- Tournament graph: either beats , so we have or we have beats so we have the edges for every
[image at 49:00 from video math for comp sci lecture 10 ]
- Example:
A -> B -> D -> E -> C. Wait,C -> A. It's unclear how to talk about the best player!
§ directed Hamiltonian path
A directed walk that visits every vertex exactly once.
§ Theorem: every tournament graph contains a directed hamiltonian path
Induction on the number of nodes. When we start thinking of the problem, we have both nodes and edges as parameters. But edges are directly related to nodes, so it makes sense we induct on nodes.
§ Induction
If we are done. In the inductive step, assume it holds for . For , let's take out one node and see what happens. In the remaining graph, we still have a tournament graph on nodes. By the induction hypothesis we have a directed hamiltonian path . We want to create a bigger path that includes .
§ Case 1
If then we will get a path .
§ Case 2
If , then it is harder! Now what do we do? Ideally we want to plug somewhere in the sequence . Let's consider the smallest such that . We know that as we are in case 2.
v1 -> ...v[i-1] -> v[i] -> ... vn ^ vIf we have we are done because we get to insert into the path as . Because is the smallest index that beats, we must that beats --- otherwise is no longer the smallest index!
§ Chicken tournament
Either a chicken pecks a chicken then or the other direction, . We say that virtually pecks if there's a patch of pecking for to peck .
The chicken king is the chicken that virtually pecks all other chickens.
We can have multiple king chickens. We want to find at least one chicken king. We may want to show that the vertex with the most number of outgoing edges is going to be a king.
§ Theorem: chicken with highest out degree is the king
Proof by contradiction: assume has highest out degree and is not the king. So there is some vertex such that . Hence we have that . In the other case, we have that .