STANDARD DEFINITIONS FOR RINGS

KEITH CONRAD

1. Introduction
Rings generalize systems of numbers and of functions that can be added and multiplied.
Definition 1.1. A ring is a set R equipped with two operations + (addition) and ×
(multiplication) such that R is an abelian group under addition (with identity denoted 0
and the inverse of a denoted −a), while multiplication is associative with an identity element
1 (meaning 1 · x = x · 1 = x for all x in R). Finally, multiplication distributes over addition:
x(y + z) = xy + xz and (x + y)z = xz + yz for all x, y, and z in R.
We say R is a commutative ring if multiplication on R is commutative, and otherwise we
say R is a noncommutative ring.1
This says a ring is a commutative group under addition, it is a “group without inverses”
under multiplication, and multiplication distributes over addition. Examples of rings are Z,
Q, all functions R → R with pointwise addition and multiplication, and M2 (R) – the latter
being a noncommutative ring – but 2Z is not a ring since it does not have a multiplicative
identity.
Some abstract algebra books do not insist rings have a multiplicative identity, leading to
the result that 2Z is considered a subring of Z. This is really, really bad. Below we will
give the correct definitions of subring, ring homomorphism, and ideal. Our definitions will
be the right ones even in the case of noncommutative rings, but little will be lost if you try
to get your bearings by supposing throughout that R is commutative. In an appendix we
will discuss what “rings without a multiplicative identity” should be called.

2. Subrings
Definition 2.1. A subring of a ring R is a subset R0 ⊂ R that is a ring under the same +
and × as R and shares the same multiplicative identity.
Example 2.2. The ring Z is a subring of Q. The ring Z/(m) for m > 0 has no subrings
besides itself: 1 additively generates Z/(m), so a subring contains 1 and thus contains
everything. The same argument (using m = 0) shows Z has no subrings other than itself.
It might seem odd to insist in the definition of a subring that it has the same multiplicative
identity as the original ring. Should that follow from the rest of the definition? After all, a
subgroup of a group is defined to be a subset that is a group for the same operation, and its
identity element can be proved to be the identity for the original group (and inverses for the
subgroup are therefore the same as for the original group). But that proof uses cancellation
in the group law, and in a ring we might not have cancellation for multiplication. This is
made clearer with an example.
1The terms “commutative group” and “abelian group” are synonyms, but there is no analogue of the second
term in ring theory: when R has commutative multiplication, nobody calls it an “abelian ring”.
1
2 KEITH CONRAD

Example 2.3. In Z/(6), the subset {0, 3} with addition and multiplication mod 6 is a ring
in its own right with identity 3 since 32 = 9 = 3. So {0, 3} is a subset of Z/(6) “with a ring
structure”. Its multiplicative identity is not the multiplicative identity of Z/(6), so we do
not consider {0, 3} to be a subring of Z/(6).
Remark 2.4. If the ring R has cancellation for multiplication (that is, xz = yz ⇒ x = y
in R if z 6= 0) then a subset of R “with a ring structure” other than {0} has to have the
same multiplicative identity as R (and thus is a subring) because if x is the multiplicative
identity in a subset “with a ring structure” then the equation x2 = x is satisfied, which
is the same as x · x = x · 1, forcing x = 1 if x 6= 0. Thus for rings with cancellation, the
constraint on a nonzero subset that it have the same multiplicative identity as the whole
ring is automatic from the other properties of a subring.
You might be thinking: what is the big fuss about subrings having the same identity for
multiplication? One reason for wanting this has to do with invertible elements. An element
x ∈ R is called a unit if it has a 2-sided inverse: xy = yx = 1 for some y ∈ R. The set
of all units forms a group, denoted R× . For example, R× = R − {0}, Z× = {±1}, and
Mn (R)× = GLn (R) = {A ∈ Mn (R) : det A 6= 0}.
Theorem 2.5. If R is a ring and R0 is a subring then R0× is a subgroup of R× .
Proof. Let 1 be the multiplicative identity in R, so it is also the multiplicative identity in
R0 . Since R0 has the same multiplicative identity as R, if x ∈ R0× then xy = yx = 1 for
some y ∈ R0 , so x ∈ R× and the inverse of x in R0 is also its inverse in R. We have shown
R0× is a subset of R× . Since the group law (multiplication) and inversion in R0 are the same
as in R, R0× is a subgroup of R× . 
Example 2.6. We return to the nonexample of {0, 3} in Z/(6). As a subset “with a ring
structure,” {0, 3} has multiplicative identity element 3, which is not a unit in Z/(6). So the
one unit in the “ring that’s not a subring” {0, 3} is not a unit in Z/(6).
It would be weird if the units in a subring are not units in the larger ring, and insisting
that subrings have the same multiplicative identity as the whole ring means this weirdness
will not happen: units of a subring are units of the larger ring.

3. Ring homomorphisms
Definition 3.1. If R and S are rings, a ring homomorphism f : R → S is a function that
preserves addition, multiplication, and the multiplicative identity: f (x + y) = f (x) + f (y)
and f (xy) = f (x)f (y) for all x and y in R, and f (1) = 1.
The last condition, that f (1) = 1, is admittedly an awkward part of the definition, since
we don’t require in the definition that f (0) = 0 too. However, it is automatic that f (0) = 0
because f is an additive group homomorphism, and group homomorphisms always preserve
the identity. But a ring is not a group under multiplication (except for the zero ring), and
if we don’t insist that f (1) = 1 as part of a ring homomorphism then weird things can
happen. Consider the next example, which builds on the previous one.
Example 3.2. Let f : Z/(6) → Z/(6) by f (x) = 3x. Since 32 = 3 in Z/(6), we have
f (x) + f (y) = 3x + 3y = 3(x + y) = f (x + y) and also (the key point) f (x)f (y) = 3x · 3y =
32 xy = 3xy = f (xy). Thus f is additive and multiplicative, but f (1) = 3 6= 1. We do not
want to call f a ring homomorphism, and requiring f (1) = 1 rules out this example.
 STANDARD DEFINITIONS FOR RINGS 3

The only ring homomorphism Z/(6) → Z/(6) is the identity function: once 1 goes to 1
everything else is fixed too by additivity.
Here is a result involving units that would break down if a ring homomorphism did not
preserve the multiplicative identities.
Theorem 3.3. Let f : R → S be a ring homomorphism. Then f (R× ) ⊂ S × and the
function f : R× → S × is a group homomorphism.
Proof. If xy = yx = 1 in R then applying f gives us f (x)f (y) = f (y)f (x) = f (1) = 1, so f
sends units in R to units in S. Since f is multiplicative, it is a group homomorphism from
R× to S × . 
Theorem 3.4. If R0 is a subring of R then the inclusion mapping R0 ,→ R is a ring
homomorphism.
Proof. Easily the inclusion map sends sums to sums and products to products. The multi-
plicative identity goes to the multiplicative identity because R0 has the same multiplicative
identity as R. 
In group theory, the kernel and image of a group homomorphism are subgroups. For a
ring homomorphism f : R → S, we have the kernel ker f = {x ∈ R : f (x) = 0} and image
f (R). Are these subrings (of R and S respectively)?
Theorem 3.5. Let f : R → S be a ring homomorphism. The image of f is a subring of S,
but the kernel of f is not a subring of R unless S is the zero ring.
Proof. From the definition of a ring homomorphism, the sum and product of f -values are
f -values. The image also contains 1 since f (1) = 1. So the image of f is a subring of S.
The kernel of f is closed under addition and multiplication. The kernel of f is not a
subring of R unless 1 is in the kernel which means f (1) = 0. Since f (1) = 1 by definition,
we must have 1 = 0 in S, so S is the zero ring (the only ring in which 1 = 0). 
This last theorem is probably why some people do not insist that rings contain 1. Ker-
nels of ring homomorphisms have all the properties of a subring except for almost never
containing the multiplicative identity. So if we want ring theory to mimic group theory by
letting kernels of ring homomorphisms be subrings, then we should not insist that subrings
contain 1 (and thus perhaps not even insist that rings contain 1). Then kernels of ring
homomorphisms could be called subrings. The development of ring theory, particularly for
commutative rings, has shown that this is a bad idea. Kernels of group homomorphisms
are special kinds of subgroups (normal subgroups), but kernels of ring homomorphisms are
something other than subrings. What are they? That is the subject of the next section.

4. Ideals
The kernel of a ring homomorphism satisfies a stronger multiplicative condition than
being closed under multiplication: if f : R → S is a ring homomorphism and x ∈ ker f ,
so f (x) = 0, then for all r ∈ R we have f (rx) = f (r)f (x) = f (r) · 0 = 0 and f (xr) =
f (x)f (r) = 0 · f (r) = 0, so rx and xr are in the kernel too. The kernel of f is closed under
multiplication by arbitrary elements of R from either side. Contrast this with Z as a subring
of Q: multiplication of an integer by most elements of Q is not again an integer.
Definition 4.1. An ideal in a ring R is an additive subgroup I ⊂ R such that RI ⊂ I and
IR ⊂ I. That is, if x ∈ I then Rx ⊂ I and xR ⊂ I: all multiples of x in R lie in I.
4 KEITH CONRAD

Example 4.2. A basic example of an ideal in a commutative ring R is the multiples of one
element: for a ∈ R, Ra = {ra : r ∈ R} is an ideal in R since a sum and difference of two
multiples is again a multiple and (most importantly) every multiple of a multiple is again
a multiple. These ideals are called principal ideals. For instance, the even numbers 2Z are
a principal ideal in the ring Z but they are not a subring of Z.
If R is noncommutative then this attempt to construct an ideal runs into trouble when
you switch the side you multiply on. If an ideal contains a then it contains not only left
multiples of a but also right multiples, and in fact multiples from both sides taken together,
which is the set {ras : r, s ∈ R}. But this set is usually not closed under addition if R
is noncommutative. So we have to take finite sums of these two-sided products, getting
r1 as1 + · · · + rn asn for n ≥ 1 and ri , si ∈ R. Now that is an ideal. Very tedious! This is why
you should not try to learn about ideals first in noncommutative rings. It’s too complicated.
Focus on ideals in the commutative setting until you get used to them.
Example 4.3. An ideal in a commutative ring that is not of the special form Ra is the
polynomials in Z[T ] that have an even constant term: I = {f (T ) ∈ Z[T ] : f (0) is even}.
Examples of elements of I are 2, T , and T 2 + 3T + 10. Check yourself that I is an ideal in
Z[T ]. We will show by contradiction that I is not a principal ideal. Assume I is a principal
ideal, so I = Z[T ]f (T ) for some f (T ). Since 2 ∈ I, 2 = g(T )f (T ) for some g(T ), so f (T )
has to be a constant polynomial. Write f (T ) = c. Then 2 = g(T )c, so c = ±1 or ±2. Since
c is in the ideal, it must be even, so c = ±2. Because T ∈ I, T = h(T )c for some h(T ), but
T on the left side of the equation has leading coefficient 1 and h(T )c on the right side has
an even leading coefficient. That’s a contradiction, so I is not principal.
While the ideal I is not generated by a single element, it is generated by two elements.
The general element of I is a linear combination of 2 and T , with coefficients in Z[T ]. We
can write I symbolically as 2Z[T ] + T Z[T ], or as 2Z + T Z[T ].
As David Rohrlich has nicely put it, ideals are “contagious for multiplication.” That may
help you remember their defining property when you’re first working with them. I like to
say ideals swallow up multiplication.
Example 4.4. An important way ideals occur in mathematics is as kernels of ring ho-
momorphisms. The kernel of a ring homomorphism is an ideal, and conversely one can
show an ideal in a ring can be viewed as the kernel of a suitable ring homomorphism using
quotient rings (analogous to quotient groups). Thus ideals in a ring are analogous to the
normal subgroups of a group: the kernel of a group homomorphism is a normal subgroup
and the quotient group construction shows a normal subgroup is the kernel of some group
homomorphism.
While Z is an additive subgroup of R, it is not an ideal in R since real numbers times
integers are usually not integers. Similarly, Z is a subgroup of Q but is not an ideal of Q.
More generally, a subring of a ring R is not an ideal of R unless it’s all of R: if R0 is a
subring of R and also R0 is an ideal of R, then since 1 ∈ R0 (!) we get for all r ∈ R that
r = r · 1 ∈ R0 . Thus R0 = R. So except for the whole ring, which is both a subring and
ideal of itself, subrings and ideals are absolutely separate concepts.

Appendix A. Rings without identity

Having explained why rings should have a multiplicative identity (and what this implies
about the correct definitions of subring and ring homomorphism), we should admit that
“ring-like” systems without a multiplicative identity do occur, especially in analysis.
 STANDARD DEFINITIONS FOR RINGS 5

Example A.1. Consider continuous functions R → R with limit 0 as x → ±∞. Examples

are 1/(x2 + 1) and (sin x)e−|x| . The set of such functions is denoted C0 (R). See Figure 1.

Figure 1. Functions in C0 (R).

Under pointwise addition and multiplication, C0 (R) satisfies the definition of a ring
except that it does not have a multiplicative identity. If there were a multiplicative identity
in C0 (R) then it would have to be the constant function 1, which does not belong to C0 (R).
Besides pointwise multiplication of functions, another important multiplicative operation
on functions in analysis is convolution; see https://en.wikipedia.org/wiki/Convolution
for the definition. Often there is no identity for convolution, so functions under addition
and convolution form another example in analysis of a ring without an identity.
On account of these examples (in analysis), what can we call a “ring without identity” if
we don’t call it a ring? There is already an available term: a “ring without identity” is an
associative Z-algebra in the sense of the following definition (with R = Z).
Definition A.2. Let R be a commutative ring (with multiplicative identity 1). An R-
algebra is an abelian group A with operation + that admits a multiplication A × A → A
and scalar multiplication R × A → A. Denoting the multiplication of a, b ∈ A as ab and the
scalar multiplication of r ∈ R and a ∈ A as ra, the conditions on these multiplications are
(1) R-bilinearity of A × A → A:
• a(b + c) = ab + ac, and (a + b)c = ac + bc for all a, b, and c in A,
• r(ab) = (ra)b = a(rb) for all r in R and all a and b in A,
(2) Scalar multiplication R × A → A is a ring homomorphism R → End(A):2
• r(a + b) = ra + rb for all r in R and a, b in A,
• (r + s)a = ra + sa for all r, s in R and all a in A,
• (rs)(a) = r(sa) for all r and s in R and a in A,
• 1 · a = a for all a in A, where 1 is the identity in R.
An R-algebra A is called commutative or associative if multiplication A × A → A is
commutative or associative, respectively. We say A has an identity if it has a multiplicative
identity.
For an R-algebra A, R is a distinguished ring by which we can multiply elements of A,
and R need not lie inside A. This is analogous to real vector spaces, whose elements can be
2The scalar multiplication conditions say A is an R-module, if you know what that means.
6 KEITH CONRAD

scaled by real numbers even though real numbers are usually not in the vector space (an
example where they are is C thought of as a real vector space).
Example A.3. The ring Mn (R) for n ≥ 1 is an R-algebra where we multiply a scalar and
a matrix in the usual way. It is associative for all n and commutative for n = 1, and has
the n × n identity matrix In as an identity.
Example A.4. The ring Mn (C) for n ≥ 1 is a C-algebra in two ways since we can define
the product of a scalar z and matrix M in the usual way as zM or in the “twisted” way as
zM . Since multiplication in Mn (C) is matrix multiplication, both C-algebra structures on
Mn (C) are associative for all n and commutative for n = 1, with identity In .
Example A.5. The set C([0, 1], R) of continuous functions [0, 1] → R is an R-algebra
under pointwise addition and multiplication. It is commutative and associative, and it has
the constant function 1 as an identity.
Example A.6. The set C0 (R) defined above is an R-algebra under pointwise addition and
multiplication. It is commutative and associative and has no identity.
Example A.7. Banach algebras and C ∗ -algebras are special types of associative algebras
over R or C in analysis.
Example A.8. The product ring Z/(5) × Z/(5) is a Z-algebra, where scalar multiplication
is n · (x mod 5, y mod 5) = (nx mod 5, ny mod 5). Similarly, the ring Z/(5) × Z/(5) is a
Z/(d)-algebra when d ≡ 0 mod 5 by (n mod d)·(x mod 5, y mod 5) = (nx mod 5, ny mod 5).
If S is a ring containing a subring R then every S-algebra is an R-algebra, where we use
the multiplication in S to define how to multiply by elements of R. This is like a feature of
linear algebra: a complex vector space is also a real vector space.
Example A.9. Every R-algebra is also a Z-algebra and a Q-algebra.
Each integer is obtained from the integer 1 by successive addition or negation, so if an
abelian group A has a multiplication A×A → A that’s biadditive (meaning a(b+c) = ab+ac
and (a + b)c = ac + bc for all a, b, c ∈ A) then A is a Z-algebra in exactly one way.
A ring is the same thing as an associative Z-algebra with identity. A “ring possibly
without identity” is the same thing as an associative Z-algebra.
An example of a nonassociative R-algebra is R3 with usual addition and the cross product
as multiplication: x × (y × z) 6= (x × y) × z in general and there is no cross product identity
vector. Another nonassociative R-algebra is Mn (R) using usual addition and the bracket
operation [A, B] = AB − BA as multiplication. This doesn’t have an identity either: for
no A is [A, B] = B for all B. (For instance, the matrix [A, B] has trace 0, so [A, B] 6= B
when B is chosen to have nonzero trace.) These nonassociative algebra structures on R3
and Mn (R) are examples of a Lie algebra.3
Poonen [1] has explained why a ring multiplication’s associativity, in a suitably general
form, implies all rings should have 1 and how the role of 1 should be related to subrings
and ring homomorphisms in the way described in Sections 2 and 3.

References
[1] B. Poonen, Why all rings should have a 1, Mathematics Magazine 92 (2019), 55–62, URL https://math.
mit.edu/∼poonen/papers/ring.pdf.

3In contrast to footnote 1, a Lie algebra with commutative multiplication can be called abelian, This is per-
haps due to Lie algebras developing out of group theory (they were originally called “infinitesimal groups”).