D-MATH Algebra I HS18

Prof. Rahul Pandharipande

Solution 2
Category theory, first definitions on Rings

1. Prove that a morphism in the category of sets is an isomorphism if and only if it

is a bijective map.
Solution: A morphism in the category of sets is a map f : X −→ Y . The identity
morphism idZ : Z −→ Z is the identity map.
Suppose that f is an isomorphism. Then there exists a map g : Y −→ X such that
g ◦ f = idX and f ◦ g = idY . Let x1 , x2 ∈ X be such that f (x1 ) = f (x2 ). Then

x1 = idX (x1 ) = g(f (x1 )) = g(f (x2 )) = idX (x2 ) = x2 .

This means that f is injective. Moreover, for every y ∈ Y we can write

y = idY (y) = f (g(y)),

so that f is surjective. Hence f is a bijective map.

Conversely, assume that f is a bijective map. For each y ∈ Y , the set

f −1 (y) := {x ∈ X : f (x) = y}

is non-empty because f is surjective. For each x, x0 ∈ f −1 (y), we notice that

f (x) = y = f (x0 ), so that injectivity of f implies x = x0 . This means that for
each y ∈ Y there exists xy ∈ X such that f −1 (y) = {xy }. Define g : Y −→ X as
g(y) := xy . Then ∀x ∈ X, (g ◦ f )(x) = g(f (x)) = xf (x) = x because f : x 7→ f (x).
On the other hand, ∀y ∈ Y, (f ◦ g)(y) = f (xy ) = y. This means that g is an
inverse of the morphism f , so that f is an isomorphism of sets.
2. Let C be a category and A an object of C. Define FA from C to sets by

∀B object of C, FA (B) := HomC (A, B)

 
HomC (A, B) −→ HomC (A, C)
∀f ∈ HomC (B, C), FA (f ) := .
g 7→ f ◦ g
Prove that FA is a functor (it is called the functor represented by A).
Solution: First, notice that FA is well-defined. Indeed HomC (A, B) is defined to
be a set for all objects A and B in C. Moreover, for each f ∈ HomC (B, C) and
g ∈ HomC (A, B), composition in C gives f ◦ g ∈ HomC (A, C).
In order to prove that FA is a functor, we need to check that it maps identity
morphisms to identity morphisms and that it respects compositions.

1
 • Let B be an object of C. Then idB ◦g = g for each morphism g ∈ HomC (A, B)
by definition of identity morphism. This implies that the map

FA (idB ) : HomC (A, B) −→ HomC (A, B)

g 7−→ idB ◦ g

is the identity of HomC (A, B). Hence FA (idB ) = idFA (B) for each object B of
C.
• Let B, C and D be three objects in C and take morphisms f1 ∈ HomC (C, D)
and f2 ∈ HomC (B, C). Then FA (f1 ◦ f2 ) and FA (f1 ) ◦ FA (f2 ) are both maps
HomC (A, B) −→ HomC (A, D). For g ∈ HomC (A, B), we notice that

FA (f1 ◦ f2 )(g) = (f1 ◦ f2 ) ◦ g = f1 ◦ (f2 ◦ g) = f1 ◦ (FA (f2 )(g))

= FA (f1 ) FA (f2 )(g) ,

so that FA (f1 ◦ f2 ) = FA (f1 ) ◦ FA (f2 ) for each pair of composable morphisms

f1 and f2 in C.

3. We want to define a category C as follows:

• An object (X, Y, f ) of C is given by two sets X and Y and a map f : X −→ Y .

• A morphism (u, v) ∈ HomC ((X, Y, f ), (X 0 , Y 0 , f 0 )) is given by maps u : X −→ X 0
and v : Y −→ Y 0 such that the following diagram commutes:
f
X - Y
u v

f0
? ?
X0 - Y0
(a) Define composition of morphisms so that C is indeed a category.
(b) Prove that F from C to sets defined by F ((X, Y, f )) = X and F ((u, v)) = u
is a functor.

Solution:

(a) Notice that the morphisms between two objects (X, Y, f ) and (X 0 , Y 0 , f 0 ) in
C form a set, as they form a subclass of the set HomC ((X, Y, f ), (X 0 , Y 0 , f 0 )).
Given three objects (X, Y, f ), (X 0 , Y 0 , f 0 ) and (X 00 , Y 00 , f 00 ) in C and mor-
phisms (u, v) : (X, Y, f ) −→ (X 0 , Y 0 , f 0 ) and (u0 , v 0 ) : (X 0 , Y 0 , f 0 ) −→ (X 00 , Y 00 , f 00 ),
that is, maps

u : X −→ X 0 , v : Y −→ Y 0 , u0 : X 0 −→ X 00 , v 0 : Y 0 −→ Y 00

2
 such that

f0 ◦ u = v ◦ f (1)
f 00 ◦ u0 = v 0 ◦ f 0 , (2)

we define

(u0 , v 0 ) ◦ (u, v) := (u0 ◦ u, v 0 ◦ v).

This definition is well-given because

(2)
f 00 ◦ (u0 ◦ u) = (f 00 ◦ u0 ) ◦ u = (v 0 ◦ f 0 ) ◦ u
(1)
= v 0 ◦ (f 0 ◦ u) = v 0 ◦ (v ◦ f ) = (v 0 ◦ v) ◦ f.

In order to conclude that C is a category, we need to check existence of iden-

tities and associativity of composition. Those properties follow immediately
from the same property in the category of sets, since we have defined com-
position coordinate-wise. Let us see this very explicitly:
• For each object (U, V, g) of C, consider the morphism e(U,V,g) := (idU , idV ) ∈
HomC ((U, V, g), (U, V, g)). This is an identity of (U, V, g). Indeed, for each
object (X, Y, f ) of C and morphism (u, v) from (X, Y, f ) to (U, V, g), one
has

e(U,V,g) ◦ (u, v) = (idU , idV ) ◦ (u, v) = (idU ◦ u, idV ◦ v) = (u, v),

so that e(U,V,g) is a left unit. Similarly, one can prove that e(U,V,g) is a
right unit.
• Let (X, Y, f ), (X 0 , Y 0 , f 0 ), (X 00 , Y 00 , f 00 ), (u, v) : (X, Y, f ) −→ (X 0 , Y 0 , f 0 )
and (u0 , v 0 ) : (X 0 , Y 0 , f 0 ) −→ (X 00 , Y 00 , f 00 ) be as above, and take a fourth
object (X 000 , Y 000 , f 000 ) and a morphism (u00 , v 00 ) : (X 00 , Y 00 , f 00 ) −→ (X 000 , Y 000 , f 000 ).
Then

((u00 , v 00 ) ◦ (u0 , v 0 )) ◦ (u, v) = (u00 ◦ u0 , v 00 ◦ v 0 ) ◦ (u, v)

= ((u00 ◦ u0 ) ◦ u, (v 00 ◦ v 0 ) ◦ v) = (u00 ◦ (u0 ◦ u), v 00 ◦ (v 0 ◦ v))
= (u00 , v 00 ) ◦ ((u0 ◦ u, v 0 ◦ v)) = (u00 , v 00 ) ◦ ((u0 , v 0 ) ◦ (u, v))

and by arbitrarity of all objects and morphisms involved we can conclude

that the composition law we defined is associative.
(b) Clearly, F associates a set to an object of C and for objects (X, Y, f ), (X 0 , Y 0 , f 0 )
of C and a morphism (u, v) between them, it associates the map u : X =
F ((X, Y, f )) → X 0 = F ((X 0 , Y 0 , f 0 )) of sets. Thus we need to check that F
respects the identity and respects compositions. But indeed, for (U, V, g) an
object of C and e(U,V,g) = (idU , idV ) the identity morphism of (U, V, g), we have

3
 F ((idU , idV )) = idU , the identity of U = F ((U, V, g)). On the other hand, gi-
ven three objects (X, Y, f ), (X 0 , Y 0 , f 0 ) and (X 00 , Y 00 , f 00 ) in C and morphisms
(u, v) : (X, Y, f ) −→ (X 0 , Y 0 , f 0 ) and (u0 , v 0 ) : (X 0 , Y 0 , f 0 ) −→ (X 00 , Y 00 , f 00 ),
we have
F ((u, v) ◦ (u0 , v 0 )) = F ((u ◦ u0 , v ◦ v 0 )) = u ◦ u0 = F ((u, v)) ◦ F ((u0 , v 0 )).
Thus F also respects compositions and hence is a functor.
4. Let R and S be two rings and f : R −→ S a map between them. Prove that f is
a ring isomorphism if and only if it is ring homomorphism and it is bijective.
Solution: Suppose that f : R −→ S is a ring isomorphism. Then, by definition, it is
a ring homomorphism and there exists an inverse ring homomorphism g : S −→ R.
In particular, at level of sets, g is an inverse map, so that, by exercise 1., f is
bijective.
Now suppose that f : R −→ S is a bijective ring homomorphism. Then, by Exercise
1., there exists a map of sets g : S −→ R such that f ◦ g = idS and g ◦ f = idR .
We need to check that g is itself a ring homomorphism. First, notice that
g(1S ) = g(f (1R )) = 1R ,
because f is a ring homomorphism so that f (1R ) = 1S . Now, for s1 , s2 ∈ S, let
r1 , r2 ∈ R be such that f (r1 ) = s1 and f (r2 ) = s2 . Notice that g(s1 ) = r1 and
g(s2 ) = r2 . Then
(∗)
g(s1 + s2 ) = g(f (r1 ) + f (r2 )) = g(f (r1 + r2 )) = r1 + r2 = g(s1 ) + g(s2 )
(∗)
g(s1 · s2 ) = g(f (r1 ) · f (r2 )) = g(f (r1 · r2 )) = r1 · r2 = g(s1 ) · g(s2 )
which allows us to conclude that g is a ring homomorphism. (In the equalities (∗)
above we used the fact that f is a ring homomorphism).
5. (a) Compute the units of Z[i].
(b) (Euclidean division in Z[i]) Let z, w ∈ Z[i] r {0}. Prove that there exist
q, r ∈ Z[i] such that z = q · w + r and |r| < |w|. [Hint: Define q ∈ Z[i] such
that it is a good approximation of wz ∈ C.]
Solution: We will make use of the complex norm N : C −→ R>0 defined by
N (z) = zz. For a + ib ∈ Z[i], this gives N (a + ib) = a2 + b2 ∈ N. Notice that for
each x, y ∈ C
N (xy) = xyxy = xxyy = N (x)N (y). (3)

(a) Let x = a + ib ∈ Z[i]. If x is a unit, then xy = 1 for some y ∈ Z. Then, by

(3),
1 = N (1) = N (x)N (y),

4
 and since N (x), N (y) ∈ N we deduce that N (x) = 1. This means that a2 +b2 =
1. This implies that a2 6 1 and b2 6 1, so that a, b ∈ {−1, 0, 1}. The only
possibilities are (a, b) = (±1, 0) and (a, b) = (0, ±1), which implies that
Z[i]× ⊆ {±1, ±i}. Since those four elements are all units since 12 = (−1)2 =
i · (−i) = 1, which allows us to conclude that Z = {±1, ±i}.
(b) Let u, v ∈ R be such that
z
= u + iv.
w
There exist u0 , v0 ∈ Z such that |u − u0 | 6 21 and |v − v0 | 6 21 . Define
q := u0 + iv0 and r := z − qw. In order to conclude, we need to check that
|r| < |w|, or, equivalently, that | wr | < 1. This is done by noticing that
r z − qw
= = (u − u0 ) + i(v − v0 )
w w
which implies, by definition of complex absolute value, that
 r 2 1 1 1
  = |u − u0 |2 + |v − v0 |2 6 + = < 1.
 
w 4 4 2
6. Let F (R, C) the set of functions R −→ C. Denote by C(R, C) the subset of conti-
nuous functions and by C0 (R, C) the subset of continuous bounded functions.
(a) Check that F (R, C), endowed with pointwise sum and multiplication, is a
commutative ring. Find F (R, C)× .
(b) Prove that C0 (R, C) and C(R, C) are subrings of F (R, C).
(c) Determine C(R, C)× and C0 (R, C)× .
(d) Is C0 (R, C) an integral domain?
(e) Which of the following maps are ring homomorphisms?
i. ϕ : C0 (R, C) −→ C, sending f 7→ f (1);
ii. ψ : C0 (R, C) −→ R, sending f 7→ supx∈R |f (x)|;
iii. η : C(R, C) −→ R, sending f 7→ Re(f (0));
iv. θ : Z −→ F (R, C) sending n ∈ Z to the constant function with value n.
Solution:

(a) The operations + and · on F (R, C) are defined pointwise, that is,
(f + g)(x) := f (x) + g(x)
(f · g)(x) := f (x)g(x).
With a notation abuse, we denote by 0 and 1 the functions R −→ C with
constant value 0 and 1 respectively. Let − : F (R, C) −→ F (R, C) be defined
by (−f )(x) := −f (x). Then the (F (R, C), +, −, ·, 0, 1) satisfies all the axioms
for a commutative. Indeed, for all a, b, c ∈ F (R, C) the following hold:

5
 • ∀x ∈ R, (a + (b + c))(x) = a(x) + (b + c)(x) = a(x) + b(x) + c(x) =
(a + b)(x) + c(x) = ((a + b) + c)(x), so that a + (b + c) = (a + b) + c (sum
is associative);
• ∀x ∈ R, (a + b)(x) = a(x) + b(x) = b(x) + a(x) = (b + a)(x), so that
a + b = b + a (sum is commutative);
• ∀x ∈ R, (0 + a)(x) = 0(x) + a(x) = 0 + a(x) = a(x), so that 0 + a = a (0
is neutral for the sum on the left);
• ∀x ∈ R, (a + (−a))(x) = a(x) + (−a)(x) = a(x) + (−a(x)) = 0 = 0(x),
so that a + (−a) = 0 (the map “−” is an inversion for the sum);
• ∀x ∈ R, (a · (b · c))(x) = a(x)(b · c)(x) = a(x)b(x)c(x) = (a · b)(x)c(x) =
((a · b) · c)(x), so that a · (b · c) = (a · b) · c (product is associative);
• ∀x ∈ R, (a · b)(x) = a(x)b(x) = b(x)a(x) = (b · a)(x), so that a · b = b · a
(product is commutative);
• ∀x ∈ R, (1 · a)(x) = 1(x) · a(x) = 1 · a(x) = a(x), so that 1 · a = a (1 is
neutral for the product on the left);
• ∀x ∈ R, (a · (b + c))(x) = a(x)(b + c)(x) = a(x)(b(x) + c(x)) = a(x)b(x) +
a(x)c(x) = (a·b)(x)+(a·c)(x), so that a·(b+c) = a·b+a·c (distributivity).
Let f ∈ F (R, C)× , with inverse g. This means that

∀x ∈ R, f (x)g(x) = 1.

Then f (x) 6= 0 for every x ∈ R. On the other hand, every non-zero complex
number z ∈ C r {0} is invertible, so that if f ∈ F (R, C) is nowhere zero,
then we can define
1
g(x) :=
f (x)

and this is an inverse of f in F (R, C). Hence

F (R, C)× = {f : R −→ C : ∀x ∈ R, f (x) 6= 0}.

(b) Given a subset R of a ring S, we say that R is a subring of S if the ring

operations on S restrict to R, and R is a ring when endowed with those
restrictions, with 0R = 0S and 1R = 1S . Clearly, if R is closed under the
operations +, − and · of S and it contains 0S and 1S , then the ring axioms
hold for R, since they hold for the whole superset S.
First, notice that C0 (R, C) ⊆ C(R, C) by definition. The constant functions 0
and 1 are continuous and bounded, hence they both belong to C0 (R, C) and
C(R, C). Basic calculus tells us moreover that for f, g continuous functions
the functions f + g, −f and f g are continuous. Hence C(R, C) is a subring
of F (R, C).

6
 Let f, g be bounded functions, that is, suppose there are numbers Mf , Mg ∈
R>0 such that |f (x)| < Mf and |g(x)| < Mg for each x ∈ R. Then for each
x∈R

|(f + g)(x)| 6 |f (x)| + |g(x)| < Mf + Mg

|(−f )(x)| = | − f (x)| = |f (x)| < Mf
|(f · g)(x)| = |f (x)g(x)| = |f (x)| · |g(x)| < Mf Mg

which means that f + g, −f and f · g are bounded functions. This means

that the ring operations on F (R, C) restrict to continuous functions and to
bounded functions and hence to C0 (R, C), which is a subring of F (R, C).
(c) If f ∈ C(R, C)× or f ∈ C0 (R, C)× , then there exists an inverse in the relevant
subring and hence in F (R, C). This implies that

C(R, C)× ⊆ C(R, C) ∩ F (R, C)× , C0 (R, C)× ⊆ C0 (R, C) ∩ F (R, C)× ,

so that by part a) we can restrict our attention to functions f such that

f (x) 6= 0 for all x ∈ R. By basic calculus, when such a function is continuous,
so is the function f1 . This means that

C(R, C)× = C(R, C) ∩ F (R, C)×

= {f : R −→ C|∀x ∈ R f (x) 6= 0 and f is continuous}.

Now let f ∈ C0 (R, C) ∩ F (R, C)× . Since the inverse of an element is unique,
f is invertible in C0 (R, C) if and only if f1 is in C0 (R, C), which is the case
if and only if f1 is bounded (since it is always continuous, as we have just
noticed). Notice that, for all x ∈ R
 
1 
 (x) < Nf ⇐⇒ |f (x)| > 1 ,
f  Nf

so that f is invertible if and only if there exists ε > 0 such that |f (x)| > ε
for all x ∈ X. Hence
  
×
 ∃ε > 0, ∃N > 0 : ∀x ∈ X : ε < |f (x)| < N
C0 (R, C) = f : R −→ C  .
and f is continuous

(d) C0 (R, C) is not an integral domain. Indeed, considering the functions f1 , f2 :

R −→ C

0 if x < 0 or x > 1
f1 (x) := 2
x − x if x ∈ [0, 1]

0 if x < −1 or x > 0
f2 (x) := 2
−x − x if x ∈ [−1, 0],

7
 we see that they are continuous (since f1 (0) = f1 (1) = 0 and f2 (−1) =
f2 (0) = 0) and bounded, since they both have image in R>0 and maximum
value 41 = f1 ( 12 ) = f2 (− 12 ), which shows moreover that f1 6= 0 6= f2 . But
f1 · f2 = 0 because f1 (x) = 0 for x 6 0 and f2 (x) = 0 for x > 0, so that
f1 (x)f2 (x) = 0 for all x ∈ R. Hence f1 and f2 are zero-divisors and C0 (R, C)
is not an integral domain.
(e) i. ϕ is a ring homomorphism. Indeed ϕ(1) = 1(1) = 1, whereas for f, g ∈
C0 (R, C) we observe that

ϕ(f + g) = (f + g)(1) = f (1) + g(1) = ϕ(f ) + ϕ(g)

ϕ(f · g) = (f · g)(1) = f (1)g(1) = ϕ(f )ϕ(g).

ii. ψ is not a ring homomorphism, because it does not respect the sum. For
example, let f, g : R −→ C be defined by

f (x) = sin(x), g(x) = − sin(x).

Then |f (x)| = |g(x)| = | sin(x)| 6 1 for all x ∈ R (so that f, g ∈ C0 (R, C),
and since |f (π/2)| = |g(π/2)| = 1 we see that sup |f | = sup |g| = 1. This
means that ψ(f ) = ψ(g) = 1. Clearly, f + g = 0, so that

ψ(f + g) = 0 6= 2 = ψ(f ) + ψ(g)

and ψ is not a ring homomorphism.

iii. η is not a ring homomorphism, because it does not respect the product.
For example, let f = g : R −→ C be the constant function with value
i. Then f · g = −1, the constant function with value −1. Then, since
Re(i) = 0 and Re(−1) = −1,

η(f g) = −1 6= 0 = 0 · 0 = η(f )η(g)

and η is not a ring homomorphism.

iv. θ is a ring homomorphism. Indeed, 1Z is mapped to the constant function
of value 1, and for each n, m ∈ Z, the sum (resp., the product) of the
function of constant value n with the function of constant value m is the
function of constant value n + m (resp., nm).

7. Let F2 ∼
= Z/2Z be the field with two elements 0, 1. Define
  
a b
R := : a, b ∈ F2 .
b a+b

(a) Prove that R is a commutative ring under the usual matrix sum and multi-
plication.

8
 (b) Prove that R is a field with exactly four elements.

Solution:
   
0 0 1 0
(a) As usual, the matrices and (obtained for (a, b) = (0, 0)
0 0 0 1
and for (a, b) = (1, 0) respectively) are seen to be neutral elements for + and
· respectively. Moreover, for each a, b, a0 , b0 ∈ F2 , we see that
  0
b0 a + a0 b + b0
   
a b a
+ =
b a+b b 0 a0 + b 0 b + b0 (a + a0 ) + (b + b0 )
 0
b0 aa0 + bb0 ab0 + a0 b + bb0
   
a b a
=
b a+b b 0 a0 + b 0 a0 b + ab0 + bb0 bb0 + aa0 + ab0 + ba0 + bb0

and both results still belong to R. As can be proven in general, sum of matrices
is commutative and associative, whereas multiplication is associative. This
proves that R is a ring. Moreover, one can check the commutativity from
the above equation by noticing that the result of the multiplication does not
change after switching a with a0 and b with b0 .
(b) There are four choices of parameters (a, b) ∈ F22 . Since the first row of the
matrix is (a, b), eachchoice   matrix. Hence |R| = 4. Those
gives adifferent
0 1 1 1
matrices are 0R , 1R , and . Notice that 1 · 1 = 1 and
1 1 1 0
     
0 1 1 1 1 0
· = = 1R ,
1 1 1 0 0 1

so that each non-zero matrix is invertible in the commutative ring R. Hence

R is a field with 4 elements.

8. Let R be a finite integral domain. Prove that R is a field. [Hint: For each x ∈
R r {0}, consider the map R −→ R sending a 7→ ax. Is it injective/surjective?]
Solution: Let x ∈ R r {0}. Call fx : R −→ R the map a 7→ ax. Suppose that
fx (a) = fx (b) for a, b ∈ R. Then (a − b)x = ax − bx = fx (a) − fx (b) = 0 and
since R is an integral domain and x 6= 0 we deduce that a − b = 0, so that a = b.
This implies that fx is injective. Since R is a finite set, fx is also surjective. In
particular, there exists y ∈ R such that yx = fx (y) = 1R , meaning that x has a
left inverse. Being R commutative, x has a right inverse as well. By arbitrarity of
x ∈ R r {0}, we can conclude that R is a field.