Вы находитесь на странице: 1из 6

AI

Hw-1

Bharath Doddadunnasandra
Narayana Reddy
Graph search

Order of Expansion for


1. a. .BFS : [a,b,c,d,e,g]
b. UCS : [a,b,c,e,g] path cost : 5
c. DFS : [a,c,e,g]
d. Greedy Search: [a,b,c,d,g]
e. A* : [a,b,c,d,e,g]
2. MiniMax problem
a.

b. The Minimax says the maximizer should take column 3 to get its maximum value
3. (Russel and Norvig, Exercise 3.3) Two friends Problem
a. State: PosiXon of friends a and b, P(pa,pb) and the acXon taken by each: A (a,b)
Ini(al state: Player a and Player b at two dierent ciXes starXng simultaneously
Ac(on: at each turn both the players move to a neighboring city
Terminal test: whether pa=pb
Path Cost: maximum Xme taken by either friend to travel from Time from city
max(d(pa1,pa2) , d(pb1,pb2) )
b. A heurisXc is admissible when it does not overesXmate the actual path cost.
D(i,j) and D(i,j)/2 are admissible as the path cost is not overesXmated.
whereas 2D(i,j) is not admissible as it overesXmates the cost of reaching the goal.
c. Consider a map of two posiXons connected by a single link. When the two friends start
over, they move to the other city in the next turn. So at each turn they end staying at
opposite ciXes. This will become an innite loop with no soluXons.
d. Adding a loop to one of the two states in the two-node single link graph, when both the
friends start, they reach the same city in just one step. But, one of the friends will visit
the same city twice.

4. A: when playing opXmally MIN always
chooses the minimum values.
B: when playing subopXmally MIN
chooses a value greater than the
minimum values available.
C: MAX will always be the root node,
which always chooses the maximum of
the available values.
Hence, from B it is inferred that, MAX
always gets to choose a greater value
while MIN plays subopXmally, compared
to when MIN plays opXmally.
b) As it can be seen from the trees in the
adjacent gure, that MAX sXll does beaer
when it plays subopXmally against a
subopXmal MIN.




5. a) adjacent diagram
b) whenever a player is given a chance to choose
between a win and a ?, the player always goes
for the win rather than staying in the ? or
limbo state
c) As MiniMax uses depth rst search, it keeps
repeaXng the already visited states
and might enter a loop, that would fail the
algorithm. As seen in the gure of
two-players game, for a game with four
posiXons, we have two repeated states in
the tree.
If we apply the answer b we can avoid the
repeated states and hence avoid the innite loop.
d)Whenever n is even, A always is ahead of B and
has a chance of going forward or backward of B.
This will prevent B from Winning in a even n
board. Hence A always has a winning chance.
Whereas in odd n board game, B will have the
advantage which A had in even n board game.

Вам также может понравиться