Hatcher dump
by yayups, Sep 27, 2020, 8:15 PM
Section 1.1
As a general convention,
will be the path variable and
will be the homotopy variable.
Solution
Solution
Solution
Solution
Solution
Solution
Solution
As a general convention,


1.1.1 wrote:
Show that the composition of paths satisfies the following cancellation property: If
and
, then
.



Trivial by group properties.
1.1.2 wrote:
Show that the change-of-basepoint homomorphism
depends only on the homotopy class of
.


Suppose we have a path
from
to
, so we have the map
This only depends on
, so we're done.



![\begin{align*} \beta_h:\pi_1(X,x_1)&\to \pi_1(X,x_0) \\ [f]&\mapsto [h\cdot f\cdot \overline{h}] = [h]\cdot[f]\cdot[h]^{-1}. \end{align*}](http://latex.artofproblemsolving.com/4/7/4/4747e0b0715c43482201733e2800fb679404f357.png)
![$[h]$](http://latex.artofproblemsolving.com/5/5/8/5584515cc49cb2b664ff811b0d429147eb37f5df.png)
1.1.3 wrote:
For a path connected space
, show that
is abelian iff all basepoint-change homomorphisms depend only on the endpoints of the path
.



As before, suppose we have a path
from
to
, so we have the map
Fix any paths
from
to
and
from
to
, and let
. Note that
and since we can attain any
this way, we're done.



![\begin{align*} \beta_h:\pi_1(X,x_1)&\to \pi_1(X,x_0) \\ [f]&\mapsto [h]\cdot[f]\cdot[h]^{-1}. \end{align*}](http://latex.artofproblemsolving.com/d/6/2/d62f57d28d8ca576663e71e544eeebdef8a8b997.png)







![\begin{align*} \beta_{\bar{h_1}} = \beta_{h_2} &\iff [h_1]^{-1}\cdot[f]\cdot[h_1] = [h_2]^{-1}\cdot[f]\cdot[h_2]\quad\forall f\in\pi_1(X,x_1) \\ &\iff [f]\cdot [g] = [g]\cdot [f], \end{align*}](http://latex.artofproblemsolving.com/f/5/6/f56227e4d15f571660dd985a73c49f38cb80e58f.png)

1.1.4 wrote:
A subspace
is said to be star-shaped if there is a point
such that, for each
, the line segment from
to
lies in
. Show that if a subspace
is locally star-shaped, in the sense that every point of
has a star-shaped neighborhood in
, then every path in
is homotopic to a piecewise linear path, that is, a path consisting of a finite number of straight line segments traversed at constant speed. Show this applies in particular when
is open or when
is a union of finitely many closed convex sets.












We first prove the problem in the case where
is itself star-shaped.
Lemma: Suppose
is star-shaped. Then, any path
is homotopic to a piecewise linear path with at most
pieces. Note that in this homotopy of paths, the endpoints remain fixed.
Proof: First, scale the path so it is from
. Also WLOG let
be the center of the star of
. Now, consider the homotopy
Note this stays in
since it is a scaling of points in
by a factor
. This takes our original
to a composition of two paths, one from some point to
, and one from
to some other point. Thus, it suffices to show that if
is a path from
to some point, then it is homotopic to a linear path.
Indeed, suppose we have
such that
. Consider the homotopy
(which is a homotopy for the same reason) which sends
to a linear path from
to
. This completes the proof of the lemma. 
The finish is a standard compactness argument which I will write out in detail for practice. Suppose we have our path
where
is locally star-shaped. For each
, let
denote the open neighborhood of
that is star-shaped. We see that
is an open set in
, so for each
, pick
an open interval that contains
and is contained within
.
We see that the
s cover
, so by compactness of
, there is some finite collection
of them such that the
cover
. Between any two consecutive
s, pick a
that is between them and contained in
. Then, including all the half integer
s and re-indexing, we have
such that
is fully contained in some
. Thus, within
, we use the lemma to fix the part of
in
to piecewise linear while fixing the endpoints, so we have a homotopy from
to a piecewise linear path of at most
pieces.

Lemma: Suppose



Proof: First, scale the path so it is from
![$[-1,1]\to U$](http://latex.artofproblemsolving.com/9/5/8/9583814894ff6a632453e34f20c50fa241f08f87.png)


![\[F(s,t) = [1-s(1-|t|)]\cdot g(t).\]](http://latex.artofproblemsolving.com/e/b/1/eb1d3c0917fe9722394ec9560c703352a44c0a5a.png)








Indeed, suppose we have
![\[g:I\to U\]](http://latex.artofproblemsolving.com/e/f/8/ef87fb8aed5c13ae5bd14e49c5c978135965d7a6.png)

![\[F(s,t) = t\cdot g(s+(1-s)t),\]](http://latex.artofproblemsolving.com/d/d/b/ddb2827345060efad0efd136b65d087f0b121f94.png)




The finish is a standard compactness argument which I will write out in detail for practice. Suppose we have our path
![\[f:I\to X,\]](http://latex.artofproblemsolving.com/2/2/7/227fbfbb6ab5e9d445902e771fa4f01e45f0be4e.png)










We see that the










![\[0=t_0<t_1<t_2<\cdots<t_{m'}=1,\]](http://latex.artofproblemsolving.com/3/6/7/367d9537badc5068d533c5cf884fe853df18638e.png)
![$[t_i,t_{i+1}]$](http://latex.artofproblemsolving.com/9/a/b/9aba5be221b8c6726708a61dabf49e970e6978f1.png)



![$[t_i,t_{i+1}]$](http://latex.artofproblemsolving.com/9/a/b/9aba5be221b8c6726708a61dabf49e970e6978f1.png)


1.1.5 wrote:
Show that for a space
, the following three conditions are equivalent:
is simply connected iff all maps
are homotopic.

- Every map
is homotopic to a constant map, with image a point.
- Every map
extends to a map
.
for all
.


We may assume that
is path connected, since any map
lands in only one component of
, as
has a single component. We'll view
as the unit circle in
. Fix
. For all
, let
be a path from
to
.
For any map
, define
in the following way:
where
is the loop given by
.
For any loop
, define
by
. Since
, we see that this map is continuous (one can prove this formally by noting that
is continuous where
is the quotient map).
We'll first show (a)
(c). Suppose we have
. Let
be a homotopy such that
and
(constant). Now, define
by
We see that this a homotopy from
to
, which is the identity loop. Thus, every element of
is homotopic to the identity, so
.
We'll now show (c)
(a). Suppose we have
, which we want to show is homotopic to a constant map. Indeed, consider the homotopy
with
and
. We then have the homotopy
given by
This is a homotopy from
to a constant loop at
, so
is homotopic to a constant loop.
Finally, we'll show (a)
(b). The key is that
is homeomorphic to
where
is identified to a point, and furthermore, this homeomorphism identically maps
to the boundary of
. Now, we have a map
such that
for a fixed map
if and only if we have a continuous map
such that
and
is constant. This is exactly equivalent to saying that
is homotopic to a constant loop, so we're done! This shows all the desired equivalences.











For any map


![\[[\tilde{f}] = [\eta_{f(1)}]\cdot[g]\cdot[\eta_{f(1)}],\]](http://latex.artofproblemsolving.com/7/1/7/717be21cf5f988b83b241743129940e14fbf9c9d.png)


For any loop






We'll first show (a)






![\[\tilde{F}(t,s) = \widetilde{F(\bullet,s)}(t).\]](http://latex.artofproblemsolving.com/0/e/2/0e29b63964a7144648ad77fe157c1a67333abdee.png)

![$[\eta_{x_1}]\cdot[\eta_{x_1}]^{-1}$](http://latex.artofproblemsolving.com/6/9/3/693f7ba6b948d0364cb25b3dfe0a1281e761ba22.png)


We'll now show (c)






![\[\tilde{F}(\omega,s) = \widetilde{F(\bullet,s)}(\omega)\]](http://latex.artofproblemsolving.com/1/3/0/13096561d44b117daf8dfc9b50c5f717437dfc2a.png)



Finally, we'll show (a)






![\[F:S^1\times I/\sim\,\,\to X\]](http://latex.artofproblemsolving.com/0/1/d/01dc8001c9fa32e7d8304db406fadd08fc82ca28.png)


![\[\tilde{F}:S^1\times I\to X\]](http://latex.artofproblemsolving.com/7/0/9/709f0ba9ad799e29fd4679f282e97bace67fbd27.png)



1.1.6 wrote:
We can regard
as the set of basepoint-preserving homotopy classes of maps
. Let
be the set of homotopy classes of maps
with no conditions on basepoints. Thus there is a natural map
obtained by ignoring basepoints. Show that
is onto if
is path connected, and that
iff
and
are conjugate in
. Hence
induces a one-to-one correspondence between
and the set of conjugacy classes in
, when
is path connected.


![$[S^1,X]$](http://latex.artofproblemsolving.com/c/1/1/c11e4e8e083dbe4a02d442d3e41f63e74e5e3fdf.png)

![$\Phi:\pi_1(X,x_0)\to [S^1,X]$](http://latex.artofproblemsolving.com/5/0/0/500056a0ab5daca828eda0bfbe701170c74eb16a.png)


![$\Phi([f])=\Phi([g])$](http://latex.artofproblemsolving.com/2/a/3/2a3aef364a8d2ce1e27513617b70bab92925a3ff.png)
![$[f]$](http://latex.artofproblemsolving.com/f/4/4/f44bfb37b68791e7968284f5b7237bb59749656a.png)
![$[g]$](http://latex.artofproblemsolving.com/f/e/6/fe6f53448f516922ffc8265065784c6aa7db8525.png)


![$[S^1,X]$](http://latex.artofproblemsolving.com/c/1/1/c11e4e8e083dbe4a02d442d3e41f63e74e5e3fdf.png)


Keep the
and
notation from the previous solution. For any
, it is easy to see that
is homotopic to
where
is a path from
to
, so
is onto.
For one direction, it suffices to show that
for all
, and this is indeed true as we can compose
with a rotation to get
.
For the other direction, suppose that
. Let the image of the basepoint
under the homotopy
be
. Let the path from
to
along
(so the image of
for some
) be
, and the rest of
be
, so
. However, if
is a path from
to
, then the homotopy tells us that
It is not hard to check that this rearranges to
and since
, we see that
and
are conjugate in
, as desired.



![$\Phi([a]\cdot[\tilde{f}]\cdot[a]^{-1})$](http://latex.artofproblemsolving.com/b/a/7/ba7d0501fc86af993ad0d629057ad5f5227d7c96.png)





For one direction, it suffices to show that
![$\Phi([f]\cdot [g]) = \Phi([g]\cdot [f])$](http://latex.artofproblemsolving.com/4/c/4/4c44b623825d730a4bab5fbe921071c778bbedb6.png)

![$\Phi([f]\cdot[g])$](http://latex.artofproblemsolving.com/f/6/b/f6bf1cbb3bcff3bb71af57deb4bf9134dac67a28.png)
![$\Phi([g]\cdot[f])$](http://latex.artofproblemsolving.com/6/2/9/62996a727e65b2c6fb3b0ad43c9ee7b28e25c2e1.png)
For the other direction, suppose that
![$\Phi([f])=\Phi([g])$](http://latex.artofproblemsolving.com/2/a/3/2a3aef364a8d2ce1e27513617b70bab92925a3ff.png)

![$\Phi([f])\to\Phi([g])$](http://latex.artofproblemsolving.com/9/c/f/9cfcc27492cf550020f60f14cab830f1686d8a59.png)




![$g([0,t])$](http://latex.artofproblemsolving.com/3/5/5/3556ba065dabc6d1e47a98716ae36d835b51e932.png)




![$[g]=[g_2]\cdot[g_1]$](http://latex.artofproblemsolving.com/e/3/c/e3cba19e3466b3db94314bbc887d0cadc8f2bc4b.png)



![\[[a]\cdot[g_1]\cdot[g_2]\cdot[a]^{-1} = [f].\]](http://latex.artofproblemsolving.com/8/1/8/8181ee80c51213d2453b40acd190a6030dfcb519.png)
![\[[g] = ([a]\cdot[g_1])^{-1}\cdot[f]\cdot([a]\cdot[g_1]),\]](http://latex.artofproblemsolving.com/a/8/8/a88c060d9a54353a1a5f1c73214892d4f6f326a2.png)
![$[a]\cdot[g_1]\in\pi_1(X,x_0)$](http://latex.artofproblemsolving.com/e/9/8/e985e079ccbbe1ad18f2ea53a0ba81c28a7a0c65.png)
![$[g]$](http://latex.artofproblemsolving.com/f/e/6/fe6f53448f516922ffc8265065784c6aa7db8525.png)
![$[f]$](http://latex.artofproblemsolving.com/f/4/4/f44bfb37b68791e7968284f5b7237bb59749656a.png)

1.1.7 wrote:
Define
by
, so
restricts to the identity on the two boundary circles of
. Show that
is homotopic to the identity by a homotopy that is stationary on one of the boundary circles, but not by any homotopy that is stationary on both.





The homotopy
works just fine.
Suppose we had a homotopy
(here
is the homotopy variable) that sent
to the identity, while being stationary on both boundary circles. The key is to now consider what this homotopy does to the induced map
on the quotient
given by gluing the two boundary circles together. Since
is stationary on the boundary circles, it is a continuous map from
.
Now, for fixed
, and fixed
,
is a loop in
, and by varying
, we get a homotopy from this loop to the identity loop. If we project this loop and the homotopy onto the
that was not formed during the quotient process, we see that we have a have a homotopy between a loop with winding number
and the identity loop, which is a contradiction by
. Thus,
cannot exist.

Suppose we had a homotopy







Now, for fixed









This post has been edited 2 times. Last edited by yayups, Sep 27, 2020, 8:18 PM