I've noticed that in the tutorial the author mentioned that there exist an linear time solution, but I don't know what the solution is ...
# | User | Rating |
---|---|---|
1 | tourist | 3856 |
2 | jiangly | 3747 |
3 | orzdevinwang | 3706 |
4 | jqdai0815 | 3682 |
5 | ksun48 | 3591 |
6 | gamegame | 3477 |
7 | Benq | 3468 |
8 | Radewoosh | 3462 |
9 | ecnerwala | 3451 |
10 | heuristica | 3431 |
# | User | Contrib. |
---|---|---|
1 | cry | 167 |
2 | -is-this-fft- | 162 |
3 | Dominater069 | 160 |
4 | Um_nik | 158 |
5 | atcoder_official | 157 |
6 | Qingyu | 155 |
7 | djm03178 | 151 |
7 | adamant | 151 |
9 | luogu_official | 150 |
10 | awoo | 147 |
I've noticed that in the tutorial the author mentioned that there exist an linear time solution, but I don't know what the solution is ...
Name |
---|
Find all the biconnected components, let there be $$$C$$$ such components.
Let $$$G$$$ be a graph with $$$C$$$ nodes. For some color $$$c$$$ let the indices of edges with such color be $$$c_1, c_2, \ldots, c_x$$$, let's add edge $$$c_i-c_{i-1}$$$ for every $$$x \geq i>1$$$ to $$$G$$$. If we do this for all colors we will have some connected components in $$$G$$$. It can be proven that for every connected component in $$$G$$$: we can have all the colors (in this component) in an optimal tree iff this component has a cycle or contains a vertex which corresponds to a bridge BCC in the original graph.
All of this can be done in linear time.