def finish(self, zeta1=0.1, zeta2=0.1):
'''
Args:
zeta1: for ip, op
zeta2: for decision layers
'''
# update layer mass
global layer_graph_table
nodes = self.get_nodes()
ipop = []
dl = []
pl_lm = 0
for node in nodes:
if self.get_node_attr(node) in self.iop_layers:
ipop.append(node)
elif self.get_node_attr(node) in self.decision_layers:
dl.append(node)
else:
total_filters = 0
for n in self._graph.predecessors(node):
total_filters += self._graph.node[n]['num_of_filters']
k = 0.1 if self._graph.node[node]['type'] == LAYERS.fc else 1
self._graph.node[node]['layer_mass'] = k * total_filters * self._graph.node[node]['num_of_filters']
pl_lm += self._graph.node[node]['layer_mass']
self.total_lm = pl_lm
for node in ipop:
self._graph.node[node]['layer_mass'] = zeta1 * pl_lm
self.total_lm += self._graph.node[node]['layer_mass']
for node in dl:
self._graph.node[node]['layer_mass'] = zeta2 * pl_lm / len(dl)
self.total_lm += self._graph.node[node]['layer_mass']
# update distance
for graph in layer_graph_table:
get_distance(graph, self, update=True)
def K_single(x1, x2):
first_term = self.alpha * np.exp(
sum([
-self.betas[i] * get_distance(x1, x2, self.v_str[i])[0]
for i in range(self.num_beta)
]))
second_term = self.alpha_bar * np.exp(
sum([
-self.beta_bars[i] *
(get_distance(x1, x2, self.v_str[i])[1])**2
for i in range(self.num_beta)
]))
return first_term + second_term
def estimate_dist():
directions = ['right', 'left']
dist_dict = {}
for direction in directions:
servo.change_pos(pos=direction)
dist = distance.get_distance()
if not math.isnan(dist):
dist_dict[direction] = distance.get_distance()
##back to neutral
servo.change_pos(pos='neutral')
return dist_dict
def cluster_aspects(existing_clusters, sim_dict, word_tagfreq_dict, delta):
phi = deepcopy(existing_clusters)
processed_cluster_flag = False
while not processed_cluster_flag:
clusters_to_loop_thro = deepcopy(phi)
for i, cluster_i in enumerate(clusters_to_loop_thro):
cluster_i_dist_map = {}
for j, cluster_j in enumerate(clusters_to_loop_thro[i + 1:]):
cluster_dist = get_distance(cluster_i, cluster_j, sim_dict,
word_tagfreq_dict)
cluster_i_dist_map[(i, i + 1 + j)] = cluster_dist
if cluster_i_dist_map:
index_tup_of_closest_clusters, dist_bw_closest_cluster = \
sorted(cluster_i_dist_map.items(),key=operator.itemgetter(1))[0]
del cluster_i_dist_map
if dist_bw_closest_cluster < delta:
phi[index_tup_of_closest_clusters[0]].\
extend(phi[index_tup_of_closest_clusters[1]])
del phi[index_tup_of_closest_clusters[1]]
break
else:
continue
else:
processed_cluster_flag = True
break
return phi
def test_get_distance():
points = generate_points(5)
for a in points:
for b in points:
distance = abs(a[0] - b[0]) + abs(a[1] - b[1])
assert get_distance(a, b) == distance, \
"get_distance({a}, {b}) != {distance}".format(a=a, b=b, distance=distance)
def __init__(self, fruit):
self.box = fruit.box
self.score = fruit.score
self.cls = 'Apple' if fruit.class_id == 48 else 'Banana'
distance = get_distance(fruit.box)
self.distance = distance if fruit.class_id == 48 else distance * 1.25
self.size = get_size(fruit.box)
def verification(pos_list, neg_list, dist_type = 'L2'):
'''
[
[feat1, feat2, ..],
...
[feat1, feat2, ..]
]
'''
# distance measure
if isinstance(dist_type, str):
dist_func = get_distance(dist_type)
else:
dist_func = dist_type
# get dist
pos_dist = []
for i in pos_list:
dist = dist_func(i[0], i[1])
diff = i[0] - i[1]
pos_dist.append([dist])
neg_dist = []
for i in neg_list:
dist = dist_func(i[0], i[1])
neg_dist.append([dist])
precision, threshold, accu_list = test_kfold(pos_dist, neg_dist)
pos = sorted(pos_dist, key=lambda x: x[0])
neg = sorted(neg_dist, key=lambda x: x[0], reverse=True)
pos = [x[0] for x in pos]
neg = [x[0] for x in neg]
acc, std = np.mean(accu_list), np.std(accu_list)
return acc, std, threshold, pos, neg
def main():
directions = load_directions('directions.txt')
starting_point = (0, 0)
starting_orientation = 'N'
ending_point, _ = follow_directions(starting_point, starting_orientation,
*directions)
print(get_distance(starting_point, ending_point))
def added_distance(self, delivery_start_time,
delivery_to_measure: Delivery):
"""
Complexity: Big O(n)
Calculate HOw many miles would be added if appending this delivery to this route
"""
if len(self.deliveries) == 0:
distance = get_distance(self.starting_location.address,
delivery_to_measure.location.address)
delivery_ETA = get_ETA(delivery_start_time,
[delivery_to_measure]).time()
if delivery_to_measure.earliest_deadline() > delivery_ETA:
return distance, 0
min_distance = None
add_index = None
for index in range(0, len(self.deliveries) + 1):
deliveries_copy = self.deliveries.copy()
deliveries_copy.insert(index, delivery_to_measure)
added_distance_for_delivery = get_miles_of_delivery_route(
self.starting_location, deliveries_copy)
checks_pass = check_all_deliveries_arrive_by_deadline(
delivery_start_time, deliveries_copy)
if checks_pass is False:
continue
if (min_distance is None
or added_distance_for_delivery < min_distance):
min_distance = round(added_distance_for_delivery, 2)
add_index = index
return (min_distance, add_index)
def distance():
result = None
if request.method == 'POST':
location_a = request.form['location_a']
location_b = request.form['location_b']
result = get_distance(location_a, location_b)
return render_template('distance.html', title='Afstand', result=result)
def get_nearest_to_hub(input_route):
# bigO O(n)-linear
cost_matrix = []
for location in get_unvisited_locations(input_route):
cost = get_distance('WGU_HUB', location)
cost_matrix.append((location, cost))
nearest_to_hub = heapq.nsmallest(1, cost_matrix, key=lambda x: float(x[1]))
return nearest_to_hub
def get_nearest_location(last_visited, unvisited):
# bigO O(n)-linear
cost_matrix = []
for destination in unvisited:
cost = get_distance(last_visited, destination)
cost_matrix.append((destination, cost))
nearest_location = heapq.nsmallest(1,
cost_matrix,
key=lambda x: float(x[1]))
return nearest_location
def get_miles_to_next_location(self):
"""
Complexity: Big O(n)
return the amount of miles to the next delivery location
"""
next_location = self.get_next_location()
if next_location is None:
return None
return get_distance(self._current_location.address,
next_location.address)
def get_nearby_items(self, user_id, point, R=0.5):
items = self.get_all(user_id)
nearby_items = []
for item in items:
if not item[3]:
continue
lat, lon = [float(c.strip()) for c in item[3][1:-1].split(',')]
if get_distance(lat, lon, *point) <= R:
nearby_items.append(item)
return nearby_items
def callback(self, data):
try:
cv_image = self.bridge.imgmsg_to_cv2(data)
depth_matrix = get_distance(cv_image)
print depth_matrix.max()
print coor_trans(depth_matrix)
except CvBridgeError as e:
print e
def distance_filter(row):
"""
:param row:
:return row:
"""
if row[1] != 0.0 and row[2] != 0.0 and row[3] != 0.0 and row[4] != 0.0 and row[5] != 0.0 and row[6] != 0.0: # filters out rows with zero elements
plong,plat,dlong,dlat=row[-4:]
if 100 > get_distance(plat,plong,dlat,dlong) > 0: # one of the many cool features of python
return True
return False
def k_means(dataset, k_value):
sample_count = len(dataset)
is_centroids_changed = True
# set the start centroids
centroids, centroids_index = set_centroids(dataset, k_value)
cluster_result = np.zeros(sample_count, dtype=int).tolist()
centroids_change_record = []
centroids_change_record.append(centroids)
loop_count = 0
while is_centroids_changed == True:
loop_count += 1
print("Loop :", loop_count)
is_centroids_changed = False
# calculate the min distance to the every centroids
for i in range(sample_count):
min_distance = distance.get_distance("E", dataset[i].data,
centroids[0])
min_cluster_index = 0
for j in range(k_value):
temp_distance = distance.get_distance("E", dataset[i].data,
centroids[j])
if min_distance > temp_distance:
min_distance = temp_distance
min_cluster_index = j
# refresh the cluster
if cluster_result[i] != min_cluster_index:
is_centroids_changed = True
cluster_result[i] = min_cluster_index
# refresh the centorid
for m in range(k_value):
temp_cluster = []
for n in range(sample_count):
if cluster_result[n] == m:
temp_cluster.append(dataset[n].data)
# temp_cluster = np.array(temp_cluster)
new_centroid = get_matrix_mean(temp_cluster)
centroids[m] = new_centroid
return centroids, cluster_result, loop_count
def get_distance():
# Print results to LCD
mylcd.lcd_display_string("Calculating Distance", 1)
sleep(1)
mylcd.lcd_display_string("...", 2)
sleep(1)
mylcd.lcd_display_string(dst.get_distance(), 4)
sleep(10)
mylcd.lcd_clear()
def get_ETA_back_at_depot(self):
"""
Complexity: Big O(1)
"""
last_delivery = self.deliveries[-1]
miles_to_deport_from_last_location = get_distance(
last_delivery.location.address, self.starting_location.address)
minutes_to_depot_from_last_location = miles_to_minutes(
miles_to_deport_from_last_location)
return (datetime.combine(datetime.today(), last_delivery.ETA) +
timedelta(minutes=minutes_to_depot_from_last_location)).time()
def return_to_base(self, truck_id):
"""
Complexity: Big O(n)
When the route is complete instuct return the Truck back to base
"""
drive_distance = get_distance(self._current_location.address,
self.starting_location.address)
self._miles_driven += drive_distance
timestamp = self.miles_traveled_time_delivered(self._miles_driven)
self._current_location = self.starting_location
self.time_route_complete = timestamp
print(f'Truck {truck_id} back at base at', timestamp)
def trainJump(save, save_as=None, curr_checkpoint=None):
model.train()
global variance
for episode in range(num_episodes):
print("-----------------------------------------")
print("Episode:", episode)
# Get state
state = get_distance()
prev_score = getScore()
print("Distance:", state)
state = np.array([state])
state = torch.from_numpy(state)
state = state.float()
# Calculate mean and variance
mean = model(state)
variance = final_variance + (initial_variance - final_variance) * \
math.exp(-1. * episode / variance_decay)
print("Mean:", float(mean), "Deviation:", float(variance))
# Construct normal distribution based off of mean and variance and sample from it
m = Normal(mean, variance)
action = m.sample()
# Perform action
print("Action:", action)
os.system("adb shell input swipe 500 500 500 500 " + str(int(action)))
# Get reward and optimize model
time.sleep(0.5)
reward = getReward(prev_score)
if reward >= 2:
reward = 10
elif reward == 1:
reward = 1
elif reward < 0:
onDeath()
print("Reward:", reward)
loss = -m.log_prob(action) * reward
optimizer.zero_grad()
loss.backward()
optimizer.step()
if save:
if (episode + 1) % 501 == 0:
save_file = {
'state_dict': model.state_dict(),
'optimizer': optimizer.state_dict(),
}
file_name = save_as + str((episode // 1000) +
curr_checkpoint) + ".pth"
torch.save(save_file, file_name)
def get_miles_to_location(self, location):
"""
Complexity: Big O(n)
return the amount of miles to a specified location
"""
if location is self.starting_location:
return 0
try:
end_location = self._route.index(location)
except ValueError:
print('Location not in route')
return
miles_to_location = 0
for loc_idx, location_iter in enumerate(self._route[:end_location]):
if loc_idx == len(self._route[:end_location]):
miles_to_location += get_distance(location_iter.address,
location.address)
miles_to_location += get_distance(location_iter.address,
self._route[loc_idx + 1].address)
return round(miles_to_location, 2)
def getReply(message, from_number):
message = message.lower().strip()
answer = ""
if "transaction" in message:
answer = "here is your latest transaction"
elif " near " in message:
typeOfResource = message.split("near", 1)[0]
location = message.split("near", 1)[1]
catagory = ""
if "food pantry" in typeOfResource:
catagory = "Food Pantry"
elif "summer meal" in typeOfResource:
catagory = "Summer Meals"
elif "education" in typeOfResource or "school" in typeOfResource:
catagory = "Educational Resource"
elif "free meal" in typeOfResource or "meal" in typeOfResource:
catagory = "Free Meal"
elif "convenience strore" in typeOfResource:
catagory = "Convenience Store"
elif "garden" in typeOfResource:
catagory = "Community Garden"
elif "backpack" in typeOfResource:
catagory = "BackPack Program"
elif "grocery" in typeOfResource or "coop" in typeOfResource or "co-op" in typeOfResource:
catagory = "Grocery/Co-Op"
elif "farm" in typeOfResource or "csa" in typeOfResource:
catagory = "CSA Program"
elif "mobile" in typeOfResource:
catagory = "Mobile market"
else:
catagory = "all"
latlng = database.get_home_address_by_phone(from_number)
result = distance.get_distance(latlng, catagory)
print(catagory)
answer = "Closest " + result[1][3] + " is " + result[1][
1] + " on " + result[1][8] + ", " + result[1][9] + ". "
# if result[1][7] != "":
# answer += "Phone # is " +str(result[1][7]) + ". "
# if result[1][12] != "":
# answer += "Open on " +str(result[1][12]) + " "
# if result[1][12] != "":
# answer += " " +str(result[1][13])
else:
answer = "please try another question?"
return answer
def start_play():
garbage_dis = 0
try:
garbage_dis = distance.get_distance()
print ("Distance 1 : %.1f " % garbage_dis)
if garbage_dis < 3:
pass
elif garbage_dis < length_threshold:
playwav.play(0)
pass
#ws2812.light(0)
else:
pass
#ws2812.light(1)
except Exception:
pass
def main():
directions = load_directions('directions.txt')
starting_point = point = (0, 0)
orientation = 'N'
visited = []
for step in directions:
next_point, orientation = follow_directions(point, orientation, step)
expanded = expand_path(point, next_point)
for intermediate_point in expanded[1:]:
if intermediate_point in visited:
print(get_distance(starting_point, intermediate_point))
return
else:
visited.append(intermediate_point)
point = next_point
print('No points visited twice.')
def irdar_sweep(channel, start, end, incr=5, distances=None):
""" Sweep IRDAR and record distances
:param channel: channel the servo is connected to
:param start: starting angle
:param end: ending angle
:param incr: increment by this many degrees
:param distances: add data to this dict
"""
angle = start
if not distances: distances = {}
while True:
setAngle(channel, angle, incr, min_delay=MIN_DELAY)
distances[angle] = distance.get_distance()
angle += incr
if start > end and angle < end: break
elif start < end and angle > end: break
return distances
def get_current_location(self):
"""
Complexity: Big O(n)
Get the current location of truck within the route
"""
if self._miles_driven == 0:
return self._route[0]
if self.route_complete is True:
return self._route[-1]
miles_driven_accu = 0
for loc_idx, location in enumerate(self._route):
delivered_location = self._route[loc_idx + 1]
driven_distance = get_distance(location.address,
delivered_location.address)
if miles_driven_accu + driven_distance > self._miles_driven:
return delivered_location
miles_driven_accu += driven_distance
def advance_by_miles(self, new_miles_driven):
"""
Complexity: Big O(n)
Move the route forward the amount of the new miles driven.
"""
print(f'have driven a total of {self._miles_driven} miles')
# if there is no next delivery we have completed our route
if self.get_next_delivery() is None:
self._current_location = self.deliveries[-1].location
self.route_complete = True
while self.get_next_delivery():
miles_driven_to_current_location = self.get_miles_to_location(
self._current_location)
# if we are not in the starting point
unused_miles_driven = self._miles_driven - miles_driven_to_current_location
print(
f'driven {miles_driven_to_current_location} miles from starting point'
)
# Get the next delivery and the distance from here to there.
next_delivery = self.get_next_delivery()
drive_distance = get_distance(self._current_location.address,
next_delivery.location.address)
# if we drove less miles than the distance to the next delivery
# Break since we haven't reached our destination yet
if self._miles_driven < miles_driven_to_current_location + drive_distance:
break
total_miles_to_next_location = miles_driven_to_current_location + drive_distance
timestamp = self.miles_traveled_time_delivered(
total_miles_to_next_location)
next_delivery.mark_as_delivered(timestamp)
self._current_location = next_delivery.location
miles_driven_to_current_location += drive_distance
if self.get_next_delivery() is None:
self._current_location = self.deliveries[-1].location
self.route_complete = True
self._miles_driven += new_miles_driven
def trainJump(save, save_as=None, curr_checkpoint=None):
model.train()
for episode in range(num_episodes):
prev_score = getScore()
print("-----------------------------------------")
print("Episode:", episode)
# Get state
state = torch.Tensor(get_distance())
# Select action based off state
node_activated, action = select_action(state)
os.system("adb shell input swipe 500 500 500 500 " + str(action))
# Optimize model
optimize_model() # Optimize here to save time
# Get reward and push state, node_activated, and actual_reward to memory
time.sleep(0.5)
actual_reward = getReward(prev_score)
print("Node Activated:", node_activated, "Action:", action)
if actual_reward >= 2:
actual_reward = 10
elif actual_reward < 0:
onDeath()
memory.push(state, node_activated, actual_reward)
# Print difference between predicted and actual rewards
predicted_reward = model(state).view(16)[node_activated]
print("Predicted Reward:", float(predicted_reward), "Actual Reward:",
actual_reward)
print("-----------------------------------------")
if save:
if (episode + 1) % 1001 == 0:
save_file = {
'state_dict': model.state_dict(),
'optimizer': optimizer.state_dict(),
}
file_name = save_as + str((episode // 1000) +
curr_checkpoint) + ".pth"
torch.save(save_file, file_name)
def cluster_features(existing_clusters, candi_features, sim_dict,
word_tagfreq_dict, delta):
phi = deepcopy(existing_clusters)
for word in candi_features:
clusters_to_loop_thro = deepcopy(phi)
word_dist_map = {}
for i, cluster_i in enumerate(clusters_to_loop_thro):
cluster_dist = get_distance(cluster_i, [word], sim_dict,
word_tagfreq_dict)
word_dist_map[i] = cluster_dist
if word_dist_map:
index_of_closest_cluster, dist_of_closest_cluster = \
sorted(word_dist_map.items(),key=operator.itemgetter(1))[0]
del word_dist_map
if dist_of_closest_cluster < delta:
phi[index_of_closest_cluster].append(word)
break
else:
continue
return phi
def irdar_sweep(channel, start, end, incr=5, distances=None):
""" Sweep IRDAR and record distances
:param channel: channel the servo is connected to
:param start: starting angle
:param end: ending angle
:param incr: increment by this many degrees
:param distances: add data to this dict
"""
angle = start
if not distances:
distances = {}
while True:
setAngle(channel, angle, incr, min_delay=MIN_DELAY)
distances[angle] = distance.get_distance()
angle += incr
if start > end and angle < end:
break
elif start < end and angle > end:
break
return distances
def add_destination(self, name, latitude, longitude, bing_api_key):
"""
Will append the given destination to the destination dictionary attribute
with name and distance from Location object to destination.
:param name: str, the name used to identify the destination.
:param latitude: float, the latitude of the destination.
:param longitude: float, the longitude of the destination.
:param bing_api_key: str, the Bing Maps API key used to make the request
to Bing Maps' distance matrix API
:return: float, the distance in miles from the Location to the given
coordinates.
"""
dest_distance = distance.get_distance(origin_lat=self.latitude,
origin_lon=self.longitude,
dest_lat=latitude,
dest_lon=longitude,
bing_api_key=bing_api_key)
self.destinations[name] = dest_distance
return dest_distance
if __name__ == '__main__':
# Usage: python filtereddistance.py 'data.csv' nameOfOutputDataFile
import sys
import pandas as pd
from distance import get_distance
from pprint import pprint
distances = []
error_count = 0
# create a pandas dataframe with the location data
df = pd.read_csv(sys.argv[1])
dfWithLocData = df[['pickup_longitude','pickup_latitude','dropoff_longitude', 'dropoff_latitude']]
for index, plong, plat, dlong, dlat in dfWithLocData.itertuples():
plong = float(plong)
plat = float(plat)
dlong = float(dlong)
dlat = float(dlat)
try:
distances.append(get_distance(plat,plong,dlat,dlong))
except:
distances.append('ERROR')
error_count += 1
print plat, plong, dlat, dlong
print error_count
filterData(distances, sys.argv[1], sys.argv[2])
