The Euler characteristic of a triangulated manifold in terms of even-dimensional faces
Alexey V. Gavrilov · 2025
Read The Euler characteristic of a triangulated manifold in terms of even-dimensional faces by Alexey V. Gavrilov (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2510-23623.txt.
Introduction
Alexey V. Gavrilov proves a formula for the Euler characteristic of a triangulated manifold whose dimension is even. That number is a classic one-number summary of a shape, often built by adding and subtracting counts of faces of every dimension. Gavrilov's formula uses only the counts of faces of even dimension. Its weights stay the same for every such manifold, no matter how large the even dimension is.
Catalog record
| CREATOR | Alexey V. Gavrilov |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 7 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2510-23623
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
The Euler characteristic of a triangulated manifold in terms of even-dimensional faces
Alexey V. Gavrilov
Abstract
For a triangulated manifold ℳ {\mathcal{M}} of even dimension d d we show that χ ( ℳ ) = ∑ m = 0 d / 2 c m f 2 m \chi({\mathcal{M}})=\sum_{m=0}^{d/2}c_{m}f_{2m} where χ ( ℳ ) \chi({\mathcal{M}}) is the Euler characteristic and f n f_{n} is the number of n n -dimensional faces in the triangulation. The coefficients c m c_{m} in this formula do not depend on d d .
1
Introduction
The Euler characteristic of a finite simplicial complex
{\mathcal{M}}
is given by the classical formula
\chi({\mathcal{M}})=\sum_{n=0}^{\infty}(-1)^{n}f_{n}.
On the right hand side are
f
- numbers of the complex, that is
f_{n}
denotes the number of faces of dimension
n
. We are interested in the case when
{\mathcal{M}}
is a triangulated closed manifold
where by a triangulation we mean a homeomorphism to a finite simplicial complex
(which is not necessarily a PL manifold). The Euler characteristic of a closed manifold of odd dimension is zero so we only consider manifolds of even dimension
d=\dim{\mathcal{M}}
.
In this situation the above is not the only possible formula for the Euler characteristic because there exist linear relations between
f
- numbers known as Dehn-Sommerville relations. (For example,
f_{d-1}=\frac{d+1}{2}f_{d}
as any
(d-1)
- face is shared between two
d
- faces.) In this note we prove a formula whose right hand side only includes even-dimensional
f
- numbers. Of course, in each particular dimension the problem is trivial as long as the relations are known: for example, if
{\mathcal{M}}
is a surface then
f_{1}=\frac{3}{2}f_{2}
hence
\chi({\mathcal{M}})=f_{0}-\frac{1}{2}f_{2}
. The same way, for a fourfold we have
f_{3}=\frac{5}{2}f_{4}
and (less obviously)
2f_{1}-3f_{2}+4f_{3}-5f_{4}=0
, so
\chi({\mathcal{M}})=f_{0}-\frac{1}{2}f_{2}+f_{1}
. The Dehn-Sommerville relations for all dimensions were described by Victor Klee
[
3
]
which reduces our task to a certain combinatorial problem.
Let
\beta_{n}\,\,(n\geq-1)
be a sequence defined by
\beta_{n-2}=\frac{4(2^{n}-1)B_{n}}{n}
where
B_{n}
are the Bernoulli numbers; note that
\beta_{n}=0
when
n
is odd with the exception of
\beta_{-1}=-2
.
A semi-Eulerian complex is a pure simplicial complex such that the link of any face of dimension
n
has the same Euler characteristic
1-(-1)^{d+n}
as the sphere
{\mathbb{S}}^{d-n-1}
[
3
,
5
,
1
]
(Klee called it an Eulerian manifold).
Theorem 1
.
If
{\mathcal{M}}
is a semi-Eulerian complex of even dimension
d
then
\chi({\mathcal{M}})=\sum_{n=0}^{d}\beta_{n}f_{n}.
(1)
It is known that if a simplicial complex is homeomorphic to a closed manifold then
the link of any face has the same homology groups as the sphere
of the corresponding dimension.
(Klee himself refers to Lefschetz’s “Topology”. Apparently, the first modern proof of this fact is given in
[
2
, Proposition 1.2]
.)
Consequently, it is always a semi-Eulerian complex. Today it is known that some topological manifolds admit no triangulations at all but we do not discuss this interesting topic here.
The formula itself is not particularly surprizing: the Dehn-Sommerville relations reduce the dimension of the space of
f
- vectors from
d+1
to
\frac{d}{2}+1
, so it is natural to expect
for even
f
- numbers to be sufficient. What is remarkable is that the coefficients of this formula do not depend on the dimension
d
. (The author does not know if this fact has some topological explanation or is merely a kind of coincidence.) It should be possible to generalize this result using more general Dehn-Sommerville relations (see
[
1
]
and references therein) but in the end the author decided to leave the formulation as it is. On the other hand, extending the formula to a manifold with a boundary is not difficult.
Corollary 1
.
If
{\mathcal{M}}
is an even-dimensional PL manifold with a boundary then
\chi({\mathcal{M}})=\sum_{n=0}^{d}\beta_{n}(f_{n}({\mathcal{M}})-\tfrac{1}{2}f_{n}(\partial{\mathcal{M}})).
Indeed, we have
2\chi({\mathcal{M}})=2\chi({\mathcal{M}})-\chi(\partial{\mathcal{M}})=\chi(\widetilde{{\mathcal{M}}})=\sum_{n=0}^{d}\beta_{n}f_{n}(\widetilde{{\mathcal{M}}})=\sum_{n=0}^{d}\beta_{n}(2f_{n}({\mathcal{M}})-f_{n}(\partial{\mathcal{M}}))
where
\widetilde{{\mathcal{M}}}
is the double of
{\mathcal{M}}
.
2
A proof of the formula
It is now customary to formulate Dehn-Sommerville relations in terms of
h
- numbers which are the coefficients in the expansion
F(x-1)=\sum_{n=0}^{d+1}h_{n}x^{d+1-n}
where
F(x)=x^{d+1}+\sum_{n=0}^{d}f_{n}x^{d-n}
.
We have
[
5
]
h_{d+1-n}-h_{n}=(-1)^{n}\binom{d+1}{n}(\chi({\mathcal{M}})-\chi({\mathbb{S}}^{d})),\,\,\,(0\leq n\leq d+1).
For our purposes it is convenient to reformulate this result as follows.
Lemma 1
.
If
{\mathcal{M}}
is a semi-Eulerian complex of dimension
d
then the polynomial
p(z)=\frac{1}{2}\chi({\mathcal{M}})+\sum_{n=0}^{d}f_{n}z^{n+1}
satisfies the identity
p(z)+(-1)^{d}p(-1-z)=0.
(2)
Remark
The equality (
2
) has already appeared in the literature
[
4
, Lemma 3]
but, for some reasons, it was formulated
in a way excluding the case when
d
is even and
\chi({\mathcal{M}})\neq 2
.
Proof.
The above relations can be written in the form
x^{d+1}F\left(\frac{1-x}{x}\right)-F(x-1)=(\chi({\mathcal{M}})-\chi({\mathbb{S}}^{d}))(x-1)^{d+1}.
By definition,
p(z)=\frac{1}{2}\chi({\mathcal{M}})-1+z^{d+1}F(z^{-1})
.
When
d
is even
\chi({\mathbb{S}}^{d})=2
, and taking
x=1+z^{-1}
we have
p(z)+p(-1-z)=\chi({\mathcal{M}})-2+(x-1)^{-d-1}F(x-1)-\left(\frac{x}{x-1}\right)^{d+1}F\left(\frac{1-x}{x}\right)=0.
For an odd
d
we have
\chi({\mathcal{M}})=\chi({\mathbb{S}}^{d})=0
and the same calculation gives
p(z)-p(-1-z)=0
.
∎
Lemma 2
.
Let
\theta:{\mathbb{R}}[z]\to{\mathbb{R}}
be a linear map defined by
\theta(z^{n})=\beta_{n-1}\,(n\geq 0).
If
p\in{\mathbb{R}}[z]
is a polynomial
satisfying the identity
p(z)+p(-1-z)=0
then
\theta(p)=0
.
If
q
is an even polynimial then
\theta(q)=-2q(0)
.
Moreover,
\theta
is the only linear functional on
{\mathbb{R}}[z]
with this properties.
Proof.
The equality
\theta(q)=-2q(0)
is obvious so we
only have to prove that
\theta(p)=0
. Without loss of generality we can assume that
p(z)=(1+2z)^{m}
where
m
is odd. Then
\theta(p)=\sum_{n=0}^{m}2^{n}\binom{m}{n}\theta(z^{n})=4\sum_{n=1}^{m+1}\binom{m}{n-1}\frac{2^{n-1}(2^{n}-1)B_{n}}{n}=2m!S_{m}
where
S_{m}=\sum_{n=1}^{m+1}\frac{(4^{n}-2^{n})B_{n}}{n!(m+1-n)!}
. The number
S_{m}
is equal to the coefficient at
z^{m+1}
in the Taylor
series of the product
\left(\sum_{n=1}^{\infty}\frac{(4^{n}-2^{n})B_{n}}{n!}z^{n}\right)\cdot e^{z}=-\frac{z}{\cosh z}.
But this is an odd function so the coefficient at an even power of
z
is zero.
Additionally, any polynomial
P\in{\mathbb{R}}[z]
can be presented in the form
P=p+q
so a functional with the properties required by the Lemma is unique.
∎
With this two lemmas the proof of (
1
) becomes obvious. If
d
is even then
p(z)+p(-1-z)=0
for the polynomial
defined in Lemma 1. Applying
\theta
to this polynomial we have
\theta(p)=-\chi({\mathcal{M}})+\sum_{n=0}^{d}\beta_{n}f_{n}=0.
Remark
If we did not want to compute the coefficients explicitely then the proof could be a few lines shorter. It is easy to show (e.g. by induction in the degree) that a decomposition
P=p+q
from Lemma 2 is unique. This fact by itself implies existence and uniqueness of
\theta