FINITE CATEGORIES
NOTES FOR THE REU: JULY 29
SCRIBE: NILES JOHNSON
For clarity, we state rst the denition of homotopy equivalence:
Denition 0.1. Given two spaces, X and Y , and two maps f : X Y and g :
Y X, we say that f and g are inverse homotopy equivalences if both composites
are homotopic to the identity maps. That is,
g f id
X
and f g id
Y
Previously we dened a series of functors
opaces oimplicial oets oimplicial /belian (roups (hain (omplexes
taking
(X
f
Y ) (S
X
f
S
Y ) (A
X
f
A
Y ) (C
X
f
C
Y )
and we noted that the product X I gives
X I S
(X I) A
X I C
X I
A chain homotopy is dened precisely so that a homotopy of topological spaces
produces, under these functors, a chain homotopy of complexes. Then the remain-
der of the result is given by a theorem.
Theorem 0.2. If f
: C
and g : D
are chain homotopic maps, then
f
and g
induce inverse isomorphisms on the homology groups H
(C) and H
(D).
Example 0.3. Suppose that the unique map X pt. and an inclusion pt
X are inverse homotopy equivalences, where pt is the one-point space (i.e. X is
contractible). Then H
n
(X) = H
n
(pt) = 0 for n ,= 0, and H
0
(X) = H
0
(pt) = Z.
Exercise 0.4. Show that D
n
= x R
n
[x[ 1 is contractible.
Hint: consider the map h : D
n
I D
n
given by h(x, t) = tx.
The following proposition is one of the foundational tools for computing homol-
ogy.
Proposition 0.5. A short exact sequence of chain complexes
0 C
f
C
g
C
0
gives rise to a long exact sequence of homology groups
H
n
(C
)
f
H
n
(C)
g
H
n
(C
)
H
n1
(C
) .
Proof. The only dicult part of this theorem is paying attention long enough to
check all of the details. The major rst step is the denition of , which we give
now.
1
2 NOTES FOR THE REU: JULY 29 SCRIBE: NILES JOHNSON
Suppose x
n
. By the surjectivity of g
n
, x C
n
such that g
n
(x) = x
.
Now since g
commutes with the dierentials, g
n1
d
n
(x) = d
n
g
n
(x) = d
n
(x
) = 0.
Therefore, by exactness at C
, x
n1
such that f
n1
(x
) = d
n
(x). Now since
f
commutes with the dierentials, f
n1
d
n1
(x
) = d
n
f
n1
(x
) = d
2
n
(x) = 0. So
d
n1
(x
) Z
n1
. We dene (x
) to be the homology class of d
n1
(x
).
What remains now is to check that is a well-dened homomorphism of apelian
groups, and that the sequence above is exact at all stages. None of these are
particularly dicult, they do require patience and care and are (as always) left to
the reader.
As an example of the usefulness of a long exact sequence, we note that one can
immediately compute the homology of all spheres inductively. For this we introduce
the reduced homology groups. For any space X, the reduced homology
H
n
(X) is
dened so that H
n
(X) =
H
n
(X) Z, where Z is in degree zero. Then there is
a similar long exact sequence for quotients; if A X is a sub-simplicial complex,
then we have a long exact sequence
H
n
(A)
H
n
(X)
H
n
(X/A)
H
n1
(A)
Example 0.6. The n-sphere S
n
is a quotient D
n
/S
n1
, and applying the sequence
above we have
H
q
(S
n1
)
H
q
(D
n
)
H
q
(S
n
)
H
q1
(S
n1
)
H
q
1(D
n
)
now recalling that
H
q
(D
n
) = 0 n, we have that
H
q
(S
n
)
=
H
q1
(S
n1
) for all q
Thus, inductively,
H
n
(S
n
)
= Z and
H
q
(S
n
) = 0 for q ,= 0.
Remark 0.7. This same method can be applied to any space X and its suspension,
X, to nd
H
q
(X)
=
H
q1
(X) for all q.
