Exercises
1.3 Homotopy
A)→Z and g:Bn(A)→Z be maps such thatfn−1◦hα|Snα−1=g|Sαn−1. Then there is a unique mapfn : (Xn∪A)→Zsuch thatfn|Xn−1∪A=fn−1
andfn◦hα=g|Bαn.
Proof. By 1.2.7,Xn∪Ais obtained fromXn−1∪Aby attachingn-cells. The result follows from the properties of the quotient topology stated in Sect. 1.1.
Remark 1.2.24.A Hausdorff spaceZ is paracompact if every open coverU of Z has a locally finite open refinementV; this means:V is a locally finite open cover ofZand every element ofV is a subset of some element ofU. It is a fact that every CW complex is paracompact; for a proof see [105]. This arises, for example, in the proof that a fiber bundle whose base space is a CW complex has the homotopy lifting property.
Historical Note:CW complexes were introduced by J.H.C. Whitehead in [154].
In his exposition primacy was given to the “open cells”e◦α rather than the “closed cells” eα, presumably because each ◦eα is homeomorphic to an open ball while eα
need not be homeomorphic to a closed ball. The shift of primacy to “closed cells”
has become standard.
Exercises
1. IfX0=∅thenX =∅; why?
2. Show that a compact subset of a CW complex lies in a finite subcomplex.
3. In Example 1.2.10(a) and (b) it is asserted that some familiar spaces have partic- ular CW complex structures. Prove that these spaces (endowed with the topol- ogy inherited from the Euclidean spaces in which they live) are homeomorphic to the indicated CW complexes.
1.3 Homotopy
Let X and Y be spaces and let f0, f1 : X → Y be maps. One says that f0
is homotopic to f1, denoted f0 f1, if there exists a map F : X×I → Y such that for allx∈X,F(x,0) =f0(x) and F(x,1) =f1(x). The mapF is called a homotopy fromf0 tof1. One often writes Ft:X →Y for the map x→F(x, t); thenF0=f0andF1=f1. IfF is a homotopy fromf0tof1, one writesF :f0f1.
Proposition 1.3.1.Homotopy is an equivalence relation on the set of maps fromX toY.
Proof. Given f : X → Y, define F : f f byF(x, t) = f(x) for all t ∈ I;
F = f ◦ (projection: X ×I → X), the composition of two maps. So F is a map. Thus reflexivity. GivenF :f g, define F :g f byF(x, t) =
F(x,1−t). The functionq:I→I,t→1−t, is a map.F=F◦(id×q), soFis a map. Thus symmetry. GivenF:f gandG:gh, defineH :X×I→Y by H(x, t) =F(x,2t) when 0 ≤t ≤ 12 and byH(x, t) = G(x,2t−1) when
1
2 ≤ t ≤ 1. Let q1 : [0,12] → I be the map t → 2t. Let q2 : [12,1] → I be the mapt→2t−1. OnX×[0,12],H agrees with the mapF ◦(id×q1); on X×[12,1],H agrees with the mapG◦(id×q2). SinceX×[0,12] andX×[12,1]
are closed in X×I, and the restriction ofH to each is a map, H is a map.
Thus transitivity.
A map f : X → Y is a homotopy equivalence if there exists a map g : Y → X such that g ◦f id : X → X and f ◦ g id : Y → Y. Such a map g is called ahomotopy inverse of f. The spaces X and Y have the same homotopy type if there exists a homotopy equivalence from X to Y. If g is a homotopy inverse of the homotopy equivalence f, then g is also a homotopy equivalence. A map homotopic to a homotopy equivalence is a homotopy equivalence. Any two homotopy inverses of f are homotopic, and any map homotopic to a homotopy inverse of f is a homotopy inverse of f. The map idX is a homotopy equivalence. A homeomorphism is a homotopy equivalence.
Definitions parallel to those of the last paragraph hold for maps of pairs.
A map f : (X, A)→ (Y, B) is ahomotopy equivalence if there exists a map g : (Y, B)→(X, A) such that g◦f id : (X, A)→(X, A) and f◦g id : (Y, B)→(Y, B). Such a mapg is called a homotopy inverse off. The pairs (X, A) and (Y, B) have the same homotopy type if there exists a homotopy equivalence from (X, A) to (Y, B), etc.
Let (X, A) and (Y, B) be pairs of spaces, let f0, f1 : (X, A) →(Y, B) be maps of pairs, and letX⊂X. One says thatf0ishomotopic tof1relative to X, denotedf0f1relX, if there exists a mapF : (X×I, A×I)→(Y, B) such that for all x ∈ X, F(x,0) = f0(x) and F(x,1) = f1(x), and for all x ∈ X and t ∈ I, F(x, t) = f0(x). Then F is a homotopy relative to X fromf0tof1. By a proof similar to that of 1.3.1 we have:
Proposition 1.3.2.Homotopy relative to X is an equivalence relation on
the set of maps from (X, A)to(Y, B).
It is customary not to distinguish between the spaceXand the pair (X,∅), so that homotopy between maps of spaces is considered to be a special case of homotopy between maps of pairs.
Proposition 1.3.3.Let f0, f1 : (X, A) → (Y, B) be homotopic rel X, let g0, g1: (Y, B)→(Z, C)be homotopic relY, wheref0(X)⊂Y. Then g0◦f0
andg1◦f1 are homotopic relX.
Proof. Let F : f0 f1 and G : g0 g1 be homotopies which behave as required on A, B, C and X. Let p: X×I → I be projection. Let (F, p) : X×I → Y ×I denote the function (x, t) →(F(x, t), p(x, t)) = (F(x, t), t).
(F, p) is a map. The required homotopy isG◦(F, p) :X×I→Z.
1.3 Homotopy 25 The spaceX iscontractible if it has the same homotopy type as the one- point space. Equivalently,X is contractible if idX is homotopic to some con- stant map fromX toX.
Proofs of 1.3.4–1.3.7 below are left as exercises:
Proposition 1.3.4.Any two maps from a space to a contractible space are
homotopic.
Proposition 1.3.5.Any two maps from a contractible space to a path con-
nected space are homotopic.
Proposition 1.3.6.The product of two contractible spaces is contractible.
Example 1.3.7.The spacesR,R+, andIare contractible. Hence also (by 1.3.6) Rn,Rn+ and Bn are contractible. Ifp∈Sn, thenSn− {p} is homeomorphic toRn; henceSn− {p} is contractible.
The subspaceA⊂X is aretract ofX if there is a mapr:X →A such thatA→i X−→r Ais the identity. Such a mapr is a retraction of X toA.
The subspace A ⊂ X is a strong deformation retract of X if there is a retraction r : X → A and a homotopy F : X ×I → X relative to A such that F0 = idX and F1 = i◦r. Such a homotopy F is a strong deformation retraction of X to A. A pair (A, A) is a strong deformation retract of the pair (X, X) if (A, A)⊂(X, X) and the retractionrand the homotopyF in the previous sentence are maps of pairs.13
Proposition 1.3.8.If A is a strong deformation retract of X then A →X is a homotopy equivalence.
Proof. In the notation above,ris a homotopy inverse for inclusion.
For building homotopies involving CW complexes the following example is fundamental.
Lemma 1.3.9.The subspace(Bn× {0})∪(Sn−1×I)is a strong deformation retract ofBn×I.
Proof. One standard proof involves “radial projection” of Bn×I onto the subspace(Bn× {0})∪(Sn−1×I) from the “light source” (0, . . . ,0,2)∈Rn+1. We give a variation which avoids complicated formulas.
Let f : Bn → I be the map f(x) = 0 if |x| ≤ 12 and f(x) = 2|x| −1 if
1
2 ≤ |x| ≤1. LetY ={(x, t)∈Bn×R|f(x)≤t≤1}. LetY0={(x, t)∈Y |
13IfF is not required to be relA whiler remains a retraction, one says thatAis adeformation retract ofX. If this is further weakened by only requiringr◦ito be homotopic to idAthenAis aweak deformation retract ofX. Note thatAis a weak deformation retract ofX iffi:A →Xis a homotopy equivalence. WhenA is a subcomplex of the CW complexX the three notions of deformation retract coincide; see Exercises 2 and 3 below.
t=f(x)}. There is a strong deformation retractionF : Y ×I→ Y of Y to Y0, namely F(x, t, s) = (x,(1−s)t+sf(x)). And there is a homeomorphism h: (Bn×I,(Bn×{0})∪(Sn−1×I))→(Y, Y0), namelyh(x, t) = (12(t+ 1)x, t).
We are preparing for the important Theorem 1.3.15. The building blocks of the proof are Lemma 1.3.9 and the following (compare 1.2.23):
Proposition 1.3.10.Let (X, A) be a CW pair and let the n-cells of X which are not in A be indexed by A. Let {hα : Bnα → X | α ∈ A}
be a set of characteristic maps for those n-cells. Let Z be a space and let Fn−1 : (Xn−1∪A)×I → Z and g : Bn(A)×I → Z be maps such that Fn−1◦(hα×id) | Sαn−1×I = g | Sαn−1×I. Then there is a unique map Fn : (Xn ∪A)×I → Z such that Fn | (Xn−1 ∪ A)×I = Fn−1 and Fn◦(hα×id) =g|Bnα×I.
The proof of 1.3.10 requires a non-obvious lemma from general topology;
see [51, p. 262]:
Lemma 1.3.11.LetZ andW be Hausdorff spaces. Ifq:Z →W is a quotient map thenq×id :Z×I→W×I is a quotient map.
Proof (Proof of 1.3.10). By 1.2.7, Xn ∪A is obtained from Xn−1∪A by attaching n-cells. Thus there is a quotient mapp: (Xn−1∪A)
Bn(A)→ Xn∪Awhich agrees with inclusion onXn−1∪Aand withhαonBαn⊂Bn(A).
By 1.3.11,p × id : ((Xn−1 ∪ A) × I)
(Bn(A) × I)→(Xn∪A)×I is a quotient map agreeing with inclusion on the first summand and withhα×id on theBαn×Ipart of the second summand, for eachα∈ A. The desired result therefore follows from the properties of the quotient topology stated in Sect.
1.1.
Proposition 1.3.12.IfY is obtained fromAby attachingn-cells, then(Y× {0})∪(A×I) is a strong deformation retract of Y ×I. Hence any map (Y × {0})∪(A×I)→Z extends to a map Y ×I→Z.
Proof. By 1.3.9, there is a strong deformation retractionF : Bn×I×I → Bn×IofBn×Ito (Bn×{0})∪(Sn−1×I). LetAindex then-cells. Consider the diagram
(A
Bn(A))×I×I
p×id×id
f
((Q
QQ QQ QQ QQ QQ Q
Y ×I×I
f˜
//
_ _ _ _
_ Y ×I
Here, p : A
Bn(A) → Y is a quotient map, and f agrees with A×I×I projection // A×I inclusion //Y ×I on A × I × I and with Bnα×I×I F //Bnα×I (p|)×id //Y ×I on Bαn ×I×I for each α ∈ A.
1.3 Homotopy 27 By 1.3.11,p×id×id is a quotient map. By the properties of quotient maps stated in Sect. 1.1, the map ˜f exists and is the desired strong deformation retraction. For the last part, any mapg: (Y× {0})∪(A×I)→Z is extended
byg◦f˜1to a well-defined mapY ×I→Z.
The last sentence of 1.3.12 illustrates the following basic property. A pair of spaces (Y, A) has thehomotopy extension property with respect to a space Z if every map (Y× {0})∪(A×I)→Z extends to a mapY×I→Z. In the presence of this property, a map onAcan be extended toX if it is homotopic to a map which can be extended toX.
By 1.2.7 we have:
Proposition 1.3.13.IfAis a subcomplex ofX, thenXn∪Ais a subcomplex ofX. MoreoverXn∪Ais obtained from Xn−1∪Aby attaching n-cells.
Proposition 1.3.14.Let Z =
n≥1
Zn be a CW complex, where each Zn is a subcomplex ofZ. For all n >1, assumeZn−1 is a strong deformation retract ofZn (in particular Zn−1⊂Zn). Then Z1 is a strong deformation retract of Z.
Proof. Letn≥2. LetF(n):Zn×[1n,n−11]→Zn satisfyFt(n)(x) =xift= n1 or ifx∈ Zn−1, and F(n)1
n−1(Zn) =Zn−1. Let ˆF(n)1
n−1 :Zn → Zn−1 be the map induced byF(n)1
n−1. DefineG(n):Zn×I→Znto agree with projection (x, t)→ xon Zn×[0,n1], with F(n) on Zn×[n1,n−11], with F(n−1)◦Fˆ(n)1
n−1 ×id on Zn×[n−11,n−12], withF(n−2)◦Fˆ(n1−1)
n−2 ×id
◦Fˆ(n)1
n−1×id
onZn×[n1−2,n1−3], . . ., and withF(2)◦Fˆ(3)1
2 ×id
◦. . .◦Fˆ(n)1
n−1×id
onZn×[12,1]. ThenG(n)is a strong deformation retraction ofZn toZ1, andG(n)agrees withG(n−1)onZn−1×I.
DefineG:Z×I→Z to agree withG(n) onZn×I for alln≥2. By 1.2.12, Gis continuous. Clearly,Gis the required strong deformation retraction ofZ
toZ1.
Now we are ready for the Homotopy Extension Theorem for CW com- plexes:
Theorem 1.3.15. (Homotopy Extension Theorem) If (X, A) is a CW pair, then(X× {0})∪(A×I)is a strong deformation retract ofX×I. Hence (X, A)has the homotopy extension property with respect to every space.
Proof. Let Zn = (X × {0})∪((Xn ∪A)×I) and let Z = X ×I. Then Z−1= (X× {0})∪(A×I), andZ=
n≥−1
Zn. By 1.3.13 and 1.2.19, eachZnis a subcomplex ofZ, and ofZn+1. By 1.3.13,Xn∪Ais obtained fromXn−1∪A by attachingn-cells. Hence, by 1.3.12, ((Xn∪A)× {0})∪((Xn−1∪A)×I) =
(Xn× {0})∪((Xn−1∪A)×I) is a strong deformation retract of (Xn∪A)×I.
It follows immediately thatZn−1 is a strong deformation retract of Zn. By 1.3.14,Z−1 is a strong deformation ofZ, as claimed. The last part is proved
as in the proof of 1.3.12.
Examination of the proof of 1.3.15 shows a little more which can be useful:
Addendum 1.3.16.Let(X, A)be a CW pair and letF : (X×{0})∪(A×I)→ Z be a map. The mapF extends to a mapF˜ :X×I→Z such that, for every cell eαof X−A,F˜1(eα) =F0(eα)∪F˜(•eα×I).
The definition of the homotopy extension property has nothing to do with CW complexes. Therefore the last part of 1.3.15 is more powerful than it might appear; it says that the existence of a CW pair structure on the pair of spaces (X, A) ensures that (X, A) has the homotopy extension property with respect to every space.
Corollary 1.3.17.If(X, A)is a CW pair andZ is a contractible space, every
map A→Z extends to a map X→Z.
Exercises
1. Prove 1.3.4–1.3.7.
For the next two exercises (X, A) is a CW pair.
2. Prove that if A is a weak deformation retract of X then A is a deformation retract ofX.
3. Prove that ifA is a deformation retract of X thenA is a strong deformation retract ofX.
Hint for 2. and 3.: Apply 1.3.15 appropriately.