7.图着色
五色定理的失败告诉我们,很多我们看上去觉得简单的证明,其实我们完全不会。 ——题记
点着色一些性质:
- \(\chi(G)=v(G)\iff G为完全图\)
证:\(G=Kv,\chi(G)=v\);
\(G≠Kv,chi(G)<v;\)
\(G\neq K_{v+1},chi(G)<(v+1)\) - \(\chi(G)\leq\Delta(G)+1\)
- \(\chi(C_v)=3(v为偶数);2(v=2k)\)
- \(\chi(H)\leq\chi(G)\)
- \(Brooks定理:G_v不是完全图和奇圈,\chi(G)\leq\Delta(G)\)
推论: Peterson图的\(\chi(G)=3\)
[========]
边着色:
- 穷举法可知Peterson图着色数=4,目前不确定求图边着色数的算法
- 连通图G不是奇圈,则存在2-边着色使得任意满足deg(v)>1的v,都关联这两种颜色
PS:c(v)=顶点v关联的边中出现的颜色数,最佳k-边着色make \(\sum c(v)min\),which maximum=\(\sum deg(v)\)
Ps2:最佳k-边着色未必是k-边着色的 - 最佳k染色C,G=(E1,E2,...,Ek),在v0的边中,i appears twice,j appears none,\(G{[E_iUE_j]}\)含v0分支为奇圈
p1: \(E_iUE_j\),c(v')>=c(v),c'(v0)=c(v0)+1 - 二分图不含奇圈,consider best\(\Delta(G)\)-着色,
PS:X'(G)为最少着色数
if vi has two edge with same color,must has a j appear none,wrong;verse same;
so best \(\Delta\)-paint make every vi has all the color,so X'(G)<=\(\Delta(G)\);
it's apparent that \(\Delta(G)\)<=X'(G),so them equal; - as to simple graph,\(\Delta(G)\)<=X'(G);
if(X'(G)>\(\Delta(G)\)+1) ,we lost another chance,
- G是二分图,\(\chi(G)=\Delta(G)\)
- Vizing's Law:G是简单图,则X'(G)=\(\Delta(G)/\Delta(G)+1\)
证:
结合最佳\(\Delta+1\)最大性、最佳匹配中i2次,j0次,G[EiUEj]必有奇圈,
逐步反证搞到一点邻点的性质,努力地找出一个更大\(\Delta+1\)-着色
[========]
平面图着色:
面着色可转化为点着色,与对偶图着色数相同
四色定理的证明:
- Kample的证明:
- Appel和Haken的思路
- 不可避免集
- 可约性
五色定理
五色定理的失败告诉我们,很多我们看上去觉得简单的证明,其实我们完全不会
G是连通的,但G的点、边导出子图,最大可匹配轨迹、圈的交并未必是连通的,
颜色多项式——正常顶点着色方法数
- pk(G)=k^v,当且仅当G是v阶0图
- pk(G)=k(k-1)...(k-v+1),当且仅当G=Kv
- p4>0,当且仅当四色定理成立
- G1,...,Gn为连通片,\(p_k(G)=\prod_{i=1}^{n} p_i(G)\)
\(p_k(G)=p_k(G-e)-p_k(G·e)\)
根据该公式,将G的着色方案数转变成Kv或K0的着色数
愿偿少年泪,犹趁未老时!
本文来自博客园,作者:clfire,转载请注明原文链接:https://www.cnblogs.com/sky1water/p/16749989.html