We can cut the ordinary torus with two cuts both ends of each of which meet at a point. We can then flatten the torus onto a square in the plane. A neighborhood of the point where the cuts met is cut into four parts which end up in the corners of the square. We can extend the pattern in the plane from the square with more squares and define a map from the plane back into the torus which includes the inverse of out flattening map and maps each new square back onto the torus. Squares will meet four at a time at their corners. This inverse map restores the continuity that was lost when the torus was cut.
The TH torus can be cut and unrolled onto an octagonal
figure in the (hyperbolic) plane.
Each of the four cuts terminate at
both end at one point on the TH torus.
A neighborhood of this point is
thus split into eight regions, each destined to a different corner of the
octagon.
A similar extension can be made in the hyperbolic plane with new
eight sided figures patched together as were the squares above.
This is
hard to imagine but this is tiling {8, 8} as described here
and drawn here.
More commentary & Images
Now imagine a circle mapped into a TH torus. Unroll that TH torus and follow the image of the circle. It will, in general cross the cuts into adjacent octagonal patches and eventually end up at a similar point in a different patch. If we take a different circle image that is homotopic to the first and do a similar map we have an arc on the plane connecting two other points in the hyperbolic plane. The collection of all homotopic circle images thus constitutes a mapping of the hyperbolic plane onto itself. This mapping is a congruence if we have made the octagonal patches congruent.
There are two groups now, the homotopy group and the translation group in the hyperbolic plane. Are they related?
Relying on the simple torus case as an example, I think that the following is the case:
Consider the group of motions of the tiled plane that takes tiles to tiles. There is a subgroup that is homomorphic to the (circle) homotopy group of the corresponding torus, if any. There is also a subgroup which is homomorphic to the homotopies of ...
I suspend that thought while I follow another lead. How many ways can a simple torus T map into itself? I think that if I know the image of two generating circles on the torus then I know the T to T map within a homotopy.