COMP 221 Tutorial 3

1
COMP 221 Tutorial 3

• Solving Problems by Searching
• A Search Assignment 1

2
A Search

• Is basically Best First Search
• Greedy Search Uniform Search
• Is optimal with admissibility
• Is complete because of monotonically

3
15-puzzle
4
Design Hints

• v     Game States
•          A start state 8,7,6,5,4,3,2,1,0
•          Goal state 1,2,3,4,5,6,7,8,0

5
Design Hints

• nt moves GAME_SIZE MAX_N_MOVES
• / From 0 / 1, 3, -1, -1 ,
• / From 1 / 0, 2, 4, -1 ,
• / From 2 / 1, 5, -1, -1 ,
• / From 3 / 0, 4, 6, -1,
• / From 4 / 1, 3, 5, 7,
• / From 5 / 2, 4, 8, -1,
• / From 6 / 3, 7, -1, -1,
• / From 7 / 4, 6, 8, -1,
• / From 8 / 5, 7, -1, -1
•  

6
Design Hints

• Heuristic Function.
• The sum of distances that each tile is from where
  it should be.
  
• The number of displaced tile.

7
Design Hints

• Open and Close List.
• The first list is the OPEN list. This stores
  nodes, which we have generated by using the rules
  from an existing node, but we don't yet know
  where they lead.
• The second list is nodes that we have generated
  and we have also explored where they go.
  

8
Why Open and Close List

• Every time when were expanding a node A to get a
  new node B. we must test whether there is a node
  who has the same state as B stored in the Open or
  Close List. If so and the rating of node B is
  larger than the rating of that node. We can
  decide not to expand to the node B. On the other
  hand, if the rating of node B is smaller, we can
  just remove the node who has already been in the
  Open or Close List.

9
Priority Queue

• Why heap-based?
• . In the demo program weve written, a Priority
  Queue based on a heap is used. The goodness of a
  heap is that it can always give you a smallest
  (or largest ) element immediately.

10
The Close List

• The close list may have a large mount of nodes in
  it. So the best strategy to implement it is a
  harsh table, the reason is that whether a node is
  in the list or not can be known very fast with
  using a harsh table.