I'm reading a text book "Artificial Intelligence - A Modern Approach (3rd Edition)".
In page 95, the book mentions that "A* has the following properties: the tree-search version of A* is optimal if h(n) is admissible, while the graph-search version is optimal if h(n) is consistent."
But there's no clear proof why admissible h(n) only guarantees optimality in the tree search version of A*.
Can someone explain why the graph version of A* with an admissible heuristic doesn't guarantee optimality?
