§ John Conway: The symmetries of things


Original way to classify wallpaper groups: think of geometric transforms that fix the pattern. Thurston's orbifold solution: think of quotients of R2\mathbb R^2 by groups --- this gives you an orbifold (orbit manifold).
Take a chair, surround it around by a sphere. The symmetries of a physical object fixes the center of gravity. So we pick the center of the sphere to be the center of gravity. The "celestial sphere" (the sphere around the chair) is a nice manifold (We only have the surface of the sphere). The vertical line that divides the chair also divides the sphere into two parts.

Conway came up with his notation for wallpaper groups/orbifolds. There are only four types of features.

An orbifol can be made to carry some amount of metrical information. The cone point only has 90 degrees, so it is in some sense, "a quarter of a point".

Let us try to work out the euler characteristic of the rectangular table orbifold by using VE+FV - E + F. The orbifold as one face. The wrong thing to say is that the orbifold has two edges and two vertices. It is untrue because the edge of the orbifold is only half an edge --- let's say that lines have thickness. In this case, we will have V=2/4V = 2/4, E=2/2E = 2/2, and F=1F = 1. The euler characteristic works out to be a half. This is appropriate, because the orbifold is a type of divided manifold.

Now, see that we started with positive euler characteristic (2), and we divide it by some n (the order of the group). So we end up with a positive euler characteric. By a sort of limiting argument, the euler characteristic of the wallpaper groups, which are infinite, is zero. However, see that we must get to the zero by starting with two dollars and buying things off the menu! If we try and figure out what all the possible ways are to start with 2 dollars and buy things till we are left with exactly 0 dollars, we get that there are 17 possible ways of buying things on the menu! Thus, this the reason for there being 17 wallpaper groups.

Because we can completely enumerate 2-manifolds, we can completely enumerate 2-orbifolds, which are essentially the same thing as symmetry groups. The real power is in the 3D case. We don't have a full classification of 3-manifolds. But we maybe able to go the other way. This is the metrization theorem.