forked from cs-oak/Fibonacci-Heaps
-
Notifications
You must be signed in to change notification settings - Fork 0
/
dijkstra.py
651 lines (563 loc) · 20.5 KB
/
dijkstra.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
# most of this code was written by Dr. Timo Heister,
# Clemson University for a demonstration of a simple
# implementation of Dijkstra's algorithm. This code
# includes extra code with the Fibonacci Heap logic
# to compare and verify correctness. The original is
# part of a class assignment where we had to fill in
# various functions of the Priority Queue
%matplotlib inline
import matplotlib.pyplot as plt
import math
from collections import deque
from pylab import rcParams
# this is my homework assignment
class PriorityQueue():
'''
The arguments passed to a PriorityQueue must consist of
objects than can be compared using <.
Use a tuple (priority, item) if necessary.
'''
def __init__(self):
self._pq = []
self.icount = 0
def _parent(self,n):
return (n-1)//2
def _leftchild(self,n):
return 2*n + 1
def _rightchild(self,n):
return 2*n + 2
def push(self, obj):
# append at end and bubble up
self._pq.append( obj )
self.icount += 1
n = len(self._pq)
self._bubble_up(n-1)
def _bubble_up(self, index):
while index>0:
cur_item = self._pq[index]
parent_idx = self._parent(index)
parent_item = self._pq[parent_idx]
self.icount += 1
if cur_item < parent_item:
# swap with parent
self._pq[parent_idx] = cur_item
self._pq[index] = parent_item
index = parent_idx
else:
break
def pop(self):
n = len(self._pq)
if n==0:
return None
if n==1:
return self._pq.pop()
# replace with last item and sift down:
obj = self._pq[0]
self._pq[0] = self._pq.pop()
self.icount += 1
self._sift_down(0)
return obj
def heapify(self, items):
""" you can assume that the PQ is empty! """
self._pq = items
# TODO: restore heap property of _pq
pass
def _sift_down(self,index):
n = len(self._pq)
while index<n:
self.icount += 1
cur_item = self._pq[index]
lc = self._leftchild(index)
if n <= lc:
break
# first set small child to left child:
small_child_item = self._pq[lc]
small_child_idx = lc
# right exists and is smaller?
rc = self._rightchild(index)
if rc < n:
r_item = self._pq[rc]
if r_item < small_child_item:
# right child is smaller than left child:
small_child_item = r_item
small_child_idx = rc
# done: we are smaller than both children:
if cur_item <= small_child_item:
break
# swap with smallest child:
self._pq[index] = small_child_item
self._pq[small_child_idx] = cur_item
# continue with smallest child:
index = small_child_idx
def size(self):
return len(self._pq)
def is_empty(self):
return len(self._pq) == 0
# this is the Fibonacci Heap
class FhObject(object):
'''
This is a node object for the Fibonacci Heap class
Data entered here must be comparable with < or >
'''
def __init__(self):
self.parent = None
self.child = None
self.data = None
self.mark = 0
self.lsib = None
self.rsib = None
def add_child(self, new):
'''
Adds a child node to a node. Maintains min-heap property if child key is greater than parent key
Very useful for auto-updating the Fibonacci heap for min-heap during the consolidate operation
'''
if new.data < self.data:
self.__balance(self, new)
if self.child == None:
self.child = new
new.parent = self
else:
x = self.child
while x.rsib != None:
x = x.rsib
x.rsib = new
new.lsib = x
new.parent = self
if new.data < self.child.data:
self.child = new
def detach(self):
'''
Special delete method for Fibonacci heap implementation - detaches whole branch below the node
Does not actually delete the node or it's children - that is done by the Fibonacci Heap class
'''
if(self.parent != None):
#if this node was the min-child of parent, fix parent's min-child
if(self.parent.child == self):
minx = None
if(self.lsib != None):
x = self.lsib
if (minx == None):
minx = x
else:
if(x.data < minx.data):
minx = x
if(self.rsib != None):
x = self.rsib
if (minx == None):
minx = x
else:
if(x.data < minx.data):
minx = x
self.parent.child = minx
self.parent = None
#reset the siblings' connections before exiting the tree
if(self.lsib != None):
if(self.rsib != None):
self.lsib.rsib = self.rsib
self.rsib.lsib = self.lsib
self.rsib = None
else:
self.lsib.rsib = None
self.lsib = None
if(self.rsib != None):
self.rsib.lsib = None
self.rsib = None
def find_rank(self):
''' Finds the rank of a given node in the heap based on it's position '''
x = self
rank = 0
while x.child != None:
if x.child != None:
rank += 1
x = x.child
temp = x
if x.lsib != None:
while x.lsib != None:
x = x.lsib
if x.child != None:
break
if x.child != None:
continue
x = temp
if x.rsib != None:
while x.rsib != None:
x = x.rsib
if x.child != None:
break
if x.child != None:
continue
return rank
def __balance(self, parent, child):
''' Private method - responsible for maintaining balance of the heap '''
if child.data < parent.data:
temp = child.data
child.data = parent.data
parent.data = temp
if(parent.parent != None):
if parent.data < parent.parent.child.data:
parent.parent.child = parent
self.__balance(parent.parent, parent)
if(child.child != None):
if child.data > child.child.data:
self.__balance(child, child.child)
else:
x = child.child
while x.rsib != None:
x = x.rsib
if x.data < child.data:
child.child = x
self.__balance(child,x)
temp = x
while x.lsib != None:
x = x.lsib
if x.data < child.data:
child.child = x
self.__balance(child,x)
class FibonacciHeap(object):
'''
This is the class for a Fibonacci Heap
'''
def __init__(self):
self.roots = [] #this list will only hold roots
self.nodes = [] #this list will hold all nodes in the F-heap, useful for keeping find operation simple
self.minroot = None
self. icount = 0
def find_min(self):
'''
Finds the root with minimum value
Used in other functions to keep the pointer to the smallest value up to date
'''
index = 0
for i in range (0, len(self.roots)):
if self.minroot == None:
self.minroot = self.roots[i]
index = i
else:
if self.roots[i].data < self.minroot.data:
self.minroot = self.roots[i]
index = i
return index
def insert(self, item):
'''
Inserts new data into the sequence
Updates the minimum root automatically
Inserted data must be comparable with < or >
Tuples of the form (priority, key) accepted
'''
#algorithm is 'lazy' - we just insert the new item for now, we do not
#bother updating the heap for balance like we do in classic min-heap
new = FhObject()
new.data = item
self.roots.append(new)
self.icount += 1
self.nodes.append(new)
if self.minroot == None:
self.minroot = new
else:
if new.data < self.minroot.data:
self.minroot = new
def union(self, obj):
'''
Concatenates root lists of two Fibonacci Heaps
Updates minimum root automatically
'''
self.roots += obj.roots
self.nodes += obj.nodes
self.icount += 1
self.find_min()
def decrease_key(self, item, value):
'''
Decreases the value of a given node in the Fibonacci Heap
If multiple keys with the same value exists, it will perform this operation on the first node with the given value
Operation is return error message if key is not found
'''
#again, we take the 'lazy' approach - just re-attach the decreased nodes to root list
#we keep the order of the heap in check by marking nodes who've had more than 1 child
#detached due to this function, and we recursively detach marked parents
x = self.__find(item)
if x == None:
print "Error: Could not find a node with the given value"
return ''
x.data = value
if x not in self.roots:
if x.parent.mark == 0:
x.parent.mark = 1
x.detach()
self.roots.append(x)
else:
y = x.parent
x.detach()
self.roots.append(x)
while y.mark != 0:
z = y.parent
if z == None:
break
y.mark = 0
y.detach()
self.roots.append(y)
y = z
self.icount += 1
self.find_min()
def fh_pop(self):
'''
Pops the minimum element from the Fibonacci Heap
Functions similar to pop in a classic Priority Queue
'''
#first extarct the min-root and append all it's children to the root list
if self.is_empty():
print "Underflow",
return ''
else:
x = self.minroot
y = x.data
self.roots.remove(self.minroot)
self.nodes.remove(self.minroot)
self.icount += 1
if x.child:
x = x.child
self.roots.append(x)
temp = x
while x.lsib != None:
x = x.lsib
self.roots.append(x)
self.icount += 1
x = temp
while x.rsib != None:
x = x.rsib
self.roots.append(x)
self.icount += 1
for i in range(0,len(self.roots)):
if self.roots[i].parent != None:
self.roots[i].detach()
#consolidate - perform all the steps which the 'lazy' algorithm has been postponing so far
#cur - current node in the loop, checks nodes for same rank, if there is a match, attaches itself
#to it, and starts over, searching for root nodes matching it's new rank, once this is exhausted,
#it moves on to the next node
#prev - basically, to check for the algorithm's termination, we check if the root list has nodes with all
#unique ranks : this is the only condition which will not reset this variable and cause the loop to break
#We cannot directly loop this using the bound len(roots) because that bound changes during the loop
if self.is_empty():
self.minroot = None
return y
cur = self.roots[0]
i = self.roots.index(cur)
prev = 0
while True:
if prev == len(self.roots):
break
i = (i+1)%(len(self.roots))
if i == self.roots.index(cur):
cur = self.roots[(i+1)%(len(self.roots))]
i = self.roots.index(cur)
prev += 1
continue
if cur.find_rank() == self.roots[i].find_rank():
cur.add_child(self.roots[i])
self.icount += 1
self.roots.remove(self.roots[i])
i = self.roots.index(cur)
prev = 0
continue
self.minroot = None
self.find_min()
return y
def is_empty(self):
'''
Returns True if the Fibonacci Heap is empty
'''
if len(self.roots) == 0:
return True
else:
return False
def __find(self, item):
'''
Private method - find an item in Fibonacci heap which matches a given key
Used only by decrease key method
'''
for i in range (0, len(self.nodes)):
if self.nodes[i].data == item:
return self.nodes[i]
return None
# Dr. Heister's code begins here
class Graph(object):
'''Represents a graph'''
def __init__(self, vertices, edges):
'''A Graph is defined by its set of vertices
and its set of edges.'''
self.V = set(vertices) # The set of vertices
self.E = set([]) # The set of edges
self.Adj = {} # A dictionary that will hold the list
# of adjacent vertices for each vertex.
self.Vcoord = {} # A dictionary that can hold coordinates
# for the vertices.
self.edge_labels = {}
self.add_edges(edges) # Note the call to add_edges will also
# update the Adj dictionary
print '(Initializing a graph with %d vertices and %d edges)' % (len(self.V),len(self.E))
def add_vertices(self,vertex_list):
''' This method will add the vertices in the vertex_list
to the set of vertices for this graph. Since V is a
set, duplicate vertices will not be added to V. '''
for v in vertex_list:
self.V.add(v)
self.build_Adj()
def add_edges(self,edge_list):
''' This method will add a list of edges to the graph
It will insure that the vertices of each edge are
included in the set of vertices (and not duplicated).
It will also insure that edges are added to the
list of edges and not duplicated. '''
for s,t in edge_list:
if (s,t) not in self.E and (t,s) not in self.E:
self.V.add(s)
self.V.add(t)
self.E.add((s,t))
self.build_Adj()
def build_Adj(self):
self.Adj = {}
for v in self.V:
self.Adj[v] = []
for e in self.E:
s,t = e
self.Adj[s].append(t)
self.Adj[t].append(s)
def degree_of(self,vertex):
if vertex in self.V:
return len(self.Adj[vertex])
else:
return None
def get_a_vertex(self):
if 0 < len(self.V):
v = self.V.pop()
self.V.add(v)
return v
else:
return None
def plot(self):
nV = len(self.V)
if len(self.Vcoord) != nV:
# Coordinates have not been specified for every vertex
dTheta = 2*math.pi/nV
k = 0
for v in self.V:
self.Vcoord[v] = (10*math.cos(math.pi/2-k*dTheta),10*math.sin(math.pi/2-k*dTheta))
k += 1
px = []
py = []
for v in self.V:
px.append(self.Vcoord[v][0])
py.append(self.Vcoord[v][1])
plt.plot(px,py,'bo',hold=True)
for vertex in self.V:
p = self.Vcoord[vertex]
pq = max(0.1,math.sqrt(p[0]**2 + p[1]**2))
rx = p[0]/pq
ry = p[1]/pq
plt.text(p[0]+0.2*rx, p[1]+0.2*ry, str(vertex))
for s,t in self.E:
plt.plot([self.Vcoord[s][0], self.Vcoord[t][0]],
[self.Vcoord[s][1], self.Vcoord[t][1]],
'b',hold=True)
if (s,t) in self.edge_labels:
label = self.edge_labels[(s,t)]
plt.text((self.Vcoord[s][0]+self.Vcoord[t][0])/2-0.1,
(self.Vcoord[s][1]+self.Vcoord[t][1])/2-0.1, label)
plt.xlim(min(px)-1.0,max(px)+1.1)
plt.ylim(min(py)-1.0,max(py)+1.1)
def get_a_component_spanning_tree(self, root):
# This routine uses a breadth-first search
# to obtain a tree that spans the component
# containing
spanning_tree = []
visited = {}
for v in self.V:
visited[v] = False
Q = deque()
visited[root] = True
Q.append(root)
while 0 < len(Q):
v = Q.popleft()
for u in self.Adj[v]:
if not visited[u]:
visited[u] = True
Q.append(u)
spanning_tree.append((v,u))
return spanning_tree
def is_connected(self):
# If the graph is connected then if the tree
# returned by get_a_component_spanning_tree has
# nV-1 edges - that is, it spans the graph.
root = self.get_a_vertex()
tree = self.get_a_component_spanning_tree(root)
if len(tree) == len(self.V)-1:
return True
else:
return False
class Network(Graph):
def __init__(self, vertices, edge_weights):
''' Initialize the network with a list of vertices
and weights (a dictionary with keys (E1, E2) and values are the weights)'''
edges = []
for e1,e2 in edge_weights:
edges.append((e1,e2))
Graph.__init__(self, vertices, edges)
self.weights = {}
for e1,e2 in edge_weights:
weight = edge_weights[(e1,e2)]
self.weights[(e1,e2)] = weight
self.weights[(e2,e1)] = weight
self.edge_labels = self.weights
V = ['A', 'B', 'C', 'D', 'E']
W = {('A','C'):3, ('A','B'):6, ('A','D'):7, ('B','C'):1, ('B','D'):2, ('C','E'):10, ('B','E'):4}
G1 = Network(V,W)
G1.Vcoord = {'A':(1,0),'C':(3,5),'B':(4,0),'D':(4,-5),'E':(7,0)}
G1.plot()
print G1.weights
# original dijkstra code plus Fibonacci Heap code
def dijkstra(network, source):
dist = {source:0}
prev = {}
done = {}
pq = PriorityQueue()
pq.push((0,source))
while not pq.is_empty():
dist_u, u = pq.pop()
if u in done:
continue
done[u] = True
for v in network.Adj[u]:
new_dist_to_v = dist_u + network.weights[(u,v)]
if not v in dist or dist[v]>new_dist_to_v:
dist[v] = new_dist_to_v
prev[v] = u
pq.push((new_dist_to_v, v))
print pq.icount
return dist, prev
print "For Normal Priority Queue:"
dist, prev = dijkstra(G1,'A')
print "distance:", dist
print "prev:", prev
print ''
def dijkstra1(network, source):
dist = {source:0}
prev = {}
done = {}
pq = FibonacciHeap()
pq.insert((0,source))
while not pq.is_empty():
dist_u, u = pq.fh_pop()
if u in done:
continue
done[u] = True
for v in network.Adj[u]:
new_dist_to_v = dist_u + network.weights[(u,v)]
if not v in dist or dist[v]>new_dist_to_v:
dist[v] = new_dist_to_v
prev[v] = u
pq.insert((new_dist_to_v, v))
print pq.icount
return dist, prev
print "For Fibonacci Heap:"
dist, prev = dijkstra1(G1,'A')
print "distance:", dist
print "prev:", prev