7

I heard that two objects are homeomorphic if one could be deformed into the other by continuous transformation. For example in this link, it is shown

a sphere and a torus are not homeomorphic

"Proof" Removing a circle from a sphere always splits it into two parts -- not so for the torus.

However, I may imagine the following operations

enter image description here

to let the points around the inner circle of the continuous torus merge to a sphere. I see no reason merging is not continuous. Why not this transformation follow the definition of homeomorphic?

user26143
  • 475
  • 4
  • 10
  • 6
    "Removing a(n embedded) circle from a sphere always splits it into two parts" is nontrivial, FYI: it's the Jordan curve theorem. –  Jun 26 '14 at 06:01

3 Answers3

10

Because when you merge many points into a single one, you do not have a bijection; a homeomorphism is a continuous map with a continuous inverse, and a non-bijective map cannot have a (two-sided) inverse.

Besides, if this operation was a homeomorphism, then its inverse --tearing a circle to turn it into a torus would be a homeomorphism.

user99680
  • 6,708
6

Intuitively, homeomorphisms do not allow tearing or gluing. This is because they and their inverse need to be continuous :

  1. Gluing breaks injectivity, hence it breaks continuity.
  2. Tearing breaks continuity.
0

Even ignoring the invertibility issues, your map doesn't turn the torus into a sphere. It turns the torus into a sphere with a wall in it.

hunter
  • 29,847