Scribd
Upload
7 views3 pages

Section3 GameTrees v2 Solutions

Game Tree solution

Uploaded by

yakool008
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
7 views3 pages

Section3 GameTrees v2 Solutions

Game Tree solution

Uploaded by

yakool008
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Section 3 Solutions

1 Games
(a) Consider the zero-sum game tree shown below. Triangles that point up, such as at the top node (root),
represent choices for the maximizing player; triangles that point down represent choices for the minimizing
player. Assuming both players act optimally, fill in the minimax value of each node.

3 2 4

10 8 3 2 15 7 6 5 4

(b) Which nodes can be pruned from the game tree above through alpha-beta pruning? If no nodes can be
pruned, explain why not. Assume the search goes from left to right; when choosing which child to visit
first, choose the left-most unvisited child.

(c) Again, consider the same zero-sum game tree, except that now, instead of a minimizing player, we have a
chance node that will select one of the three values uniformly at random. Fill in the expectimax value of
each node. The game tree is redrawn below for your convenience.

1
8

7 8 5

10 8 3 2 15 7 6 5 4

(d) Which nodes can be pruned from the game tree above through alpha-beta pruning? If no nodes can be
pruned, explain why not. No nodes can be pruned. There will always be the possibility that an as-yet-
unvisited leaf of the current parent chance node will have a very high value, which increases the overall
average value for that chance node. For example, when we see that leaf 4 has a value of 2, which is much
less than the value of the left chance node, 7, at this point we cannot make any assumptions about how
the value of the middle chance node will ultimately be more or less in value than the left chance node. As
it turns out, the leaf 5 has a value of 15, which brings the expected value of the middle chance node to 8,
which is greater than the value of the left chance node. In the case where there is an upper bound to the
value of a leaf node, there is a possibility of pruning: suppose that an upper bound of +10 applies only to
the children of the rightmost chance node. In this case, after seeing that leaf 7 has a value of 6 and leaf 8
has a value of 5, the best possible value that the rightmost chance node can take on is 6+5+10
3 = 7, which
is less than 8, the value of the middle chance node. Therefore, it is possible to prune leaf 9 in this case.

2
2 Nonzero-sum Games
1. Let’s look at a non-zero-sum version of a game. In this formulation, player A’s utility will be represented
as the first of the two leaf numbers, and player B’s utility will be represented as the second of the two leaf
numbers. Fill in this non-zero game tree assuming each player is acting optimally.

2. Which nodes can be pruned from the game tree above through alpha-beta pruning? If no nodes can be
pruned, explain why not. No nodes can be pruned. Because this game is non-zero-sum, there can exist
a leaf node anywhere in the tree that is good for both player A and player B.

You might also like