Ideals and Quotient Rings
In group theory the road to quotient
groups runs through normal subgroups: those special subgroups you are allowed to
collapse, folding a whole group down onto a smaller one. Rings tell the same story one operation
louder. A ring carries
two operations, so the subobject you are allowed to collapse must respect both. That subobject
is an ideal — and the collapsed ring you get is a quotient
ring.
You have already met a quotient ring without knowing its true name. The clock ring
\mathbb{Z}_{12} — where 7 + 8 = 3 and everything
wraps at 12 — is not really a separate invention. It is the integers
\mathbb{Z} with the ideal 12\mathbb{Z} (the
multiples of twelve) crushed down to a single point called 0. Doing
arithmetic "mod 12" is arithmetic in the quotient ring
\mathbb{Z}/12\mathbb{Z}. This page pins down what makes a subset
quotient-able, builds the quotient ring, and shows how two flavours of ideal — prime
and maximal — control whether the quotient is an integral domain or a full field.
The absorbing subset
A subring is a subset
closed under addition and multiplication — a ring sitting inside a ring. That is not enough to quotient
by. The extra ingredient is absorption: multiplying any element of the ring
by an element of the ideal must land you back inside the ideal. The ideal is a black hole for
multiplication.
Let R be a commutative ring. A subset
I \subseteq R is an ideal if:
-
Additive subgroup. I is a subgroup of
(R, +): it contains 0, and is closed under
addition and negation (so x - y \in I whenever
x, y \in I).
-
Absorption. For every r \in R and every
x \in I, the product rx \in I. Multiplying by
anything in the ring keeps you inside I.
Compare this with a normal subgroup N \trianglelefteq G,
the exact thing you are allowed to quotient a group by. There the closure condition is conjugation
(gng^{-1} \in N); here it is multiplication. In both cases the rule is
"the ambient structure cannot knock you out of the special subset," and in both cases the pay-off is
the same: you can build a quotient. Absorption is genuinely stronger than the closure a
subring asks for — a subring only multiplies elements of the subring together and stays inside, whereas
an ideal must swallow products with everything in R.
Ideals of the integers are the multiples of n
The cleanest example is \mathbb{Z} itself. Fix an integer
n and take
n\mathbb{Z} = \{\,\dots, -2n, -n, 0, n, 2n, \dots\,\} = \{\, rn : r \in \mathbb{Z}\,\}.
This is an ideal: the difference of two multiples of n is a multiple of
n (subgroup), and any integer times a multiple of n
is again a multiple of n (absorption). The remarkable fact is that
every ideal of \mathbb{Z} looks like this — there are no
others. (Proof sketch: an ideal I \neq \{0\} contains a smallest positive
element n, and dividing any other element by n
with remainder forces the remainder — which lies in I — to be zero.)
Two ideals sit at the extremes, and they exist in every ring:
-
The zero ideal \{0\} = 0\mathbb{Z}, the smallest ideal,
containing nothing but the additive identity.
-
The unit ideal R = 1\mathbb{Z}, the whole ring, the
largest ideal.
An ideal (a) = \{\, ra : r \in R \,\} generated by a single element is
called a principal ideal; in \mathbb{Z} that is exactly
a\mathbb{Z}. Since every ideal of \mathbb{Z} is
principal, we call \mathbb{Z} a principal ideal domain. Here is a
watertight consequence of absorption you will lean on constantly:
If an ideal I contains a unit (an invertible
element u with u^{-1} \in R), then
I = R.
Why: absorption gives u^{-1} u = 1 \in I, and then for
any r \in R we get r \cdot 1 = r \in I.
So the only ideal that touches the units at all is the whole ring. In particular, in a
field — where every nonzero element is a unit — the only ideals are
\{0\} and the field itself.
All the ideals of \mathbb{Z}_{12}
Because every ideal of \mathbb{Z} is n\mathbb{Z},
the ideals of the quotient \mathbb{Z}_{12} correspond exactly to the
divisors of 12. There are six of them — one for each
divisor d \in \{1, 2, 3, 4, 6, 12\}:
- (1) = \mathbb{Z}_{12} = \{0,1,\dots,11\} — the whole ring, 12 elements.
- (2) = \{0,2,4,6,8,10\} — 6 elements.
- (3) = \{0,3,6,9\} — 4 elements.
- (4) = \{0,4,8\} — 3 elements.
- (6) = \{0,6\} — 2 elements.
- (12) = (0) = \{0\} — the zero ideal, 1 element.
Containment reverses divisibility: (d) \supseteq (e) exactly when
d \mid e. So the bigger the ideal (more elements), the smaller the
generator. Drawing the six ideals as nodes and joining each to those it contains gives a
Hasse diagram — and it is precisely the divisor lattice of 12,
with the whole ring (1) at the top and the zero ideal
(0) at the bottom.
Notice the two maximal proper ideals (2) and (3)
hanging directly below the top: nothing sits strictly between them and the whole ring. Those two, we
will see, are precisely the ideals whose quotients are fields — and their generators
2 and 3 are exactly the primes dividing
12.
Building the quotient ring R/I
Now the pay-off. Given an ideal I, glue together any two elements that
differ by something in I. The resulting bundles are the
cosets
a + I = \{\, a + x : x \in I \,\},
and the set of all cosets is the quotient ring R/I. It becomes a ring under
the natural rules:
-
Elements are the cosets a + I, with
a + I = b + I exactly when a - b \in I.
-
Addition:
(a + I) + (b + I) = (a + b) + I.
-
Multiplication:
(a + I)(b + I) = ab + I.
-
Well-defined precisely because of absorption. If you nudge the representatives to
a + x and b + y with
x, y \in I, the product is
ab + (ay + xb + xy); every term in that bracket has a factor from
I, so absorption drops it back into I and the
coset ab + I is unchanged.
That last bullet is the whole reason ideals are defined the way they are. A mere subring would keep
addition well-defined but wreck multiplication — the product could jump to a different coset depending
on which representatives you picked. Absorption is exactly the condition that makes multiplication of
cosets consistent.
The headline example: \mathbb{Z}/n\mathbb{Z} \cong \mathbb{Z}_n. The
cosets 0 + n\mathbb{Z}, 1 + n\mathbb{Z}, \dots, (n-1) + n\mathbb{Z} are the
n "remainder classes," and adding or multiplying them is just
arithmetic mod
n. The clock ring was a quotient ring all along.
Worked example: computing in \mathbb{Z}/6\mathbb{Z}
Take R = \mathbb{Z} and I = 6\mathbb{Z}. The
quotient \mathbb{Z}/6\mathbb{Z} has six cosets, which we write
\bar{0}, \bar{1}, \bar{2}, \bar{3}, \bar{4}, \bar{5} (each
\bar{a} = a + 6\mathbb{Z}). Watch multiplication of cosets in action:
-
\bar{4} + \bar{5} = \overline{4 + 5} = \bar{9} = \bar{3}, because
9 - 3 = 6 \in 6\mathbb{Z}.
-
\bar{4} \cdot \bar{5} = \overline{20} = \bar{2}, because
20 = 3 \cdot 6 + 2.
-
Representative independence: \bar{4} = \bar{10} and
\bar{5} = \bar{-1}, and indeed
\bar{10} \cdot \bar{-1} = \overline{-10} = \bar{2} — same answer,
different representatives.
Notice something uncomfortable: \bar{2} \cdot \bar{3} = \bar{6} = \bar{0},
yet neither \bar{2} nor \bar{3} is zero. The
quotient \mathbb{Z}/6\mathbb{Z} has zero divisors. That is
the fingerprint of a quotient by a non-prime ideal — which brings us to how the ideal shapes
the quotient.
Prime and maximal ideals
The single deepest idea on this page: the type of the ideal is read off from the type of the
quotient ring. Two conditions matter.
Let I be a proper ideal of a commutative ring R with unity.
-
I is prime \iff
R/I is an integral domain (no zero divisors).
Equivalently, ab \in I \Rightarrow a \in I or
b \in I.
-
I is maximal \iff
R/I is a field. (Maximal = no ideal squeezes strictly
between I and R.)
-
Since every field is an integral domain, every maximal ideal is prime — but not
conversely (the zero ideal (0) \subset \mathbb{Z} is prime, because
\mathbb{Z} is a domain, yet not maximal).
In \mathbb{Z} this becomes crisp. The ideal (n)
is prime iff n is 0 or a prime number, and
maximal iff n is prime. That is why
\mathbb{Z}/p\mathbb{Z} = \mathbb{Z}_p is a field exactly when
p is prime: in \mathbb{Z}_5 every nonzero element
has an inverse (2 \cdot 3 = 6 = 1), whereas in
\mathbb{Z}_6 the element \bar{2} has no inverse at
all. Prime generates a domain; prime-and-therefore-maximal generates a field.
The commonest error is to treat "subring" and "ideal" as synonyms. They are not — absorption is
strictly stronger than closure. The cleanest counterexample:
\mathbb{Z} is a subring of \mathbb{Q} (the
integers are closed under + and \times and contain
1), but \mathbb{Z} is not an ideal
of \mathbb{Q}. Absorption fails: take r = \tfrac{1}{2} \in
\mathbb{Q} and x = 3 \in \mathbb{Z}; then
rx = \tfrac{3}{2} \notin \mathbb{Z}. Multiplying an integer by a non-integer
knocked us out of the subset — so you cannot quotient \mathbb{Q} by
\mathbb{Z}.
In fact this is unavoidable: \mathbb{Q} is a field, and a field's only
ideals are \{0\} and \mathbb{Q}. The moment an
ideal contains any nonzero element of a field, it contains a unit, and — as we proved — it swells to the
whole ring. Subrings can be small and interesting; ideals in a field cannot.
The name is a fossil of one of the great near-disasters of number theory. In the 1840s several
mathematicians, chasing Fermat's Last Theorem, tried to factor expressions like
x^p + y^p inside rings of "cyclotomic integers." The strategy only works if
those rings have unique factorization into primes — and Ernst Kummer
discovered, to everyone's dismay, that they often do not. Numbers could split into primes in genuinely
different ways, and the whole approach threatened to collapse.
Kummer's rescue was to invent ideal numbers — phantom factors that restored unique
factorization even when no actual number would. A generation later Richard Dedekind
realised these ghostly objects could be made perfectly concrete: each "ideal number" is really a
set of numbers — an absorbing subset — and he named the sets ideals.
Unique factorization returns, not for elements, but for ideals: in a ring of algebraic integers every
ideal factors uniquely into prime ideals. A failed attack on Fermat thus handed modern algebra one of
its central objects, and prime ideals are its direct descendants.