9$ from Example 3(d).
Ring Homomorphisms and the Isomorphism Theorems
Typical homomorphisms: reduction $\mathbb{Z}\to\mathbb{Z}_n$; evaluation $R[x]\to R$, $f\mapsto f(c)$; complex conjugation $\mathbb{C}\to\mathbb{C}$; $\mathbb{Z}[i]\to\mathbb{Z}_5$, $x+yi\mapsto x+3y$; the projections $R\times S\to R$; and the non-unital embedding $R\to R\times S$, $a\mapsto(a,0)$, which sends $1$ to the idempotent $(1,0)$. The map $x\mapsto2x$ on $\mathbb{Z}$ is a group homomorphism but not a ring homomorphism ($2(xy)\ne(2x)(2y)$ in general).
Let $\phi:R\to S$ be a ring homomorphism.
- $\phi(0)=0$, $\phi(-a)=-\phi(a)$, $\phi(na)=n\phi(a)$, and $\phi$ is injective iff $\ker\phi=\{0\}$.
- $\ker\phi$ is an ideal of $R$ and $\phi(R)$ is a subring of $S$; more generally preimages of ideals are ideals.
- If $R$ has identity, $e=\phi(1)$ is an idempotent of $S$ and an identity of the ring $\phi(R)$. If $S$ is an integral domain and $\phi\ne0$, then $\phi(1)=1$.
- If $\phi$ is unital, it maps units to units: $\phi(u^{-1})=\phi(u)^{-1}$.
- If $R$ is a field and $S\ne0$, every unital homomorphism $R\to S$ is injective; more generally every nonzero homomorphism from a field (or from any ring with no proper nonzero ideals) is injective.
- For every ring $S$ with identity there is exactly one unital homomorphism $\mathbb{Z}\to S$, namely $n\mapsto n\cdot1_S$. Its kernel is $c\,\mathbb{Z}$ with $c=\operatorname{char}S$, and its image (the prime subring) is isomorphic to $\mathbb{Z}$ if $c=0$ and to $\mathbb{Z}_c$ if $c\gt0$.
Proof. (1) $\phi$ is a homomorphism of the additive groups, so this is §6.2. (2) $\ker\phi$ is an additive subgroup, and if $\phi(a)=0$ then $\phi(ra)=\phi(r)\cdot0=0$ and $\phi(ar)=0$. If $J\trianglelefteq S$ and $\phi(a)\in J$, then $\phi(ra)=\phi(r)\phi(a)\in J$; so $\phi^{-1}(J)$ absorbs, and it is an additive subgroup. The image is closed under subtraction and products: $\phi(a)\phi(b)=\phi(ab)$. (3) $e^2=\phi(1\cdot1)=e$ and $e\phi(a)=\phi(1a)=\phi(a)=\phi(a)e$. In a domain, $e(e-1)=0$ gives $e=0$ or $e=1$, and $e=0$ would force $\phi(a)=\phi(1)\phi(a)=0$ for all $a$. (4) $\phi(u)\phi(u^{-1})=\phi(1)=1$ and similarly on the other side. (5) $\ker\phi$ is an ideal of the field $R$, so it is $\{0\}$ or $R$ (Theorem 7.11); it is not $R$ because $\phi\ne0$ (for unital $\phi$, $\phi(1)=1\ne0$ as $S\ne0$). (6) A unital $\phi$ must send $1$ to $1_S$ and hence $n=1+\dots+1$ to $n\cdot1_S$ (and $-n$ to $-n\cdot1_S$); this map is a homomorphism by Theorem 7.1(3). Its kernel is the set of $n$ with $n\cdot1_S=0$, which is $c\mathbb{Z}$ by Theorem 7.8(1). The description of the image follows from the first isomorphism theorem below. ∎
Proof. Put $K=\ker\phi$. By the first isomorphism theorem for groups (§6.2), applied to the additive groups, $\bar\phi$ is well defined ($a+K=b+K\Rightarrow a-b\in K\Rightarrow\phi(a)=\phi(b)$), additive, injective and onto $\phi(R)$. It also respects products: $\bar\phi((a+K)(b+K))=\bar\phi(ab+K)=\phi(ab)$, and $\phi(ab)=\phi(a)\phi(b)=\bar\phi(a+K)\,\bar\phi(b+K)$. ∎
- Guess the answer $S$ from the relations: in $R/I$ every generator of $I$ becomes $0$ (in $\mathbb{Z}[x]/\langle x^2+1\rangle$, "$x^2=-1$"; in $\mathbb{Z}[x]/\langle 2,x\rangle$, "$2=0$ and $x=0$").
- Write down a homomorphism $\phi:R\to S$ that kills the generators — usually an evaluation $f\mapsto f(c)$, a reduction mod $n$, or a pair of such maps into a product.
- Check that $\phi$ is onto (often obvious; for a product use the Chinese remainder theorem).
- Show $\ker\phi=I$: the inclusion $I\subseteq\ker\phi$ is step 2; for $\ker\phi\subseteq I$ use a division algorithm (divide by a monic polynomial) or a norm argument.
- Conclude $R/I\cong S$ by Theorem 7.16. Then read off properties: $R/I$ is a domain, a field, has nilpotents… exactly when $S$ does.
- Second. If $S$ is a subring and $I$ an ideal of $R$, then $S+I$ is a subring of $R$, $S\cap I$ is an ideal of $S$, and $(S+I)/I\cong S/(S\cap I)$.
- Third. If $I\subseteq J$ are ideals of $R$, then $J/I$ is an ideal of $R/I$ and $(R/I)/(J/I)\cong R/J$.
- Correspondence. $J\mapsto J/I$ is an inclusion-preserving bijection from the ideals of $R$ containing $I$ to the ideals of $R/I$; its inverse sends an ideal $\bar J$ of $R/I$ to $\pi^{-1}(\bar J)$.
Proof. (1) $(s+x)-(s'+x')=(s-s')+(x-x')$ and $(s+x)(s'+x')=ss'+(sx'+xs'+xx')$ with the bracket in $I$, so $S+I$ is a subring; it contains $I$ as an ideal. The map $S\to(S+I)/I$, $s\mapsto s+I$, is a homomorphism; it is onto, because $(s+x)+I=s+I$; and its kernel is $S\cap I$. Theorem 7.16 gives the isomorphism (and shows $S\cap I$ is an ideal of $S$, being a kernel). (2) The map $R/I\to R/J$, $a+I\mapsto a+J$, is well defined because $I\subseteq J$, is an onto homomorphism, and has kernel $\{a+I:a\in J\}=J/I$. Apply Theorem 7.16. (3) For an ideal $\bar J$ of $R/I$, $\pi^{-1}(\bar J)$ is an ideal of $R$ (Theorem 7.15(2)) containing $\pi^{-1}(0)=I$, and $\pi(\pi^{-1}(\bar J))=\bar J$ since $\pi$ is onto. If $J\supseteq I$ is an ideal, $\pi(J)=J/I$ is an ideal of $R/I$ ($\pi$ is onto, so absorption by $R/I$ comes from absorption by $R$), and $\pi^{-1}(J/I)=J+I=J$. These two maps are inverse to each other and clearly preserve inclusions. ∎
The correspondence theorem re-proves Theorem 7.10(2): ideals of $\mathbb{Z}_n=\mathbb{Z}/n\mathbb{Z}$ correspond to ideals $d\mathbb{Z}\supseteq n\mathbb{Z}$, i.e. to divisors $d$ of $n$. It also shows that $\mathbb{Z}[x]/\langle2,x\rangle$ has exactly two ideals, because $\langle2,x\rangle$ is contained in no ideal other than itself and $\mathbb{Z}[x]$ — which we will recognise in §7.2 as "$\langle2,x\rangle$ is maximal".
Let $R$ be a commutative ring with identity and $I,J$ ideals with $I+J=R$. Then $r\mapsto(r+I,\ r+J)$ induces an isomorphism $$R/(I\cap J)=R/IJ\;\cong\;R/I\times R/J.$$ Consequences, for $\gcd(m,n)=1$: $\mathbb{Z}_{mn}\cong\mathbb{Z}_m\times\mathbb{Z}_n$ as rings; $U(mn)\cong U(m)\times U(n)$; and if $n=p_1^{a_1}\cdots p_k^{a_k}$, then $\mathbb{Z}_n\cong\mathbb{Z}_{p_1^{a_1}}\times\dots\times\mathbb{Z}_{p_k^{a_k}}$, so $\mathbb{Z}_n$ has exactly $2^k$ idempotents.
Proof. The map $\psi(r)=(r+I,r+J)$ is a homomorphism with kernel $I\cap J$, which equals $IJ$ by Theorem 7.12(2). It is onto: write $1=i+j$ with $i\in I$, $j\in J$; given $a,b\in R$ put $r=aj+bi$. Modulo $I$, $j=1-i\equiv1$ and $i\equiv0$, so $r\equiv a$; modulo $J$, similarly $r\equiv b$. Theorem 7.16 finishes. For $\mathbb{Z}$: $m\mathbb{Z}+n\mathbb{Z}=\mathbb{Z}$ and $m\mathbb{Z}\cap n\mathbb{Z}=mn\mathbb{Z}$ when $\gcd(m,n)=1$ (Theorem 7.12(1)). A ring isomorphism restricts to an isomorphism of unit groups, and the units of $R\times S$ are the pairs of units. Induction on $k$ gives the decomposition of $\mathbb{Z}_n$. Finally $(e_1,\dots,e_k)$ is idempotent iff each $e_i$ is, and $\mathbb{Z}_{p^a}$ has only the idempotents $0,1$: from $p^a\mid e(e-1)$ and $\gcd(e,e-1)=1$, $p^a$ divides $e$ or $e-1$. ∎
This is the Chinese remainder theorem of §5.2 restated: solving $x\equiv a\pmod m$, $x\equiv b\pmod n$ is computing $\psi^{-1}(a,b)$, and the formula $r=aj+bi$ is exactly the usual "$x=a\,n\,n'+b\,m\,m'$" construction, with $i=mm'$ and $j=nn'$ where $mm'+nn'=1$.
- The ring homomorphisms $\phi:\mathbb{Z}_m\to\mathbb{Z}_n$ are exactly the maps $\phi(\bar x)=\overline{xe}$, where $e\in\mathbb{Z}_n$ satisfies $e^2=e$ and $me\equiv0\pmod n$; distinct $e$ give distinct maps.
- Their number is $2^{k}$, where $k$ is the number of primes $p\mid n$ such that the full power $p^{v_p(n)}$ dividing $n$ also divides $m$. In particular the homomorphisms $\mathbb{Z}\to\mathbb{Z}_n$ number $2^{\omega(n)}$, $\omega(n)$ being the number of distinct prime factors of $n$.
- A unital homomorphism $\mathbb{Z}_m\to\mathbb{Z}_n$ exists iff $n\mid m$, and then it is unique (reduction mod $n$).
Proof. (1) $\phi$ is determined by $e=\phi(\bar1)$, because $\phi(\bar x)=x\phi(\bar1)$. Necessarily $e^2=\phi(\bar1)^2=\phi(\bar1)=e$ and $me=\phi(\bar m)=\phi(\bar0)=0$. Conversely, given such $e$, $\bar x\mapsto\overline{xe}$ is well defined (if $x\equiv x'\pmod m$ then $(x-x')e$ is a multiple of $me\equiv0$), additive, and multiplicative: $\overline{xye}=\overline{xye^2}=\overline{xe}\cdot\overline{ye}$.
(2) By Theorem 7.18 an idempotent $e$ of $\mathbb{Z}_n$ is the same as a choice of $e\equiv0$ or $e\equiv1$ modulo each prime power $p^a\,\|\,n$. The condition $n\mid me$ holds iff $p^a\mid me$ for every such $p^a$: automatic where $e\equiv0\pmod{p^a}$, and equivalent to $p^a\mid m$ where $e\equiv1\pmod{p^a}$. So "$e\equiv1$" is allowed exactly at the $k$ primes with $p^a\mid m$, and there are $2^k$ choices. For $\mathbb{Z}\to\mathbb{Z}_n$ there is no condition $me=0$. (3) Unital means $e=1$, allowed iff $n\mid m\cdot1$. ∎
- Rings need not have an identity. $2\mathbb{Z}$ is a commutative ring without $1$; the ideal it generates by $2$ is not $\{2r\}$.
- A subring can have a different identity. $\{(a,0)\}\subset\mathbb{Z}\times\mathbb{Z}$ has identity $(1,0)\ne(1,1)$; $\mathbb{Z}_6$ contains $\{0,3\}\cong\mathbb{Z}_2$ with identity $3$ and $\{0,2,4\}\cong\mathbb{Z}_3$ with identity $4$.
- Homomorphisms need not be unital. $a\mapsto(a,0)$ and $\bar x\mapsto\overline{21x}$ ($\mathbb{Z}_{12}\to\mathbb{Z}_{28}$) send $1$ to a nontrivial idempotent. Into an integral domain, every nonzero homomorphism is unital.
- Reduction mod $n$ is defined on $\mathbb{Z}_m$ only when $n\mid m$. "$\mathbb{Z}_4\to\mathbb{Z}_6$, $\bar x\mapsto\bar x$" is not even well defined ($\bar0=\bar4$ in $\mathbb{Z}_4$ but $0\ne4$ in $\mathbb{Z}_6$).
- Subrings are not ideals. $\mathbb{Z}\subset\mathbb{Q}$, the diagonal of $\mathbb{Z}\times\mathbb{Z}$, the upper triangular matrices in $M_2(\mathbb{R})$.
- Unions of ideals are not ideals. $2\mathbb{Z}\cup3\mathbb{Z}$ contains $2,3$ but not $5$. Intersections, sums and products are ideals.
- Zero divisors do not form an ideal. In $\mathbb{Z}_6$, $2$ and $3$ are zero divisors and $2+3=5$ is a unit. Nilpotents in a commutative ring do form an ideal.
- Commutativity matters for nilpotents and inverses. In $M_2(\mathbb{R})$, $E_{12}+E_{21}$ is a sum of nilpotents that is a unit. In the ring of linear maps on sequences, the shifts $L(a_1,a_2,\dots)=(a_2,a_3,\dots)$ and $R(a_1,a_2,\dots)=(0,a_1,a_2,\dots)$ satisfy $LR=1$ but $RL\ne1$: a one-sided inverse need not be two-sided.
- "Every nonzero element is a unit or a zero divisor" is false in general. It holds in finite commutative rings and in $M_n(F)$, but $2\in\mathbb{Z}$ and $x\in\mathbb{Z}[x]$ are neither.
- Characteristic says nothing about size. $\mathbb{Z}_p[x]$ is infinite of characteristic $p$; a domain has characteristic $0$ or prime, but $\mathbb{Z}_2\times\mathbb{Z}_3$ has characteristic $6$.
- Simple does not mean division ring. $M_2(\mathbb{R})$ has no proper nonzero two-sided ideals, yet $E_{11}$ is a nonzero non-unit. Theorem 7.11(2) needs commutativity.
- Products and images of domains. $\mathbb{Z}\times\mathbb{Z}$ is not a domain though $\mathbb{Z}$ is; the image $\mathbb{Z}_6$ of $\mathbb{Z}$ is not a domain. A quotient of a domain can have zero divisors, and a quotient of a ring with zero divisors can be a field ($\mathbb{Z}_6/\langle\bar2\rangle\cong\mathbb{Z}_2$).
- $\mathbb{Z}[i]/\langle\alpha\rangle$ is $\mathbb{Z}_{N(\alpha)}$ only when $\gcd(a,b)=1$. $\mathbb{Z}[i]/\langle3\rangle$ is a field of order $9$ and $\mathbb{Z}[i]/\langle2\rangle$ has a nonzero nilpotent; neither is $\mathbb{Z}_9$ or $\mathbb{Z}_4$.
Exercises
- 1 Which of these are rings under the usual operations: (a) the odd integers; (b) $\{a+b\sqrt2:a,b\in\mathbb{Q}\}$; (c) $\{a+b\sqrt[3]{2}:a,b\in\mathbb{Z}\}$; (d) the real matrices $\begin{pmatrix}a&b\\-b&a\end{pmatrix}$?
- 2 List the units, zero divisors, nilpotent and idempotent elements of $\mathbb{Z}_{18}$.
- 3 How many units, zero divisors, nilpotents and idempotents does $\mathbb{Z}_{100}$ have? List the idempotents.
- 4 Show that $5+2\sqrt6$ is a unit of $\mathbb{Z}[\sqrt6]$ and $2+\sqrt5$ is a unit of $\mathbb{Z}[\sqrt5]$, and find their inverses.
- 5 Show that $1+2x$ is a unit of $\mathbb{Z}_4[x]$, and deduce that $\mathbb{Z}_4[x]$ has infinitely many units. Why does this not happen in $\mathbb{Z}[x]$?
- 6 In the Boolean ring $\mathcal{P}(X)$, $X=\{1,2,3\}$ ($A+B=A\triangle B$, $AB=A\cap B$), find the identity, the units, the zero divisors, the nilpotents and the idempotents.
- 7 Prove that the only idempotents of an integral domain are $0$ and $1$, and that an integral domain has no nonzero nilpotents.
- 8 Find $|GL(2,\mathbb{Z}_2)|$ and $|GL(2,\mathbb{Z}_3)|$ (the unit groups of $M_2(\mathbb{Z}_2)$ and $M_2(\mathbb{Z}_3)$), and show that $GL(2,\mathbb{Z}_2)\cong S_3$.
- 9 Find all nilpotent elements of $M_2(\mathbb{Z}_2)$.
- 1 Prove that $\mathbb{Q}(\sqrt2)=\{a+b\sqrt2:a,b\in\mathbb{Q}\}$ is a field.
- 2 Find the characteristic of $\mathbb{Z}_4\times\mathbb{Z}_6$, $\mathbb{Z}_3[x]$, $M_2(\mathbb{Z}_6)$, $\mathbb{Z}\times\mathbb{Z}_5$ and $\mathcal{P}(X)$ ($X\ne\varnothing$).
- 3 Show that $(a+b)^9=a^9+b^9$ in every commutative ring of characteristic $3$, and expand $(1+x)^9$ and $(1+x)^{6}$ in $\mathbb{Z}_3[x]$.
- 4 Let $R$ be an integral domain of prime characteristic $p$. Show that $F(a)=a^p$ is an injective ring homomorphism $R\to R$, that it is an automorphism if $R$ is finite, and that it is not onto for $R=\mathbb{Z}_p[x]$.
- 5 Prove that an integral domain with only finitely many ideals is a field.
- 6 Give an example of an infinite integral domain of characteristic $3$ that is not a field, and of an infinite field of characteristic $3$.
- 7 Let $R$ be a ring with $x^3=x$ for all $x\in R$. Prove that $6x=0$ for all $x$. Verify that $\mathbb{Z}_6$ is such a ring.
- 8 How many idempotents does $\mathbb{Z}_6\times\mathbb{Z}_{10}$ have? How many does $\mathbb{Z}_{60}$ have? Are the rings isomorphic?
- 1 List all ideals of $\mathbb{Z}_{20}$ with the number of elements of each.
- 2 Compute $12\mathbb{Z}+18\mathbb{Z}$, $12\mathbb{Z}\cap18\mathbb{Z}$ and $(12\mathbb{Z})(18\mathbb{Z})$.
- 3 In $\mathbb{Z}[x]$, show that $\langle2\rangle+\langle x\rangle\ne\mathbb{Z}[x]$ and $\langle2\rangle\cap\langle x\rangle=\langle2x\rangle$.
- 4 Is the set of non-units of $\mathbb{Z}_8$ an ideal? Of $\mathbb{Z}_{12}$?
- 5 Identify $\mathbb{Z}[x]/\langle3,x\rangle$, $\ \mathbb{Z}[x]/\langle x-2\rangle$ and $\mathbb{Z}[x]/\langle2,x^2+x+1\rangle$. Which are fields?
- 6 Prove that in a commutative ring the set of nilpotent elements is an ideal (the nilradical).
- 7 Identify $\mathbb{Z}[i]/\langle4+i\rangle$, and find the number of elements and the characteristic of $\mathbb{Z}[i]/\langle6\rangle$ and of $\mathbb{Z}[i]/\langle4+2i\rangle$.
- 8 Let $R,S$ be rings with identity. Show that every ideal of $R\times S$ has the form $I\times J$ with $I\trianglelefteq R$, $J\trianglelefteq S$. How many ideals has $\mathbb{Z}_4\times\mathbb{Z}_6$?
- 9 Let $I=\{f\in C[0,1]:f(x)=0\text{ for }0\le x\le\tfrac12\}$. Show that $I$ is an ideal and that it is not principal.
- 1 Count the ring homomorphisms $\mathbb{Z}_6\to\mathbb{Z}_6$, $\ \mathbb{Z}_{15}\to\mathbb{Z}_{10}$ and $\mathbb{Z}_6\to\mathbb{Z}_4$.
- 2 Show that the only ring homomorphisms $\mathbb{Z}\to\mathbb{Z}$, and $\mathbb{Q}\to\mathbb{Q}$, are the zero map and the identity.
- 3 Show that the only ring homomorphisms $\mathbb{R}\to\mathbb{R}$ are $0$ and the identity. (Hint: nonnegative reals are squares.)
- 4 Show that $2\mathbb{Z}$ and $3\mathbb{Z}$ are isomorphic as groups but not as rings, and that $\mathbb{Z}[\sqrt2]\not\cong\mathbb{Z}[\sqrt3]$ as rings.
- 5 Let $\phi:R\to S$ be an onto unital homomorphism of commutative rings. Show that $\phi(R^\times)\subseteq S^\times$, and give an example where $\phi(R^\times)\ne S^\times$.
- 6 Under $\mathbb{Z}_{60}\cong\mathbb{Z}_4\times\mathbb{Z}_3\times\mathbb{Z}_5$, which element corresponds to $(1,2,3)$? Which idempotent of $\mathbb{Z}_{60}$ corresponds to $(1,0,1)$?
- 7 Let $R$ be commutative with identity and $e$ an idempotent. Show that $Re$ is a ring with identity $e$ and that $R\cong Re\times R(1-e)$.
- 8 Let $T$ be the ring of upper triangular $2\times2$ integer matrices and $I$ the set of matrices $\begin{pmatrix}0&b\\0&0\end{pmatrix}$. Show that $I\trianglelefteq T$ and $T/I\cong\mathbb{Z}\times\mathbb{Z}$.
- 9 Secret sharing. A vault code $x$ with $0\le x\lt1001$ is split among three officers, who receive $x\bmod7$, $x\bmod11$ and $x\bmod13$. They report $3$, $5$ and $9$. Recover $x$, and explain with Theorem 7.18 why two officers alone cannot determine it.
E1. CSIR Part B GATE MA The number of nilpotent elements of the ring $\mathbb{Z}_{360}$ (including $0$) is (A) $2$; (B) $6$; (C) $12$; (D) $30$.
E2. CSIR Part B The number of ring homomorphisms (not necessarily sending $1$ to $1$) from $\mathbb{Z}_{20}$ to $\mathbb{Z}_{30}$ is (A) $1$; (B) $2$; (C) $4$; (D) $10$.
E3. CSIR Part B The quotient ring $\mathbb{Z}[i]/\langle1+3i\rangle$ is (A) a field; (B) isomorphic to $\mathbb{Z}_{10}$; (C) a ring with $4$ elements; (D) an integral domain that is not a field.
E4. CSIR Part C Which of the following rings are integral domains? (One or more options may be correct.) (A) $\mathbb{Z}[x]/\langle x^2+1\rangle$; (B) $\mathbb{Z}_3[x]/\langle x^2+1\rangle$; (C) $\mathbb{Z}_5[x]/\langle x^2+1\rangle$; (D) $\mathbb{R}[x]/\langle x^2\rangle$.
E5. CSIR Part C Let $R$ be a commutative ring with $1\ne0$. Which of the following are always true? (A) If $a$ is nilpotent, then $1+a$ is a unit. (B) The nilpotent elements form an ideal. (C) The zero divisors together with $0$ form an ideal. (D) The units together with $0$ form a subring.
E6. CSIR Part C Which of the following statements are true? (A) There is a ring homomorphism $\mathbb{Z}_{12}\to\mathbb{Z}_4$ sending $1$ to $1$. (B) There is a ring homomorphism $\mathbb{Z}_4\to\mathbb{Z}_{12}$ sending $1$ to $1$. (C) Every nonzero ring homomorphism from a field to a ring is injective. (D) $\mathbb{Z}\times\mathbb{Z}$ and $\mathbb{Z}[i]$ are isomorphic rings.
Suggested plan — 3 lectures × 60 min.
- Rings and their elements. Definition 7.1 with the gallery of examples (ask the class which have an identity, which are commutative); Theorem 7.1; subrings and the "different identity" phenomenon. Units, zero divisors, nilpotents, idempotents (Definition 7.3, Theorem 7.3) and Example 1 on the board. Close with domains and fields (Theorems 7.4, 7.5) and the norm on $\mathbb{Z}[\sqrt d]$ (Theorem 7.6, Example 2(a)).
- Characteristic and ideals. Theorem 7.7 (determinant criterion) in five minutes; characteristic and the freshman's dream (Theorem 7.8); Example 3 as a prove-or-disprove session. Ideals (Definition 7.7, Theorems 7.9–7.12) and Example 4; quotient rings (Theorem 7.13). Project Interactive 1 with $\alpha=2+i$, then $\alpha=2$ and $\alpha=3$, and let students conjecture when the quotient is $\mathbb{Z}_N$ (Theorem 7.14, Example 5).
- Homomorphisms. Definition 7.9 and Theorem 7.15; the first isomorphism theorem and the strategy box, with Example 6 (a), (c), (e) on the board; second/third isomorphism and correspondence theorems in outline; CRT for rings (Theorem 7.18). Finish with Theorem 7.19 and Interactive 2 (start with $m=12$, $n=28$, then $m=20$, $n=30$) and Exam-corner E2 and E6 done live.
Discussion questions. (1) Why must a kernel absorb multiplication by all ring elements, while a subring need only be closed under multiplication by its own elements? (2) $\mathbb{Z}_{12}$, $\mathbb{Z}_2\times\mathbb{Z}_6$ and $\mathbb{Z}_3\times\mathbb{Z}_4$ all have $12$ elements. Which two are isomorphic rings, and which quick invariant (characteristic, idempotents, nilpotents) separates the third? (3) Is there a field with $6$ elements? Which theorem of this section rules it out?
Misconceptions to watch for. "Every subring is an ideal"; "a ring homomorphism must send $1$ to $1$" (depends on the convention — state it); "reduction $\mathbb{Z}_m\to\mathbb{Z}_n$ always exists"; counting group homomorphisms ($\gcd(m,n)$) when ring homomorphisms are asked for; "a finite ring is a field"; "$\mathbb{Z}[i]/\langle\alpha\rangle\cong\mathbb{Z}_{N(\alpha)}$ always"; using the binomial theorem or "sum of nilpotents is nilpotent" in matrix rings.
Suggested homework. I(2, 3, 5, 8), II(2, 4, 5, 7), III(1, 4, 5, 6, 7), IV(1, 3, 6, 9); Exam corner E1, E4, E5.
Research question. The ring isomorphism $\mathbb{Z}_{pq}\cong\mathbb{Z}_p\times\mathbb{Z}_q$ makes RSA decryption several times faster. Why — and how can a single hardware error during the fast computation hand an attacker the secret factorisation?
Why it matters. RSA signatures protect software updates, bank cards and web certificates worldwide, and almost every implementation decrypts "by CRT" to save time on smart cards and servers. In 1997 researchers at Bellcore in the United States (Boneh, DeMillo and Lipton) showed that one faulty CRT signature can reveal $p$ and $q$; since then fault attacks — induced by voltage glitches, lasers or overheating — have been a central topic in hardware security, and certified smart cards must defend against them.
What to do.
- With Python and
sympy.randprime, generate primes $p,q$ of about $1024$ bits, set $n=pq$, $e=65537$, $d=e^{-1}\bmod(p-1)(q-1)$, and sign a message $m$: $s=m^d\bmod n$. - Compute the same signature by CRT: $s_p=m^{d\bmod(p-1)}\bmod p$, $s_q=m^{d\bmod(q-1)}\bmod q$, and combine them with the formula $r=aj+bi$ of Theorem 7.18 (or Garner's form $s=s_q+q\,\big(q^{-1}(s_p-s_q)\bmod p\big)$). Check that both methods agree, and time each over $200$ signatures.
- Explain the speed-up from the sizes of the exponents and moduli (schoolbook multiplication of $k$-bit numbers costs about $k^2$ operations).
- Simulate a fault: flip one bit of $s_p$ before combining, to get a wrong signature $s'$. Compute $\gcd(s'^{\,e}-m,\ n)$ and explain, using the projections $\mathbb{Z}_n\to\mathbb{Z}_p$ and $\mathbb{Z}_n\to\mathbb{Z}_q$, why it equals $q$.
Tools: Python 3 with sympy (or any language with big integers); the time module.
Hand in: A one-page report: timing table, the speed-up factor and your explanation, the recovered factor, and one countermeasure (for example, verifying $s^e\equiv m$ before releasing $s$).
Going further. Read about Lenstra's observation that the attack needs only the faulty signature and the message, not a correct signature; and about the use of the ring $\mathbb{Z}_q[x]/\langle x^{256}+1\rangle$ in the post-quantum standard ML-KEM, where a ring version of CRT (the number-theoretic transform) plays the same speed-up role.
🧪 Self-check quiz
Q1. How many zero divisors does $\mathbb{Z}_{20}$ have?
Q2. Which of these rings is an integral domain?
Q3. The ring $\mathbb{Z}[i]/\langle2+i\rangle$ is isomorphic to
Q4. The characteristic of $\mathbb{Z}_6\times\mathbb{Z}_{10}$ is
Q5. Which subset of $\mathbb{Z}[x]$ is an ideal?