Chapter 6 Products and Quotients - Department of Mathematics
Chapter 6 Products and Quotients - Department of Mathematics
Chapter 6 Products and Quotients - Department of Mathematics
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
1. Introduction<br />
<strong>Chapter</strong> VI<br />
<strong>Products</strong> <strong>and</strong> <strong>Quotients</strong><br />
In <strong>Chapter</strong> III we defined the product <strong>of</strong> two topological spaces \ <strong>and</strong> ] <strong>and</strong> developed some simple<br />
properties <strong>of</strong> the product. ( See Examples III.5.10-5.12 <strong>and</strong> Exercise IIIE20. ) Using simple induction,<br />
that work generalized easy to finite products. It turns out that looking at infinite products leads to<br />
some very nice theorems. For example, infinite products will eventually help us decide which<br />
topological spaces are metrizable.<br />
2. Infinite <strong>Products</strong> <strong>and</strong> the Product Topology<br />
The set \‚] was defined as ÖÐBßCÑÀ B−\ßC −]× . How can we define an “infinite product” set<br />
\œÖ\ α Àα −E× ? Informally, we want to say something like<br />
\œÖ\ α Àα −EלÖÐBαÑÀBα −\ α×<br />
so that a point B in the product consists <strong>of</strong> “coordinates B α” chosen from the \ α's.<br />
But what exactly<br />
does a symbol like BœÐBÑmean if there are “uncountably many coordinates?”<br />
α<br />
We can get an idea by first thinking about a countable product. For sets \ " ß \ # ß ÞÞÞß \ 8,...<br />
we can<br />
informally define the product set as a certain set <strong>of</strong> sequences: \œ ∞ 8œ" \ 8 œÖÐB8ÑÀB8 −\ 8×<br />
.<br />
But if we want to be careful about set theory, then a legal definition <strong>of</strong> \ should have the form<br />
\œÖÐB8Ñ−Y ÀB8 −\ 8 × . From what “pre-existing set” Y will the sequences in \ be chosen? The<br />
answer is easy: we write<br />
\œ ∞ \ œÖB−Ð ∞<br />
\Ñ ÀBÐ8ÑœB−\×Þ 8œ" 8 8œ" 8 8 8<br />
∞ 8œ"<br />
Thus the elements <strong>of</strong> \ 8 are certain functions (sequences) defined on the index set .<br />
This idea<br />
generalizes naturally to any productÞ<br />
Definition 2.1 Let Ö\ α À α − E× be a collection <strong>of</strong> sets. We define the product set<br />
\œÖ\ À −EלÖB− E<br />
α α \ α ÀBÐαÑ−\ α× . The \ α's<br />
are called the factors <strong>of</strong> \ . For each<br />
α, the function 1α À Ö\ α À α − E× Ä \ α defined by 1αÐBÑ<br />
œ Bα<br />
is called the α - .<br />
th projection map<br />
For B−\ , we write more informally B œBÐαÑœthe α - coordinate <strong>of</strong> B<strong>and</strong> write BœÐB).<br />
th<br />
α α<br />
Note: the index set E might not be ordered. So, even though we use the informal notation B œ ÐBα Ñ,<br />
such phrases as “the first coordinate <strong>of</strong> B,” “the next coordinate in B after Bα,”<br />
<strong>and</strong> “the coordinate<br />
in B preceding Bα” may not make sense. The notation ÐBαÑis h<strong>and</strong>y but can lead you into errors if<br />
you're not careful.<br />
239
α α<br />
A point B in Ö\ À α − E× is, by definition, a function that “chooses” a coordinate B from each set<br />
in the collection Ö\ α À α − E× . To say that such a “choice function” must exist if all the \ α's<br />
are<br />
nonempty is precisely the Axiom <strong>of</strong> Choice. ( See the discussion following Theorem I.6.8.)<br />
Theorem 2.2 The Axiom <strong>of</strong> Choice (AC) is equivalent to the statement that every product <strong>of</strong><br />
nonempty sets is nonempty.<br />
Note: In ZF set theory, certain special products can be shown to be nonempty without using AC.<br />
∞<br />
For example, if \ 8 œ ß then <br />
8œ" \ 8 œ ÖB − À BÐ8Ñ − × œ . Without using AC, we can<br />
precisely describe a point ( œ function) in the product for<br />
example, the identity function<br />
3œÖÐ7ß8Ñ− ‚ À7œ8×so œ \ Ág.<br />
∞ 8œ" 8<br />
We will <strong>of</strong>ten write Ö\ α À α − E× as <br />
α−E\<br />
α.<br />
If the indexing set E is clearly understood, we may<br />
simply write \α.<br />
Example 2.3<br />
1) If E œ g, then Ö\ À − E× œ ÖÐ E<br />
α \ Ñ À BÐαÑ− \ × œ Ög×Þ<br />
α α−E<br />
α α<br />
2) Suppose \ α œ g for some α − E . Then α \ α œ ÖÐ α \ αÑ À BÐαÑ − \ α×Þ<br />
! ! E<br />
−E −E<br />
Since BÐ Ñ − \ is impossible, \ œ g.<br />
α! α! αα−E<br />
α<br />
3) Strictly speaking, we have two different definitions for a finite product \ " ‚\ # :<br />
i) \ " ‚\ # œÖÐBßBÑÀB " # " −\ßB " # −\× # (a set <strong>of</strong> ordered pairs)<br />
Ö"ß#×<br />
" # " # 3<br />
ii) \ ‚\ œ ÖB − Ð\ ∪\ Ñ À BÐ3Ñ − \ × (a set <strong>of</strong> functions)<br />
But there is an obvious way to identify these two sets: the ordered pair ÐB" ß B# Ñ corresponds to the<br />
function B œ ÖÐ"ß B" Ñß Ð#ß B# Ñ× − Ð\ " ∪ \ # Ñ Þ Ö"ß#×<br />
4) If Eœ, then Ö\ À8− ל<br />
∞ \ œÖB−Ð ∞<br />
<br />
8 8œ" 8 8œ" \ 8ÑÀB8 −\ 8×<br />
œ ÖÐB" ß B# ß ÞÞÞß B8ßÞÞÞß Ñ À B8 − \ 8× œ the set <strong>of</strong> all sequences ÐB8Ñwhere B8 − \ 8Þ<br />
5) Suppose the \ α's are identical, say \ α œ ] for all α − E. Then Ö\<br />
α À α − E×<br />
œÖB−Ð E E E<br />
α−E\ αÑÀBÐα Ñ−\ αלÖB−] ÀBα −]ל] . If lElœ7,<br />
we will sometimes<br />
7 write this product simply as ] œ “the product <strong>of</strong> 7 copies <strong>of</strong> ] ” because the number <strong>of</strong> factors 7 is<br />
<strong>of</strong>ten important while the specific index set E is not.<br />
Now that we have a definition <strong>of</strong> the set Ö\ αÀ α − E× , we can think about a product topology.<br />
We<br />
begin by recalling the definition <strong>and</strong> a few basic facts about the “weak topology.” (See Example<br />
III.8.6.)<br />
240
Definition 2.4 Let \ be a set. For each α − E, suppose Ð\ αßgαÑ is a topological space <strong>and</strong> that<br />
0α À \ Ä \ α. The weak topology on \ generated by the collection Y œ Ö0αÀ α − E× is the smallest<br />
topology on that makes all the 's continuous.<br />
\ 0α<br />
Certainly, there is at least one topology on \ that makes all the 0α's<br />
continuous: the discrete topology.<br />
Since he intersection <strong>of</strong> a collection <strong>of</strong> topologies on \ is a topology ( why? ), the weak topology<br />
exists we can describe it “from the top down” as { g : g is a topology on \ making all the 0α's<br />
continuous×Þ<br />
However, this slick description <strong>of</strong> the weak topology doesn't give a useful description <strong>of</strong> what sets are<br />
open. Usually it is more useful to describe the weak topology on \ “from the bottom up.” to make all<br />
the 0α's<br />
continuous it is necessary <strong>and</strong> sufficient that<br />
for each α −E<strong>and</strong> for each open set Y ©\ , the set 0 ÒYÓmust be open.<br />
α α α α<br />
"<br />
Therefore the weak topology g is the smallest topology that contains all such sets 0αÒYαÓ <strong>and</strong> that is<br />
" the topology for which Æ œÖ0 ÒYÓÀ α −EßY open in \ × is a subbase. ( See Example III.8.6. )<br />
α α α α<br />
Therefore a base for the weak topology consists <strong>of</strong> all finite intersections <strong>of</strong> sets from Æ.<br />
A typical<br />
" " "<br />
basic open set has form 0α ÒYα ] ∩0α ÒYα Ó∩... ∩0α ÒYα Ówhere each α − E <strong>and</strong> each Yα<br />
is<br />
" " # # 8 8 3<br />
3<br />
open in \α3.<br />
To cut down on symbols, we will use a special notation for these subbasic <strong>and</strong> basic open<br />
sets:<br />
We will write Yα for 0 ÒYαÓÞ " " "<br />
A typical basic open set is then Y œ 0 ÒY ] ∩0 ÒY Ó∩... ∩0 ÒY Ó<br />
α "<br />
α" α α# α α8<br />
α<br />
which we further abbreviate as Y œ Yα" ß Yα# ß ÞÞÞß Yα8 <br />
241<br />
" # 8<br />
So B − Y œ Y ß Y ß ÞÞÞY iff 0 ÐBÑ − Y for each 3 œ "ß ÞÞÞß 8Þ<br />
α α α α α<br />
" # 8 3 3<br />
This notation is not st<strong>and</strong>ard but it should be because it's very h<strong>and</strong>y. You should verify that to get a<br />
base for the weak topology g on \ , it is sufficient to use only the sets Yα ß Yα ß ÞÞÞß Yα where<br />
" # 8<br />
each Y is a basic (or even subbasic) open set in \ .<br />
α α<br />
3<br />
Example 2.5 Suppose Ð\ßgÑ is any topological space <strong>and</strong> E©\ÞLet 3ÀEÄ\ be the inclusion<br />
map 3Ð+Ñ œ +Þ Then the subspace topology on E is the same as the weak topology generated by<br />
"<br />
Y œÖ3×ÞTo see this, just note that a base for the weak topology is Ö3 ÒYÓÀY open in \× <strong>and</strong> that<br />
" 3 ÒYÓ œ Y ∩E which is a typical open set in the subspace topology.<br />
The following theorem tells us that a map 0 into a space \ with the weak topology is continuous iff<br />
each composition 0 ‰0 is continuous.<br />
α<br />
Theorem 2.6 For α −E, suppose 0α À\ÄÐ\ αßgαÑ <strong>and</strong> that \ has the weak topology generated by<br />
the functions 0α. Let ^ be a topological space <strong>and</strong> 0 À ^ Ä \ . Then 0 is continuous iff<br />
α<br />
0 ‰0 À ^ Ä \<br />
α α is continuous for every .<br />
Pro<strong>of</strong> If 0 is continuous, then each composition 0 α ‰0 is continuous. Conversely, suppose each<br />
"<br />
0 α ‰0 is continuous. To show that 0 is continuous, it is sufficient to show that 0 ÒZÓ is open in ^<br />
whenever Z is a subbasic open set in \ ( why? ). So let Z œ Yα with Y α open in \ α.<br />
Then<br />
" " " "<br />
0 ÒZ Ó œ 0 Ò0 ÒY ÓÓ œ Ð0 ‰ 0Ñ ÒY Ó which is open since 0 ‰ 0 is continuous.<br />
ñ<br />
α α α α α<br />
3<br />
"
Definition 2.7 For each α −Eßlet Ð\ ßg Ñbe<br />
a topological space. The g on the<br />
set \ is the weak topology generated by the collection <strong>of</strong> projection maps Y œ Ö1 À α − E×Þ<br />
α α product topology<br />
α α<br />
The product topology is sometimes called the “Tychon<strong>of</strong>f topology.” We always assume that a<br />
product space has the product topology unless something else is explicitly stated.<br />
\α<br />
Since the product topology is a weak topology, a subbase for the product topology consists <strong>of</strong> all sets<br />
"<br />
<strong>of</strong> form Yα œ1αÒYαÓ, where α −E <strong>and</strong> Y α is open in \ αÞA<br />
base consists <strong>of</strong> all finite<br />
intersections <strong>of</strong> such sets À<br />
" "<br />
α α α α α α<br />
Y ∩ ÞÞÞ ∩ Y œ 1 ÒY ] ∩ ... ∩ 1 ÒY Ó<br />
" 8 " " 8 8<br />
α α α<br />
œ ÖB − \ À B − Y for each 3 œ "ß ÞÞÞß 8×<br />
3 3<br />
<br />
α−E α α α<br />
" # 8<br />
œ Y where Y œ \ for α Á α ß α ß ÞÞÞß α (*)<br />
( It is sufficient to use only Y ' s which are basic (or even subbasic) open sets in \ Why? )<br />
α αÞ<br />
A basic open set in “depends on only finitely many coordinates” in the following sense:<br />
\α<br />
B − Yα" ß Yα# ß ÞÞÞß Yα8 iff B satisfies the finitely many restrictions Bα3 − Yα3 Ð3 œ "ß ÞÞÞß 8Ñ.<br />
Since (basic) open sets containing B are the “st<strong>and</strong>ard” for measuring closeness to B,<br />
we can<br />
say, roughly, that in the product topology “closeness depends on only finitely many<br />
coordinates.”<br />
For a finite index set E œ Ö"ß #ß ÞÞÞß 8× , the product topology on <br />
α−E\<br />
α is just the one for which the<br />
basic open sets are the open boxes <br />
α−E<br />
α:<br />
Y<br />
8<br />
Y œ Y œ Y ß Y ß ÞÞÞß Y œ Y ‚ Y ‚ ÞÞÞ ‚ Y<br />
α−E α 3œ" 3 α" α# α8<br />
" # 8<br />
Then condition (*) that “ Yα œ \ α for all but finitely many α'<br />
s” is satisfied automatically so, for finite<br />
products the product topology is exactly as we defined it in <strong>Chapter</strong> III.<br />
Definition 2.7 is probably not what was expected. Someone's “first guess” would more likely be to<br />
use all “boxes” <br />
α−EYα ( Y α open in \ αÑ<br />
as a base in defining the product topology. But in Definition<br />
2.7, a “box” <strong>of</strong> the form <br />
α−EYα<br />
might not be open because (*) may not hold. It is perfectly possible,<br />
<strong>of</strong> course, to define a different topology on the set using all boxes <strong>of</strong> the form <br />
α−E\ α α−EYαas<br />
a<br />
base. This alternate topology on <br />
α−E<br />
\ α is called the box topology.<br />
But we will see that our<br />
definition <strong>of</strong> the product topology is the “right” definition to use. (Of course, the box topology <strong>and</strong> the<br />
product topology coincide for finite products. )<br />
Theorem 2.8 Each projection map 1α À ÖÖ\<br />
α À α − E× Ä \ α is continuous <strong>and</strong> open.<br />
If Ö\ α À α − E× Á g, each 1α<br />
is onto. A function 0 À ^ Ä \<br />
α is continuous iff<br />
1α ‰0 À^ Ä\ α is continuous for every α.<br />
Pro<strong>of</strong> By definition, the product topology is one that makes all the 1α's<br />
continuous. <strong>and</strong> clearly, each<br />
1α is onto if the product is nonempty. To show that 1α<br />
is open, it is sufficient to show that the image<br />
<strong>of</strong> a basic open set Y œ Y ß Y ß ÞÞÞß Y is open. This is clearly true if Y œ g.<br />
α α α<br />
" # 8<br />
242
Yα3 For Y Á g, we get 1αÒY Ó œ 1αÒYα"<br />
ß Yα# ß ÞÞÞß Yα8 Ó œ <br />
\ α<br />
if α œ α3<br />
..<br />
if α Á α" ß ÞÞÞß α8<br />
Either way, 1αÒYα" ß Yα# ß ÞÞÞß Yα8 Ó is open.<br />
Because the product has the weak topology generated by the projections, Theorem 2.6 gives<br />
that 0 is continuous iff each composition ‰0 is continuous. ñ<br />
1α<br />
Generally, projection maps are not closed. For example, if JœÖÐBßCÑ−‘ ÀCœ ßB!× ,<br />
# "<br />
B<br />
then 1 "ÒJ Ó œ the projection <strong>of</strong> J on the B-axis ‘ is not closed in ‘ Þ<br />
Example 2.9<br />
243<br />
" "<br />
1) A subbasic open set in ‘ ‚ ‘ has form 1 " ÒYÓ œ Y ‚ ‘ or 1 # ÒZÓ œ ‘ ‚Z, where Y <strong>and</strong><br />
Z are open in ‘ Þ ( We get a subbase even if we restrict ourselves just to using basic open intervals<br />
YßZ in ‘ Þ)<br />
. So basic open sets have form ÐY ‚ ‘ Ñ ∩ Ð ‘ ‚ Z Ñ œ Y ‚ Z . Therefore the product<br />
#<br />
topology on ‘ ‚ ‘ is the usual topology on ‘ .<br />
# #<br />
The function 0À‘ Ä ‘ ‚ ‘ given by 0Ð>ÑœÐ>ß sin >Ñ is continuous because the<br />
# #<br />
compositions Ð 1 ‰ 0ÑÐ>Ñ œ > <strong>and</strong> Ð 1 ‰ 0ÑÐ>Ñ œ sin > are both continuous functions from ‘ to ‘ .<br />
" #<br />
i!<br />
2) Let \œ œ“the<br />
product <strong>of</strong> countably many copies <strong>of</strong> .”<br />
A base for the product<br />
topology consists <strong>of</strong> all sets YœY8 where finitely many Y8's<br />
are singletons <strong>and</strong> all the others are<br />
equal to Þ Each Y is infinite ( in fact lY l œ - why? ), so every nonempty open set is infinite. In<br />
particular, \ has no isolated points Þ More generally, an infinite product <strong>of</strong> (nonempty) discrete spaces<br />
is not discrete.<br />
The box topology on \ is the discrete topology. For each point B œ Ð5" ß 5# ß ÞÞÞß 58, ... Ñ − \ ,<br />
the set ÖB× œ ∞ 8œ" Ö5 8×<br />
is open in the box topology. ( For a finite product, the box <strong>and</strong> product<br />
topologies care the same: a finite product <strong>of</strong> discrete spaces is discrete. )<br />
‘ ‘<br />
< <<br />
3) Let ‘ œ Ö\ À < − ‘ × , where \ œ ‘ . Each point 0 in ‘ is a function 0 À ‘ Ä ‘ <strong>and</strong><br />
1< Ð0Ñ œ 0Ð
∞<br />
"<br />
3<br />
3œR"<br />
# 3 3 #R<br />
#<br />
#<br />
R<br />
<br />
3œ"<br />
3 3<br />
#<br />
∞<br />
<br />
3œR"<br />
3 3<br />
# "Î#<br />
# %<br />
#R<br />
# %<br />
#<br />
"Î#<br />
# #<br />
Pick R so that % %<br />
Þ If ÐB C Ñ for each 3 œ "ß ÞÞÞß R,<br />
then we have<br />
.ÐBß CÑ œ Ð ÐB C Ñ ÐB C Ñ Ñ ÐR † Ñ œ % Þ<br />
244<br />
This is the natural<br />
metric topology on the product L, <strong>and</strong> we see here that we can make “ C close to B”<br />
by<br />
requiring “closeness” in just finitely many coordinates "ß ÞÞÞß RÞ This is just what the product<br />
topology does <strong>and</strong>, in fact, the product topology on L turns out to be the topology g. Þ.<br />
In Ð!ß "Ñ we can see a similar phenomenon quite clearly: for two points B œ !ÞB" ÞÞÞB8ÞÞÞ <strong>and</strong><br />
C œ !ÞC" ÞÞÞC8ÞÞÞ ß “ C will be close to B” if C <strong>and</strong> B agree in, say, the first 8 decimal places.<br />
Roughly speaking, “closeness” depends on only “finitely many coordinates.”<br />
A h<strong>and</strong>y “rule <strong>of</strong> thumb” that has proved true every time I've used it is that if a topology on a<br />
product set makes “closeness” depend on only a finite number <strong>of</strong> coordinates, then that<br />
topology is the product topology.<br />
3) The bottom line: a mathematical definition justifies itself by the fruit it bears. The<br />
definition <strong>of</strong> the product topology will lead to some beautiful theorems. For example, we will<br />
see that compact Hausdorff spaces are (topologically) nothing other than the closed subspaces<br />
7<br />
<strong>of</strong> cubes Ò!ß "Ó with the product topology ( 7 may be infinite). For the time being, you will<br />
need to accept that things work out nicely down the road, <strong>and</strong> that by contrast, the box<br />
topology turns out to be rather ill-behaved. ( See Exercise E11.)<br />
As a simple example <strong>of</strong> nice behavior, the following theorem is exactly what one would hope for <strong>and</strong><br />
the pro<strong>of</strong> depends on having the “correct” definition for the product topology. The theorem says that<br />
convergence <strong>of</strong> sequences in a product is “coordinatewise convergence” : that is, in a product,<br />
>2 >2<br />
ÐB8ÑÄBiff for all α , the α coordinate <strong>of</strong> B 8converges<br />
(in \ αÑ<br />
to the α coordinate <strong>of</strong> B.<br />
The<br />
product topology is sometimes called the “topology <strong>of</strong> coordinatewise convergence.”<br />
Theorem 2.11 Suppose ÐB8Ñ is a sequence in \ œ Ö\ αÀ α − E× . Then ÐB8ÑÄB− \ iff<br />
( 1 ÐB Ñ) Ä 1 ÐBÑ in \ for all α − E.<br />
α 8 α α<br />
Pro<strong>of</strong> If ÐB8ÑÄB, then ( 1αÐB8Ñ) Ä 1αÐBÑ because each 1α<br />
is continuous.<br />
Conversely, suppose ( 1αÐB8Ñ) Ä 1αÐBÑ in \ α for each α <strong>and</strong> consider any basic open let<br />
Y œ Yα" ß Yα# ß ÞÞÞß Yα that contains B œ ÐBαÑ. For each 3 œ "ß ÞÞÞß 5, we have Bα − Y<br />
5 3 α3.<br />
Since<br />
( 1α ÐB ) 1α , we have 1α<br />
α for some . If max , then for<br />
3 8ÑÄ3ÐBÑ ÐB Ñ − Y 8 R R œ ÖR ß ÞÞÞß R ×<br />
3 8 3<br />
3 " 5<br />
8 R we have 1α ÐB Ñ − Yα for every 3 œ "ß ÞÞÞß 5. This means that B − Y for 8 R,<br />
so<br />
3 8 3<br />
8<br />
ÐB8 Ñ Ä B. ñ<br />
In the pro<strong>of</strong>, R is the max <strong>of</strong> a finite set. If \ has the box topology, the basic open set Y œ Yα might involve infinitely many open sets Yα Á \ α. For each such α,<br />
we could pick Rα,<br />
just as in the<br />
pro<strong>of</strong>. But the set <strong>of</strong> all Rα's might not have a max R so the pro<strong>of</strong> would collapse. Can you create a<br />
specific example with the box topology where this happens?<br />
< <<br />
‘ Example 2.12 Consider ‘ œ Ö\ À < − ‘ × , where \ œ ‘ . Each point 0 in the product is a<br />
‘<br />
function 0À‘ Ä‘ Þ Suppose that Ð0Ñ 8 is a sequence <strong>of</strong> points in ‘ . By Theorem 2.11, Ð0ÑÄ0 8<br />
iff Ð08 Ð
Question: if ‘ is given the box topology, is convergence <strong>of</strong> a sequence simply<br />
‘ Ð08Ñ uniform convergence (as defined in analysis)?<br />
The following “theorem” is stated loosely. You can easily create variations. Any reasonable version<br />
<strong>of</strong> the statement is probably true.<br />
“Theorem 2.13” Topological products are associative in any “reasonable” sense: for example, if<br />
EœF∪Gwhere F<strong>and</strong> Gare<br />
disjoint indexing sets, then<br />
Ö\ Àα −E× Ö\ À" −F× ‚ Ö\<br />
À# −G×<br />
α " #<br />
“Pro<strong>of</strong>” A point B−Ö\ α Àα −E× is a function BÀEÄ <br />
α−E\<br />
α.<br />
Define<br />
0ÀÖ\ α Àα −E×ÄÖ\ " À" −F× ‚ Ö\<br />
# À# −G× by 0ÐBÑœÐBlFßBlGÑ.<br />
Clearly 0 is one-to-one <strong>and</strong> onto.<br />
For convenience, let ] œ Ö\ " À " − F× <strong>and</strong> ^ œ Ö\<br />
# À # − G× so that<br />
1 ‰0 ÀÖ\ Àα −E×Ä] <strong>and</strong> 1 ‰0 ÀÖ\ Àα −E×Ä^Þ<br />
" α #<br />
α<br />
0 is a mapping into a product Y ‚^, so 0 is continuous iff 1" ‰0 <strong>and</strong> 1#<br />
‰0 are both continuous.<br />
But 1" ‰0 ÀÖ\ α Àα −E×Ä] œ Ö\<br />
" À −F× is also a map into a product ß so 1"<br />
‰0 is<br />
continuous iff 1" ‰ 1" ‰0 À Ö\<br />
α À α − E× Ä \ " is continuous for all " − FÞ<br />
But 1 ‰ 1 ‰0 œ 1 À Ö\<br />
À α − E× Ä \ which is continuous.<br />
" " " α "<br />
The pro<strong>of</strong> that 1# ‰0 is continuous is completely similar.<br />
" 0 À ] ‚^ Ä "<br />
Ö\ α À α − E× is given by B œ 0 ÐCßDÑ œ C∪D ( the union <strong>of</strong> two functions)<br />
<strong>and</strong><br />
" "<br />
0 is continuous iff 1α ‰0 À ] ‚^ Ä \ α is continuous for each α − E œ F ∪GÞ<br />
"<br />
Suppose α − F : then 1α ‰ 0 ÐCß DÑ œ 1αÐBÑ, where B œ C ∪ DÞ Since α − Fß Bα<br />
"<br />
œ Ð 1α ‰ 1" ÑÐCß DÑß so 1α ‰ 0 œ 1α ‰ 1" , which is continuous. The case where α − G is completely<br />
similar.<br />
Therefore 0 is a homeomorphism. ñ<br />
The question <strong>of</strong> topological commutativity for products only makes sense when the index set E is<br />
ordered in some way. Even in that case, if we view a product as a collection <strong>of</strong> functions, the question<br />
<strong>of</strong> commutativity is irrelevant the question reduces to the fact that set theoretic unions are<br />
Ö"ß#×<br />
commutative. For example, \ " ‚\ # œ ÖB − Ð\ " ∪\ # Ñ À BÐ3Ñ − \ 3 for 3 œ "ß#×<br />
Ö"ß#×<br />
œ ÖB − Ð\ # ∪ \ " Ñ À BÐ3Ñ − \ 3 for 3 œ "ß #× œ \ # ‚ \ " Þ<br />
So viewed as sets <strong>of</strong> functions, \ " ‚\ # <strong>and</strong> \ # ‚\ " are exactly the same set! The same observation<br />
applies to any product viewed as a collection <strong>of</strong> functions.<br />
But we might look at an ordered product in another way: for example, thinking <strong>of</strong> \ " ‚\ # <strong>and</strong><br />
\ # ‚\ " as sets <strong>of</strong> ordered pairs. Then generally \ " ‚\ # Á\ # ‚\ " . From that point <strong>of</strong> view, the<br />
topological spaces \ 1 ‚\ 2 <strong>and</strong> \ 2 ‚\ 1 are not literally identical, but there is a homeomorphism<br />
between them: 0ÐBßBÑœÐBßBÑÞ<br />
" # # " So the products are still topologically identical. We can make a<br />
similar argument whenever any ordered product is “commuted” by permuting the index set.<br />
The general rule <strong>of</strong> thumb is that “whenever it makes sense, topological products are commutative.”<br />
245
Exercise 2.14 Recall that for a space \ <strong>and</strong> cardinal 7 , \ denotes the product <strong>of</strong> 7 copies <strong>of</strong> the<br />
7 8 78 space \ . Prove that Ð\ Ñ is homeomorphic to \ . ( Hint: Look at the pro<strong>of</strong> <strong>of</strong> Theorem I.14.7;<br />
the bijection 9 given there is a homeomorphism. )<br />
Notice that “cancellation properties” may not be true. For example, ‚ Ö!× <strong>and</strong> ‚<br />
Ö!ß "× are<br />
homeomorphic (both are countable discrete spaces) but topologically you can't “cancel the ” À Ö!× is<br />
not homeomorphic to Ö 0,1 × !<br />
Here are a few results which are quite simple but very h<strong>and</strong>y to remember. The first states that<br />
singleton factors are topologically irrelevant in a product.<br />
Lemma 2.15 \ ‚ Ö: × \<br />
α −E α " −F " α−E<br />
α<br />
Pro<strong>of</strong> <br />
" −FÖ: " × is itself a one-point space Ö: }, so we only need to prove that<br />
\ ‚Ö:× \ . The map 0ÐBß:Ñ œ Bis clearly a homeomorphism Þ ñ<br />
α−E α α−E<br />
α<br />
Lemma 2.16 Suppose . For any , <br />
α−E\ α Á g F © E α−F\<br />
α is homeomorphic to a subspace ^<br />
<strong>of</strong> that is, can be embedded in <br />
α−E\ α α−F\ α α−E\<br />
αÞ In fact, if all the \ α's<br />
are X"<br />
-spaces,<br />
then \ is homeomorphic to a closed subspace ^ <strong>of</strong> \ .<br />
α−F α α−E<br />
α<br />
α <br />
α−E<br />
α<br />
Pro<strong>of</strong> Pick a point :œÐ: Ñ− \ . Then by Lemma 2.15,<br />
\ \ ‚ Ö: × œ ^ © \<br />
α−F α α−F α α−EF α α−E<br />
α<br />
Now suppose all the \ α's are X" . If C − Ð <br />
α−E\<br />
α Ñ^, then for some # − EFß C# Á : # Þ Since<br />
\ is a X space, there is an open set Y in \ that contains C but not : . Then C − Y <strong>and</strong><br />
# "<br />
# # # # #<br />
Y ∩Ð \ ‚ Ö: ×Ñœg.<br />
# α−F α α−EF<br />
α<br />
Therefore ^ is closed in <br />
α−E\<br />
α.<br />
ñ<br />
Note: 1) Assume all the \ α 's œ ‘ . Go through the preceding pro<strong>of</strong> step-by-step when E œ Ö"ß ÞÞÞß 5×<br />
<strong>and</strong> when Eœ <strong>and</strong><br />
2) In the case FœÖ α! × , Lemma 2.16 says that each factor \ α! is homeomorphic to subspace<br />
<strong>of</strong> <br />
α−E \ α (a closed subspace if all the \ α's<br />
are X"<br />
).<br />
3) But Lemma 2.16 does not say that if all the \ α 's are X"<br />
, then every embedded copy <strong>of</strong><br />
\ α \<br />
! α α in is closed: only that there a closed homeomorphic copy. It is very easy to show<br />
<br />
−E exists (<br />
# #<br />
a copy <strong>of</strong> ‘ embedded in ‘ that is not closed in ‘ , for example ... ?)<br />
α α " #<br />
Lemma 2.17 Suppose \œ Ö\ À −E×Ág . Then \ is a X space (or, X space) iff every<br />
factor \ is X (or, X ).<br />
α " #<br />
246<br />
7
Pro<strong>of</strong> for Hausdorff Suppose all the \ α's are Hausdorff. If B Á C − \ , then Bα! Á Cα!<br />
for some<br />
α!. Pick disjoint open sets Yα <strong>and</strong> Z α in \ α containing Bα <strong>and</strong> Cα<br />
. Then<br />
! ! ! ! !<br />
Y <strong>and</strong> Z are disjoint (basic) open sets in \<br />
that contain B<strong>and</strong> C.<br />
α α α<br />
! !<br />
Conversely, suppose \Ág . By Lemma 2.16, any factor \ α is homeomorphic to a subspace <strong>of</strong> \ .<br />
Since a subspace <strong>of</strong> a Hausdorff space is Hausdorff ( why? Ñ , \ is Hausdorff. ñ<br />
Exercise 2.18 Prove Lemma 2.17 if “Hausdorff” is replaced by “ X" .” ( The pro<strong>of</strong> is similar but<br />
easier.)<br />
Theorem 2.19 The product <strong>of</strong> countably many 2-point discrete spaces is homeomorphic to the Cantor<br />
set G.<br />
Pro<strong>of</strong> We show that if \ 8 œ Ö!ß #× , then ∞<br />
8œ" \ 8 G.<br />
<br />
To construct G we defined, for each sequence ÐB" ß B# ß ÞÞÞß B8ßÞÞÞÑ − Ö!ß #× , a descending<br />
sequence <strong>of</strong> closed sets J B" ª J B" B# ª ÞÞÞ ª J B" B# ÞÞÞB8<br />
ª ÞÞÞ in Ò!ß "Ó whose intersection gave a unique<br />
point :−GÀ Ö:ל∞ 8œ" JB" B# ÞÞÞB8<br />
(s ee Section IV.10).<br />
For each 8, we can write Gas<br />
a union <strong>of</strong><br />
8 # disjoint clopen sets: G œ <br />
ÐB ßÞÞÞßB Ñ−Ö!ß#× 8 ÐG ∩JBBÞÞÞB Ñ.<br />
" 8<br />
" # 8<br />
Define 0 À G Ä ∞ 8œ" \ 8 by 0Ð:Ñ œ B œ ÐB" ß B# ß ÞÞÞß B8ßÞÞÞÑÞ Clearly 0 is one-to-one <strong>and</strong><br />
onto. To show that 0 is continuous at : − G, it is sufficient to show that for each 8,<br />
the function<br />
18 ‰ 0 À G Ä Ö!ß #× is continuous at : , so pick any 8. One <strong>of</strong> the clopen sets G ∩ J B" B# ÞÞÞB8<br />
contains :<br />
(call it Y) <strong>and</strong> Ð 18 ‰0ÑlY œ B 8. Thus, 18 ‰ 0 is constant on a neighborhood <strong>of</strong> : in G , so 18<br />
‰0 is<br />
continuous at : .<br />
By Lemma 2.17, ∞ 8œ" \ 8 is Hausdorff. Since 0 is a continuous bijection <strong>of</strong> a compact space<br />
onto a Hausdorff space, 0 is a homeomorphism ( why? ). ñ<br />
Corollary 2.20 Ö!ß #× i! is compact.<br />
This corollary is a very special case <strong>of</strong> the Tychon<strong>of</strong>f Product Theorem which states that any product<br />
<strong>of</strong> compact spaces is compact. The Tychon<strong>of</strong>f Product Theorem is much harder <strong>and</strong> will be proved in<br />
<strong>Chapter</strong> IX.<br />
Corollary 2.21 The Cantor set is homeomorphic to a product <strong>of</strong> countably many copies <strong>of</strong> itself.<br />
i i i i †i i<br />
Pro<strong>of</strong> By Exercise 2.14 above, G<br />
8 G G for 8 − is similar. ñ<br />
! ÐÖ!ß #× ! Ñ ! Ö!ß #× ! ! Ö!ß #× ! GÞ The case <strong>of</strong><br />
Example 2.22 Convince yourself that each assertion is true:<br />
1) If \ is the Sorgenfrey line, then \ ‚\ is the Sorgenfrey plane ( see Examples<br />
III.5.3 <strong>and</strong> III.5.4).<br />
" # "<br />
2) Let W be the unit circle in ‘ . Then W ‚ Ò!ß "Ó is homeomorphic to the<br />
cylinder ÖÐBß Cß DÑ − ‘ À B C œ "ß D − Ò!ß "Ó×Þ<br />
$ # #<br />
" "<br />
3) W ‚W is homeomorphic to a torus ( œ “the surface <strong>of</strong> a doughnut”).<br />
247<br />
α
Exercises<br />
E1. Does it ever happen that \‚Ö!× open in \‚‘ ? If so, what is a necessary <strong>and</strong> sufficient<br />
condition on \ for this to happen?<br />
E2. a) Suppose \ <strong>and</strong> ] are topological spaces <strong>and</strong> that E © \ß F © ]Þ Prove that<br />
int \‚] ( E‚FÑœint\ E‚ int ] F:<br />
that is, “the interior <strong>of</strong> the product is the product <strong>of</strong> the interiors.”<br />
( By induction, the same result holds for any finite product.)<br />
Give an example to show that the<br />
statement may be false for infinite products.<br />
b) Suppose E © \ for all α − E . Prove that in the product \ œ \<br />
,<br />
α α α<br />
cl( EÑœcl E.<br />
α α<br />
Note: When the 's are closed, this shows that is closed: so “any product <strong>of</strong> closed sets is<br />
Eα Eα<br />
closed.” Can you see any plausible reason why products <strong>of</strong> closures are better behaved than products<br />
<strong>of</strong> interiors?<br />
c) Suppose \œ \ α Ág<strong>and</strong> that E α ©\ α. Prove that E<br />
α is dense in \ iff Eα<br />
is dense in<br />
\α for each α. Note: Part c) implies that a finite product <strong>of</strong> separable spaces is separable.<br />
Part c) doesn't tell us whether or not an infinite product <strong>of</strong> separable spaces is separable: why not?<br />
d) For each α , let ; α − \ α. Prove that F œ ÖB − \<br />
α À Bα œ ; α for all but at most finitely many<br />
α× is dense in <br />
\ α. ( Note: Suppose \œ‘ œ <br />
8−\ 8 where each \ 8 œ‘ Þ Suppose each ; 8 is<br />
chosen to be a rational say ; 8 œ ! . What does d) say about ‘ ? )<br />
e) Let α! −E . Prove that ] α œÖB− <br />
! \ αÀBα œ: α for all α Á α!<br />
× is homeomorphic<br />
to \α! . Note: So any factor <strong>of</strong> a product has a “copy” <strong>of</strong> itself inside the product in a “natural” way.<br />
For example, in ‘ , the set <strong>of</strong> points where all coordinates except the first are is homeomorphic to<br />
8 !<br />
the first factor, ‘ . Þ<br />
f) Give an example <strong>of</strong> infinite spaces \ß] ß^ such that \ ‚ ] is homeomorphic to \ ‚ ^ but ]<br />
is not homeomorphic to ^.<br />
E3. Let \ be a topological space <strong>and</strong> consider the “diagonal” set<br />
? œÖÐBßBÑÀB−\ש\‚\Þ<br />
a) Prove that ? is closed in \ ‚\ iff \ is Hausdorff.<br />
b) Prove that ? is open in \ ‚\ iff \ is discrete.<br />
E4. Suppose \ is a Hausdorff space <strong>and</strong> that \ α © \ for each α − E..<br />
Show that<br />
] œ Ö\ À α − E× is homeomorphic to a closed subspace <strong>of</strong> the product Ö\<br />
À α − E×.<br />
α α<br />
248
E5. For each 8−, suppose H is a countable dense set in Ð\ ßg ÑÞ<br />
8 8 8<br />
a) Prove that HœH is dense in \<br />
.<br />
8 8<br />
b) Prove that \8 is separable. ( Caution: H may not be countable (when is H countable?<br />
. Hint: In a product “closeness depends on only finitely many coordinates.” )<br />
E6. a) Suppose a3ß4 − Ö!ß #× for all 3ß 4 − . Prove that there exists a sequence Ðn5Ñin such that,<br />
for each 3, lim a38<br />
5Ä∞<br />
exists.<br />
, 5<br />
Hint: “picture” the + 3ß4 in an infinite matrix. For each fixed 4,<br />
the “j-th column" <strong>of</strong> the matrix is a<br />
i!<br />
point in the Cantor set G œ Ö!ß #× , <strong>and</strong> G is a compact metric space.<br />
∞<br />
8<br />
8œ"<br />
∞<br />
8œ"<br />
8 w<br />
w<br />
8<br />
∞<br />
w<br />
8 or 8 is called a subseries <strong>of</strong> <br />
8œ"<br />
WœÖ=−‘ À=<br />
8.<br />
∞<br />
+× 8<br />
‘<br />
8œ"<br />
∞<br />
8œ"<br />
8 w<br />
i i<br />
b) Let + be an absolutely convergent sequence <strong>of</strong> real numbers. A series +<br />
where each<br />
+ œ + + œ ! +<br />
Prove that is the sum <strong>of</strong> a subseries <strong>of</strong> is closed in .<br />
Hint: “Absolute convergence” just guarantees that every subseries converges. Each subseries +<br />
can be associated in a natural way with a point B − Ö!ß "× Þ Consider the mapping 0 À Ö!ß "× Ä<br />
given by 0ÐBÑœ ∞<br />
+ − ‘ Þ Must 0 be a homeomorphism?<br />
8œ"<br />
8 w<br />
249<br />
! ! ‘<br />
c) Suppose K is an open dense subset <strong>of</strong> the Cantor set G. Must Fr K be countable?<br />
Hint: Consider ÖÐ+ß ,Ñ − G ‚ G À , Á !×Þ<br />
E7. Let have the c<strong>of</strong>inite topology.<br />
a) Does the product have the c<strong>of</strong>inite topology? Does the answer depend on ?<br />
7 7<br />
b) Prove is separable (<br />
7 Hint: When 7 is infinite, consider the simplest possible points in<br />
the product.Ñ<br />
Part b) implies that an arbitrarily large product <strong>of</strong> X" spaces with more than one point can<br />
be separable. According to Theorem 3.8, this statement is false for X#<br />
spaces.
E8. In any set \ , we can define a topology by choosing a nonempty family <strong>of</strong> subsets Y <strong>and</strong> defining<br />
closed sets to be all sets which can be written as an intersection <strong>of</strong> finite unions <strong>of</strong> sets from Y .<br />
Y is called a subbase for the closed sets <strong>of</strong> \ . ( This construction is “complementary” to generating a<br />
topology on \ by using a collection <strong>of</strong> sets as a subbase for the open sets. )<br />
a) Verify that this procedure does give a topology on \ .<br />
b) Suppose \ α is a topological space. Give \<br />
α the topology for which collection <strong>of</strong> “closed<br />
boxes” Y œÖJα ÀJ αclosed in \ α×<br />
is a subbase for the closed sets. Is this topology the product<br />
topology?<br />
E9. Prove or disprove:<br />
i i<br />
! !<br />
There exists a bijection 0 À \ œ Ö!ß "× Ä œ ] such that for all B − \ <strong>and</strong> for all 8,<br />
the first 8 coordinates <strong>of</strong> C œ 0ÐBÑare determined by the first 7 coordinates <strong>of</strong> B.<br />
Here, 7 depends on 8 <strong>and</strong> B.<br />
More formally, we are asking whether there exists a bijection<br />
0 such that:<br />
aB − \ a8 − b7 − such that changing B4 for any 4 7 does not change C5<br />
for 5Ÿ8<br />
(Hint: Think about continuity <strong>and</strong> the definition <strong>of</strong> the product topology.)<br />
250
3. Productive Properties<br />
We want to consider how some familiar topological properties behave with respect to products.<br />
Definition 3.1 Suppose that, for each α −E , the space \ α has a certain property T.<br />
We say that<br />
productive<br />
if \<br />
α must have property T<br />
the property T is countably productive if \<br />
α must have property T when Eis<br />
countable<br />
finitely<br />
productive if \ must have property T when Eis<br />
finite<br />
For example, Lemma 2.17 shows that the X" <strong>and</strong> X#<br />
properties are productive.<br />
α<br />
Some topological properties behave very badly with respect to products. For example, the Lindel<strong>of</strong> ¨<br />
property is a “countability property” <strong>of</strong> spaces, <strong>and</strong> we might expect the Lindel<strong>of</strong> ¨ property to be<br />
countably productive. Unfortunately, this is not the case.<br />
Example 3.2 The Lindel<strong>of</strong> ¨ property is not finitely productive; in fact if \ is a Lindel<strong>of</strong> ¨ space, then<br />
\‚\ may not be Lindel<strong>of</strong>. ¨ Let \ be the set <strong>of</strong> real numbers with the topology for which a<br />
neighborhood base at + is U+ œ ÖÒ+ß ,Ñ À +ß , − W , , +× . ( Recall that \ is called the Sorgenfrey<br />
Line: see Example III.5.3.) We begin by showing that f is Lindel<strong>of</strong>. ¨<br />
It is sufficient to show that a collection h <strong>of</strong> basic open sets covering \ has a countable<br />
subcover. Given such a cover h, Let i œÖÐ+ß,ÑÀ Ò+ß,Ñ− h× <strong>and</strong> define E œi.<br />
For a<br />
moment, we can think <strong>of</strong> E as a subspace <strong>of</strong> ‘ with its usual topology. Then E is Lindel<strong>of</strong> ¨<br />
w<br />
( why? ), <strong>and</strong> i is a covering <strong>of</strong> E by usual open sets, so there is a countable subfamily i with<br />
i w œE.<br />
Replace the left endpoints <strong>of</strong> the intervals in i to get h œÖÒ+ß,ÑÀÐ+ß,Ñ− i ש h. If h<br />
covers \ we are done, so suppose \ w Á g. For each B − \ w<br />
h h , pick a set Ò+ß,Ñ in<br />
h that contains B. In fact, B must be the left endpoint <strong>of</strong> Ò+ß ,Ñ for if B − Ò+ß ,Ñ − h <strong>and</strong><br />
BÁ+ , then B−i œ i © h Þ So, for each BÂh we can pick a set [ B ) −h.<br />
251<br />
w w w w<br />
w w w , ,B<br />
h w<br />
If B <strong>and</strong> C are distinct points not in , then ÒB, , B) <strong>and</strong> ÒC,<br />
, C)<br />
must be disjoint ( why? ). But<br />
w w<br />
there can be at most countably many disjoint intervals ÒBß , BÑÞ<br />
So h ∪ ÖÒB, ,B)<br />
À B Â h<br />
} is<br />
a countable subcollection <strong>of</strong> h that covers W.<br />
However the Sorgenfrey plane W‚Wis not Lindel<strong>of</strong>. ¨ If it were, then its closed subspace<br />
HœÖÐBßCÑÀBCœ"× would also be Lindel<strong>of</strong> ¨ (Theorem III.7.10). But that is impossible<br />
since H is uncountable <strong>and</strong> discrete in the subspace topology. ( See the figure on the following<br />
page.)
Fortunately, many other topological properties interact more nicely with products. Here are several<br />
topological properties T to which the “same” theorem applies. We combine these into one large<br />
theorem for efficiency.<br />
Theorem 3.3 Suppose \œÖ\ α Àα −E×Ág. Let T be any one <strong>of</strong> the properties “first<br />
countable,” “second countable,” “metrizable,” or “completely metrizable.” Then \ has property T iff<br />
1) all the \ α's<br />
have property T,<br />
<strong>and</strong><br />
2) there are only countably \α's<br />
that do not have the trivial topology.<br />
This theorem “essentially” is a statement about countable products because:<br />
1) The nontrivial \α's<br />
are the “interesting” factors, <strong>and</strong> 2) says there are only countably<br />
many <strong>of</strong> them. In practice, one hardly ever works with trivial spaces; <strong>and</strong> if we completely exclude<br />
trivial spaces from the discussion, then the theorem just states that \ has property T iff \ is a<br />
countable product <strong>of</strong> spaces with property T .<br />
2) A nonempty X " space \ α has the trivial topology iff l\ αl<br />
œ "Þ So, if we deal<br />
only with X " spaces (as is most <strong>of</strong>ten the case) the theorem says that \ has property T iff all the \ α's<br />
have property T <strong>and</strong> all but countably many <strong>of</strong> the \α's<br />
are “topologically irrelevant” singletons.<br />
Of course, in the case <strong>of</strong> metrizability (or complete metrizability), the X" condition is automatically<br />
satisfied.<br />
Pro<strong>of</strong><br />
Throughout, let Fœthe set <strong>of</strong> “interesting” indices œÖα−EÀ \ α is a nontrivial space × .<br />
We begin with the case Tœ“first<br />
countable.”<br />
Suppose 1) <strong>and</strong> 2) hold <strong>and</strong> B−\. We need to produce a countable neighborhood base at B.<br />
8<br />
For each α −F, let ÖY À8− ×<br />
be a countable open neighborhood base at B −\ Þ Let<br />
α α α<br />
U œÖYÀYis a finite intersection <strong>of</strong> sets <strong>of</strong> the form Y ßα −Fß8− ×<br />
B<br />
Since F is countable, UB is a countable collection <strong>of</strong> open sets containing B <strong>and</strong> we claim that<br />
UB is a neighborhood base at B − \. To see this, suppose Z œ Zα ß ÞÞÞß Zα is a basic<br />
" 5<br />
open set containing B. ( We may assume that all α3's<br />
are in FÀwhy?<br />
). For each 3 œ "ß ÞÞÞß 5ß<br />
83 83 8" 8 8<br />
#<br />
5<br />
pick Y so that B − Y © Z Þ Then Y œ Y ß Y ß ÞÞÞß Y − U<br />
<strong>and</strong> B − Y © Z Þ<br />
α α α α α α α<br />
3 3 3<br />
3 " #<br />
252<br />
k<br />
8<br />
α<br />
B
Conversely, suppose \ is first countable.<br />
We need to prove that 1) <strong>and</strong> 2) hold.<br />
Since \Ág , Lemma 2.16 gives that each \ α is homeomorphic to a subspace <strong>of</strong> \ . Therefore<br />
each \α is first countable, so 1) holds.<br />
We prove the contrapositive for 2): assuming F is uncountable, we find a point B − \ at<br />
which there cannot be a countable neighborhood base. Pick a point : œÐ: αÑ−\<br />
For each " −Fßpick an open set S " ©\ " for which gÁS" Á\ " . Choose<br />
<strong>and</strong> Define using the coordinates<br />
B −S C ÂS Þ BœÐB Ñ−\<br />
" " " " α<br />
B" if α œ " − F<br />
Bαœ <br />
: − \ if α Â F<br />
α α<br />
Suppose UBis a countable collection <strong>of</strong> neighborhoods <strong>of</strong> B. For each R − UBchoose<br />
a basic<br />
open set Y with B − Y œ Yα" ß ÞÞÞß Yα © R.<br />
There are only finitely many α3's<br />
involved<br />
5<br />
in the expression for the chosen Y, <strong>and</strong> there are only countably many R's in UB.<br />
So, since F<br />
is uncountable, we can pick a " −Fthat is not one <strong>of</strong> the α3's<br />
involved in the expression for<br />
any <strong>of</strong> the chosen sets Y . Then<br />
i) S" is an open set that contains Bbecause B" −S"<br />
ii) for all R−UB ßR©Î S" : to see this, consider the point A œ ÐAαÑ which is the same as B except in the " coordinate:<br />
Bαif α Á "<br />
Aαœ <br />
C if α œ "<br />
"<br />
For each R − FB, we chose Y œ Yα" ß ÞÞÞß Yα © R. Since B − Y <strong>and</strong> A has the same<br />
5<br />
α" ,..., α8<br />
coordinates as B, we have A−Y also. But AÂ S" because A" œC" ÂS" Þ<br />
Since Y©Î S , certainly R ©Î S Þ So is not a neighborhood at B.<br />
" " UB base<br />
We are now able to see immediately that if the product \œ \ α has any <strong>of</strong> the other properties T,<br />
then conditions 1) <strong>and</strong> 2) must hold.<br />
If \ is second countable, metrizable or completely metrizable, then \ is first countable <strong>and</strong>,<br />
by the first part <strong>of</strong> the pro<strong>of</strong>, condition 2) must hold.<br />
If \ is second countable or metrizable then every subspace has these same properties in<br />
particular each \ α is second countable or metrizable respectively. If \ is completely<br />
metrizable, then \ <strong>and</strong> all the subspaces \ α are X" . By Lemma 2Þ 16, \ α is homeomorphic to<br />
a closed therefore complete subspace <strong>of</strong> \ . Therefore \αis<br />
completely metrizable.<br />
It remains to show that if Tœ“second<br />
countable,” “metrizable,” or “completely metrizable” <strong>and</strong><br />
conditions 1) <strong>and</strong> 2) hold, then \œ \ α<br />
also has property T.<br />
253
Suppose Tœ“ second countable.”<br />
For each α in the countable set F, let UαœÖSαßSαßÞÞÞß<br />
SαßÞÞÞ } be a countable base for the<br />
open sets in \α <strong>and</strong> let<br />
254<br />
" # 8<br />
U œÖSÀSis a finite intersection <strong>of</strong> sets <strong>of</strong> form S , where α −F<strong>and</strong> 8−}<br />
U is countable <strong>and</strong> we claim U is a base for the product topology on \ .<br />
Suppose B − Z œ Zα ß ÞÞÞß Zα , a basic open set in \ . For each 3 œ "ß ÞÞÞß 5ß<br />
" 5<br />
83<br />
Bα3 − Z α3 so we can choose a basic open set in \ α3 such that Bα3 − S α © Z 3 α3.<br />
Then<br />
8" 8# 85 8" 8# 85<br />
B − Sα ß Sα ß ÞÞÞß Sα © Z <strong>and</strong> Sα ß Sα ß ÞÞÞß Sα − U . Therefore Z can be<br />
" # 5 " #<br />
5<br />
written as a union <strong>of</strong> sets from U, so U is a base for \ .<br />
Suppose Tœ“ metrizableÞ”<br />
Since all the \ α's are X " , condition 2) implies that all but countably many \ α's<br />
are singletons<br />
<strong>and</strong> dropping the singleton factors from the product does not change \ topologically.<br />
Therefore it is sufficient to prove that if each space \ " ß \ # ß ÞÞÞß \ 8ßÞÞÞ<br />
is metrizable, then<br />
∞<br />
\œ \ 8 is metrizable.<br />
8œ"<br />
Let . 8 be a metric for \ 8ß where without loss <strong>of</strong> generality, we can assume each . 8 Ÿ "<br />
( why? ). For points BœÐBÑ, CœÐCÑ−\ , define<br />
8 8<br />
∞ .ÐBßCÑ<br />
.ÐBßCÑ œ 8 8 8<br />
# 8<br />
8œ"<br />
Then . is a metric on \ ( check ), <strong>and</strong> we claim that g. is the product topology g . Because \<br />
is a countable product <strong>of</strong> first countable spaces, g is first countable, so g can be described<br />
using sequences: so it is sufficient to show that ÐD5ÑÄDin Ð\ß g Ñ iff ÐD5 Ñ Ä D in Ð\ß g.<br />
ÑÞ<br />
But ÐD5ÑÄDin Ð\ß g Ñ iff the ÐD5Ñconverges “coordinatewise.” Therefore it is sufficient to<br />
show that:<br />
( D5 Ñ Ä D in Ð\ß g.<br />
Ñ iff a8 ÐD5Ð8ÑÑ Ä DÐ8Ñ in Ð\ 8ß . 8Ñ,<br />
that is,<br />
a8 . 8ÐD5Ð8Ñß DÐ8ÑÑ Ä !<br />
i) Suppose .ÐD ß DÑ Ä ! . Let % ! <strong>and</strong> consider a particular 8 . We can choose O<br />
5<br />
5<br />
∞<br />
<br />
8œ"<br />
#<br />
!<br />
.ÐDÐ8ÑßDÐ8ÑÑ<br />
8 5<br />
# 8<br />
%<br />
8!<br />
. 8 ÐD Ð8 ÑßDÐ8 ÑÑ<br />
! 5 ! !<br />
#<br />
%<br />
# 8 5 ! !<br />
so that 5 O implies .ÐDßDÑœ . Therefore, if 5 O<br />
certainly 8 ! 8 , <strong>and</strong> therefore so . ÐD Ð8 Ñß DÐ8 ÑÑ % .<br />
! !<br />
So . ÐD Ð8 Ñß DÐ8 ÑÑ Ä ! .<br />
8 5 ! !<br />
!<br />
ii) On the other h<strong>and</strong>, suppose . ÐD Ð8Ñß DÐ8ÑÑ Ä ! for every 8. Let ! . Choose<br />
∞<br />
"<br />
# 8 #<br />
8œR"<br />
α 8<br />
8 5 %<br />
R so that %<br />
<strong>and</strong> then choose O so that if 5 O<br />
. " ÐD5Ð"Ñß DÐ"ÑÑÎ# <br />
# %<br />
. # ÐD5Ð#Ñß DÐ#ÑÑÎ# #R<br />
ÞÞÞ<br />
R<br />
. ÐD ÐRÑß DÐRÑÑÎ# <br />
R 5<br />
Then for 5 O we have .ÐDßDÑœ<br />
5<br />
"<br />
%<br />
#R<br />
%<br />
#R<br />
∞<br />
<br />
8œ"<br />
.ÐDÐ8ÑßDÐ8ÑÑ<br />
8 5<br />
# 8<br />
.
R<br />
.ÐDÐ8ÑßDÐ8ÑÑ<br />
∞<br />
.ÐDÐ8ÑßDÐ8ÑÑ<br />
8œ"<br />
# 8<br />
8œR"<br />
# 8<br />
#R<br />
œ 8 5<br />
8 5<br />
% % R † œ % Þ<br />
Therefore .ÐD ß DÑ Ä ! .<br />
Suppose Tœ“ completely metrizableÞ”<br />
5<br />
Just as for Tœ “metrizable,” we can assume \œ \ 8 <strong>and</strong> that . 8 is a complete metric<br />
on \ 8 with . 8 Ÿ " . Using these . 8 's, we define . in the same way. Then g.<br />
is the product<br />
topology on \ . We only need to show that Ð\ß.Ñis<br />
complete.<br />
255<br />
∞ 8œ"<br />
Suppose ÐD5Ñ is a Cauchy sequence in Ð\ß .Ñ . From the definition <strong>of</strong> . it is easy to see that<br />
ÐD5Ð8ÑÑ is Cauchy in Ð\ 8ß . 8Ñfor each 8, so that ÐD5Ð8ÑÑ Ä some point + 8 − \ 8Þ<br />
Let +œÐ+ 8Ñ−\ . Since ÐD5Ð8ÑÑÄ+ 8 for each 8, we have ÐD5ÑÄ+ in the product topology<br />
œ g.. Therefore Ð\ß.Ñ is complete. ñ<br />
What is the correct formulation <strong>and</strong> pro<strong>of</strong> <strong>of</strong> the theorem for Tœ“pseudometrizable”<br />
?<br />
We might wonder why Tœ“separable”<br />
is not included with Theorem 3.3. Since separability is a<br />
“countability property,” we might hope that separability is preserved in countable products although<br />
our experience Lindel<strong>of</strong> ¨ spaces could make us hesitate. The reason for the omission is that separability<br />
is better behaved for products than the other properties. ( You should try to prove directly that a<br />
countable product <strong>of</strong> separable spaces is separable remembering that in the product topology,<br />
“closeness depends on finitely many coordinates.” If necessary, first look at finite products.) But<br />
surprisingly, the product <strong>of</strong> as many as - separable spaces must be separable, <strong>and</strong> the product <strong>of</strong> more<br />
than - nontrivial separable spaces can sometimes be separable.<br />
We begin the treatment <strong>of</strong> separability <strong>and</strong> products with a simple lemma which is really nothing but<br />
set theory.<br />
Lemma 3.4 Suppose lEl Ÿ -Þ There exists a countable collection e <strong>of</strong> subsets <strong>of</strong> E with the<br />
following property: given distinct α" , α# , ... , α8 − E, there are disjoint sets E" ß E# ß ÞÞÞß E8 − e such<br />
−E 3<br />
that for each .<br />
α3 3<br />
Pro<strong>of</strong> (Think about how you would prove the theorem when Eœ‘ Þ The general case is just a<br />
“carry over” on the same idea.)<br />
"<br />
Since lElŸ-, there is a one-to-one map 9 ÀEÄ‘ . Let e œÖ9 ÒÐ+ß,ÑÓÀ+ß,− ×Þ<br />
For distinct α" , α# ,..., α8 − E, 9Ðα" Ñ, 9Ðα# Ñ, ..., 9Ðα8Ñare distinct real numbers; we can choose<br />
+ 3, , 3 − so that 9Ðα3Ñ − Ð+ 3ß, 3Ñ <strong>and</strong> so that the intervals Ð+ 3ß, 3Ñ<br />
are pairwise disjoint. Then the sets<br />
" E œ 9 ÒÐ+ ß, ÑÓ − e are the ones we need. ñ<br />
3 3 3<br />
Theorem 3.5<br />
<br />
α−E α α<br />
1) Suppose \œ \ Ág . If \ is separable, then each \ is separable.<br />
α <br />
α−E<br />
α<br />
2) If each \ is separable <strong>and</strong> lEl Ÿ - , then \ œ \ is separable.<br />
Part 2) <strong>of</strong> Theorem 3.5 is attributed (independently) to several people. In a slightly more general<br />
version, it is sometimes called the “Hewitt-Marczewski-Pondiczery Theorem.” Here is an amusing<br />
#
sidelight, written by topologist Melvin Henriksen online in the Topology Atlas.<br />
A few words have<br />
been modified to conform with our notation:<br />
Most topologists are familiar with the Hewitt-Marczewski-Pondiczery theorem. It states that if<br />
7<br />
7 is an infinite cardinal, then a product <strong>of</strong> # topological spaces, each <strong>of</strong> which has a dense<br />
set <strong>of</strong> cardinality Ÿ7ß also has a dense set with Ÿ7points. In particular, the product <strong>of</strong> -<br />
separable spaces is separable (where - is the cardinal number <strong>of</strong> the continuum). Hewitt's<br />
pro<strong>of</strong> appeared in [Bull. Amer. Math. Soc. 52 (1946), 641-643], Marczewski's pro<strong>of</strong> in [Fund.<br />
Math. 34 (1947), 127-143], <strong>and</strong> Pondiczery's in [Duke Math. 11 (1944), 835-837]. A pro<strong>of</strong><br />
<strong>and</strong> a few historical remarks appear in <strong>Chapter</strong> 2 <strong>of</strong> Engelking's General Topology.<br />
The<br />
spread in the publication dates is due to dislocations caused by the Second World War; there<br />
is no doubt that these discoveries were made independently.<br />
Hewitt <strong>and</strong> Marczewski are well-known as contributors to general topology, but who was (or<br />
is) Pondiczery? The answer may be found in Lion Hunting & Other Mathematical Pursuitsß<br />
edited by G. Alex<strong>and</strong>erson <strong>and</strong> D. Mugler, Mathematical Association <strong>of</strong> America, 1995. It is a<br />
collection <strong>of</strong> memorabilia about Ralph P. Boas Jr. (1912-1992), whose accomplishments<br />
included writing many papers in mathematical analysis as well as several books, making a lot<br />
<strong>of</strong> expository contributions to the American Mathematical Monthly, being an accomplished<br />
administrator (e.g., he was the first editor <strong>of</strong> Mathematical Reviews (MR) who set the tone for<br />
this vitally important publication, <strong>and</strong> was the chairman for the <strong>Mathematics</strong> <strong>Department</strong> at<br />
Northwestern University for many years <strong>and</strong> helped to improve its already high quality), <strong>and</strong><br />
helping us all to see that there is a lot <strong>of</strong> humor in what we do. He wrote many humorous<br />
articles under pseudonyms, sometimes jointly with others. The most famous is “A Contribution<br />
to the Mathematical Theory <strong>of</strong> Big Game Hunting” by H. Petard that appeared in the Monthly<br />
in 1938. This book is a delight to read.<br />
In this book, Ralph Boas confesses that he concocted the name from Pondicheree (a place in<br />
India fought over by the Dutch, English <strong>and</strong> French), changed the spelling to make it sound<br />
Slavic, <strong>and</strong> added the initials E.S. because he contemplated writing spo<strong>of</strong>s on extra-sensory<br />
perception under the name E.S. Pondiczery. Instead, Pondiczery wrote notes in the Monthly,<br />
reviews for MR, <strong>and</strong> the paper that is the subject <strong>of</strong> this article. It is the only one reviewed in<br />
MR credited to this pseudonymous author.<br />
One mystery remains. Did Ralph Boas have a collaborator in writing this paper? He certainly<br />
had the talent to write it himself, but facts cannot be established by deduction alone. His son<br />
Harold (also a mathematician) does not know the answer to this question...<br />
Pro<strong>of</strong> 1) Let H be a countable dense set in \ . For each ß ÒHÓ is countable <strong>and</strong> dense in \<br />
(because \ œ1 Ò\Óœ1 Òcl HÓ© cl 1 ÒHÓ ). Therefore each \ is separable.<br />
α α α α α<br />
256<br />
α 1α α<br />
1 2 8<br />
2) Choose a family e as in Lemma 3.4 <strong>and</strong> for each α, let Hα œ ÖBα, BαßÞÞÞß BαßÞÞÞ × be a<br />
countable dense set in \α.<br />
Define a countable set f by<br />
f œ ÖÐE ß ÞÞÞß E ß 6 ß ÞÞÞß 6 Ñ À 8 − ß6−, E − e with the E 's pairwise disjoint×Þ<br />
" 8 " 8 3 3 3<br />
Pick a point : α − \ α for each α. For each #8 -tuple = œ ÐE" ß ÞÞÞß E8 ß 6" ß ÞÞÞß 68Ñ−f, define a point<br />
B= − \ with coordinates
6<br />
3 B if − E3<br />
BÐ = αÑœ<br />
α α<br />
<br />
: if α Â E<br />
α<br />
3<br />
Let HœÖB= À=− f×<br />
. His countable <strong>and</strong> we claim that H is dense in \ . To see this, consider any<br />
nonempty basic open set Y œ Y ß ÞÞÞß Y À we will show that Y ∩ H Á gÞ<br />
For Y œ Y ß ÞÞÞß Y Á g,<br />
α α<br />
" 5<br />
α α<br />
" 5<br />
i) Choose disjoint sets E ß ÞÞÞß E in e so that α − E ß ÞÞÞß α − E<br />
" 5<br />
257<br />
" " 5 5<br />
8<br />
α α α α<br />
3<br />
ii) For each 3 œ "ß ÞÞÞß 5ß H is dense in \ <strong>and</strong> we can pick a point Bα in H ∩ Y<br />
3<br />
" # 8<br />
œ ÖB , B ß ÞÞÞß B ß ÞÞÞ × ∩ Y Þ<br />
α α α α<br />
3 3 3 3<br />
3 3 3 3<br />
3<br />
Let = œ ÐE" ß ÞÞÞß E3ßÞÞÞß E5 ß 8" ß ÞÞÞ83ßÞÞÞß 85Ñ−fÞBecause α3 − E3, we have B= Ðα3ÑœBα−Yα. 3<br />
3<br />
Therefore B − Y ∩H. ñ<br />
=<br />
Example 3.6 The rather abstract construction <strong>of</strong> a dense set H in the pro<strong>of</strong> <strong>of</strong> Theorem 3.5 can be<br />
‘<br />
nicely illustrated with a concrete example. Consider ‘ Ðœ <br />
Theorem 3.8 shows that for Hausdorff spaces, a nonempty product with more than - factors cannot be<br />
separable.<br />
Example 3.7 For each α −E , let \ α be a set with l\ αl"Þ Choose : α −\ α <strong>and</strong> let<br />
gα œ ÖS © \ α À : α − S× ∪ Ög×Þ The singleton set Ö: α× is dense in Ð\ αßgαÑ , so \ α is separable.<br />
If :œÐ: αÑ−\œ <br />
α−E\<br />
α,<br />
then singleton set Ö:× is dense ( why? look at a nonempty basic<br />
open set Y . ) So \ is separable, no matter how large the index set E is.<br />
The \ α 's in this example are not X"<br />
. But Exercise E7 shows that an arbitrarily large<br />
product <strong>of</strong> separable X" spaces can turn out to be separable.<br />
α α α<br />
Theorem 3.8 Suppose \œ −E \ Ág where each \ is a X#<br />
space with more than one point.<br />
If \ is separable, then lEl Ÿ -Þ<br />
Pro<strong>of</strong> For each α, we can pick a pair <strong>of</strong> disjoint, nonempty open sets Yα<strong>and</strong> Z α in \ αÞ<br />
Let H be a<br />
countable dense set in \ <strong>and</strong> let Hα œ Yα ∩H for each α. If α Á " − Eß there is a point<br />
:− Yα ßZ" ∩Hbecause H is dense. Then :−H α but :ÂH" œ Y" ∩Hsince<br />
B"  Y" À therefore Hα Á H" . Therefore the map 9 À E Ä cÐHÑ given by 9ÐαÑ œ Hαis<br />
one-toi!<br />
one, so lElŸlcÐHÑlœ# œ-Þñ We saw in Corollary V.2.19 that a finite product <strong>of</strong> connected spaces is connected. In fact, the<br />
following theorem shows that connectedness behaves very nicely with respect to any product. The<br />
pro<strong>of</strong> <strong>of</strong> the theorem is interesting because, unlike our previous pro<strong>of</strong>s about products, this pro<strong>of</strong> uses<br />
the theorem about finite products to prove the general case.<br />
<br />
α−E α α<br />
Theorem 3.9 Suppose \œ \ Ág . \ is connected iff each \ is connected.<br />
Pro<strong>of</strong> \ is nonempty so each map 1α À \ Ä \ α is onto. If \ is connected, then each \ α is a<br />
continuous image <strong>of</strong> a connected space so \α is connected (see Example V.2.6).<br />
Conversely, suppose each \ α is connected Þ For each α, pick a point : α − \ α.<br />
For each finite<br />
set J©E , let \ J œ \ ‚ <br />
−J −EJ Ö:× . \ J is homeomorphic to the finite product <br />
α α α α α−J<br />
\ α,<br />
so each \ J is connected. Let H œ Ö\<br />
J À J is a finite subset <strong>of</strong> E× . Each \ J contains the point<br />
:œÐ: α Ñ, so Corollary V.2.10 tells us that His connected. Unfortunately, HÁ\ (except in trivial<br />
cases; why?).<br />
But we claim that H is dense in \ . If Y œ Yα ß ÞÞÞß Yα is any nonempty basic open set in \ ,<br />
" 8<br />
we need to show that H ∩ Y Á g. Choose a point Bα3 − Yα3 for each 3 œ "ß ÞÞÞß 8, <strong>and</strong> define B − \<br />
having coordinates<br />
B if α œ α<br />
BÐαÑ œ <br />
: if α Á α , α ,..., α<br />
α3<br />
3<br />
α<br />
" # 8<br />
If we let J œ Ö+ ß α ß ÞÞÞß α × , then B − \ ∩ Y © H ∩ Y , so H ∩ Y Á gÞ<br />
" # 8 J<br />
Therefore \œcl His connected (Corollary V.2.20). ñ<br />
Question: is the analogue <strong>of</strong> Theorem 3.9 true for path connected spaces?<br />
Just for reference, we state one more theorem here. We will not prove Theorem 3.10 until <strong>Chapter</strong> IX,<br />
but we may use it in examples. (Of course, the pro<strong>of</strong> in <strong>Chapter</strong> IX will not depend on any <strong>of</strong> these<br />
258
examples! )<br />
Theorem 3.10 (Tychon<strong>of</strong>f Product Theorem) Suppose \œ <br />
α−E\<br />
α Ág . Then \ is compact iff<br />
each \α is compact.<br />
The pro<strong>of</strong> “ ” in Tychon<strong>of</strong>f's Theorem is quite easy ( And the easy theorem that a<br />
Ê why?). finite<br />
product <strong>of</strong> compact spaces is compact was in Exercise IV.E.26. .<br />
259
Exercises<br />
E10. Let \œÒ!ß"Ó have the product topology.<br />
Ò!ß"Ó<br />
a) Prove that the set <strong>of</strong> all functions in \ with finite range ( sometimes called step functions)<br />
is<br />
dense in \ .<br />
in \ .<br />
b) By Theorem 3.5, \ is separable. Describe a countable set <strong>of</strong> step functions which is dense<br />
c) Let E œ ÖB − \ À 0 is the characteristic function <strong>of</strong> a singleton set Ö
4. Embedding Spaces in <strong>Products</strong><br />
It is <strong>of</strong>ten possible to embed a space \ in a product \ αÞ<br />
Such embeddings will give us some nice<br />
theorems for example, we will see that there is a separable metric space \ that contains all other<br />
separable metric spaces topologically a “universal” separable metric space.<br />
To illustrate the general method we will use, consider two maps 0" À Ò!ß"Ó Ä ‘ <strong>and</strong> 0# À Ò!ß"Ó Ä ‘<br />
# B #<br />
given by 0ÐBÑœB " <strong>and</strong> 0ÐBÑœ/Þ # Using these maps we can define /ÀÒ!ß"ÓÄ ‘ ‚ ‘ œ‘<br />
by<br />
# B<br />
using 0" <strong>and</strong> 0 # as “coordinate functions”: /ÐBÑ œ Ð0" ÐBÑß 0# ÐBÑÑ œ ÐB ß / ÑÞ This map / is called the<br />
evaluation map defined by the set <strong>of</strong> functions Ö0" , 0 # × . In this example, / is an embedding that is,<br />
# # / is a homeomorphism <strong>of</strong> Ò!ß "Ó into ‘ so that Ò!ß "Ó ran Ð/Ñ © ‘ ( see the figure).<br />
An evaluation map does not always give an embedding: for example, the evaluation map<br />
# / À Ò!ß "Ó Ä ‘ defined by the family Ö cos # 1Bß sin # 1B×<br />
is not a homeomorphism between Ò!ß "Ó <strong>and</strong><br />
ran Ð/Ñ © ‘ ( )<br />
# why? what is ran Ð/Ñ?<br />
We want to generalize the idea <strong>of</strong> an evaluation map / into a product <strong>and</strong> to find some conditions under<br />
which / will be an embedding.<br />
Definition 4.1 Suppose \ <strong>and</strong> \ α ( α − EÑare topological spaces <strong>and</strong> that 0α À \ Ä \ α for each αÞ<br />
The evaluation map defined by the family Ö0 À α − E× is the function / À \ Ä \ given by<br />
α <br />
α−E<br />
α<br />
/ÐBÑÐαÑ œ 0αÐBÑ α α α<br />
th<br />
Then /ÐBÑ is the point in the product<br />
coordinate notation, /ÐBÑ œ Ð0αÐBÑÑÞ \ whose coordinate function is 0 ÐBÑ:\.<br />
In more informal<br />
α α α<br />
Exercise 4.2 Suppose \œ −E\ÞFor each α there is a projection map 1 À\Ä\ . What is<br />
α<br />
the evaluation map defined by the family Ö1 À α − E× ?<br />
α<br />
261
Definition 4.3 Suppose \ <strong>and</strong> \ α( α − EÑare topological spaces <strong>and</strong> that 0α À \ Ä \ αÞ<br />
We say that<br />
the family Ö0α À α − E× separates points if, for each pair <strong>of</strong> points B Á C − \ , there exists an α − E<br />
for which 0ÐBÑÁ0ÐCÑ.<br />
α α<br />
Clearly, the evaluation map /À\Ä <br />
α−E\<br />
α is one-to-one Í for all BÁC−\ß/ÐBÑÁ/ÐCÑ<br />
Í for all BÁC−\ there is an α −E for which /ÐBÑÐαÑœ0αÐBÑÁ0αÐCÑœ/ÐCÑÐαÑ Í the family Ö0 À − E× separates points.<br />
α α<br />
Theorem 4.4 Suppose \ has the weak topology generated by the maps 0α À \ Ä Ð\ αßgαÑ <strong>and</strong> that<br />
the family Ö0α À α − E× separates points. Then / is an embedding that is, / a homeomorphism<br />
between \ <strong>and</strong> /Ò\Ó © \ α.<br />
Pro<strong>of</strong> Since Ö0 α×<br />
separates points, / is one-to-one <strong>and</strong> / is continuous because each composition<br />
1α ‰/œ0α is continuous.<br />
/ preserves unions <strong>and</strong> (since / is one-to-one) intersections. So to check that / is an open map<br />
from \ to /Ò\Ó , it is sufficient to show that / takes subbasic open sets in \ to open sets in /Ò\Ó.<br />
Because \ has the weak topology, a subbasic open set has the form Y œ 0αÒZÓ, where Z is open in<br />
"<br />
" "<br />
\ α. But then /ÒY Ó œ /Ò0α ÒZ ÓÓ œ 1α ÒZ Ó ∩ /Ò\Ó is an open set in /Ò\ÓÞ ñ<br />
( Note: /ÒY Ó might not be open in \α,<br />
but that is irrelevant. See the earlier example where<br />
# B /ÐBÑ œ ÐB ß / )Þ<br />
The converse <strong>of</strong> Theorem 4.4 is also true: if e is an embedding, then the 0 α's<br />
separate points <strong>and</strong> \<br />
has the weak topology generated by the 0α's.<br />
However, we do not need this fact <strong>and</strong> will omit the pro<strong>of</strong><br />
(which is not very hard).<br />
Example 4.5<br />
Let Ð\ß .Ñ be a separable metric space. We can assume, without loss <strong>of</strong> generality, that . Ÿ "Þ \ is<br />
second countable so there is a countable base U œ ÖY" ß ÞÞÞß Y8ßÞÞÞ× for the open sets. For each 8,<br />
let<br />
08ÐBÑ œ .ÐBß \ Y8ÑÞ Then 08 À \ Ä Ò!ß "Ó is continuous <strong>and</strong>, since \ Y8<br />
is closed, we have<br />
08ÐBÑ! iff B−Y8ÞIf BÁC−\ , there is an 8such that B−Y8 <strong>and</strong> CÂY8ÞThen 08ÐCÑœ!Á08ÐBÑ, so 08's<br />
separate points.<br />
We claim that the topology g. on \ is actually just the weak topology gAgenerated<br />
by the 08's.<br />
Because the functions 0 8 are continuous if \ has the topology g. , we know that g. ª gAÞ<br />
To show<br />
g. © gA , suppose B−Y −g. Þ For some 8, we have B−Y 8 ©Y<strong>and</strong> therefore 08ÐBÑœ-!ÞBut " -<br />
Z8 œ08 ÒÐ# ß"ÓÓis a subbasic open set in the weak topology <strong>and</strong> B−Z 8 ©Y 8 ©YÞTherefore<br />
Y−gA.<br />
i i<br />
! !<br />
By Theorem 4.4, /À\ ÄÒ!ß"Ó is an embedding, so \ /Ò\Ó©Ò!ß"Ó Þ We sometimes write this<br />
top<br />
i! i!<br />
as \ © Ò!ß "Ó . From Theorems 3.3 <strong>and</strong> 3.5, we know that Ò!ß "Ó is itself a separable metrizable<br />
space <strong>and</strong> therefore all its subspaces are separable <strong>and</strong> metrizable. Putting all this together, we get<br />
that topologically, the separable metrizable spaces are nothing more <strong>and</strong> nothing less than the<br />
subspaces <strong>of</strong> Ò!ß "Ói! .<br />
We can view this fact with a “half-full” or “half-empty” attitude:<br />
262
i) separable metric spaces must not be very complicated since topologically they are nothing<br />
more than the subspaces <strong>of</strong> a single very nice space: the “cube” Ò!ß "Ó i!<br />
ii) separable metric spaces can get quite complicated, so the subspaces <strong>of</strong> a cube Ò!ß "Ó i!<br />
are more complicated than we imagined.<br />
i<br />
∞ "<br />
! Since Ò!ß "Ó <br />
8œ" Ò!ß 8 Ó œ L (the “Hilbert cube”) © j#<br />
, we can also say that topologically<br />
the separable metrizable spaces are precisely the subspaces <strong>of</strong> j# . This is particularly amusing<br />
because <strong>of</strong> the almost “Euclidean” metric on :<br />
. j#<br />
∞<br />
8 8<br />
8œ"<br />
.ÐBß CÑ œ Ð ÐB C Ñ Ñ<br />
In some sense, this elegant “Euclidean-like” metric is adequate to describe the topology <strong>of</strong> any<br />
separable metric space.<br />
We can summarize by saying that each <strong>of</strong> L <strong>and</strong> j# is a “universal separable metric space” ( But they<br />
are not homeomorphic: L is compact but j# is not. )<br />
Example 4.6<br />
Suppose Ð\ß .Ñ is a metric space, not necessarily separable, <strong>and</strong> that ÖY À − E× is a base for the<br />
topology g. , where 7 œ lElÞ An argument completely analogous to the reasoning in Example 4.5<br />
(just replace “ 8 ” everywhere with “ α ”) shows that \ © Ò!ß"Ó . Therefore topologically<br />
top<br />
7 every metric<br />
7 space is a subspace <strong>of</strong> some sufficiently large “cube.” Of course, if 7 i! , the cube Ò!ß "Ó is not<br />
itself metrizable ( why? ); in general this cube will have many subspaces that are nonmetrizable. So the<br />
result is not quite as dramatic as in the separable case.<br />
263<br />
" # #<br />
α α<br />
The weight AÐ\Ñ <strong>of</strong> a topological space Ð\ß g Ñ is defined as min ÖlUlÀ U is a base for g × i! Þ<br />
We are assuming here that the “min” in the definition exists: see Example 5.22 in <strong>Chapter</strong> VIII.<br />
J or some very simple spaces, the “min” could be finite in which case the “ i! ” guarantees<br />
that AÐ\Ñ i Ðconvenient<br />
for purely technical reasons that don't matter in these notes).<br />
!<br />
The density $ Ð\Ñ is defined as min ÖlHl À H is dense in Ð\ß g Ñ× i! Þ For a metrizable space, it is<br />
not hard to prove that AÐ\Ñ œ $ Ð\Ñ.<br />
The pro<strong>of</strong> is just like our earlier pro<strong>of</strong> (in Theorem III.6.5) that<br />
separability <strong>and</strong> second countability are equivalent in metrizable spaces. Therefore we have that for<br />
any metric space Ð\ß .Ñß<br />
top<br />
AÐ\Ñ $ Ð\Ñ<br />
\ © Ò!ß "Ó œ Ò!ß "Ó (*)<br />
Notice that, for a given space Ð\ß .Ñ,<br />
the exponents in this statement are not necessarily the smallest<br />
top<br />
AБ Ñ i!<br />
possible. For example, (*) says that ‘ © Ò!ß "Ó œ Ò!ß "Ó , but in fact we can do much better than<br />
"<br />
the exponent i À ‘ Ð!ß "Ñ © Ò!ß "Ó œ Ò!ß "Ó !<br />
!<br />
We add one additional comment, without pro<strong>of</strong>: For a given infinite cardinal 7,<br />
it is possible to define<br />
a metrizable space L7with weight 7 such that every metric space \ with weight 7 can be<br />
i! i!<br />
embedded in H 7Þ In other words, L7 is a metrizable space which is “universal for all metric spaces<br />
with AÐ\Ñ œ 7.” The price <strong>of</strong> metrizability, here, is that we need to replace Ò!ß "Ó by a more<br />
complicated space L7.
Without going into all the details, you can think <strong>of</strong> L7as a “star” with 7 different copies <strong>of</strong><br />
Ò!ß "Ó (“rays”) all placed with ! at the center <strong>of</strong> the star. For two points Bß C on the same “ray”<br />
<strong>of</strong> the star, .ÐBß CÑ œ lB Cl; if Bß C are on different rays, the distance between them is<br />
measured “via the origin” À .ÐBß CÑ œ lBl lClÞ<br />
The condition in Theorem 4.4 that “ \ has the weak topology generated by a collection <strong>of</strong> maps<br />
0α À \ Ä \ α” is sometimes not so easy to check. The following definition <strong>and</strong> theorem can<br />
sometimes help.<br />
Definition 4.7 Suppose \ <strong>and</strong> \ α( α − EÑare topological spaces <strong>and</strong> that, for each α,<br />
0α À \ Ä \ αÞ<br />
We say that the collection Y œÖ0αÀ α −E× separates points from closed sets if whenever Jis<br />
a<br />
closed set in \ <strong>and</strong> B Â J , there is an such that 0 ÐBÑ Â cl 0 ÒJ ÓÞ<br />
α α α<br />
Example 4.8 Let \œ‘ <strong>and</strong> Y œGБ Ñ. Suppose J is a closed set in ‘ <strong>and</strong>
Note: the more open (closed) sets there are in \ , the harder it is for a given family Ö0 À − E× to<br />
succeed in separating points <strong>and</strong> closed sets. In fact ß the lemma shows that a family <strong>of</strong> continuous<br />
functions Ö0α À α − E× succeeds in separating points <strong>and</strong> closed sets only if g is the smallest<br />
topology that makes the 0α's<br />
continuous.<br />
"<br />
Pro<strong>of</strong> Suppose U is a base. Then the sets 0αÒZÓ in U are open, so the 0α's<br />
are continuous <strong>and</strong><br />
i) holds.<br />
To prove ii), suppose J is closed in \ <strong>and</strong> B Â J. For some α <strong>and</strong> some Z open in \ α,<br />
we<br />
" "<br />
have B − 0αÒZÓ © \ JÞ Then 0αÐBÑ − Z <strong>and</strong> Z ∩0αÒJÓ œ g Ðsince 0αÒZÓ∩J œ gÑ.<br />
Therefore 0ÐBÑÂcl 0ÒJÓso<br />
ii) also holds.<br />
α α<br />
Conversely, suppose i) <strong>and</strong> ii) hold. If B−S<strong>and</strong> S is open in \ , we need to find a set<br />
" "<br />
0α ÒZÓ − U such that B − 0α ÒZÓ © SÞ Since B  J œ \ S,<br />
condition ii) gives us an α for which<br />
" "<br />
0αÐBÑ Â cl 0αÒJ Ó. Then 0αÐBÑ − Z œ \ α cl 0αÒJ ÓÞ Then B − 0α ÒZ Ó <strong>and</strong> we claim 0α ÒZ Ó © S:<br />
"<br />
Suppose D − 0α ÒZ Ó so 0αÐDÑ−Zœ \ α cl 0αÒJ ÓÞ<br />
But if D−J, then 0ÐDÑ−0ÒJÓœ0Ò α α α cl JÓ© cl 0ÒJÓÞ α<br />
Therefore D−\JœSÞñ Theorem 4.10 Suppose 0 À \ Ä \ is continuous for each α − E. If the collection Ö0 À α − E×<br />
265<br />
α α<br />
α α α<br />
i) separates points <strong>and</strong> closed sets, <strong>and</strong><br />
<br />
ii) separates points<br />
then the evaluation map /À\Ä \ α is an embedding.<br />
Pro<strong>of</strong> Since the 0 α's<br />
are continuous, Lemma 4.9 gives us that \ has the weak topology. Then<br />
Theorem 4.4 implies that / is an embedding. ñ<br />
If the space \ we are trying to embed is a X" -space ( as is most <strong>of</strong>ten the case),<br />
then<br />
i) Ê ii) in Theorem 4.10 , so we have the simpler statement given in the following corollary.<br />
Corollary 4.11 Suppose 0α À \ Ä \ α is continuous for each α − E . If \ is a X"<br />
-space <strong>and</strong><br />
Ö0 À α − E× separates points <strong>and</strong> closed sets, then evaluation map / À \ Ä \<br />
is an embedding.<br />
α α<br />
Pro<strong>of</strong> By Theorem 4.10, it is sufficient to show that the 0α's separate points, so suppose B Á C − \ .<br />
Since B  the closed set ÖC× , there is an α for which 0αÐBÑ Â cl 0αÒÖC×Ó. Therefore 0αÐBÑ Á 0αÐCÑß so<br />
Ö0α À α − E× separates points.<br />
ñ
Exercises<br />
E14. A space Ð\ß g Ñ is called a X! space if whenever B Á C − \ , then aB Á aC<br />
(equivalently, either<br />
there is an open set Y containing B but not C, or vice-versa). Notice that the X! condition is weaker<br />
than X ( see example III.2.6.4). Clearly, a subspace <strong>of</strong> a X -space is X .<br />
" ! !<br />
α α ! α !<br />
a) Prove that a nonempty product \œ Ö\ À −E× is X iff each \ is X.<br />
b) Let W be “Sierpinski space” that is, W œ Ö!ß "× with the topology g œ Ögß Ö 1 ×ß Ö!ß "×× . Use<br />
7<br />
the embedding theorems to prove that a space \ is X ! iff \ is homeomorphic to a subspace <strong>of</strong> W for<br />
some cardinal 7. ( Nearly all interesting spaces are X! , so nearly all interesting spaces are<br />
7 (topologically) just subspaces <strong>of</strong> W for some cardinal 7.<br />
)<br />
Hint Ê : for each open set Y in \ , let ;Y be the characteristic function <strong>of</strong> YÞ Use an embedding<br />
theorem.<br />
E15. a) Let Ð\ß .Ñ be a metric space. Prove that GÐ\Ñ ( œ the collection <strong>of</strong> continuous real-valued<br />
functions on \ ) separates points from closed sets Þ ( Since \ is X" , GÐ\Ñ therefore also separates<br />
points).<br />
b) Suppose \ is any X! topological space for which GÐ\Ñ separates points <strong>and</strong> closed sets.<br />
Prove that \ can be embedded in a product <strong>of</strong> copies <strong>of</strong> ‘ .<br />
E16. A space \ satisfies the countable chain condition (CCC) if every collection <strong>of</strong> nonempty<br />
pairwise disjoint open sets must be countable. ( For example, every separable space satisfies CCC. )<br />
Suppose that \α is separable for each α −E . Prove that \œÖ\ α:<br />
α −E× satisfies CCC.<br />
( There isn't much to prove when | A | Ÿ c; why? )<br />
Hint: Let ÖU > À > − X× be any such collection. We can assume all the U> 's are basic open sets (why?).<br />
Prove that if W© T <strong>and</strong> | W| Ÿ c, then | W| Ÿ i! <strong>and</strong> hence Xmust<br />
be countable.)<br />
E17. There are several ways to define dimension for topological spaces. One classical method is the<br />
following inductive definition.<br />
Define dim gœ 1.<br />
For :−\ , we say that \ has dimension Ÿ0 at p if there is a neighborhood base at p<br />
consisting <strong>of</strong> sets with 1 dimensional (that is, empty) frontiers. Since a set has empty frontier iff<br />
it is clopen, \ has dimension Ÿ 0 at : iff : has a neighborhood base consisting <strong>of</strong> clopen sets.<br />
We say \ has dimension Ÿ 8 at : if : has a neighborhood base consisting <strong>of</strong> sets with n 1<br />
dimensional frontiers.<br />
We say \ has dimension Ÿ n, <strong>and</strong> write dimÐ\Ñ Ÿ n, if \ has dimension Ÿ 8 at p for each<br />
p − X <strong>and</strong> that dim(X) œ n if dim( \ ) Ÿ 8 but dim Ð\Ñ ŸÎ8" . We say dimÐ\Ñ<br />
œ ∞ if<br />
dim Ð\Ñ Ÿ 8 is false for every 8 − Þ<br />
While this definition <strong>of</strong> dim Ð\Ñ makes sense for any topological space \ , it turns out that<br />
“dim” produces a nicely behaved dimension theory only for separable metric spaces. The<br />
dimension function “dim” is sometimes called small inductive dimension to distinguish it from<br />
other more general definitions <strong>of</strong> dimension. The classic discussion <strong>of</strong> small inductive dimension<br />
is in Dimension Theory (<br />
Hurewicz <strong>and</strong> Wallman).<br />
266
It is clear that “dim Ð\Ñ œ 8 ” is a topological property <strong>of</strong> \ . There is a theorem stating that<br />
8 8 7<br />
dim( ‘ ) œ8, from which it follows that ‘ is not homeomorphic to ‘ if m Án.<br />
In proving the<br />
8 8<br />
theorem, showing dim( Б Ñ Ÿ n is easy; the hard part is showing that dim Б Ñ ŸÎ n 1.<br />
a) Prove that dim( ‘ ) œ" .<br />
b) Let G be the Cantor set. Prove that dim( G)<br />
œ !Þ<br />
c) Suppose Ð\ß .Ñ is a ! -dimensional separable metric space. Prove that \ is homeomorphic<br />
to a subspace <strong>of</strong> C. ( Hint: Show that \ has a countable base <strong>of</strong> clopen sets. View G as Ö!ß #× i!<br />
<strong>and</strong> apply the embedding theorems.)<br />
E18. Suppose \ <strong>and</strong> ] are X ! -spaces. Part a) outlines a sufficient condition enabling ] to be<br />
top<br />
embedded in a product <strong>of</strong> \ 's, i.e., that ] © \ for some cardinal m. Parts b) <strong>and</strong> c) look at some<br />
7<br />
corollaries.<br />
a) Theorem Let \ <strong>and</strong> ] be X ! spaces. Then ] can be topologically embedded in \ for<br />
some cardinal 7 if, for every closed set J ©] <strong>and</strong> every point C ÂJ,<br />
there exists a<br />
continuous function 0À]Ä\ such that f( y) cl f[ F] (here, n is finite <strong>and</strong> depends on f).<br />
8 Â<br />
Pro<strong>of</strong> Let X œ Ö> œ ÐCßJÑ À J is closed in ] <strong>and</strong> C  J× <strong>and</strong>, for each such pair<br />
8<br />
>œÐCß0Ñ, let 0 > be the function given in the hypothesis. Let \ > œ\ (the space<br />
containing the values <strong>of</strong> 0> ). Then clearly <br />
7<br />
Ö\ > À > − X× is homeomorphic to \ for some<br />
7 , so it suffices to show ] can be embedded in Ö\ > À > − X× . Let<br />
2À] Ä Ö\ > À>−X× be defined as follows: for C−] , 2Ð>Ñ has for its > -th coordinate<br />
0> ÐCÑß i.e., 2ÐCÑÐ>Ñ œ 0> ÐCÑÞ<br />
i) Show 2 is continuous.<br />
ii) Show 2 is one-to-one.<br />
iii) Show that 2 is a closed mapping onto its range 2Ò]Ó to complete the pro<strong>of</strong> that 2 is<br />
+ homeomorphism between ] <strong>and</strong> 2Ò]Ó © Ö\ > À > − X× . ( Note: the converse <strong>of</strong> the<br />
theorem is also true. Both the theorem <strong>and</strong> converse are due to S. Mrowka.)<br />
b) Let F denote Sierpinski space {{0,1},{ ,{0},{0,1}}. Use the theorem to show that every T<br />
7<br />
space Y can be embedded in F for some m.<br />
g !<br />
c) Let D denote the discrete space Ö 0,1 × . Use the theorem to show that every X " -space ]<br />
7<br />
satisfying dim Ð] Ñ œ ! (see Exercise E14) can be embedded in D for some m.<br />
Parts b) <strong>and</strong> c) are due to Alex<strong>and</strong>r<strong>of</strong>f.<br />
Mrowka also proved that there is no T" -space \ such that every T" -space ] can be<br />
7<br />
embedded in X for some m.<br />
7<br />
E19 Þ Let \ œ Ö!ß "× <strong>and</strong> consider the product space \ , where 7 is an infinite cardinal.<br />
267<br />
7
7<br />
a) Show that \ contains a discrete subspace <strong>of</strong> cardinality 7.<br />
7 b) Show that AÐ\ Ñ œ 7 Ðsee<br />
Example 4.6).<br />
E20. A space \ is called totally disconnected every connected subset E satisfies lEl Ÿ "Þ Prove that<br />
a totally disconnected compact Hausdorff space is homeomorphic to a closed subspace <strong>of</strong> Ö!ß "× for 7<br />
some 7. ( Hint: see Lemma V.5.6).<br />
E21. Suppose \ is a countable space that does not have a countable neighborhood base at the point<br />
+−\Þ ( For instance, let +œÐ!ß!Ñin the space P,<br />
Example III.9.8. )<br />
i! Let E4 œ ÖB − \ À B3 œ + for 3 4× <strong>and</strong> E œ ∞ i!<br />
4œ! E 4 © \ . Prove that no point in the<br />
(countable) space E has a countable neighborhood base. ( Note: it is not necessary that \ be<br />
countable. That condition simply forces E to be countable <strong>and</strong> makes the example more dramatic. )<br />
268
5. The Quotient Topology<br />
Suppose that for each α −Ewe have a map 1α À\ α Ä] , where \ α is a topological space <strong>and</strong> ] is a<br />
set. Certainly there is a topology for ] that will make all the 1α's<br />
continuous: for example, the trivial<br />
topology on ] . But what is the largest topology on ] that will do this? Let<br />
"<br />
α<br />
g œÖS©] Àfor all α −E, 1 ÒSÓ is open in \ ×<br />
It is easy to check that g is a topology on ] . For each α,<br />
by definition, each set 1αÒSÓ is open in \ α<br />
so g makes all the 1α's continuous. Moreover, if F © ] <strong>and</strong> F Â g , then for at least one α,<br />
" 1αÒFÓ is not open in \ α so adding F to g would “destroy” the continuity <strong>of</strong> at least one 1αÞ Therefore g is the largest possible topology on ] making all the 1α's<br />
continuous.<br />
Definition 5.1 Suppose 1α À \ α Ä ] for each α − E.<br />
The strong topology g on ] generated by the<br />
"<br />
maps 1 α is the largest topology on ] making all the maps 1α continuous ß <strong>and</strong> S − g iff 1αÒSÓis open<br />
in \α for every α.<br />
The strong topology generated by a collection <strong>of</strong> maps Ö1 À − E× is “dual” to the weak topology in<br />
α α<br />
the sense that it involves essentially the same notation but with “all the arrows pointing in the opposite<br />
direction.” For example, the following theorem states that a map 0 out <strong>of</strong> a space with the strong<br />
topology is continuous iff each map 0‰1αis continuous; but a map 0 into a space with the weak<br />
topology generated by mappings 0αis continuous iff all the compositions 0 α ‰0 are continuous ( see<br />
Theorem 2.6)<br />
Theorem 5.2 Suppose ] has the strong topology generated by a collection <strong>of</strong> maps Ö1αÀ α − E×Þ If<br />
^is a topological space <strong>and</strong> 0À] Ä^, then 0 is continuous iff 0‰1α À\ α Ä^ is continuous for<br />
each α −E.<br />
1α0 Pro<strong>of</strong> For each α −E , we have \ α Ä] Ä^, <strong>and</strong> the 1 α's<br />
are continuous since ] has the strong<br />
topology.<br />
If 0 is continuous, so is each composition 0 ‰1αÞ Suppose each 0‰1αis continuous <strong>and</strong> that Yis open in ^. We want to show that 0 ÒYÓ<br />
" " "<br />
is open in ] . But ] has the strong topology, so 0 ÒY Ó is open in ] iff each 1α Ò0 ÒY ÓÓ<br />
" " "<br />
is open in \ αÞBut 1α Ò0 ÒYÓÓœ Ð0 ‰1αÑÒYÓwhich is open because 0 ‰1α<br />
is<br />
continuous. ñ<br />
We introduced the idea <strong>of</strong> the strong topology as a parallel to the definition <strong>of</strong> weak topology.<br />
However, we are going to use the strong topology only in a special case:<br />
when there is only one map<br />
1α œ 1 <strong>and</strong> 1 is onto.<br />
Definition 5.3 Suppose Ð\ß g Ñ is a topological space <strong>and</strong> 1 À \ Ä ] is onto.<br />
The strong topology on<br />
] generated by 1 is also called the quotient topology<br />
on ] . If ] has the quotient topology from 1,<br />
we<br />
269<br />
α<br />
"<br />
"
say that 1À\Ä] is a quotient mapping <strong>and</strong> we say ] is a quotient <strong>of</strong> \ . We also say the ] is a<br />
quotient space <strong>of</strong> \ <strong>and</strong> sometimes as ] œ \Î1.<br />
From our earlier discussion we know that if \ is a topological space <strong>and</strong> 1 À \ Ä ] , then the quotient<br />
" "<br />
topology on ] is g œ ÖY À 1 ÒY Ó is open in \× . Thus Y − g if <strong>and</strong> only if 1 ÒY Ó is open in \ :<br />
notice in this description that<br />
“only if” guarantees that 1 is continuous <strong>and</strong><br />
“if” guarantees that g is the largest topology on ] making 1 continuous.<br />
<strong>Quotients</strong> <strong>of</strong> \ are used to create new spaces ] by “pasting together” (“identifying”) several points <strong>of</strong><br />
\ to become a single new point. Here are two intuitive examples:<br />
i) Begin with \œÒ!ß"Ó <strong>and</strong> identify ! with " (that is, “paste” ! <strong>and</strong> " together to become a<br />
" "<br />
single point). The result is a circle, WÞ This identification is exactly what the map 1ÀÒ!ß"ÓÄW<br />
given by 1ÐBÑ œ Ð cos # 1Bß sin # 1BÑ accomplishes. It turns out ( see below)<br />
that the usual topology on<br />
" W is the same as quotient topology generated by the map 1. Therefore we can say that 1 is a quotient<br />
"<br />
map <strong>and</strong> that W is a quotient <strong>of</strong> Ò!ß "Ó<br />
"<br />
ii) If we take the space \œW <strong>and</strong> use a mapping 1to<br />
“identify” the north <strong>and</strong> south poles<br />
together, the result is a “figure-eight” space ] . The usual topology on ] (from ‘ ) turns out to be the<br />
#<br />
same as quotient topology generated by 1 ( see below).<br />
Therefore we can say that 1 is a quotient<br />
mapping <strong>and</strong> the the “figure-eight” is a quotient <strong>of</strong> W . "<br />
Suppose we are given an onto map 1ÀÐ\ßg ÑÄÐ]ßg Ñ. How can we tell whether 1is<br />
a quotient<br />
w<br />
map that is, how can we tell whether g is the quotient topology? By definition, we must check that<br />
w<br />
w "<br />
Y−g iff 1 ÒYÓ−g Þ Sometimes it is fairly straightforward to do this. But the following theorem<br />
can sometimes make things much easier.<br />
Theorem 5.4 Suppose 1ÀÐ\ßg ÑÄÐ]ßg Ñis continuous <strong>and</strong> onto. If 1is<br />
open (or closed), then<br />
w<br />
g w is the quotient topology, so 1 is a quotient map. In particular,<br />
if \ is compact <strong>and</strong> ] is Hausdorff,<br />
1 is a quotient mapping.<br />
Note: Whether the map 1À\Ä] is continuous depends, <strong>of</strong> course, on the topology g , but if g<br />
makes 1 continuous, then so would any smaller topology on ] . The theorem tells us that if 1 is<br />
w w<br />
both continuous <strong>and</strong> open (or closed), then g is completely determined by 1:<br />
g is the largest<br />
topology making 1 continuous.<br />
270<br />
w " w "<br />
w w<br />
Pro<strong>of</strong> Suppose Y © ] Þ W / must show Y − g iff 1 ÒY Ó − g Þ If Y − g , then 1 ÒY Ó − g<br />
" "<br />
because 1 is continuous. On the other h<strong>and</strong>, suppose 1 ÒY Ó − g . Since 1 is onto, 1Ò1 ÒY ÓÓ œ Y <strong>and</strong><br />
" w<br />
so, if 1 is open, Y œ 1Ò1 ÒY ÓÓ − g Þ<br />
" "<br />
For J © ] ß 1 Ò] J Ó œ \ 1 ÒJ ÓÞ It follows easily that J is closed in the quotient<br />
"<br />
space if <strong>and</strong> only if 1 ÒJÓ is closed in \Þ With that observation, the pro<strong>of</strong> that a continuous, closed,<br />
onto map 1is a quotient map is exactly parallel to the pro<strong>of</strong> when 1 is open.<br />
If \ is compact <strong>and</strong> ] is X , then 1 must be closed, so 1 is a quotient mapping. ñ<br />
#<br />
"<br />
Note: Theorem 5.4 implies that the map 1ÐBÑ œ Ð cos # 1Bß sin # 1 BÑ from Ò!ß "Ó to W is a quotient<br />
"<br />
map, but 1 is not open. The same formula 1 can be used to define a quotient map 1 À ‘ Ä W<br />
which is not closed (why?). Exercise E25 gives examples <strong>of</strong> quotient maps 1 that are neither
open nor closed.<br />
Suppose µ is an equivalence relation on a space \ . For each B − \ , the equivalence class <strong>of</strong> B is<br />
ÒBÓœÖD−\ÀDµB× . The equivalence classes partition \ into a collection <strong>of</strong> nonempty pairwise<br />
disjoint sets. (Conversely, it is easy to see that any partition <strong>of</strong> \ is the collection <strong>of</strong> equivalence<br />
classes for some equivalence relation namely, B µ D iff B<strong>and</strong> D are in the same set <strong>of</strong> the partition. )<br />
The set <strong>of</strong> equivalence classes, ] œÖÒBÓÀB−\× , is sometimes denoted \ÎµÞ There is a natural<br />
onto map 1À\Ä] given by 1ÐBÑœÒBÓ . We can think <strong>of</strong> the elements <strong>of</strong> ] as “new points” which<br />
arise from “identifying together as one” all the members <strong>of</strong> each equivalence class in \ . If \ is a<br />
topological space, we can give the set <strong>of</strong> equivalence classes ] the quotient topology. Conversely,<br />
whenever 1À\Ä] is any onto mapping, we can think <strong>of</strong> ] as the set <strong>of</strong> equivalence classes for<br />
"<br />
some equivalence relation on \ namely B µ C Í C − 1 ÐBÑ Í 1ÐCÑ œ 1ÐBÑ .<br />
Example 5.5 For + , , − ,<br />
define + µ , iff ,+ is even. There are two equivalence classes<br />
Ò!Ó œ ÖÞÞÞß %ß #ß !ß #ß %ß ÞÞÞ× <strong>and</strong> Ò"Ó œ ÖÞÞÞ $ß "ß "ß $ß ÞÞÞ× so Î<br />
µ œ ÖÒ!Óß Ò"Ó×Þ<br />
Define 1À Äε by 1Ð+ÑœÒ+Ó<strong>and</strong> give ε<br />
the quotient topology. A set Y is<br />
" open in ε iff 1 ÒYÓis open in , <strong>and</strong> that is true for every Y © ε because is discrete.<br />
Therefore the quotient Î µ is a two point discrete space.<br />
Example 5.6 Let Ð\ß .Ñ be a pseudometric space <strong>and</strong> define a relation µ in \ by B µ D iff<br />
.ÐBßDÑœ! . Let ] œ\ε , define 1À\Ä] by 1ÐBÑœÒBÓ . Give ] the quotient topologyÞ<br />
Points at distance ! in \ have been “identified with each other” to become one point (the equivalence<br />
class) in ]Þ<br />
w w<br />
For ÒBÓß ÒDÓ − ] , define . ÐÒBÓß ÒDÓÑ œ .ÐBß DÑ . In order to see that . is well-defined,<br />
we need to check that the definition is independent <strong>of</strong> the representatives chosen from the equivalence<br />
classes:<br />
w w w w<br />
If ÒBÓœÒBÓ<strong>and</strong> ÒDÓœÒDÓ , then .ÐBßBÑœ! <strong>and</strong> .ÐDßDÑœ!ÞTherefore w w w w w w<br />
.ÐBßDÑ Ÿ .ÐBßB Ñ .ÐB ß D Ñ .ÐD ß DÑ œ .ÐB ß D Ñ,<br />
<strong>and</strong> similarly<br />
w w w w w w w w<br />
.ÐB ß D Ñ Ÿ .ÐBß DÑ . Thus .ÐB ß D Ñ œ .ÐBß DÑ so . ÐÒB Óß ÒD ÓÑ œ . ÐÒBÓß ÒDÓÑÞ<br />
w w w<br />
It is easy to check that . is a pseudometric on ] . In fact, . is a metric:<br />
if . ÐÒBÓßÒDÓÑ œ ! , then<br />
.ÐBß DÑ œ ! , which means that B µ D <strong>and</strong> ÒBÓ œ ÒDÓÞ<br />
We now have two definitions for topologies on ] : the quotient topology g <strong>and</strong> the metric topology<br />
g w. We claim that g œ g wÞ<br />
To see this, first notice that<br />
. .<br />
w . w .<br />
% %<br />
ÒCÓ − F ÐÒBÓÑ iff . ÐÒCÓß ÒBÓÑ % iff .ÐCß BÑ % iff C − F ÐBÑ<br />
w w<br />
" . . . .<br />
% % % %<br />
Therefore 1 ÒF ÐÒBÓÑÓ œ F ÐBÑ <strong>and</strong> 1 ÒF ÐBÑÓ œ F ÐÒBÓÑ.<br />
But then we have<br />
Y−g. w iff Y is a union <strong>of</strong> . -balls<br />
" iff 1 ÒYÓ is a union <strong>of</strong> . -balls<br />
" iff 1 ÒYÓ is open in \<br />
iff Y−g Þ<br />
271<br />
w
w The metric space Ð] ß . Ñ is called the metric identification <strong>of</strong> the pseudometric space Ð\ß .Ñ.<br />
In effect,<br />
we turn the pseudometric space into a metric space by agreeing that points in \ at distance ! are<br />
“identified together” <strong>and</strong> treated as one point.<br />
Note: In this particular example, it is easy to verify that the quotient mapping 1À\Ä] is open,<br />
so 1 would be a homeomorphism if only 1 were "" . If the original pseudometric . is actually a<br />
metric, then 1 is "" <strong>and</strong> a homeomorphism: the metric identification <strong>of</strong> a metric space Ð\ß.Ñ is<br />
itself.<br />
Example 5.7 Exactly what does it mean if we say “identify the endpoints <strong>of</strong> Ò!ß "Ó <strong>and</strong> get a circle”?<br />
Of course, one could simply take this to be the definition <strong>of</strong> a (topological) “circle.” Or, it could mean<br />
that we already know what a circle is <strong>and</strong> are claiming that a certain quotient space is homeomorphic to<br />
a circle. We take the latter point <strong>of</strong> view.<br />
" Let 1ÀÒ!ß"ÓÄW be given by 1ÐBќРcos # 1Bß sin # 1BÞ ) This map is onto <strong>and</strong> "" except<br />
that 1Ð!Ñœ1Ð"Ñ, so 1 corresponds to the equivalence relation on \ for which !µ" (<strong>and</strong> there are no<br />
other equivalences except that BµB for every BÑÞ We can think <strong>of</strong> the equivalence classes as<br />
corresponding in a natural way to the points <strong>of</strong> W . "<br />
" "<br />
Here W has its usual topology <strong>and</strong> 1 is continuous. However, since \ is compact <strong>and</strong> W is Hausdorff,<br />
" Theorem 5.4 gives that the usual topology on W is the quotient topology <strong>and</strong> 1is<br />
a quotient map.<br />
When it “seems apparent” that the result <strong>of</strong> making certain identifications produces some familiar<br />
space ] , we need to check that the familiar topology on ] is actually the quotient topology. Example<br />
5.7 is reassuring: if we believed, intuitively, that the result <strong>of</strong> identifying the endpoints <strong>of</strong> Ò!ß "Ó should<br />
" "<br />
be W but then found that the quotient topology on the set W turned out to be different from the usual<br />
topology, we would be inclined to think that we had made the “wrong” definition for a “quotient.”<br />
Example 5.8 Suppose we take a square Ò!ß "Ó <strong>and</strong> identify points on top <strong>and</strong> bottom edges using the<br />
#<br />
equivalence relation ÐBß !Ñ µ ÐBß "ÑÞ<br />
We can schematically picture this identification as<br />
272
The arrows indicate that the edges are to be identified as we move along the top <strong>and</strong> bottom edges in<br />
# $<br />
the same direction. We have an obvious map 1 from Ò!ß "Ó to a cylinder in ‘ which identifies points<br />
in just this way, <strong>and</strong> we can think <strong>of</strong> the equivalence classes as corresponding in a natural way to the<br />
points <strong>of</strong> the cylinder.<br />
The cylinder has its usual topology from <strong>and</strong> the map is (clearly) continuous <strong>and</strong> onto. Again,<br />
‘ $ 1<br />
Theorem 5.4 gives that the usual topology on the cylinder is, in fact, the quotient topology.<br />
273
Example 5.9 Similarly, we can show that a torus (the “surface <strong>of</strong> a doughnut”) is the result <strong>of</strong> the<br />
#<br />
following identifications in Ò!ß"Ó : ÐBß!ѵÐBß"Ñ<strong>and</strong> Ð!ßCѵÐ"ßCÑ<br />
Thinking in two steps, we see that the identification <strong>of</strong> the two vertical edges produces a cylinder; the<br />
circular ends <strong>of</strong> the cylinder are then identified (in the same direction) to produce the torus.<br />
The two circles darkly shaded on the surface represent the identified edges.<br />
We can identify the equivalence classes naturally with the points <strong>of</strong> this torus in <strong>and</strong> just as before<br />
‘ $<br />
we see that the usual topology on the torus is in fact the quotient topology.<br />
274
Example 5.10 Define an equivalence relation µ in Ò!ß "Ó by setting ÐBß !Ñ µ Ð" Bß "ÑÞ Intuitively,<br />
the idea is to identify the points on the top <strong>and</strong> bottom edges with each other as we move along the<br />
edges in opposite directions. We can picture this schematically as<br />
Physically, we can think <strong>of</strong> a strip <strong>of</strong> paper <strong>and</strong> glue the top <strong>and</strong> bottom edges together after making a<br />
“half-twist.” The quotient space \Î µ is called a Mobius ¨ strip.<br />
We can take the quotient \Î µ as the definition <strong>of</strong> a Mobius ¨ strip, or we can consider a “real” Mobius ¨<br />
strip Q in ‘ <strong>and</strong> define a map 1 À Ò!ß "Ó Ä Q that accomplishes the identification we have in mind.<br />
$ #<br />
In that case there is a natural way to associate the equivalence classes to the points <strong>of</strong> the torus in ‘ $<br />
<strong>and</strong> again Theorem 5.4 guarantees that the usual topology on the Mobius ¨<br />
strip is the quotient topology.<br />
275<br />
#
Example 5.11 If we identify the vertical edges <strong>of</strong> Ò!ß "Ó (to get a cylinder) <strong>and</strong> then identify its<br />
#<br />
circular ends with a half-twist (reversing orientation): Ð!ß CÑ µ Ð"ß CÑ <strong>and</strong> ÐBß !Ñ µ Ð" Bß "Ñ.<br />
We get<br />
a quotient space which is called a Klein bottle.<br />
It turns out that a Klein bottle cannot be embedded in ‘ the physical construction would require a<br />
$ <br />
“self-intersection” (additional points identified) which is not allowed. A pseudo-picture looks like<br />
In these pictures, the thin “neck” <strong>of</strong> the bottle actually intersects the main body in order to re-emerge<br />
“from the inside” in a “real” Klein bottle (in ‘ , say), the slef-intersection would not happen.<br />
%<br />
In fact, you can imagine the Klein bottle as a subset <strong>of</strong> ‘ using color as a 4 dimension. To each<br />
% >2<br />
point ÐBßCßDÑon the “bottle” pictured above, add a 4th coordinate to get ÐBßCßDßÑwhere BßCßD depend on time > <strong>and</strong> > is recorded as a 4 -coordinate. A point on the<br />
surface then has coordinates <strong>of</strong> form ÐBßCßDß>Ñ. At “a point” where we see a self-intersection in ‘ ß $<br />
there are really two different points (with different time coordinates ><br />
).<br />
276
Example 5.12<br />
" "<br />
1) In W, identify antipodal points that is, in vector notation, TµTfor each T−W.<br />
" "<br />
Convince yourself that the quotient Wε is W.<br />
# #<br />
2) Let H be the unit disk ÖT − ‘ À lTl Ÿ "× . Identify antipodal points on the boundary <strong>of</strong><br />
# " #<br />
H : that is, T µ T if T − W © H Þ The quotient HÎ µ is called the projective plane,<br />
a space<br />
which, like the Klein bottle, cannot be embedded in ‘ . $<br />
3) For any space \ , we can form the product \ ‚ Ò!ß "Ó <strong>and</strong> let ÐBß "Ñ µ ÐCß "Ñ for all<br />
Bß C − \ . The quotient Ð\ ‚ Ò!ß "ÓÑÎ µ is called the cone over \ . ( Why? )<br />
4) For any space \ , we can form the product \‚Ò"ß"Ó<strong>and</strong> define ÐBß"ѵÐCß"Ñ<strong>and</strong><br />
ÐBß"ѵÐCß"Ñfor all BßC−\ . The quotient Ð\‚Ò"ß"ÓÑε is called the suspension <strong>of</strong> \ .<br />
( Why? )<br />
There is one other very simple construction for combining topological spaces. It is <strong>of</strong>ten used in<br />
conjunction with quotients.<br />
Definition 5.13 For each α −E, let Ð\ αßgαÑ be a topological space, <strong>and</strong> assume that the sets \ α are<br />
pairwise disjoint. The topological sum (or “free sum”) <strong>of</strong> the \ α's is the space Ð <br />
α−E\<br />
αßgÑ<br />
where<br />
g œÖS© <br />
α−E \ α ÀS∩\ α is open in \ α for every α −E}.<br />
We denote the topological sum by<br />
\ . In the case lEl œ # , we use the simpler notation \ \ Þ<br />
α−E<br />
α<br />
277<br />
" #<br />
In \ , each \ is a clopen subspace. Any set open (or closed) in \ is open (or closed) in the sum.<br />
α−E<br />
α α α<br />
The topological sum \ can be pictured as a union <strong>of</strong> the disjoint pieces \ , all “far apart” from<br />
α−E<br />
α α<br />
each other so that there is no topological “interaction” between the pieces.<br />
Example 5.14 In , let E" <strong>and</strong> E # be open disks with radius " <strong>and</strong> centers at Ð!ß !Ñ <strong>and</strong> Ð$ß !ÑÞ<br />
Then toppological sum E" E# is the same as E" ∪E#<br />
with subspace topology<br />
Let F " be an open disk with radius " centered at Ð!ß!Ñ <strong>and</strong> let F#<br />
be a closed disk<br />
with radius " centered at Ð#ß !Ñ. Then F" F# is not the same as F" ∪ F#<br />
with the subspace topology<br />
(why? ÑÞ<br />
‘ #<br />
Exercise 5.15 Usually it is very easy to see whether properties <strong>of</strong> the \α's<br />
do or do not carry over to<br />
\ . For example, you should convince yourself that:<br />
α−E<br />
α<br />
1) If the \ 's are nonempty <strong>and</strong> separable, then \ is separable iff lEl Ÿ i .<br />
α α<br />
α−E<br />
2) If the \ 's are nonempty <strong>and</strong> second countable, then \ is second countable iff lEl Ÿ i .<br />
α α<br />
α−E<br />
3) A function 0À \ Ä^is continuous iff each 0l\ is continuous.<br />
α−E<br />
α α<br />
!<br />
!
4) If . Ÿ 1 is a metric for \ , then the topology on \ is the same as g where<br />
α α α<br />
α−E<br />
. ÐBß CÑ if Bß C − \<br />
.ÐBßCÑ œ <br />
" otherwise<br />
α α<br />
so \ is metrizable if all the \ 's are metrizable. You should be able to find similar statements for<br />
α−E<br />
α α<br />
other topological properties such as first countabe, second countable, Lindel<strong>of</strong>, ¨ compact, connected,<br />
path connected, completely metrizable, ... .<br />
Definition 5.16 Suppose \ <strong>and</strong> ] are disjoint topological spaces <strong>and</strong> 0 À E Ä ] , where E © \ . In<br />
the sum \] , define BµC iff Cœ0ÐBÑÞIf we form Ð\]Ñε , we say that we have attached<br />
\ to ] with 0 <strong>and</strong> write this space as \ ] .<br />
0<br />
For each : − 0ÒEÓß its equivalence class under µ is Ö:× ∪ 0 ÒÖ:×ÓÞ You may think <strong>of</strong> the function<br />
“attaching” the two spaces by repeatedly selecting a group <strong>of</strong> points in \ , identifying them together,<br />
<strong>and</strong> “sewing” them all onto a single point in ]just<br />
as you might run a needle <strong>and</strong> thread through<br />
several points in the fabric \ <strong>and</strong> then through a point in ] <strong>and</strong> pull everything tight.<br />
Example 5.17<br />
1) Consider disjoint cylinders \ <strong>and</strong> ] . Let E be the circle forming one end <strong>of</strong> \ <strong>and</strong> F the<br />
circle forming one end <strong>of</strong> ] . Let 0ÀEÄ0ÒEÓœF©] be a homeomorphism. Then 0 “sews<br />
together” \ <strong>and</strong> ] by identifying these two circles. The result is a new cylinder.<br />
2) Consider a sphere W © ‘ . Excise from the surface W two disjoint open disks H" <strong>and</strong> H#<br />
<strong>and</strong> let E" ∪E # be union <strong>of</strong> the two circles that bounded those disks. Let ] be a cylinder whose<br />
ends are bounded by the union <strong>of</strong> two circles, F" ∪F# . Let 0 be a homeomorphism carrying the<br />
points <strong>of</strong> E" <strong>and</strong> E" clockwise onto the points <strong>of</strong> F" <strong>and</strong> F#<br />
respectively..<br />
# Then W ] is a “sphere with a h<strong>and</strong>le.”<br />
0<br />
278<br />
"<br />
# $ #<br />
# $ #<br />
3) Consider a sphere W © ‘ . Excise from the surface W an open disk <strong>and</strong> let E be the<br />
#<br />
circular boundary <strong>of</strong> the hole in the surface W . Let Q be a Mobius ¨ strip <strong>and</strong> let F be the curve that<br />
"<br />
bounds it. Of course, FW . Let 0ÀEÄFbe a homeomorphism. The we can use 0 to join the<br />
spaces by “sewing” the edge <strong>of</strong> the Mobius ¨ strip to the edge <strong>of</strong> the hole in W . The result is a “<br />
# sphere<br />
with a crosscap.”<br />
There is a very nice theorem, which we will not prove here, which uses all these ideas. It is a<br />
“classification” theorem for certain surfaces.<br />
Definition 5.18 A Hausdorff space \ is a 2- manifold if each B − \ has an open neighborhood Y that<br />
is homeomorphic to ‘ . Thus, a 2-manifold looks “locally” just like the Euclidean plane. A is<br />
# surface<br />
a Hausdorff 2-manifold.<br />
Theorem 5.19 Let \ be a compact, connected surface. The \ is homeomorphic to a sphere W or to #<br />
W # with a finite number <strong>of</strong> h<strong>and</strong>les <strong>and</strong> crosscaps attached.<br />
You can read more about this theorem <strong>and</strong> its pro<strong>of</strong> in (William<br />
Algebraic Topology: An Introduction<br />
Massey).<br />
.
Exercises<br />
E22. a) Let µ be the equivalence relation on given by ÐBßCѵÐBßCÑ " " # # iff C" œC#<br />
. Prove that<br />
# ‘ ε is homeomorphic to ‘ .<br />
‘ #<br />
b) Find a counterexample to the following assertion: if µ is an equivalence relation on a<br />
space \ <strong>and</strong> each equivalence class is homeomorphic to the same space ] , then Ð\Î µ Ñ‚] is<br />
homeomorphic to \ .<br />
Why might someone ever wonder whether this assertion might be true? In part a), we have<br />
\œ ‘ ‚ ‘ , each equivalence class is homeomorphic to ‘ <strong>and</strong> ( \Î µ Ñ‚ ‘ ‘ ‚ ‘ \ . In this<br />
example, you “divide out” equivalence classes that all look like ‘ , then “multiply” by ‘ , <strong>and</strong> you're<br />
back where you started.<br />
# # # #<br />
c) Let 1À‘ Ä‘ be given by 1ÐBßCÑœB C. Then the quotient space ‘ Î1 is<br />
homeomorphic to what familiar space?<br />
d) On ‘ , define an equivalence relation ( iff . Prove<br />
# # #<br />
BßCѵÐBßCÑ " " # # B" C" œB# C#<br />
that ‘ / is homeomorphic to some familiar space.<br />
# µ<br />
E23. For Bß C − Ò!ß "Ó, define B µ C iff B C is rational. Prove that the corresponding quotient<br />
space Ò!ß "ÓÎ µ is trivial.<br />
E24. Prove that a 1-1 quotient map is a homeomorphism.<br />
E25. a) Let ] " œ Ò!ß "Ó with its usual topology <strong>and</strong> ] # œ Ò!ß "Ó with the discrete topology. Define<br />
1À] " ] # Ä] by letting 1 be “the identity map” on both ] " <strong>and</strong> ]Þ # Prove that 1is<br />
a quotient map<br />
that is neither open nor closed.<br />
# b) Let ‘ have the topology g for which a subbase consists <strong>of</strong> all the usual open sets together<br />
#<br />
with the singleton set ÖÐ!ß !Ñ× . Let ‘ have the usual topology <strong>and</strong> let 0 À ‘ Ä ‘ be the projection<br />
0ÐBßCÑœ B. Prove that 0 is a quotient map which is neither open nor closed.<br />
E26. State <strong>and</strong> prove a theorem <strong>of</strong> the form:<br />
“for two disjoint subsets E<strong>and</strong> F <strong>of</strong> ‘ , EFE∪Fiff<br />
... ”<br />
#<br />
#<br />
E27. Let ‘ have the topology g for which a subbasis consists <strong>of</strong> all the usual open sets together with<br />
#<br />
the singleton set ÖÐ!ß !Ñ× . Let ‘ have the usual topology <strong>and</strong> define 0À ‘ Ä ‘ by 0ÐBß CÑ œ B.<br />
Prove that 0 is a quotient map which is neither open nor closed.<br />
E28. Let ] œ Ð ‚<br />
Ð!ß "ÑÑ <strong>and</strong> let ] œ ] Ò!ß "Ñ . Prove ] is not homeomorphic to ] but<br />
" # " " #<br />
279
that each is a continuous one-to-one image <strong>of</strong> the other<br />
E29. Show that no continuous image <strong>of</strong> ‘ can be represented as a topological sum \] , where<br />
\ß] Á g. How can this be result be strengthened?<br />
E30. Suppose \= (s −W ) <strong>and</strong> ] > ( >−X ) are pairwise disjoint spaces. Prove that \ = ‚ ]<br />
> is<br />
homeomorphic to Ð\ ‚ ] ÑÞ<br />
=−W >−X<br />
=−Wß>−X<br />
= ><br />
E31. This problem outlines a pro<strong>of</strong> (due to Ira Rosenholtz) that every nonempty compact metric space<br />
\ is a continuous image <strong>of</strong> the Cantor set G . From Example 4.5, we know that \ is homeomorphic to<br />
i!<br />
a subspace <strong>of</strong> Ò!ß "Ó Þ<br />
a) Prove that the Cantor set G©‘ consists <strong>of</strong> all reals <strong>of</strong> the form 4<br />
where<br />
each + 4 œ ! or 2.<br />
280<br />
∞<br />
+<br />
$<br />
4œ!<br />
∞ ∞<br />
+ 4<br />
b) Prove that Ò!ß "Ó is a continuous image <strong>of</strong> G. Hint: Define 1Ð Ñ œ + 4<br />
Þ<br />
$ #<br />
4œ! 4œ!<br />
4<br />
4 4<br />
i!<br />
c) Prove that the cube Ò!ß "Ó is a continuous image <strong>of</strong> G.<br />
i! i! . i!<br />
Hint: By Corollary 2.21, G Ö!ß #× G Use 1 from part b) to define 0 À G Ä Ò!ß "Ó by i!<br />
0ÐB" ß B# ß ÞÞÞß ÞÞÞÑ œ Ð1ÐB" Ñß 1ÐB# Ñß ÞÞÞß ÞÞÞÑ<br />
d) Prove that a closed set O©Gis a continuous image <strong>of</strong> G.<br />
(Hint: G is homeomorphic to the “middle two-thirds" set G consisting <strong>of</strong> all reals <strong>of</strong> the form<br />
w<br />
∞<br />
, 4<br />
'<br />
4œ!<br />
w G w BßC−G<br />
BC<br />
#<br />
w ÂG Þ w O w G<br />
w w<br />
4 . has the property that if , then If is closed in , we can map<br />
G Ä O by sending each point B to the point in O nearest to B.)<br />
e) Prove that every nonempty compact metric space \ is a continuous image <strong>of</strong> G.<br />
.
<strong>Chapter</strong> VI Review<br />
Explain why each statement is true, or provide a counterexample.<br />
i i<br />
! !<br />
1. Ð!ß "Ñ is open in Ò!ß "Ó Þ<br />
2. Suppose J is a closed set in Ò!ß "Ó ‚ ‘ . Then 1 ÒJ Ó is a closed set in ‘ Þ<br />
3. is discrete<br />
i!<br />
#<br />
4. If G is the Cantor set, then there is a complete metric on G which produces the product topology.<br />
i!<br />
‘ ‘<br />
8 8<br />
5. Let ‘ have the box topology. A sequence Ð0 Ñ Ä 0 − ‘ iff Ð0 Ñ Ä 0 uniformly.<br />
8 Ò!ß"Ó<br />
8 8 8<br />
6. Let 0 À Ò!ß "Ó Ä Ò!ß "Ó be given by 0 ÐBÑ œ B . The sequence Ð0 Ñ has a limit in Ò!ß "Ó .<br />
‘ #<br />
7. Let 1−‘ be defined by 1ÐBÑœBfor all B−‘ . Give an example <strong>of</strong> a sequence Ð08Ñ <strong>of</strong> distinct<br />
functions in ‘ that converges to .<br />
‘ 1<br />
8. Let G be the Cantor set. Then G is homeomorphic to the topological sum G G.<br />
9. The projection maps 1 <strong>and</strong> 1 from ‘ ‘ separate points from closed sets.<br />
B C<br />
# Ä<br />
10. The letter N is a quotient <strong>of</strong> the letter M .<br />
11. Suppose µ is an equivalence relation on \ <strong>and</strong> that for B − \ß ÒBÓ represents its equivalence<br />
class. If B is a cut point <strong>of</strong> \ , then ÒBÓ is a cut point <strong>of</strong> the quotient space \Î µ .<br />
12. If 1À \ Ä ] is a quotient map <strong>and</strong> ] is compact T # , then ] is compact T # .<br />
13. Suppose E is infinite <strong>and</strong>, that in each space \ α ( α − E)<br />
there is a nonempty proper open subset<br />
Sα. Then S α is not a basic open set <strong>of</strong> the product \ œ \ α, <strong>and</strong>, in fact, Sα<br />
cannot even be<br />
open in the product.<br />
14. If \œ∞ E , where the E 's are disjoint clopen sets in \ , then \z∞ 8œ" 8 8 8œ" E8 ( œthe<br />
topological sum <strong>of</strong> the 's).<br />
E8<br />
15. Let \ œ Ö!ß "× <strong>and</strong> ] œ with their usual topologies. Then \ is homeomorphic to ]<br />
Þ<br />
8 8 8 8<br />
8œ" 8œ"<br />
281<br />
∞ ∞<br />
∞<br />
16. Let \ 8 œ Ö!ß "× <strong>and</strong> ] 8 œ with their usual topologies. Then <br />
8œ" \ 8 is homeomorphic to<br />
]Þ<br />
∞ 8œ" 8<br />
$ # # $<br />
17. Let E œ ÖÐBß Cß DÑ − ‘ À B C #CD D lsin ÐBCDÑl × , <strong>and</strong> let 0À ‘ Ä ‘ be given by<br />
0ÐBßCßDÑœ B# . Then 0 ÒEÓis<br />
open but not closed in ‘<br />
.
18. Suppose E 8 is a connected subset <strong>of</strong> \ 8Ð Á gÑ <strong>and</strong> that ∞ E 8 is dense in ∞<br />
8œ" 8œ" \ 8Þ<br />
Then each<br />
\8 is connected.<br />
19. In ‘ , every neighborhood <strong>of</strong> the function sin contains a step function (that is, a function with<br />
‘<br />
finite range).<br />
20. Let X be an uncountable set with the cocountable topology. Then { ÐBß BÑ À B − \× is a closed<br />
subset <strong>of</strong> the product \‚\Þ.<br />
"<br />
8<br />
21. Let EœÖ À8− ש ‘ , <strong>and</strong> let 7be any cardinal number. E is a closed set in ‘ Þ<br />
22. If Gis the Cantor set, then G‚G‚G‚G is homeomorphic to G‚G.<br />
" " 33. W ‚W is homeomorphic to the “infinity symbol”: ∞<br />
282<br />
5 5<br />
34. 31. Let T be the set <strong>of</strong> all real polynomials in one variable, with domain restricted to Ò!ß "Ó,<br />
for<br />
Ò!ß"Ó<br />
which ranÐT Ñ © Ò!ß "ÓÞ Then T is dense in Ò!ß "Ó Þ<br />
35. Every metric space is a quotient <strong>of</strong> a pseudometric space.<br />
37. A separable metric space with a basis <strong>of</strong> clopen sets is homeomorphic to a subspace <strong>of</strong> the Cantor<br />
set.<br />
38. Let { \ α À α − E } be a nonempty product space. Each \ α is a quotient <strong>of</strong> the product space<br />
Ö\ α À α − E×<br />
39. Suppose \ does not have the trivial topology. Then \ cannot be separable.<br />
#-<br />
40. Every countable space \ is a quotient <strong>of</strong> <br />
41. ‚ is homeomorphic to the sum <strong>of</strong> i! disjoint copies <strong>of</strong> .<br />
42. Suppose BœÐBÑ−int E, where E© \ . Then for every α, B −int 1 ÒEÓ.<br />
α α α α<br />
"<br />
α α α<br />
43. The unit circle, W , is homeomorphic to a product −E \ , where each \ © Ò!ß"Ó<br />
" (i.e., W can be “factored” into a product <strong>of</strong> subspaces <strong>of</strong> Ò!ß "Ó ).<br />
i!<br />
44. is homeomorphic to ‘ .