Example Reduced Graph Powers#

Some understanding of reduced graph powers can be obtained by looking at examples.

For a simple graph $G=(V,E)$, we will sometimes write $|V(G)|$ and $|E(G)|$ for the number of vertices and edges of $G$. We assume the graph $G$ is connected, so the Betti number is given by $\beta(G)=|E(G)|-|V(G)|+1$. When the graph G is understood from context, we will often use $v(G)$ and $e(G)$, or simply $v$ and $e$, for the number of vertices and edges of G.

The number of vertices in the reduced graph power $G^{(k)}$ is

\[|V(G^{(k)})|=\binom{v+k-1}{k} \]

where $v=|V(G)|$. The number of edges in the reduced graph power $G^{(k)}$ is

(10)#\[|E(G^{(k)})|=e\binom{v+k-2}{k-1} \]

where $e=|E(G)|$. When the graph $G$ is understood, will sometimes abbreviate the number of vertices and edges in its reduced graph power by $v^{(k)} = |V(G^{(k)})|$ and $e^{(k)} = |E(G^{(k)})|$.

The path graph $P_n$#

A path graph is a graph whose $n$ vertices can be listed in the order $0, 1, \ldots , n-1$ such that the edges are $(i, i+1)$ for $i = 0, 1 \ldots , n-2$. For example,

for n in [2,4,7]:
    graphs.PathGraph(n).show(figsize=3,title='P%s:' %(n))
_images/77b2cf63e3ad66eb5eccd12e2075150cf944ae0e08e4e78f3657724044830dea.png _images/79107ac0121158b19e08d2e060582d29d2abd499cf4cae35652342714ec94bd8.png _images/d5e52c79003a61f6940e92458541877a0a80e758312e83af63946053ae28c03c.png

Evidently, $|V(P_n)|=n$, $|E(P_n)|=n-1$, and $\beta(P_n)=0$. The number of vertices of the reduced graph power $P_n^{(k)}$ is $|V(P_n^{(k)})|=\binom{n+k-1}{k}$. The number of edges is $|E(P_n^{(k)})|=(n-1)\binom{n+k-2}{k-1}$. Consequently, the Betti number of the path graph on $n$ vertices is

\[\begin{equation} \beta(P_n^{(k)}) = (n-1)\binom{n+k-2}{k-1}-\binom{n+k-1}{k}+1 \, . \end{equation}\]

Specializing to the case of a dimer ($k=2$), we have $|V(P_n^{(2)})|=\binom{n+1}{2}$, $|E(P_n^{(2)})|=n(n-1)$, and $\beta(P_n^{(2)}) = \binom{n-1}{2}$.

(11)#\[\beta(P_n^{(k)}) = (n-1)\binom{n+k-2}{k-1}-\binom{n+k-1}{k}+1 \, . \]

The cycle graph $C_n$#

The cycle graph $C_n$ is a graph whose $n$ vertices can be listed in the order $0, 1, \ldots , n$ such that the edges are $(i, i+1)$ for $i = 0, 1, \ldots , n-2$, and also $(0,n-1)$. For example, $C_5$ is

Hide code cell source

for n in [3,5,12]:
    graphs.CycleGraph(n).show(figsize=3,title='K%s:' %(n))
_images/f4ed016125b3af0f5122daf6a17eeb520e3a98fd81d559b33d80f3a5665a11fd.png _images/65c4aa62b360405a58cbae0bbf0dc6d935c212ee69fa8156f5b02d0aaa884fb3.png _images/1c51f79b144b2d533b623d9a5c3527b47f01bbf4e5d4b6768156c20456249b9b.png

For the cycle graph $|E(C_n)|=n$ and $\beta(C_n)=1$.

The number of vertices of the reduced graph power $C_n^{(k)}$ is $|V(C_n^{(k)})|=\binom{n+k-1}{k}$. The number of edges is $|E(C_n^{(k)})|=n\binom{n+k-2}{k-1}$. The Betti number is

\[\beta(C_n^{(k)}) = n\binom{n+k-2}{k-1}-\binom{n+k-1}{k}+1 \, . \]

Specializing to the case of a dimer ($k=2$), we have $v^{(2)}=\binom{n+1}{2}$, $e^{(2)}=n^2$, and

\[\beta(C_n^{(2)}) = \binom{n}{2}+1 = n(n-1)/2+1 \,.\]

The complete graph $K_n$#

The complete graph $K_n$ is a graph with $n$ vertices and $|E(K_n)|=\binom{n}{2}$ edges.
For example, $K_5$ is

Hide code cell source

for n in [2,4,7]:
    graphs.CompleteGraph(n).show(figsize=3,title='C%s:' %(n))
_images/5219182fe802d4459b40951cb0034cd73d8aaab14a448d0141657cf2e0d97820.png _images/7077748d77fb76144e3b736e3c7d8cd2fd44914979385f15e1a4f62d5c74fef7.png _images/443d11b0955afc1f9a823ff314ce2cd7dfa90d2ed4ce268c48cd10bfdb26b8cc.png

For the complete graph, $\beta(K_n)=\binom{n-1}{2}=(n-1)(n-2)/2$.

The number of vertices of the reduced graph power $K_n^{(k)}$ is $|V(K_n^{(k)})|=\binom{n+k-1}{k}$, the number of edges is $|E(K_n^{(k)})|=\binom{n}{2}\binom{n+k-2}{k-1}$, and the Betti number is

\[\beta(K_n^{(k)}) = \binom{n}{2}\binom{n+k-2}{k-1}-\binom{n+k-1}{k}+1 \, . \]

For a dimer ($k=2$), we have $v^{(2)}=\binom{n+1}{2}$, $e^{(2)}=\binom{n}{2} n = n^2(n-1)/2$, and

\[\beta(K_n^{(2)}) = (n+1)(n-1)(n-2)/2 \, . \]

The hypercube graph $Q_n$#

The hypercube graph $Q_n$ has $|V(Q_n)|=2^n$ vertices and $|E(Q_n)|=2^{n-1}n $ edges. For example, $Q_4$ is

Hide code cell source

for n in [2,4,5]:
    if n>3:
        graphs.CubeGraph(n).show(figsize=3,vertex_size=20,vertex_labels=false,title='Q%s:' %(n))
    else:
        graphs.CubeGraph(n).show(figsize=3,title='Q%s:' %(n))
_images/ca9dec13d1875d43a2eae336701817b650ac7b794b86c1c5251a6079433e990b.png _images/06d5b7fb87c59c2314cd3f909a159d99c4b64a583f975a03cd355e12ef925062.png _images/a4544f7ec70ca3a2e3bd9838bfd92048ccf703ce9539f3bdeb6e80a56718862b.png

Hide code cell source

n=3
graphs.CubeGraph(n).show3d(title='Q%s:' %(n))

For the hypercube graph, $\beta(Q_n)= 2^{n-1} (n-2)+1$.

The number of vertices of the reduced graph power $Q_n^{(k)}$ is $|V(Q_n^{(k)})|=\binom{2^n+k-1}{k}$. The number of edges is $|E(Q_n^{(k)})|=2^{n-1} n \binom{2^n+k-2}{k-1}$. The Betti number is

\[\beta(Q_n^{(k)}) =2^{n-1} n \binom{2^n+k-2}{k-1}-\binom{2^n+k-1}{k}+1 \, .\]

The reduced graph product of two hypercube graphs, has $|V(Q_n^{(2)})|=2^{n-1} (2^n+1)$ vertices, $|E(Q_n^{(2)})| = 2^{2n-1} n$ edges, and Betti number

\[\beta(Q_n^{(2)}) = 2^{2n-1} (n-1) - 2^{n-1}+1 \, . \]

In particular, $\beta(Q_2^{(2)}) = 7$ and $\beta(Q_3^{(2)}) = 61$.

The bipartite graph $K_{n,m}$#

The complete bipartite graph or biclique, denoted by $K_{n,m}$, is a bipartite graph where the vertices are partitioned into two sets. Every vertex of the first set, ${0, 1, \ldots, n-1}$, is connected to every vertex of the second set, ${n, n+1, \ldots, n+m-1}$.

Hide code cell source

graphs.CompleteBipartiteGraph(3,4).show(figsize=3,title='K%s,%s:' %(3,4))
graphs.CompleteBipartiteGraph(5,7).show(figsize=4,title='K%s,%s:' %(5,7))
_images/38ea852151dd499de62fb54ab0e15bbd44c94d5afbd804939449f2e8bb1422c7.png _images/c606886cbb9b9d5ade543af6b515860301e2de3851bb70aad76613b7cc913988.png

The complete bipartite graph has has $|V(K_{n,m})|=n+m$ vertices, $|E(K_{n,m})|=nm $ edges, and Betti number

\[\beta(K_{n,m})= nm-(n+m)+1 = (n-1)(m-1) \, . \]

The number of vertices of the reduced graph power $K_{n,m}^{(k)}$ is $|V(K_{n,m}^{(k)})|=\binom{n+m+k-1}{k}$. The number of edges is $|E(K_{n,m}^{(k)})|=nm\binom{n+m+k-2}{k-1}$. This gives the Betti number

\[\beta(K_{n,m}^{(k)}) = nm\binom{n+m+k-2}{k-1} - \binom{n+m+k-1}{k} +1 \, . \]

Specializing to the case of a dimer ($k=2$), we have $v^{(2)}= \binom{n+m+1}{2}$, $e^{(2)}=nm (n+m)$, and

\[\begin{split}\begin{eqnarray} \beta(K_{n,m}^{(2)}) & = &nm (n+m) - (n+m+1)(n+m)/2 +1 \\ & = & (n+m)[nm+(n-1)(m-1)]/2+1 \, . \end{eqnarray}\end{split}\]

References#