Lecture 2

Question 1

what is state space search?

Answer 1

use of graph to keep track of relationships between states. each node is a state and each arc (edge) is a operator to move from one state to another.

Question 2

what does open and closed list mean?

Answer 2

open - current fringe of search

closed - states already visited

Question 3

explain 4 steps to solve search problem

Answer 3

state representation (f w g c), initial state (E E E E), goal state (W W W W), operators (farmer goes alone, farmer takes wolf etc)

Question 4

explain breadth-first-search with example

Answer 4

searches state level by level, doesn’t store pathway to go to that state to store pathway at each state store a pair (current state, parent state)

Question 5

characteristics of breadth first search

Answer 5

nodes at level n are examined before preceding to level n+1, solutn guaranteed to be shortest, good when simple solutn is known to exist, if each node has large # of descendants BAD, B = avg # of children B^n states on n level, space complexity makes it hard for large problems

Question 6

explain depth first search

Answer 6

explores root then all children before moving on to sibling

Question 7

characteristics of depth first search

Answer 7

gets to deep search space quickly, if search path is long, doesn’t waste time searching “shallow” space, can get lost and miss short solutions, Bxn states to go n deep, using depth limit is good for making it a bit better.

Question 8

4 factors for evaluating search strategies

Answer 8

completeness (guaranteed to find solutn), time complexity how long to find solutn (measured in # of nodes), space complexity (usually maximum size node list becomes), optimality/admissibility (is solutn is found it is guaranteed to be optimal (minimum cost))

Question 9

evaluate bedth-first-search

Answer 9

completeness - yes, time complexity 1 + b + b^2 + b^3 + … + b^d = O(b^d), space complexity O(b^d), optimality and admissibility - yes.

Question 10

evaluate depth-first search and DFS with depth limit and iterative deepening

Answer 10

DFS/depthlimit/ID
completeness = no/yes if I > d/yes
time = O(b^n)/O(b^I)/O(b^d)
space = O(BM), O(BI), O(BD)
optimaland admissible = no/no/yes
b - branching factor, d - depth of solutn, m - maximum depth of tree, I - depth limit

Question 11

difference between heuristic search strategies and non-heuristic (depth and breadth)

Answer 11

heuristic - have some way of telling how close a state is to goal state.
non-heuristic - goal state larger = further away

Question 12

what heuristics state do we use if 2 have same value?

Answer 12

use one closer to root, depth can be added to heuristic value. f(n) = g(n) + h(n). g(n) = length of path from start to end, h(n) = heuristic estimate (length from n to goal)

Question 13

what is admissibility and informedness?

Answer 13

admissible - when a heuristic finds shortest path to goal its said to be admissible.
informedness - how much better one heuristic search is better than another

Question 14

what is f*(n)?

Answer 14

it is admissible and f(n) = g(n) + h(n). evaluates a function. g(n) - is cost of shortest path from start node to node n. h*(n) - actual cost of shortest path from node n to the goal

Question 15

what if h = 0, what type of search is made? (in heuristic search models)

Answer 15

we get a breadth first search, that is NOT very informed

Question 16

properties of A* search

Answer 16

complete(guaranteed to find solutn)
admissible(guaranteed to find solutn)
space : O(b^d) (worst case)
time: O(b^d) (worst case)

Question 17

explain best-first-search

Answer 17

still has open and closed but each state ha a heuristic # associated with it. arrange open from smallest to largest. close the state on the left of the open list until the leftmost state state in open list is goal state

Question 18

give 2 heuristic ideas and apply to 8-tile-puzzle

Answer 18

• count how many tiles are out of place in state.

- add distances of how out of place each tile is