def dijkstra_shortest_path(graph: AdjacentListGraph, start: SPVertice,
end: SPVertice) -> Tuple[int, List[SPVertice]]:
visited = set()
for v in graph.vertices:
v.shortest_path_estimate = inf
start.shortest_path_estimate = 0
q = bintrees.RBTree()
for v in graph.vertices:
q.insert(v, v.shortest_path_estimate)
while not q.is_empty():
u, current_sp_eta = q.min_item()
q.remove(u)
visited.add(u)
if u == end:
shortest_path = [u]
while u.previous:
u = u.previous
shortest_path.append(u)
return end.shortest_path_estimate, list(reversed(shortest_path))
for adj_v, w in graph[u].items():
if adj_v in visited:
continue
if adj_v.shortest_path_estimate > current_sp_eta + w:
adj_v.previous = u
q.remove(adj_v)
adj_v.shortest_path_estimate = current_sp_eta + w
q.insert(adj_v, adj_v.shortest_path_estimate)
return inf, []
def a_star(
graph: AdjacentListGraph,
start: PlaneVertice,
end: PlaneVertice,
h_function: Callable = h_euclid_dist
) -> Tuple[int, List[PlaneVertice]]:
visited = set()
start.shortest_path_estimate = 0
q = bintrees.RBTree()
for v in graph.vertices:
q.insert(v, v.shortest_path_estimate)
while not q.is_empty():
u, _ = q.min_item()
q.remove(u)
visited.add(u)
if u == end:
shortest_path = [u]
while u.previous:
u = u.previous
shortest_path.append(u)
return end.shortest_path_estimate, list(reversed(shortest_path))
for adj_v, w in graph[u].items():
if adj_v in visited:
continue
if adj_v.shortest_path_estimate > u.shortest_path_estimate + w:
adj_v.previous = u
q.remove(adj_v)
adj_v.shortest_path_estimate = u.shortest_path_estimate + w
q.insert(adj_v,
adj_v.shortest_path_estimate + h_function(adj_v, end))
return inf, []
def dijkstra_shortest_path(s, t):
# Initialization of the distance of starting point.
s.distance_with_fewest_edges = DistanceWithFewestEdges(0, 0)
node_set = bintrees.RBTree([(s, None)])
while node_set:
# Extracts the minimum distance vertex from heap.
u = node_set.pop_min()[0]
if u.id == t.id:
break
# Relax neighboring vertices of u.
for v in u.edges:
v_distance = u.distance_with_fewest_edges.distance + v.distance
v_num_edges = u.distance_with_fewest_edges.min_num_edges + 1
new_distance = DistanceWithFewestEdges(v_distance, v_num_edges)
if v.vertex.distance_with_fewest_edges > new_distance:
node_set.discard(v.vertex)
v.vertex.pred = u
v.vertex.distance_with_fewest_edges = new_distance
node_set.insert(v.vertex, None)
def output_shortest_path(v):
if v:
output_shortest_path(v.pred)
print(v.id)
# Outputs the shortest path with fewest edges.
output_shortest_path(t)
def projection_L1ball(self, v):
"""Find the projection of a vector onto the L1 ball. See for reference
'Efficient projections onto L1-ball for learning in high dimensions'
Duchi , Schwartz, Singer. SECTION 4, Figure 2.
Parameters
-------------------
v : vector to be projected
Returns
-------------------
w : projection on v onto the L1 ball of radius z
"""
vector_copy = np.abs(v)
keys_ar = np.unique(vector_copy)
r = map(lambda k: (np.where(k == vector_copy)[0]).sum(), keys_ar)
r = list(r)
# r = np.where(j == vector_copy)[0].sum()
# print("where ", r)
# r = map(lambda k: len(np.where(k == vector_copy)[0]), keys_ar)
# print("where2 ", r)
n_elements = np.cumsum(r[::-1])[::-1] # r di cui abbiamo bisogno
s_sum = keys_ar * r
cum_sum = np.cumsum(s_sum[::-1])[::-1]
dictionary = {}
for i in range(len(keys_ar)):
dictionary[keys_ar[i]] = [n_elements[i], cum_sum[i]]
rb_tree = bintrees.RBTree(dictionary)
theta = self._pivotsearch(rb_tree, rb_tree._root, 0., 0.)
return np.clip(vector_copy - theta, a_min=0, a_max=np.inf) * np.sign(v)
def OBJECTIVE(measure):
"""
produces the objective LineData()s RBTree arrangement
given the dictionary of interval measures;
N levels => N+1 LineData()s (verify?)
"""
def sweep(x):
Xminus = np.cumsum(x)
total = Xminus[-1]
Xplus = total - Xminus
X = Xplus - Xminus
return X
# prepare constants kappa and alpha
PREALPHA = np.array([0.] + [w for f, w in measure.items()])
ALPHA = sweep(PREALPHA)
PREKAPPA = np.array([0.] + [f * w for f, w in measure.items()])
KAPPA = sweep(PREKAPPA)
Cz = bintrees.RBTree()
ff = [f for f in measure] + [np.inf] # should be in order
for f, alpha, kappa in zip(ff, ALPHA, KAPPA):
Cz.insert(-f, LineData(alpha, kappa))
return Cz
def act(*args):
global t
h, card = t.max_item()
# print(card)
policy[p] = card.uid
if card.update_card_limit(p.w) == 0:
t.remove(h)
t = T.RBTree(t.items())
return h, card
def generate_first_k_a_b_sqrt2(k):
candidates = bintrees.RBTree([(ABSqrt2(0, 0), None)])
result = []
while len(result) < k:
next_smallest = candidates.pop_min()[0]
result.append(next_smallest.val)
candidates[ABSqrt2(next_smallest.a + 1, next_smallest.b)] = None
candidates[ABSqrt2(next_smallest.a, next_smallest.b + 1)] = None
return result
def generate_first_k_a_b_sqrt2(k):
tree = bintrees.RBTree()
tree.insert(AB(0, 0), None)
result = []
for i in range(k):
cand = tree.pop_min()[0]
result.append(cand.val)
tree.insert(AB(cand.a + 1, cand.b), None)
tree.insert(AB(cand.a, cand.b + 1), None)
return result
def generate_first_k_a_b_sqrt2(k: int) -> List[float]:
candidate = bintrees.RBTree([(Pairing(0, 0), None)])
res = []
while len(res) < k:
min_pair, _ = candidate.pop_min()
res.append(min_pair.val)
candidate.insert(Pairing(min_pair.a + 1, min_pair.b), None)
candidate.insert(Pairing(min_pair.a, min_pair.b + 1), None)
return res
def trapezoid_tree(trap_dict):
# for trapezoid containment query DS
tree = bintrees.RBTree()
for I in trap_dict:
a, b = I
L1, L2 = trap_dict[I]
tree[a] = (a, b, L1, L2)
return tree
def generate_first_k_a_b_sqrt2(k: int) -> List[float]:
# Initial for 0 + 0 * sqrt(2).
candidates = bintrees.RBTree([Number(0,0), None])
result: List[float] = []
while len(result) < k:
next_smallest = candidates.pop_min()[0]
result.append(next_smallest.val)
# Adds the next two numbers derived
candidates[Number(next_smallest.a + 1, next_smallest.b)] = None
candidates[Number(next_smallest.a, next_smallest.b + 1)] = None
return result
def find_closest_elements_in_sorted_arrays(sorted_arrays):
min_distance_so_far = float('inf')
# stores array iterators in each entry.
iters = bintrees.RBTree()
for idx, sorted_array in enumerate(sorted_arrays):
it = iter(sorted_array)
print("it: ", it)
first_min = next(it, None)
print("first_min: ", first_min)
print("idx: ", idx)
print("sorted_array: ", sorted_array)
if first_min is not None:
iters.insert((first_min, idx), it)
print("iters: ", iters)
print("\n")
while True:
min_value, min_idx = iters.min_key()
print("min_value: ", min_value)
print("min_idx: ", min_idx)
max_value = iters.max_key()[0]
print("max_value: ", max_value)
min_distance_so_far = min(max_value - min_value, min_distance_so_far)
print("min_distance_so_far: ", min_distance_so_far)
it = iters.pop_min()[1]
print("iters: ", iters)
next_min = next(it, None)
print("next_min: ", next_min)
# return if some array has no remaining elements.
if next_min is None:
return min_distance_so_far
iters.insert((next_min, min_idx), it)
print("iters: ", iters)
print("\n")
def generate_first_k_a_b_sqrt2(k):
# Initial for 0 + 0 * sqrt(2).
candidates = bintrees.RBTree([(ABSqrt2(0, 0), None)])
result = []
while len(result) < k:
next_smallest = candidates.pop_min()[0]
result.append(next_smallest)
# Adds the next two numbers derived from next_smallest.
candidates[ABSqrt2(next_smallest.a + 1, next_smallest.b)] = None
candidates[ABSqrt2(next_smallest.a, next_smallest.b + 1)] = None
return result
def SEGMENT(S, T):
""" sorted points into queues hanging from an indexed segment """
segment = bintrees.RBTree()
for i, s in enumerate(S):
q = ensure_key(s, segment)
q.P.append(i)
for j, t in enumerate(T):
q = ensure_key(t, segment)
q.Q.append(j)
return segment
def generate_first_k_a_b_sqrt2(k):
tree = bintrees.RBTree()
tree.insert(0, (0, 0))
res = []
def f(a, b):
return a + b * 2**0.5
while len(res) < k:
v, pair = tree.pop_min()
res.append(v)
tree.insert(f(pair[0] + 1, pair[1]), (pair[0] + 1, pair[1]))
tree.insert(f(pair[0], pair[1] + 1), (pair[0], pair[1] + 1))
return res
def get_most_visited_pages(it, k):
next_page = next(it, None)
rbt = bintrees.RBTree()
num_elements = 0
while next_page:
if next_page in rbt:
rbt[next_page] += 1
else:
if num_elements == k:
rbt.min_key() #remove the min key
else:
rbt[next_page] = 1
num_elements += 1
next_page = next(it, None)
return rbt
def find_closest_in_three_sorted_arrays(sorted_arrays):
min_distance_so_far = float('inf')
iters = bintrees.RBTree()
for idx, sorted_array in enumerate(sorted_arrays):
it = iter(sorted_array)
first_min = next(it, None)
if first_min is not None:
iters.insert((first_min, idx), it)
while True:
min_value, min_idx = iters.min_key()
max_value = iters.max_key()[0]
min_distance_so_far = min(max_value - min_value, min_distance_so_far)
it = iters.pop_min()[1]
next_min = next(it, None)
if next_min is None:
return min_distance_so_far
iters.insert((next_min, min_idx), it)
def generate_first_k_a_b_sqrt2(k):
# initial for 0 + 0 * sqrt(2).
candidates = bintrees.RBTree([(Number(0, 0), None)])
result = []
while len(result) < k:
next_smallest = candidates.pop_min()[0]
print("candidates: ", candidates)
print("next_smallest: ", next_smallest.val)
result.append(next_smallest.val)
# adds the next two numbers derived from next_smallest.
candidates[Number(next_smallest.a + 1, next_smallest.b)] = None
candidates[Number(next_smallest.a, next_smallest.b + 1)] = None
return result
def find_closest_elements_in_sorted_arrays(sorted_arrays):
min_so_far = float("inf")
iters = bintrees.RBTree()
for idx, sorted_array in enumerate(sorted_arrays):
it = iter(sorted_array)
first_min = next(it, None)
if first_min is not None:
iters.insert((first_min, idx), it)
while True:
min_value, min_idx = iters.min_key()
max_value, _ = iters.max_key()
min_so_far = min(min_so_far, max_value - min_value)
it = iters.pop_min()[1]
next_min = next(it, None)
if next_min is None:
return min_so_far
iters.insert((next_min, min_idx), it)
def find_closest_elements_in_sorted_arrays(sorted_arrays):
iters = bintrees.RBTree()
for i, s in enumerate(sorted_arrays):
it = iter(s)
v = next(it, None)
if not v is None:
iters.insert((v, i), it)
dist = float('inf')
while True:
pair, it = iters.pop_min()
min_v, arr_i = pair
max_v = iters.max_key()[0]
dist = min(dist, max_v - min_v)
next_v = next(it, None)
if next_v is None:
return dist
else:
iters.insert((next_v, arr_i), it)
return 0
def calculate_view_from_above(A):
class Endpoint:
def __init__(self, is_left, line):
self.is_left = is_left
self.line = line
def __lt__(self, other):
return self.value() < other.value()
def value(self):
return self.line.left if self.is_left else self.line.right
sorted_endpoints = sorted([Endpoint(True, a)
for a in A] + [Endpoint(False, a) for a in A])
result = []
prev_xaxis = sorted_endpoints[0].value() # Leftmost end point.
prev = None
active_line_segments = bintrees.RBTree()
for endpoint in sorted_endpoints:
if active_line_segments and prev_xaxis != endpoint.value():
active_segment = active_line_segments.max_item()[1]
if prev is None: # Found first segment.
prev = LineSegment(prev_xaxis, endpoint.value(),
active_segment.color, active_segment.height)
else:
if (prev.height == active_segment.height
and prev.color == active_segment.color
and prev_xaxis == prev.right):
prev = prev._replace(right=endpoint.value())
else:
result.append(prev)
prev = LineSegment(prev_xaxis, endpoint.value(),
active_segment.color,
active_segment.height)
prev_xaxis = endpoint.value()
if endpoint.is_left: # Left end point.
active_line_segments[endpoint.line.height] = endpoint.line
else: # Right end point.
del active_line_segments[endpoint.line.height]
# Output the remaining segment (if any).
return result + [prev] if prev else result
def find_closest_elements_in_sorted_arrays(sorted_arrays):
min_distance_so_far = float('inf')
# Stores array iterators in each entry.
iters = bintrees.RBTree()
for idx, sorted_array in enumerate(sorted_arrays):
it = iter(sorted_array)
first_min = next(it, None)
if first_min is not None:
iters.insert((first_min, idx), it)
while True:
min_value, min_idx = iters.min_key()
max_value = iters.max_key()[0]
min_distance_so_far = min(max_value - min_value, min_distance_so_far)
it = iters.pop_min()[1]
next_min = next(it, None)
# Return if some array has no remaining elements.
if next_min is None:
return min_distance_so_far
iters.insert((next_min, min_idx), it)
def dijkstra_shortest_paths(graph: AdjacentListGraph, start: SPVertice):
visited = set()
for v in graph.vertices:
v.shortest_path_estimate = inf
start.shortest_path_estimate = 0
q = bintrees.RBTree()
for v in graph.vertices:
q.insert(v, v.shortest_path_estimate)
while not q.is_empty():
u, current_sp_eta = q.min_item()
q.remove(u)
visited.add(u)
for adj_v, w in graph[u].items():
if adj_v in visited:
continue
if adj_v.shortest_path_estimate > current_sp_eta + w:
adj_v.previous = u
q.remove(adj_v)
adj_v.shortest_path_estimate = current_sp_eta + w
q.insert(adj_v, adj_v.shortest_path_estimate)
def find_closest_elements_in_sorted_arrays(sorted_arrays):
min_dist_so_far = float('inf')
iters = bintrees.RBTree()
# you need i, to tell you what idx you're coming from down below
for i, sorted_arr in enumerate(sorted_arrays):
it = iter(sorted_arr)
first_min = next(it, None)
if first_min is not None:
it.insert((first_min, i), it)
while True:
min_val, min_idx = iters.min_key()
max_val = iters.max_key()[0]
min_dist_so_far = min(max_val - min_val, min_dist_so_far)
it = iters.pop_min()[1]
next_min = next(it, None)
# if one is done, you've already exhausted finding the min between the 3
if next_min is None:
return min_dist_so_far
iters.insert((next_min, min_idx), it)
return 0
def MEASURE(segment, length, rbound=None):
if rbound is not None:
lbound = length
else:
lbound = 0.
rbound = length
# bintree instead of dict so that it is enumerated in sorted order
measure = bintrees.RBTree()
posts = [lbound] + [y for y, q in segment.iter_items()] + [rbound]
intervals = zip(posts[:-1], posts[1:])
deltas = [0] + [len(q.P) - len(q.Q) for y, q in segment.iter_items()]
F = np.cumsum(deltas)
for (a, b), f in zip(intervals, F):
measure.setdefault(f, 0.)
measure[f] += b - a
return measure
def find_closest_elem_in_sorted_arrays(sorted_arrays):
min_distance_so_far = float('inf')
iters = bintrees.RBTree()
for idx, s_arr in enumerate(sorted_arrays):
it = iter(s_arr)
first_min = next(it, None)
if first_min is not None:
iters.insert((first_min, idx), it)
while True:
min_val, min_idx = iters.min_key()
max_val = iters.max_key()[0]
min_distance_so_far = min(max_val - min_val, min_distance_so_far)
min_itr = iters.pop_min()[1]
next_min = next(min_itr, None)
if next_min is None:
return [min_val] + [key[0] for key in iters.keys()]
iters.insert((next_min, min_idx), min_itr)
def find_closest_elements_in_sorted_arrays(sorted_arrays):
min_distance_so_far = float('inf')
tree = bintrees.RBTree()
for idx, sorted_array in enumerate(sorted_arrays):
tree.insert((sorted_array[0], idx), 0)
result = []
while len(tree) == 3:
result = []
min_value, aidx = tree.min_key() # O(k)
max_value, aidx = tree.max_key() # O(k)
if min_distance_so_far > max_value - min_value:
min_distance_so_far = max_value - min_value
result = []
for node in tree:
result.append(node[0])
key, idx = tree.pop_min() # O(k)
next_min, arr_idx = key
if idx + 1 < len(sorted_arrays[arr_idx]):
tree.insert((sorted_arrays[arr_idx][idx + 1], arr_idx),
idx + 1) # O(k)
return (min_distance_so_far, result)
def closest_entry_in_3_arrays(sorted_arrays):
bst = bintrees.RBTree()
for idx, sorted_array in enumerate(sorted_arrays):
array_iter = iter(sorted_array)
min_value = next(array_iter, None)
if min_value is not None:
bst.insert((min_value, idx), array_iter)
best_min_distance_so_far = float('inf')
while True:
min_key, min_iter = bst.min_item()
max_key, max_iter = bst.max_item()
min_distance_now = max_key[0] - min_key[0]
best_min_distance_so_far = min(best_min_distance_so_far,
min_distance_now)
bst.pop_min()
next_value = next(min_iter, None)
if next_value is None:
return best_min_distance_so_far
bst.insert((next_value, min_key[1]), min_iter)
return best_min_distance_so_far
def find_closest_elements_in_sorted_arrays(sorted_arrays):
min_dist_so_far = float('inf')
iters = bintrees.RBTree()
for idx, array in enumerate(sorted_arrays):
it = iter(array)
first_min = next(it, None)
if first_min:
# first_min is inserted with idx as tuple
# to differentiate duplicate elements among sorted arrays
iters.insert((first_min, idx), it)
while True:
print_tree(iters)
min_value, min_idx = iters.min_key()
max_value = iters.max_key()[0]
min_dist_so_far = min(max_value - min_value, min_dist_so_far)
it = iters.pop_min()[1]
next_min = next(it, None)
if next_min is None:
return min_dist_so_far
# next_min is inserted with min_idx as tuple
# to differentiate duplicate elements among sorted arrays
iters.insert((next_min, min_idx), it)
def Astar(start, end):
open = bintrees.RBTree()
in_open = {}
d[start] = Nod(heuristic(start))
d[start].f = 0
d[start].g = 0
open.insert((d[start].f, start), d[start].f)
in_open[start] = 1
while len(open) > 0:
best = open.pop_min()
current = best[0][1]
del in_open[current]
if current == end:
return make_path(current)
successors = expand(current)
#For any in the expansion list
for adj in successors:
if adj not in d:
d[adj] = Nod(heuristic(adj))
now_g = d[current].g + 1
if now_g < d[adj].g:
d[adj].last = current
d[adj].g = now_g
d[adj].f = d[adj].g + d[adj].h
if adj not in in_open:
open.insert((d[adj].f, adj), d[adj].f)
in_open[adj] = 1
return "Nu solution"
