9

It is easy to compute the fundamental group of a solid torus. You easily get $\mathbb{Z}$ just because the torus is the cartesian product of a circumference and a closed disk.

The next step is imagine two solid torus one inside of the other one. Now consider the topological space that is the difference of the bigger one with the one that is inside. You get a torus with a hole that is a $3$-manifold with borders. Now you identify the interior border with the exterior border (in an obvious way). I worked a bit and I saw that quotient is homeomorphic to $S^1 \times S^1 \times S^1$, so the fundamental group is $\mathbb{Z}^3$.

What I am unable to compute is the fundamental group in this case that is similar to the one before. You have again bigger torus and what it is inside is homeomorphic to a torus but doing two loops inside the bigger torus (not just only one like the previous case). You do again the difference of sets and get a torus with a two loops hole inside. Now you identify borders again and you obtain a $3$-manifold with no borders. However I see myself unable to compute the fundamental group this time...

Here goes an image of the third picture (the torus with the hole looping twice). (remember that later you identify borders). The blue tube represents the hole inside the torus.

https://www.dropbox.com/s/x2jm7gg9f9hhkw2/20150416_003300.jpg?dl=0

DCao
  • 653
  • can you explain the 3rd pic properly?? it is not clear to me – Anubhav Mukherjee Apr 15 '15 at 22:25
  • 1
    Imagine that the solid torus is an apple and a worm starts moving and eating the apple (of course leaving a hole).

    In the previous case the worm just was doing a loop inside the torus and reaching the same place where it began.

    In the third case the worm does a loop again the torus but does not finish at the same place, so it does another loop and now it finishes at the place it began. I'll post a picture if I'm able to.

    – DCao Apr 15 '15 at 22:28
  • well, to get the idea of the fundamental group, before gluing the inside to the outside, the solid torus with a torus-shaped hole winding twice inside it, deformation retracts to the torus ($S^1 \times S^1$) with a Möbius strip inside it, and the fundamental group becomes (via Van-Kampen) $$ \frac{(\mathbb Z a + \mathbb Z b) \times \mathbb Z c}{2(a + b) \sim 2c} $$ – John C Apr 15 '15 at 22:40
  • I thought about using Van Kampen, but it is not clear for me the open sets you choose.

    I also can ''imagine'' the retraction you tell me about, but I do not see why it is usefull if you identify borders later. I mean, the fundamental group of $X$ is the same that on a retract of $X$, but in our case we do a quotient after so I do not see the relationship. Maybe there is some staff I do not know like ''a quotient of a retract of $X$ is a retract of the quotient of $X$''? (I do not mean the last sentence to be true, just an example).

    – DCao Apr 15 '15 at 23:01
  • 2
    Is it clear that the space doesn't depend on the torus homeomorphism used to identify the boundaries? The torus mapping class group is $SL(2,\mathbb Z)$ and there's no cannonical choice for the identification anymore – spitespike Apr 15 '15 at 23:31
  • I assume that you just identify it in the ''easiest way'' the circumferences of the interior border glue with the exterior ones and they do it in a proportional way. When you have glued half of the interior, you have glued half of the exterior, etc. – DCao Apr 15 '15 at 23:40
  • @Marcuswood I don't understand your identifications. The longitudal circles of the torus are identified with the longitudal circles of the inner torus, that is okay, but what are you doing with the great circles? – Balarka Sen Apr 16 '15 at 06:06
  • 1
    Every point on the boundary of the torus is on a longitudinal circle. It is like that any point of a sphere is on a meridian and on a parallel. – DCao Apr 16 '15 at 06:25

2 Answers2

1

This answer will be a bit incomplete, but I think you can work out the full details from it (feedback is welcome).

Consider you have the solid torus with the torus shaped hole winding ally around it twice, and call this space $X$, then you want to calculate $\pi_1(Z)$, where $\frac{X}{Y_1\sim_f Y_2}$, $Y_1,Y_2$ are the connected components of $\partial X$ and $f:Y_1\to Y_2$ is a homeomorphism. Now, with the notation out of the way, cover $Z$ with two open sets (Van Kampen in the making) $U,V$, where $U = \mathrm{Int} X$, and $V$ is the quotient (by the relation $\sim_f$) of an open neighborhood of $\partial X$; further, since $U\cap V$ is not connected, we are gonna have to use the groupoid version of the Van Kampen theorem.

So choose two points in $U \cap V$, which I will call $0$ and $1$ because I don't want to use more letters, and we calculate the fundamental groupoids of $U,V$ and $U\cap V$, with respect to the set of basepoints $\{0,1\}$. Each of these groupoids has two elements, and all we need to determine is the automorphism group of these two elements (corresponding to $0$ and $1$) and the morphisms between them (if they exist).

  • For $\pi_1(U\cap V,\{0,1\})$, each element only has the fundamental group of the torus (\mathbb Z \oplus \mathbb Z) as it's automorphism group, and since $0$ and $1$ lie in different components of $U\cap V$ there are no morphisms (which correspond to paths) between them.

  • For $\pi_1(V,\{0,1\})$, you can see that since $V$ deformation-retracts to the boundary of $X$ under the identification $\sim_f$. Further, you get that, once again, under this identification the boundary of $X$ is homeomorphic to the torus, and so $\hom(0,0) = \hom(1,1) = \pi_1(V) = \mathbb Z \times \mathbb Z$. Also, by choosing the base-points $0$ and $1$ such that $1 \in r^{-1}(f\circ r (0))$, one can see that all elements of $\hom(0,1)$ are of the form $\circ \gamma_{0,1} \circ k_0$, where $\gamma_{0,1}$ is a path from $0$ to $1$ and $k_0$ is an element of $\hom(0,0)$ (note that $\gamma_{0,1}\circ k_0 = h(k_0)\circ\gamma_{0,1}$ where $h$ is the "identity" $\hom(0,0) = \hom(1,1)$.

  • Finally (we still have to see what are the maps, but, we're almost there) for $\pi_1(U,\{0,1\})$, we can see that $U$ deformation-retracts to a torus with a Möbius band inside it, and so has its fundamental group given by $\pi_1(X) = \frac{\mathbb Za + \mathbb Zb + \mathbb Zc}{2a + b \sim 2c}$ (there is a mistake in my comment before, sorry about that). And once again, $\hom(0,0) = \hom(1,1) = \pi_1(X)$ and $\hom(0,1)$ has elements of the form $\delta_{0,1}\circ k_0$.

Now, all that is left is to assemble the pieces by choosing a point, (either $0$ or $1$, I will choose $0$).

  • The map $\pi_1(U\cap V,0)\to \pi_1(V,0)$ is the identitiy

  • The map $\pi_1(U\cap V,0)\to \pi_1(U,0)$ takes one generator to $a$ and the other generator to $b$, as per the representation given above (note that I am assuming $0$ is the point "close to the outer boundary of $X$")

  • Finally, we need to consider the group elements that arise from composing elemnts of $\hom(0,1)$ with elements of $\hom(1,0)$, these will have the form $\delta^{-1}\circ\gamma$ which is not homotopy equivalent to any of the other generators.

So, we get $\pi_1(Z) = \frac{\mathbb Za + \mathbb Zb + \mathbb Zc + \mathbb Z(\delta^{-1}\gamma)}{2a + b \sim 2c}$

John C
  • 821
0

Here's a computation using a CW-structure on $Z$. I'm not sure if it agrees or disagrres with John C. Perhaps somebody can see the error or isomorphism?

I computed the presentation $\pi_1(Z)=\left \langle a,b,c \ \middle | \ [a,b], a[a,c], b[b,c] \right \rangle$ using the homotopic CW complex: http://tinyurl.com/mk94d82

First deformation retract a Mobius band bounded by the inner torus so that it becomes a circle (twice the inner torus longitude). Pick a base point on the torus. The 1-skeleton is the meridian $a$, the longitude $b$, and a curve $c$ between the inner and outer torus. Three two cells are attached: the torus giving $[a,b]=1$, a meridian cross-section giving $a[a,c]=1$, and a longitudinal cross-section giving $b[b,c]=1$. Cutting along 2-cells in $Z$ we obtain a 3-ball so that is the total cell-decomposition.

spitespike
  • 1,168