Created: 2026-03-06 07:59:30
Updated: 2026-03-06 08:18:27

2.1 Simplexes

:
r-simplex σr=<p0p1pr>\sigma _r=\left<p_{0}p_{1}\dots p_{r}\right> is defined as

σr={xRmx=i=0rcipi,ci0,ici=1}\begin{equation} \sigma _{r}=\left\{x\in \mathbb{R}^{m}\mid x = \sum_{i=0}^{r} c_{i} p_{i}, c_{i}\geq 0, \sum_{i}c_{i}=1\right\} \end{equation}

For example ,00-simplex is a point, 11-simplex <p0p1>\left<p_{0}p_{1}\right> is an edge, <p0p1p2>\left<p_{0}p_{1}p_{2}\right> is a triangle with its interior included, 33-simplex is a solid tetrahedron.
(c0,,cr)(c_{0},\dots,c_{r}) is called barycentric coordinate of xx. Since σr\sigma_{r} is bounded and closed subset of Rm\mathbb{R}^{m}, it is compact.

Let qrq\leq r a nonnegative integer. We choose q+1q+1 points out of p0,,prp_{0},\dots,p_{r}, define a qq-simplex σq=<pi0piq>\sigma_{q}=\left<p_{i_{0}}\dots p_{i_{q}}\right>, which is called qq-face of σr\sigma_{r}. Write σqσr\sigma_{q}\leq\sigma_{r} if σq\sigma_{q} is a face of σr\sigma_{r}, and moreover a proper face of σr\sigma_{r} if σqσr\sigma_{q}\neq\sigma_{r}.

The number of qq-faces of an rr-simplex is (r+1q+1){r+1 \choose q+1}

2.2 Simplicial complexes, polyhedra

Let KK be a set of finite number of simplexes in Rm\mathbb{R}^{m}. If these simplexes are nicely fitted together, KK is a simplicial complex. By nicely we mean:

Homology groups of simplicial complexes

3.1 Oriented Simplexes

Assign an orientation to an rr-simplex for r1r\geq 1. Use ()(\dots) for an oriented simplex. (p0p1)(p_{0}p_{1}) is a directed line segment from p0p_{0} to p1p_{1}. (p1p0)=(p0p1)(p_{1}p_{0})=-(p_{0}p_{1}). Similarly, for σ2=(p0p1p2)\sigma_{2}=(p_{0}p_{1}p_{2}) is a triangular rigion with a prescribed orientation along the edges. Any swap of the points produces a -1:

(pipjpk)=sgn(P)(p0p1p2)\begin{equation} (p_{i}p_{j}p_{k})=\text{sgn}(P)(p_{0}p_{1}p_{2}) \end{equation}

More generally, for rr-simplex, select a permutation PSr+1P\in S_{r+1}, we have that

(pP(0)pP(r))=sgn(P)(p0p1pr)\begin{equation} (p_{P(0)}\dots p_{P(r)}) = \text{sgn}(P)(p_{0}p_{1}\dots p_{r}) \end{equation}

3.2 Chain group, cycle group, boundary group

Let K={σα}K=\{\sigma_{\alpha}\} be an nn-dimensional simplical complex. We regard σα\sigma_{\alpha} as orinted, denote by σα\sigma_{\alpha} as before.

(Cr(K)C_{r}(K))
The rr-chain group Cr(K)C_{r}(K) is a free Abelian group generated by oriented rr-simplexes of KK. Cr(K)=0C_{r}(K)=0 if r>dimKr>\text{dim}K. An element of Cr(K)C_{r}(K) is called rr-chain.
Let IrI_{r} be number of rr-simplexes in K. cCr(K),\forall c\in C_{r}(K),

c=i=1Irciσr,i ciZc=\sum_{i=1}^{I_{r}}c_{i}\sigma_{r,i}\ c_{i}\in \mathbb{Z}

Addition of r-chains:

c+c=i(ci+ci)σr,ic+c'=\sum_{i}(c_{i}+c_{i}')\sigma_{r,i}

We see that Cr(K)C_{r}(K) is a free Abelian group of rank IrI_{r}, Cr(K)ZIrC_{r}(K)\cong \mathbb{Z}^{I_{r}}

boundary operator is a map r:Cr(K)Cr1(K)\partial_{r}:C_{r}(K)\to C_{r-1}(K) satisfying that for rr-simplex σr=(p0pr)\sigma_{r}=(p_{0}\dots p_{r}),

r(p0p1pr)=i=0r()i(p0p1p^ipr)\partial_{r}(p_{0}p_{1}\dots p_{r})= \sum_{i=0}^{r}(-)^{i}(p_{0}p_{1}\dots \hat{p}_{i}\dots p_{r})

where p^i\hat{p}_{i} is omitted. r\partial_{r} is linear.

  • 0p0=0\partial_{0}p_{0}=0
  • 1(p0p1)=p1p0\partial_{1}(p_{0}p_{1})=p_{1}-p_{0}
  • 2(p0p1p2)=(p1p2)(p0p2)+(p0p1)\partial_{2}(p_{0}p_{1}p_{2})=(p_{1}p_{2})-(p_{0}p_{2})+(p_{0}p_{1})

It is easy to see the boundary operator is a homomorphism. There is a sequence of free Abelian groups and homomorphisms

0iCn(K)nCn1(K)1C0(K)000 \overset{ i }{ \to }C_{n}(K)\overset{ \partial_{n} }{ \to }C_{n-1}(K)\to\dots\overset{ \partial_{1} }{ \to }C_{0}(K)\overset{ \partial_{0} }{ \to }0

where i:0Cn(K)i:0\to C_{n}(K) is an inclusion map, 0 is regarded as unit element of Cn(K)C_{n}(K)

(Zr(K)Z_{r}(K))
If cCr(K)c\in C_{r}(K), rc=0\partial_{r}c=0, then cc is an rr-cycle. Define Zr(K)Z_{r}(K) to be the set of all rr-cycles:

Zr(K)={cc is an r-cycle}=ker rZ_{r}(K)=\left\{c\mid c\text{ is an r-cycle}\right\}=\text{ker }\partial_{r}

(Br(K)B_{r}(K)):
If cCr(K)c\in C_{r}(K), dCr+1(K)\exists d\in C_{r+1}(K) such that c=r+1dc=\partial_{r+1}d, then cc is called rr-boundary. Br(K)B_{r}(K) is the rr-boundary group, Br(K)=Imr+1B_{r}(K)=\text{Im} \partial_{r+1}

\fcolorbox{pink}{}{Lemma 3.3} rr+1\partial_{r}\circ \partial_{r+1} is a zero map, rr+1c=0\partial_{r}\partial_{r+1}c=0 for any cCr+1(K)c\in C_{r+1}(K)
\fcolorbox{pink}{}{End Lemma}
You can prove the lemma by calculating rr+1c\partial_{r}\partial_{r+1}c directly for some given cc. This means that Br(K)Zr(K)B_{r}(K)\subset Z_{r}(K), or Imr+1kerr\mathrm{Im} \partial_{r+1} \subset \ker \partial_{r}.

3.3 Homology groups are topological invariants

We have defined 3 groups, how are they related to topological properties of K or to the topological space whose triangularization is K? Clearly Cr(K)C_{r}(K) is not conserved under homeomorphism: If K is a triangle, C1(K)Z3C_{1}(K)\cong \mathbb{Z}^{3}; for square C1(K)Z4C_{1}(K)\cong \mathbb{Z}^{4}.

Define Hr(K)=Zr(K)/Br(K)H_{r}(K)=Z_{r}(K) / B_{r}(K) is the rrth homology group. Use Hr(K;F)H_{r}(K;F) for definition of different field of coefficients.

Since Br(K)B_{r}(K) is a subgroup of Zr(K)Z_{r}(K),

Hr(K){[z]zZr(K)}H_{r}(K)\equiv \left\{[z]\mid z\in Z_{r}(K)\right\}

where [z][z] denotes equivalence classes(homology class). zzz\sim z' iff zzBr(K)z-z'\in B_{r}(K)

Homology groups are topological invariants: Let XX be homeomorphic to YY, let (K,f),(L,g)(K,f),(L,g) be triangulations of X,YX,Y respectively. Then

Hr(K)Hr(L),r=0,1,2,H_{r}(K)\cong H_{r}(L),r=0,1,2,\dots

In particular, if (K,f)\left(K,f\right) and (L,g)(L,g) are to particular triangulation of XX, then

Hr(K)Hr(L),r=0,1,2,H_{r}(K)\cong H_{r}(L),r=0,1,2,\dots

(We are using == as \cong)

  • Let K={p0}K=\left\{p_{0}\right\}, C0(K)={ip0iZ}ZC_{0}(K)=\{i p_{0}\mid i\in \mathbb{Z}\}\cong \mathbb{Z}, Z0(K)=C0(K)Z_{0}(K)=C_{0}(K),B0(K)={0}B_{0}(K)=\{0\}
  • K={p0,p1}K=\{p_{0},p_{1}\}, H0(K)=Z2H_{0}(K)=\mathbb{Z}^{2}, Hr(K)={0},r1H_{r}(K)=\{0\},r\geq 1
  • K={p0,p1,(p0p1)}K=\{p_{0},p_{1},(p_{0}p_{1})\}, Z0=C0Z2,Z1={0},B0={i(p1p0)iZ}Z1Z_{0}=C_{0}\cong \mathbb{Z}^{2},Z_{1}=\{0\},B_{0}=\left\{i(p_{1}-p_{0})\mid i\in \mathbb{Z}\right\}\cong \mathbb{Z}^{1}, B1={0}B_{1}=\{0\}, So H0Z,H1={0}H_{0}\cong \mathbb{Z},H_{1}=\{0\}
  • K={p0,p1,p2,(p0p1),(p1p2),(p2p0)}K=\{p_{0},p_{1},p_{2},(p_{0}p_{1}),(p_{1}p_{2}),(p_{2}p_{0})\}, Z0=Z4,Z1=Z1,B0=Z3,B1=Z0Z_{0}=\mathbb{Z}^{4},Z_{1}=\mathbb{Z}^{1},B_{0}=\mathbb{Z}^{3},B_{1}=\mathbb{Z}^{0}, H1=H0=ZH_{1}=H_{0}=\mathbb{Z} ( triangulation of S1S^{1})
  • K={p0,p1,p2,p3,(p0p1),(p1p2),(p2p3),(p3p0)}K=\{p_{0},p_{1},p_{2},p_{3},(p_{0}p_{1}),(p_{1}p_{2}),(p_{2}p_{3}),(p_{3}p_{0})\}, H1=H0=ZH_{1}=H_{0}=\mathbb{Z} (triangulation of S1S^{1})
  • K={p0,p1,p2,(p0p1),(p1p2),(p2p0),(p0p1p2)}K=\{p_{0},p_{1},p_{2},(p_{0}p_{1}),(p_{1}p_{2}),(p_{2}p_{0}),(p_{0}p_{1}p_{2})\}, First H0ZH_{0}\cong \mathbb{Z}. Z1={i[(p0p1)+(p1p2)+(p2p0)iZ]Z_{1}=\{i[(p_{0}p_{1})+(p_{1}p_{2})+(p_{2}p_{0})\mid i\in\mathbb{Z}], if b=2cb=\partial_{2}c, c=m(p0p1p2)c=m(p_{0}p_{1}p_{2}), we have b=m[(p0p1)+(p1p2)+(p2p0)]b=m[(p_{0}p_{1})+\left(p_{1}p_{2}\right)+(p_{2}p_{0})], so H1{0}H_{1}\cong \{0\}. Moreover H2{0}H_{2}\cong \{0\}
  • K={p0,p1,p2,p3,(p0p1),(p0p2),(p0p3),(p1p2),(p1p3),(p2p3),(p0p1p2),(p0p1p3),(p0p2p3),(p1p2p3)}K=\{p_{0},p_{1},p_{2},p_{3},(p_{0}p_{1}),(p_{0}p_{2}),(p_{0}p_{3}),(p_{1}p_{2}),(p_{1}p_{3}),(p_{2}p_{3}),(p_{0}p_{1}p_{2}),(p_{0}p_{1}p_{3}),(p_{0}p_{2}p_{3}),(p_{1}p_{2}p_{3})\} be a simplical complex of a tetrahedron( triangulation of S2S^{2}). H0=Z,H1={0},H2=ZH_{0}=\mathbb{Z},H_{1}=\{0\},H_{2}=\mathbb{Z}
    where we have used Hr:=Hr(K)H_{r}:=H_{r}(K), etc.

3.4 Computation of H0H_{0}

Let KK be a connected simplicial complex, then H0ZH_{0}\cong \mathbb{Z}

3.5 More homology computations


The triangulation of Mobius strip. We can verify that H1(K)ZH_{1}(K)\cong \mathbb{Z}, H2(K)={0}H_{2}(K)=\{0\}, H0(K)ZH_{0}(K)\cong\mathbb{Z} for being connected.

The triangulation of RP2\mathbb{R}P^{2}, defined as S2S^{2} where antipodal points identified.
By calculating 2z=0\partial_{2}z=0 with z=imiΔ2iz=\sum_{i}m_{i}\Delta_{2i}, where Δ2i\Delta_{2i} being all 2 simplexes, we find that Z2(K)Z_{2}(K) is trivial, H2(K)={0}H_{2}(K)=\{0\}.
Consider z=imΔ2,iz=\sum_{i}m\Delta_{2,i} 2z=2m(p3p5)+2m(p5p4)+2m(p4p3)\partial_{2}z=2m(p_{3}p_{5})+2m(p_{5}p_{4})+2m(p_{4}p_{3}), so if zZ2(K)z\in Z_{2}(K), m=0,m=0, so Z2(K)={0}Z_{2}(K)=\{0\}. Note that any cycle is homologous to nznz, with z=(p3p5)+(p5p4)+(p4p3)z=(p_{3}p_{5})+(p_{5}p_{4})+(p_{4}p_{3}). Furthermore, even multiple of zz is a boundary of 2-chain. Thus zz is a cycle and z+zz+z is homologous to 0. Hence we find that

H1(K)={[z][z]+[z][0]}Z2\begin{equation} H_{1}(K)=\left\{[z]\mid [z]+[z] \sim [0]\right\}\cong \mathbb{Z}_{2} \end{equation}

This shows that a homology group is not necessarily free Abelian but may have the full structure of a finitely generated Abelian group.

Consider T2T^{2} torus. r-th homology group is generated by boundaryless rr-chains that are not boundaries of r+1r+1 chains. For example, the surface of torus has no boundary but it is not boundary of some 3-chain. Thus H2(T2)H_{2}(T^{2}) is freely generated by 1 generator(the surface itself), H2(T2)ZH_{2}(T^{2})\cong \mathbb{Z}.

Look at H1(T2)H_{1}(T^{2}). Two loops a,ba,b have no boundaries but are not boundaries of some 2-chains. Take another loop aa', aa' is homologous to aa since aa=2sa'-a=\partial_{2}s for some ss. So H1(T2)H_{1}(T^{2}) is freely generated by a,ba,b, H1(T2)Z2H_{1}(T^{2})\cong \mathbb{Z}^{2}. H0(T2)ZH_{0}(T^{2})\cong \mathbb{Z} for T2T^{2} connected.

Then generalizing the analysis to Σg\Sigma_{g}. Since Σg\Sigma_{g} has no boundary and there are no 3-simplexes, the surface itself freely generates H2(Σg)ZH_{2}(\Sigma_{g})\cong \mathbb{Z}. There are 2g2g independent paths, so H1(Σg)Z2gH_{1}(\Sigma_{g})\cong \mathbb{Z}^{2g}. H0(Σg)=ZH_{0}(\Sigma_{g})=\mathbb{Z} since it is connected.


For Klein bottles, let z=imσ2,iZ2(K)z=\sum_{i}m\sigma_{2,i}\in Z_{2}(K), 2z=2ma\partial_{2}z=-2ma, a=(p0p1)+(p1p2)+(p2p0)a=(p_{0}p_{1})+(p_{1}p_{2})+(p_{2}p_{0}). so ker2=Z2=H2={0}ker \partial_{2}=Z_{2}=H_{2}=\{0\}; for H1H_{1}, we have that each 1-cycle is homologous to ia+jb,i,jZia+jb,i,j\in\mathbb{Z}. For a 2-chain to have a boundary consisting a,ba,b only, all 2-simplex in KK must be added with the same coefficient. for such a 2-chain z=imσ2,iz=\sum_{i}m\sigma_{2,i}, z=2ma\partial z=2ma. this shows that 2ma02ma\sim 0. Thus H1H_{1} is generated by two cycles a,ba,b such that a+a=0a+a=0, namely

H1(K)={i[a]+j[b]i,jZ}Z2Z\begin{equation} H_{1}(K)=\left\{i[a]+j[b]\mid i,j\in \mathbb{Z}\right\}\cong \mathbb{Z}_{2}\oplus \mathbb{Z} \end{equation}

3.4 General properties

1. Connectedness

Let KK be a disjoint union of NN connected components, K=K1K2KNK=K_{1}\cup K_{2}\dots \cup K_{N}, KiKj=K_{i}\cap K_{j}=\emptyset, then

Hr(K)=i=1NHr(Ki)\begin{equation} H_{r}(K) = \bigoplus_{i=1}^{N}H_{r}(K_{i}) \end{equation}

So is true for $C_{r}(K)$, $Z_{r}(K)$ and $B_{r}(K)$

One corollary is that for disjoint union of NN connected components H0(K)ZNH_{0}(K)\cong \mathbb{Z}^{N}

2. Structure

The most general form of Hr(K)H_{r}(K) is

Hr(K)ZZfZk1Zkp\begin{equation} H_{r}(K)\cong \underbrace{ \mathbb{Z}\oplus \dots \oplus \mathbb{Z} }_{ f } \oplus \mathbb{Z}_{k_{1}}\oplus \dots \oplus \mathbb{Z}_{k_{p}} \end{equation}

From our experience Hr(K)H_{r}(K) counts the number of (r+1)(r+1)-d holes in K|K|. The first part is free and the last p part are called torsion subgroup. Projective plane has H1(K)Z2H_{1}(K)\cong \mathbb{Z}_{2}, Klein bottle H1(K)ZZ2H_{1}(K)\cong \mathbb{Z}\oplus \mathbb{Z}_{2}. The torsion subgroup detects the twisting in polyhedron K\left| K \right| in a sense. We can now clarify why use Z\mathbb{Z} coefficients for homology groups instead of Z2\mathbb{Z}2 or R\mathbb{R}. Since Z2\mathbb{Z}_{2} has no non-trivial subgroups, and R\mathbb{R} will make us not see the torsion subgroup, since R/mR{0}\mathbb{R}/ m\mathbb{R}\cong \{0\} for any mZ{0}m\in\mathbb{Z}-\{0\}. If Hr(K;Z)H_{r}(K;\mathbb{Z}) is given by above, Hr(K;R)RfH_{r}(K;\mathbb{R})\cong \mathbb{R}^{f}.

3. Betti numbers, Euler-Poincare theorem

The rr-th Betti number br(K)b_{r}(K) is defined as

br(K)dimHr(K;R)\begin{equation} b_{r}(K) \equiv \text{dim} H_{r}(K;\mathbb{R}) \end{equation}

Or the free abelian part of Hr(K;Z)H_{r}(K;\mathbb{Z})

For example, K=T2K=T^{2}, bi(K)=1,2,1b_{i}(K)=1,2,1 for i=0,1,2i=0,1,2
bi(S2)=1,0,1b_{i}(S^{2})=1,0,1 for i=0,1,2i=0,1,2

Let KK be an nn-dimensional simplicial complex and IrI_{r} the number of r-simplexes in KK. Then

χ(K)r=0n()rIr=r=0nbr(K)\begin{equation} \chi(K)\equiv \sum_{r=0}^{n}(-)^{r}I_{r}= \sum_{r=0}^{n} b_{r}(K) \end{equation}

The first equality defines the Euler characteristic of general polyhedron K|K|. This is the generalization of Euler characteristic defind on surfaces.

r:Cr(K;R)Cr1(K;R)\partial_{r}:C_{r}(K;\mathbb{R})\to C_{r-1}(K;\mathbb{R}), where C1(K;R)={0}C_{-1}(K;\mathbb{R})=\{0\}.

Ir=dimCr(K;R)=dim(kerr)+dim(im r)=dimZr(K;R)+dimBr1(K;R)\begin{equation} \begin{aligned} I_{r} & =\text{dim}C_{r}(K;\mathbb{R})=\text{dim}(\ker \partial_{r}) + \text{dim} (\text{im }\partial_{r}) \\ & =\text{dim}Z_{r}(K;\mathbb{R})+ \text{dim} B_{r-1}(K;\mathbb{R}) \end{aligned} \end{equation}

We also have

br(K)=dimHr(K;R)=dim(Zr(K;R)/Br(K;R))=dimZr(K;R)dimBr(K;R)\begin{equation} \begin{aligned} b_{r}(K) & = \text{dim}H_{r}(K;\mathbb{R})=\text{dim}(Z_{r}(K;\mathbb{R})/ B_{r}(K; \mathbb{R})) \\ & = \text{dim}Z_{r}(K;\mathbb{R})- \text{dim} B_{r}(K;\mathbb{R}) \end{aligned} \end{equation}

So

χ=r=0n()rIr=r=0n()rdimZr(K;R)r=0n()rdimBr(K;R)=r=0n()rbr(K)\begin{equation} \begin{aligned} \chi & =\sum_{r=0}^{n}(-)^{r}I_{r}= \sum_{r=0}^{n} (-)^{r} \text{dim} Z_{r}(K;\mathbb{R}) - \sum_{r=0}^{n}(-)^{r}\text{dim} B_{r}(K;\mathbb{R}) \\ & = \sum_{r=0}^{n}(-)^{r}b_{r}(K) \end{aligned} \end{equation}

Since Betti numbers are topological invariants, χ(K)\chi(K) is conserved under homeomorphism. If f:KXf:|K|\to X, g:KXg:|K'| \to X are two triangulations, χ(K)=χ(K)\chi(K)=\chi(K')

Leave a Comment

captcha
Fontsize