Scribd
Upload
28 views2 pages

Lecture 18

This document discusses discrete manifolds in the context of differential geometry, defining concepts such as discrete m-manifolds, discrete spheres, and their Euler characteristics. It presents theorems related to the properties of these manifolds, including Euler's Gem and the General Gauss-Bonnet theorem, along with proofs and examples. The document also touches on the construction of manifolds from functions and the definition of curvature in discrete settings.

Uploaded by

roger.chemoul86
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
28 views2 pages

Lecture 18

This document discusses discrete manifolds in the context of differential geometry, defining concepts such as discrete m-manifolds, discrete spheres, and their Euler characteristics. It presents theorems related to the properties of these manifolds, including Euler's Gem and the General Gauss-Bonnet theorem, along with proofs and examples. The document also touches on the construction of manifolds from functions and the definition of curvature in discrete settings.

Uploaded by

roger.chemoul86
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

DIFFERENTIAL GEOMETRY

MATH 136

Unit 18: Discrete Manifolds

18.1. A discrete m-manifold is a finite graph G = (V, E) for which every unit sphere
S(v) is a discrete (m−1)-sphere. A discrete m-sphere is a discrete m-manifold which
has the property that removing a point renders it contractible. Inductively, a graph is
called contractible, if both S(v) and S\v are contractible for some v ∈ V . The 1-point
space 1 is contractible. The empty graph is the (−1)-sphere. Let Fk denote the set
of Kk+1 subgraphs (k-simplices) and fk = |FP k |. We have F0 = V, F1 = E. The Euler
characteristic of M is defined as χ(M ) = m k
k=0 (−1) fk = f0 − f1 + f2 − f3 + · · · +
(−1)m fm . This definition of Ludwig Schläfli generalizes χ(M ) = f0 −f1 +f2 = V −E+F
for 2-manifolds.

Figure 1. This 2-manifold M of genus g = 2 has χ(M ) = 2 − 2g = −2.

18.2. A graph without edges is a 0-manifold. A 0-manifold is a 0-sphere, if V =


2, E = 0 (removing a vertex produces K1 which is contractible by definition). Every
connected 1-manifold is a 1-sphere, a circular graph Cn with n ≥ 4. Every finite
2-manifold is either a 2-sphere S 2 or a connected sum of tori or projective planes:
M = S 2 , M = T2 # · · · #T2 or M = P2 # · · · #P2 . A 2-sphere can be characterized
as 2-manifold of Euler characteristic 2. The 16 cell and the 600 cells are examples
of 3-spheres. The join of two 1-spheres is a 3-sphere. The join of a k-sphere with a
m-sphere is a (k+m+1)-sphere. The join of G with the 0-sphere is called suspension.
18.3. Euler’s formula χ(M ) = V − E + F = 2 for 2-spheres generalizes to higher
dimension. The 0-sphere has χ(M ) = V = 2, every 1-sphere has χ(M ) = V − E = 0.
Every 2-sphere has χ(M ) = V − E + F = 2. This pattern continues:
Theorem 1 (Euler’s Gem). If M is a m-sphere, then χ(M ) = 1 + (−1)m .
Differential Geometry

Proof. Use induction with respect to dimension m. For m = 0, we have χ(M ) = 2.


The induction assumption is that all (m − 1)-spheres S satisfy χ(S) = 1 + (−1)m−1 .
Pick a vertex v. As the unit sphere S(v) is a (m − 1)-sphere and S(v) = B(v) ∩ G \ v,
where both the unit ball B(v) and G \ v are contractible with Euler characteristic 1, we
have, using the induction assumption, χ(M ) = χ(G \ v) + χ(B(v)) − χ(G \ v ∩ B(v)) =
2 − (1 − (−1)m−1 ) = 1 + (−1)m . □
18.4. In the continuum, manifolds can be constructed as level surfaces of functions
like x2 + y 2 + z 2 = 1. We can do that also in the discrete. Take an arbitrary function
on vertices V which takes values in Zk = {0, . . . , k}. It defines a new graph Mf , where
the vertices are the set of complete subgraphs on which f attains all k values. Connect
two of these points by an edge, if one is contained in the other. The new graph Mf
is a sub-graph of the Barycentric refinement of M . Here is the analog of what we
have seen classically for functions on manifolds. It surprises that singularities like in
the Viviani curve (HW 1) do never occur in the discrete:
Theorem 2 (Level Sets). If M is a m-manifold and f : M → Zk is an arbitrary
function, then either Mf is empty or then Mf is a (m − k)-manifold.
Proof. Let x be a n-simplex on which f takes all values. This means f (x) = Zk . The
graph S − (x) = {y ⊂ x, y ̸= x} is a (n − 1)-sphere in the Barycentric refinement of M .
The simplices in S − (x) on which f still reaches Zk is by induction a (n−1−k)-manifold
and since we are in a simplex, it has to be a (n − 1 − k)-sphere. Every unit sphere
S(x) in the Barycentric refinement is a (m − 1)-sphere as it is the join of S − (x) with
S + (x) = {y, x ⊂ y, x ̸= y}. (The join of two spheres is always a sphere.) The sphere
Sf+ (x) in Mf is the same than S + (x) in M because every simplex z in M containing x
automatically has the property that f (z) = Zk . So, the unit sphere S(x) in Mf is the
join of a (n − k − 1)-sphere and the (m − n − 1)-sphere and so a (m − k − 1)-sphere.
Having shown that every unit sphere in Mf is a (m − k − 1)-sphere, we see that Mf is
a (m − k)-manifold. □
18.5. What about differential geometry? No problem. Define curvature as
m
X (−1)k fk−1 (S(v)) f0 (S(v)) f1 (S(v)))
K(v) = =1− − + ... .
k=0
k+1 2 3
In the case of a 2-manifold this boils down to 1 − f0 (S(v))/2 + f1 (S(v))/3 = 1 −
d(v)/6, where d(v) is the vertex degree. For odd-dimensional manifolds, the curvature
is constant zero. You experiment with this in Homework 10.
P
Theorem 3 (General Gauss-Bonnet). v∈V K(v) = χ(M )
Proof. The proof is the same as in the 2-dimensional case. Again look at the energies
ω(x) = (−1)dim(x) attached to each simplex
P x in the graph (complete subgraph with
dim(x) + 1 vertices). Then χ(M ) = x ω(x). Now distribute all these energies of a
k-simplex x equally to the k + 1 vertices contained in x. As there are fk−1 simplices in
k
S(v) which correspond do simplices containing v, this adds (−1) fk+1
k−1 (S(v))
to each vertex
v. Now just collect all up at a vertex v to get the curvature K(v). The transactions of
energies preserved the total energy = Euler characteristic. □

Oliver Knill, [email protected], Math 136, Fall, 2024

You might also like