Definition (Minimum Spanning Tree)
Given graph with weight function , we want the spanning tree where
is as small as possible.
Lemma (Constructing MST by Lightest Edge)
If is a finite weighted connected graph, and is the minimum weight edge of , then has a MST containing edge .
In fact, if is the unique lightest edge, then all MSTs contain .
Proof (1): Let be a weighted, connected graph and let be an edge with the minimum weight in . Let be any MST of .
If , the lemma holds and we are done.
If , then we can find a new spanning tree where and . If was minimal, then so is .
Then if we add to , then as were already connected (by definition of a Tree), this creates a cycle , and so we no longer have a tree. Cycle must have some other edge where
and
This is a spanning tree. We know is connected. If we remove from then we still have a connected graph. In particular, this shows is a connected graph. But,
then we must have no cycles by Lemma (Tree Edge Count). Therefore, is a tree.
Now we can work with . Next,
But then
But then was already an MST, then must also be an MST.
If was the unique lightest edge, then for any other edge . Then strictly,
indeed, all MSTs must contain .
Graph Contraction
When we construct an edge to our MST, we can treat as the same vertex, combining the previous vertices of the edge. We can repeat our construction by selecting the lightest remaining edge in lemma.
Where remaining means an edge that does not form a cycle.