Homological Algebra - Problem Set 1 - Solutions

Problem 1. Let Cn = Z/8Z, n Z. Let d : Cn Cn1 be the map induced by

multiplication by 4 considered modulo 8. Show that d2 = 0 and compute the homology
groups of the resulting complex of abelian groups.
Solution: The map d2 : Z/8Z Z/8Z corresponds to multiplication by 16 which is
congruent to 0 modulo 8, hence d2 = 0. The kernel of d is the subgroup of Z/8Z generated
by the residue class of 2, the image of d is the subgroup of Z/8Z generated by the residue
class of 4. The homology group at any n is given by the quotient group of kernel by
image which is isomorphic to Z/2Z.

Problem 2. Find a free resolution of the C[x, y]-module C[x, y]/(x, y).
Solution: We denote R = C[x, y]. The R-module M = C[x, y]/(x, y) is generated by
the class of 1, hence we obtain a surjection

R M, 1 7 1
The kernel of this surjection is the R-module given by the ideal I = (x, y) R. This
module has a set of generators given by {x, y} and hence we obtain a surjection
 

a
2
7 ax + by.
R I,
b
Let K denote the kernel of . An element (a, b) R2 is in K if ax+by = 0. Reformulating
this equality, we obtain
ax = by
which implies that a is divisible by y and b is divisible by x so that we may write a = ry
and b = sx. Substituting in the above equation, we have
rxy = sxy
which implies r = s. In other words, K R2 is the R-submodule generated by (y, x).
Therefore, we obtain a surjection
 

y
.
R K, 1 7
x
Since the R-submodule K is free of rank 1, the map has trivial kernel. Assembling the
free modules arising in this resolution process, we obtain the free resolution of M given
by



x y
x
/0
/R
/ R2
/R

equipped with the augmentation map R M, 1 7 1.


Problem 3. For each of the following simplicial complexes determine explicity the simplicial chain complex with coefficients in R and compute its homology.
(1) The simplicial complex on {0, 1, 2, 3} given by all subsets.
(2) The simplicial complex on {0, 1, 2, 3} given by all subsets of cardinality 2.

Solution: (1) The module of simplicial n-chains is the free vector space on the set Kn
of n-simplices of the simplicial complex K. Choosing bases according to the orderings
K3
K2
K1
K0

= {{0, 1, 2, 3}}
= {{1, 2, 3}, {0, 2, 3}, {0, 1, 3}, {0, 1, 2}}
= {{0, 1}, {0, 2}, {0, 3}, {1, 2}, {1, 3}, {2, 3}}
= {{0}, {1}, {2}, {3}}

we obtain the complex of R-vector spaces

/0

/R

d3

/ R4

d2

/ R6

d1

/ R4

(1)

with

1
1

d3 =
1
1

0
0
1
1
0
1
0 1

0 1 1
0
d2 =
1
0
0
1

1
0
1
0
1
1
0
0

1 1 1
0
1
0
0
1
d1 =
0
1
0
1
0
0
1
0

0
0
1
0
.
0 1
1
1

Gauss-Jordan elimination shows rank(d3 ) = 1, rank(d2 ) = 3, and rank(d1 ) = 3 where

rank(di ) is defined to be the dimension of the image of di . But, since, for any linear map
of finite dimensional vector spaces
:V W
we have V / ker()
= im(), and, in particular, dim(ker()) = dim(V ) dim(im()), we
obtain dim(ker(d2 )) = 4 3 = 1, and dim(ker(d1 )) = 6 3 = 3. Comparing dimensions,
we obtain that we only have one nontrivial homology group H0
= R. The simplicial
chain complex for part (2) is obtained from (1) by omitting the modules of 3-chains and
2-chains. We hence obtain nontrivial homology in degrees 0 and 1 where we have H0
=R
3
and H1
R
.

=
Problem 4. Let k be a field and let A be an associative k-algebra. We define, for every
n 0, the k-vector space
Cn (A) = A k A k k A .
{z
}
|
n + 1 copies

Further, we define maps d : Cn (A) Cn1 (A) by k-linearly extending the formula
d(a0 a1 an ) = (a0 a1 ) a2 an
+

n1
X

(1)i a0 (ai ai+1 ) an

i=1

+ (1)n (an a0 ) a1 an1 .

Show that d2 = 0 so that C (A) forms a chain complex of vector spaces. Explicitly
determine the complex C (k) and compute its homology.
Solution: This problem is easier to handle, and seems somewhat more natural, when
couched in the language of simplicial objects.
Definition 0.1. Let C be a category. A simplicial object in C consists of the data
({Cn }nN , {in }nN,0in , {sni }nN,0in )
where each Cn is an object of C and
in : Cn Cn1
sni : Cn Cn+1
are morphisms satisfying the simplicial identities
n
(1) in jn+1 = j1
in+1 , i < j

n1 n
(2) in+1 snj = sj1
i , i < j

(3) in+1 snj = idCn , i = j or i = j + 1

n
(4) in+1 snj = sjn1 i1
, i>j+1

n
(5) sn+1
snj = sn+1
i
j+1 si , i j

The in are called face maps, and the snj degeneracy maps.
Let R be a unital, commutative ring, and consider the category Rmod of R modules.
Let (Mn , in , sni ) be a simplicial object in R mod. We call this collection a simplicial
R-module. Define the operators dn+1 : Mn+1 Mn (we will use the convention that
Mn = 0 for n < 0),
 Pn
i n
n>0
i=0 (1) i
dn =
(2)
0
n0
Proposition 0.2.
dn dn+1 = 0
3

Proof.
n n+1

d d

=d

n+1
X

(1)j jn+1

j=0

n X
n+1
X

(1)i+j in jn+1

j1
n+1 X
X

(1)i+j in jn+1

j1
n+1 X
X

n
(1)i+j j1
in+1 +

j1
n X
X

n
(1)i+j+1 j1
in+1

i=0 j=0

j=1 i=0

n X
n
X
j=0 i=j

n
n X
X
(1)i+j in jn+1
j=0 i=j

j=1 i=0

(1)i+j in jn+1

j=0 i=0

n X
n
X

(1)i+j in jn+1 (after reindexing in the first summand)

j=0 i=j

=0

Thus a simplicial R-module gives rise to a chain complex.

...

/M

dn
n

/ Mn1 dn1 / . . .

/M

d1
1

/M

/0

To see why all this might be useful, consider the following example. Let X =
(Xn , in , snj ) be a simplicial object in Set (a simplicial set). For each n N, let R{Xn }
be the free R-module on Xn . Extend the face and degeneracy operators of {Xn }nN to
morphisms of the corresponding free R-modules by linearity. This clearly gives rise to a
simplicial R-module. The homology of the resulting complex is called the homology of
X with coefficients in R, and is denoted
H (X; R)

Without too much effort it is possible to regard an abstract simplicial complex as defined
in lectures as a simplicial set. The homology of this simplicial set will be the same as the
homology of a simplicial complex as defined in the lectures. Hence, homology of simplicial
modules provides a generalisation of the homology of simplicial complexes, and this is at
least a tentative reason to consider them.
The complex which appears in the problem sheet is also a complex arising from a simplicial R-module. It is called the Hochschild complex (of the algebra A with coefficients in
A).
More generally, let A be an associative R-algebra. Recall that an A A bimodule M is

an abelian group together with both left and right actions of A, which are compatible in
the sense that
(a.m).b = a.(m.b)
for all a, b A and m M .
Set
Mn = M An
Define face maps

i=0
ma1 a2 . . . an
m a1 . . . ai ai+1 . . . an 0 < i < n
in (m a1 . . . an ) =

an m a1 . . . an1
i=n

and degeneracy maps by

si (m a1 . . . an ) = a1 . . . ai 1 ai+1 . . . an
Proposition 0.3. With Mn , in , sni as above, (Mn , in , sni ) is a simplicial R-module.
Proof. We will only prove the identity
n
in jn+1 = j1
in+1 , i < j

(as it is actually the only one needed to answer the question on the problem sheet; note
that the map 2, and the proof of proposition 0.2 that it is in fact a differential, only use
the face maps and the relations between them).
We only need to verify the formula on pure tensors
m a1 . . . an+1
There are several cases. First let 0 < i < j < n + 1. Then
in jn+1 (m a1 . . . an+1 )
= in (m a1 . . . aj aj+1 . . . an+1 )

m a1 . . . ai ai+1 . . . aj aj+1 . . . an+1 j 6= i + 1
=
m a1 . . . aj1 aj aj+1 . . . an+1
j =i+1
n
in+1 (m a1 . . . ai ai+1 . . . an+1 )
j1
n
(m a1 a2 . . . an+1 )
= j1

m a1 . . . ai ai+1 . . . aj aj+1 . . . an j 6= i + 1
=
m a1 . . . aj1 aj aj+1 . . . an
j =i+1

Note that we have indeed used the fact that i < j.

The remaining calculations are largely similar, but we include them for completeness.
Let i = 0, 0 < j < n + 1.
0n jn+1 (m a1 . . . an+1 )
= 0n (m a1 . . . aj aj+1 . . . an+1 )

ma1 . . . aj aj+1 . . . an+1 j > 1
=
ma1 a2 . . . an+1
j=1

n
j1
0n+1 (m a1 . . . an+1 )
n
= j1
(ma1 a2 . . . an+1 )

ma1 . . . aj aj+1 . . . an+1 j > 1
=
ma1 a2 . . . an+1
j=1

Next let 0 < i n, j = n + 1

n+1
in n+1
(m a1 . . . an+1 )
n
= i (an+1 m a1 . . . aj aj+1 . . . an )

an an+1 m a1 . . . an+1
i=n
=
an+1 m a1 . . . ai ai+1 . . . an+1 i < n

nn in+1 (m a1 . . . an+1 )
= nn (ma1 a2 . . . ai ai+1 . . . an+1 )

an an+1 m a1 . . . an+1
i=n
=
an+1 m a1 . . . ai ai+1 . . . an+1 i < n
Finally i = 0, j = n + 1
n+1
0n n+1
(m a1 . . . an+1 )
n
= 0 (an+1 m a1 . . . aj aj+1 . . . an )

0
n=0
=
an+1 m a1 a2 . . . ai ai+1 . . . an+1 n > 0

nn 0n+1 (m a1 . . . an+1 )
= nn (ma1 a2 . . . ai ai+1 . . . an+1 )

0
n=0
=
an+1 m a1 . . . ai ai+1 . . . an+1 n > 0

The homology of the corresponding complex is called the Hochschild homology of A

with coefficients in M , denoted
HH (A, M )
The complex C (k) simply yields

/k

/k

/k

/k

/k

so that we have H0
= k and Hi
= 0 for i > 0.

Problem 5. Let R be a ring and let f : C D be a morphism of chain complexes of
R-modules.
(1) Show that f preserves cycles and boundaries and hence induces, for every n, a map
Hn (f ) : Hn (C) Hn (D) of homology modules. Show that this defines, for every
n, a functor from the category of chain complexes of R-modules to the category of
R-modules.
(2) Recall that f is called a quasi-isomorphism if, for every n, the map Hn (f ) is an
isomorphism. A morphism g : D C of chain complexes is called a quasi-inverse
of f if, for every n, the map Hn (g) is the inverse of Hn (f ). Give an explicit example
of a quasi-isomorphism of chain complexes of abelian groups which does not admit
a quasi-inverse (provide a proof).
Solution: Part (1) is readily verified. An example for a quasi-isomorphism which does
not have a quasi-inverse is given as follows. Consider the complex C with nonzero chains
C0 = Z, C1 = Z and differential
2

/Z

/Z

/0

and the complex D with nonzero chains D0

= Z/2Z, and hence zero differentials. Considet the morphism C D given by

/0

/Z

/Z

/0


/ Z/2Z


/0

/0

/0

where the nonzero map C0 D0 is the quotient map Z Z/2Z. This morphism is a
quasi-isomorphism. However, the only map from Z/2Z Z is the zero map and hence,
the only morphism of complexes D C is the zero map in each component which is
certainly not a quasi-inverse since both complexes have nontrivial homology H0
= Z/2Z.

Problem 6. Consider a commutative diagram of R-modules of the form
X

/Y
g

/ Z

/0

/Y
/Z
/X
0
with exact rows. Show that there is an exact sequence

ker(f ) ker(g) ker(h) coker(f ) coker(g) coker(h).

7

Solution: We have the diagram

ker(f )

/ ker(g)

/ ker(h)


/Y


/ Z




/X

/Y

/Z


/ coker(g)


/ coker(h)

coker(f )

/0

Where the vertical arrows in the top and bottom rows are well-defined by the commutativity of the original diagram. Before defining the connecting homomorphism
ker(h) coker(f )
let us check exactness at ker(g) and coker(g) Let y ker(g) and suppose (x) = 0. By
exactness at Y there is some x X such that (x ) = y . But f (x ) = g (x) = 0
and is injective by exactness. Hence x ker(f ). This shows exactness at ker(g).
Now let y + im(g) coker(g) be such that (y) + im(h) = 0, i.e. (y) = h(z ) for
some z Z . Now z = (y ) for some y Y by exactness. Then g(y ) =
h (y ) = h(z ) = (y). So (y g(y )) = 0. Hence y g(y ) = (x) for some x X,
and y + img = (x) + imf as required.
Now let z ker(h). Then z = (y ) for some y Y by exactness. Moreover
0 = h(z ) = h (y ) = g(y ). By exactness g(y ) = (x) for some unique
x X. Suppose there is y such that (y ) = z = (y ). Then (y y ) = 0,
so y = y + (x ) for some x X by exactness. Then g(y ) = g(y ) + f (x ).
Let (x) = g(y ), (x ) = g(y ). Then by injectivity of , we have x = x + f (x ), in
particular x + imf = x + imf . Thus, we get a well-defined map sending z ker(h) to
x + imf coker(f ) where (x) = g(y ) for some y with z = (y ). It is evidently a
morphism of R-modules. We denote this morphism by .
Suppose (z ) = 0. Let x be such that (x) = g(y ) for some y with z = (y ).
Then x = f (x ) for some x X . Then (x) = g (x ), so y (x ) ker(g) and
(y (x )) = z .
Suppose (x) + im(g) = 0, i.e. (x) = g(y ) for some y Y . We have h (y ) =
g(y ) = (x) = 0. So (y ) = 0. Moreover, ( (y )) = x + im(f ), as required.

Problem 7. Let R be a ring, M a right R-module, N a left R-module, and A an abelian
group. A map of sets : M N A is called bilinear if the following hold
1. for m1 , m2 M , n N , we have (m1 + m2 , n) = (m1 , n) + (m2 , n),
8

2. for m M , n1 , n2 N , we have (m, n1 + n2 ) = (m, n1 ) + (m, n2 ),

3. for r R, m M , n N , we have (mr, n) = (m, rn).
(1) Explicitly construct an abelian group M R N equipped with a bilinear map :
M N M R N which is universal in the following sense: For every bilinear
map : M N A there exists a unique homomorphism : M R N A of
abelian groups such that the diagram
M N
II
II
II

II
II

$
/A
M R N
commutes.
(2) Show that the universal property in (1) uniquely determines the abelian group
M R N up to isomorphism. We call M R N the tensor product of M and N over
R.
(3) Let m, n be integers. Determine the tensor product Z/nZ Z Z/mZ. Compute
Q Z Z/mZ.
Solution: (1) This is more or less verbatim the construction for vector spaces given in
the file background on the course website except that we have to be a bit careful with
left and right scalar multiplication over a general ring. We define M R N to be the
quotient of the free abelian group F (M N ) on the set M N (set-theoretic Cartesian
product) modulo the subgroup of relations as dictated by the bilinearity properties 1, 2,
3, i.e., the subgroup generated by all elements of the form
(m1 + m2 , n) (m1 , n) (m2 , n),
(m, n1 + n2 ) (m, n1 ) (m, n2 ),
(mr, n) (m, rn)
where mi M , ni N and r R. Part (2) follows by construction. For part (3), we
have Z/mZ Z Z/nZ
= Z/(m, n) where (m, n) is the ideal generated by m and n. From
algebra, we know that this ideal is principal with generator given by the greatest common
divisor of m and n. Consider an element of Q Z Z/mZ of the form q t. Then we have
qt=

q
q
mt=
mt = 0.
m
m

But Q Z Z/mZ is generated by elements of the above form and hence must be the 0
group.