def a_star_search(problem, heuristic):
"""
Accepts an AStarProblem; see problems/a_star_problem.py
heuristic is a function from states in the problem to real values such that heuristic(s) underestimates the cost to the nearest goal
Returns a path to goal, or None if a goal is not found.
"""
pq = PriorityQueue()
init_cost, init_node = problem.initial_state()
# put start state on the priority queue
pq.put((init_cost, init_node))
parent_map = {}
explored_nodes = set()
while not pq.empty():
# pop node off of the priority queue
cur_node_cost, cur_node = pq.get()
print "cur_node_cost, cur_node", cur_node_cost, cur_node
# check if this is the goal node
if problem.goal_test(cur_node):
return path_from_start(cur_node, parent_map)
# if not, acquire the list of actions for that state.
actions = problem.actions(cur_node) # [(2, node1), (4, node2)]
parent_heuristic_cost = heuristic(cur_node)
print "parent_heuristic_cost", parent_heuristic_cost
for action in actions:
transition_cost, child_node = action ####
print "transition_cost, child_node", transition_cost, child_node
# don't explore nodes more than once
if child_node in explored_nodes:
continue
child_heuristic_cost = heuristic(child_node)
# create a map from the child to the current node
parent_map[child_node] = cur_node
current_cost = cur_node_cost - parent_heuristic_cost + transition_cost + child_heuristic_cost
pq.put((current_cost, child_node))
# put the current node on the list of explored, once we have
# explored the available actions for this node
explored_nodes.add(cur_node)
# goal not found
return None
def uniform_cost_search(problem):
"""
Accepts a UCSProblem; see problems/ucs_problem.py.
Returns a path to goal, or None if a goal is not found.
"""
pq = PriorityQueue()
init_cost, init_node = problem.initial_state()
# put start state on the priority queue
pq.put((init_cost, init_node))
parent_map = {}
explored_nodes = set()
while not pq.empty():
# pop node off of the priority queue
cur_node_cost, cur_node = pq.get()
# check if this is the goal node
if problem.goal_test(cur_node):
return path_from_start(cur_node, parent_map)
print "tried", cur_node.name
# if not, acquire the list of actions for that state.
actions = problem.actions(cur_node) # [(2, node1), (4, node2)]
for action in actions:
transition_cost, child_node = action
# don't explore nodes more than once
if child_node in explored_nodes:
continue
# create a map from the child to the current node
parent_map[child_node] = cur_node
print "queueing", child_node.name
# place each child node on the priority queue
pq.put((cur_node_cost + transition_cost, child_node))
# put the current node on the list of explored, once we have
# explored the available actions for this node
explored_nodes.add(cur_node)
# goal not found
return None
