def compareHeuristics():
timeList = []
while (len(timeList) < 10):
time.time()
signal.signal(signal.SIGALRM, handler)
signal.alarm(1)
map = None
map = generateMap(int(100),0.15)
# Setting up map
map[0][0]=0
map[len(map)-1][len(map)-1] = 0
start_time = time.time()
time.time()
signal.signal(signal.SIGALRM, handler)
signal.alarm(1)
try:
path = astar(map,1)
except IOError:
path = None
if path is None:
print("path does not exist")
else:
print(path)
print("--- %s seconds ---" % (time.time() - start_time))
timeList.append((time.time() - start_time))
print(timeList)
print("Avg time: "+str(sum(timeList)/len(timeList)))
def get_distance_from_to_key(grid, from_key, to_key):
''' BFS to find distance from one key to another '''
global DIRECTIONS
# Locations
current_location = from_key[0]
end_location = to_key[0]
def is_goal_fn(node):
return node == end_location
def heuristic(node: Node):
return node.manhattan_distance(end_location)
def cost(a, b):
return 1
visited = set()
def get_neighbors(node):
neighbors = []
for d in DIRECTIONS:
new = node + d
if new not in visited and grid[new] != '#':
neighbors.append(new)
visited.add(new)
return neighbors
def get_key_fn(node: Node):
return node.to_tuple()
return (end_location,
astar(current_location, is_goal_fn, heuristic, cost, get_neighbors,
get_key_fn))
def calculateAvgPathLength():
i = 100
i2 = i
solvable = 0
unsolvable = 0
pathLen = []
while (i > 0):
map = None
map = generateMap(int(100),0.3)
printMap(map)
map[0][0] = 0
map[99][99] = 0
time.time()
signal.signal(signal.SIGALRM, handler)
signal.alarm(1)
try:
path = astar(map,int(1))
except IOError:
time.time()
unsolvable = unsolvable + 1
continue
print(path)
if path is None:
unsolvable = unsolvable + 1
else:
pathLen.append(len(path))
solvable = solvable + 1
i=i-1
print("Generated: "+ str(i2) +"maps and # solvable: " + str(solvable))
print(pathLen)
print("Avg Path Length: "+ str(sum(pathLen)/len(pathLen)))
def genetic_algorithm(algo):
dim = 15
prob = .4
count = 0
# Population will hold our valid hard mazes
population = []
# While loop halts when we have 100 hard mazes in the population
while count != 100:
# create a random map
map = generateMap(dim, prob)
if (algo == "dfs"):
result = dfs(map)
else:
temp = copy.deepcopy(map)
temp[0][0] = 0
temp[len(temp) - 1][len(temp) - 1] = 0
start_time = time.time()
signal.signal(signal.SIGALRM, handler)
signal.alarm(1)
try:
result = astar(temp, 0)
except IOError:
result = None
if (result == None):
continue
else:
# if w ehave a valid solvable maze then add to population
result.append(map)
population.append(result)
count = count + 1
#print("Population count "+ str(count))
count = 0
# while loop halts once we have costruced 50 valid children
while count != 50:
# sort the population by our heuristic and reverse so the highest score is first
population = sorted(population, key=itemgetter(1), reverse=True)
# dad is the best hard maze we have
dad = population[0]
# mom is the second best hard maze we have
mom = population[1]
# set up child as 2D array which is how enivroments are used in backend
kid = [[0 for x in range(dim)] for y in range(dim)]
map1 = dad[2]
map2 = mom[2]
# Thsi snippet of code perform recombination
# we take the first half of dad and second half of mom and populate the child accrodingly
go = 0
stop = (dim * dim) / 2
for x in range(dim):
for y in range(dim):
if (go < stop):
kid[x][y] = map1[x][y]
else:
kid[x][y] = map2[x][y]
go = go + 1
# We run whichever algo the user had chosen on the newly created kid maze
if algo == "dfs":
final = dfs(kid)
else:
kid[0][0] = 0
kid[len(kid) - 1][len(kid) - 1] = 0
start_time = time.time()
signal.signal(signal.SIGALRM, handler)
signal.alarm(1)
try:
final = astar(kid, 0)
except IOError:
final = None
# if child is not solvable we get rid of dad and try again with a differnt set of mazes
if (final == None):
population.pop(0)
# if child is valid we append the kid and pop the worst in popualtion
# Later the kid will be placed in the right position during out sorting phase
else:
population.pop()
population.append(final)
count = count + 1
# we return the most fit maze or hardest maze in our population as our answer
return population[0]
def main():
inputDim = input("Enter dimension of map: ")
inputP = input("Enter probability 0<p<1: ")
map=generateMap(int(inputDim),float(inputP))
mazeVisual(map)
count = 2
search = input("Choose search option: visuals, dfs, bfs, a*, bi-bfs, all \n")
if (search == "dfs"):
start_time = time.time()
result = dfs(map)[0]
print("--- %s seconds ---" % (time.time() - start_time))
if (result == None):
print("NO DFS PATH FOUND")
visual(map, result[0], "DFS", count)
elif (search == "bfs"):
start_time = time.time()
result = bfs(map)
print("--- %s seconds ---" % (time.time() - start_time))
if (result == None):
print("NO BFS PATH FOUND")
return
visual(map, result, "BFS", count)
elif (search == "a*" or search == "a" ):
# Setting up map
map[0][0]=0
map[len(map)-1][len(map)-1] = 0
flag = input("Enter number: (0) Manhattan (1) Euclidean \n")
# 1 for Euclidean
# 0 for Manhattan
start_time = time.time()
time.time()
signal.signal(signal.SIGALRM, handler)
signal.alarm(25)
try:
path = astar(map,flag)[0]
except IOError:
path = None
if path is None:
print("A* path does not exist")
return
#else:
print("--- %s seconds ---" % (time.time() - start_time))
result = path
visual(map, result, "A*", count)
elif (search == "bi-bfs"):
start_time = time.time()
result = bi_bfs(map)
print("--- %s seconds ---" % (time.time() - start_time))
if (result == None):
print("NO BI-BFS PATH FOUND")
visual(map, result, "BI-DIRECTIONAL BFS", count)
elif (search == "all"):
start_time = time.time()
path = dfs(map)[0]
print("dfs path length: "+ str(len(path)))
print("--- dfs took %s seconds ---" % (time.time() - start_time))
start_time = time.time()
path = bfs(map)
print("bfs path length: "+ str(len(path)))
print("--- bfs took %s seconds ---" % (time.time() - start_time))
# Setting up map
map[0][0]=0
map[len(map)-1][len(map)-1] = 0
start_time = time.time()
time.time()
signal.signal(signal.SIGALRM, handler)
signal.alarm(3)
try:
path = astar(map,0)[0]
except IOError:
path = None
if path is None:
print("NO PATH FOUND FOR A* MANHATTAN")
else:
print("a* manhatttan path length: "+ str(len(path)))
print("--- a* manhatttan took %s seconds ---" % (time.time() - start_time))
# Setting up map
map[0][0]=0
map[len(map)-1][len(map)-1] = 0
start_time = time.time()
time.time()
#signal.signal(signal.SIGALRM, handler)
#signal.alarm(3)
try:
path = astar(map,1)[0]
except IOError:
path = None
if path is None:
print("NO PATH FOUND FOR A* EUCLIDEAN")
else:
print("a* euclidean path length: "+ str(len(path)))
print("--- a* euclidean took %s seconds ---" % (time.time() - start_time))
start_time = time.time()
path = bi_bfs(map)
print("bi-bfs path length: "+ str(len(path)))
elif (search == "visuals"):
temp = copy.deepcopy(map)
temp[0][0]=0
temp[len(map)-1][len(map)-1] = 0
result1 = dfs(map)
result2 = bfs(map)
result3 = bi_bfs(map)
result4 = astar(temp, 0)
result5 = astar(temp, 1)
if (result1 == None):
print("NO DFS PATH FOUND")
else:
visual(map, result1[0], "DFS", count)
count = count + 1
if (result2 == None):
print("NO BFS PATH FOUND")
else:
visual(map, result2, "BFS", count)
count = count + 1
if (result3 == None):
print("NO BI-BFS PATH FOUND")
else:
visual(map, result3, "BI-BFS", count)
count = count + 1
if (result4 == None):
print("NO A* EUCLIDEAN PATH FOUND")
else:
visual(temp, result4[0], "A* EUCLIDEAN", count)
count = count + 1
if (result5 == None):
print("NO A* MANHATTAN PATH FOUND")
else:
visual(temp, result5[0], "A* MANHATTAN", count)
count = count + 1
plt.show()
empty = node[1]
neighbors = []
# move in every direction, create a new state where the empty node is at current + d
for d in DIRECTIONS:
new_empty = empty + d
if new_empty.x > 36 or new_empty.x < 0 or new_empty.y > 24 or new_empty.y < 0:
continue
if new_empty in full_nodes:
continue
# if wanted data is the one we just moved, reflect that in the new node
new_wanted = wanted
if new_empty == wanted:
new_wanted = empty
new = (new_wanted, new_empty)
neighbors.append(new)
return neighbors
s = time.time()
answer = astar(start, is_goal_fn, heuristic_fn, lambda x, y: 1,
get_neighbors_fn, get_key_fn)
print(answer)
print(time.time() - s)
# elevator cant be outside of building
if elevator + d < 0 or elevator + d > 3:
continue
# create a copy of the existing state so that we don't share references of sets within
new_floors = copy_state(floors)
# try to make the move
for c in combo:
new_floors[elevator].remove(c)
new_floors[elevator + d].add(c)
# don't add if invalid config
if not is_valid(new_floors):
continue
neighbors.append(
(new_floors, elevator + d, current_steps + 1))
return neighbors
steps = astar(start=(floors, 0, 0),
is_goal_fn=lambda x: is_done(x[0]),
heuristic_fn=heuristic,
cost_fn=lambda a, b: 1,
get_neighbors_fn=get_neighbors,
get_key_fn=lambda x: (get_string_rep(x[0]), x[2]))
print(steps)
print(time.time() - start)
def get_shortest_path(grid, starting_position, distance_map):
''' Get shortest path to getting all keys '''
####
# A* NODE IS TUPLE -> (string of current keys, current cost, current location)
# Indexed by: (0 , 1 , 2 )
####
num_keys = sum(1 for i in grid.values() if i in string.digits)
min_dist = 100000
for this_char in distance_map:
distances = distance_map[this_char]
for other_char_key in distances:
other_char_pos, distance_to_other = distance_map[this_char][
other_char_key]
if distance_to_other < min_dist:
min_dist = distance_to_other
finalNode = None
def is_goal_fn(node):
nonlocal finalNode
is_goal = len(node[0]) == num_keys + 1
if is_goal:
finalNode = node
return is_goal
def heuristic(node):
return min_dist * (num_keys - len(node[0]))
def cost(a, b):
return b[1] - a[1]
def get_neighbors(node):
keys_acquired = set(node[0])
this_char = grid[node[2]]
distances_from_here = distance_map[this_char]
keys_available = []
for other_char_key in distances_from_here:
if other_char_key[1] == '.' and len(keys_acquired) != num_keys:
continue
other_char_pos, distance_to_other = distance_map[this_char][
other_char_key]
# If all of the keys on this path have been acquired, it's a viable path
if other_char_key[1] not in keys_acquired:
keys_available.append(
(other_char_pos, distance_to_other, other_char_key[1]))
neighbors = []
for key_point, distance_to, key_char in keys_available:
new_node = ("".join(sorted(node[0])) + key_char,
node[1] + distance_to, key_point)
neighbors.append(new_node)
return neighbors
def get_key_fn(node):
return node[0]
return (astar(("", 0, starting_position), is_goal_fn, heuristic, cost,
get_neighbors, get_key_fn), finalNode)
(not is_left and direction == 0)):
tile_array = node[0][:]
new_tile = (possible_next_tile, ori_ind)
tile_array.append(new_tile)
remaining_tiles = node[1] - set([possible_next_tile])
new_node = (tile_array, remaining_tiles)
neighbors.append(new_node)
break
return neighbors
grid_layout_flat = astar(start=entry,
is_goal_fn=is_goal_fn,
heuristic_fn=heuristic_fn,
cost_fn=lambda a, b: 1,
get_neighbors_fn=get_neighbors_fn,
get_key_fn=lambda n: str(n),
include_final_node=True)[1][0]
final = array([])
gridz = [grids[i[0]][i[1]] for i in grid_layout_flat]
for y in range(width):
row = array([])
for x in range(width):
grid_no, ori = grid_layout_flat[x + (width * y)]
cell = grids[grid_no][ori]
cell = cell[1:-1]
