__init__.py 56.4 KB
Newer Older
1
#! /usr/bin/env python
2
# -*- coding: utf-8 -*-
3
#
4
5
# graph_tool -- a general graph manipulation python module
#
Tiago Peixoto's avatar
Tiago Peixoto committed
6
# Copyright (C) 2007-2012 Tiago de Paula Peixoto <tiago@skewed.de>
7
8
9
10
11
12
13
14
15
16
17
18
19
20
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program.  If not, see <http://www.gnu.org/licenses/>.

21
"""
22
23
``graph_tool.topology`` - Assessing graph topology
--------------------------------------------------
24
25
26
27
28
29
30

Summary
+++++++

.. autosummary::
   :nosignatures:

31
   shortest_distance
Tiago Peixoto's avatar
Tiago Peixoto committed
32
   shortest_path
Tiago Peixoto's avatar
Tiago Peixoto committed
33
   pseudo_diameter
34
   similarity
35
   isomorphism
36
37
   subgraph_isomorphism
   mark_subgraph
38
39
   max_cardinality_matching
   max_independent_vertex_set
40
   min_spanning_tree
41
   random_spanning_tree
42
43
44
   dominator_tree
   topological_sort
   transitive_closure
Tiago Peixoto's avatar
Tiago Peixoto committed
45
   tsp_tour
46
   sequential_vertex_coloring
47
48
   label_components
   label_biconnected_components
49
   label_largest_component
50
   label_out_component
51
   is_bipartite
Tiago Peixoto's avatar
Tiago Peixoto committed
52
   is_DAG
53
   is_planar
Tiago Peixoto's avatar
Tiago Peixoto committed
54
   edge_reciprocity
55
56
57

Contents
++++++++
58

59
60
"""

61
62
from __future__ import division, absolute_import, print_function

Tiago Peixoto's avatar
Tiago Peixoto committed
63
from .. dl_import import dl_import
64
dl_import("from . import libgraph_tool_topology")
65

66
from .. import _prop, Vector_int32_t, _check_prop_writable, \
67
     _check_prop_scalar, _check_prop_vector, Graph, PropertyMap, GraphView
68
import random, sys, numpy
69
__all__ = ["isomorphism", "subgraph_isomorphism", "mark_subgraph",
70
           "max_cardinality_matching", "max_independent_vertex_set",
71
           "min_spanning_tree", "random_spanning_tree", "dominator_tree",
Tiago Peixoto's avatar
Tiago Peixoto committed
72
           "topological_sort", "transitive_closure", "tsp_tour",
73
74
75
           "sequential_vertex_coloring", "label_components",
           "label_largest_component", "label_biconnected_components",
           "label_out_component", "shortest_distance", "shortest_path",
Tiago Peixoto's avatar
Tiago Peixoto committed
76
77
           "pseudo_diameter", "is_bipartite", "is_DAG", "is_planar",
           "similarity", "edge_reciprocity"]
78
79
80
81
82
83
84
85
86
87


def similarity(g1, g2, label1=None, label2=None, norm=True):
    r"""Return the adjacency similarity between the two graphs.

    Parameters
    ----------
    g1 : :class:`~graph_tool.Graph`
        First graph to be compared.
    g2 : :class:`~graph_tool.Graph`
Tiago Peixoto's avatar
Tiago Peixoto committed
88
        Second graph to be compared.
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
    label1 : :class:`~graph_tool.PropertyMap` (optional, default: ``None``)
        Vertex labels for the first graph to be used in comparison. If not
        supplied, the vertex indexes are used.
    label2 : :class:`~graph_tool.PropertyMap` (optional, default: ``None``)
        Vertex labels for the second graph to be used in comparison. If not
        supplied, the vertex indexes are used.
    norm : bool (optional, default: ``True``)
        If ``True``, the returned value is normalized by the total number of
        edges.

    Returns
    -------
    similarity : float
        Adjacency similarity value.

    Notes
    -----
    The adjacency similarity is the sum of equal entries in the adjacency
    matrix, given a vertex ordering determined by the vertex labels. In other
    words it counts the number of edges which have the same source and target
    labels in both graphs.

    The algorithm runs with complexity :math:`O(E_1 + V_1 + E_2 + V_2)`.

    Examples
    --------
    >>> from numpy.random import seed
    >>> seed(42)
    >>> g = gt.random_graph(100, lambda: (3,3))
    >>> u = g.copy()
    >>> gt.similarity(u, g)
    1.0
    >>> gt.random_rewire(u);
    >>> gt.similarity(u, g)
    0.03333333333333333
    """

    if label1 is None:
        label1 = g1.vertex_index
    if label2 is None:
        label2 = g2.vertex_index
    if label1.value_type() != label2.value_type():
        raise ValueError("label property maps must be of the same type")
    s = libgraph_tool_topology.\
           similarity(g1._Graph__graph, g2._Graph__graph,
                      _prop("v", g1, label1), _prop("v", g1, label2))
    if not g1.is_directed() or not g2.is_directed():
        s /= 2
    if norm:
        s /= float(max(g1.num_edges(), g2.num_edges()))
    return s
140

Tiago Peixoto's avatar
Tiago Peixoto committed
141

142
def isomorphism(g1, g2, isomap=False):
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
    r"""Check whether two graphs are isomorphic.

    If `isomap` is True, a vertex :class:`~graph_tool.PropertyMap` with the
    isomorphism mapping is returned as well.

    Examples
    --------
    >>> from numpy.random import seed
    >>> seed(42)
    >>> g = gt.random_graph(100, lambda: (3,3))
    >>> g2 = gt.Graph(g)
    >>> gt.isomorphism(g, g2)
    True
    >>> g.add_edge(g.vertex(0), g.vertex(1))
    <...>
    >>> gt.isomorphism(g, g2)
    False

161
    """
162
163
    imap = g1.new_vertex_property("int32_t")
    iso = libgraph_tool_topology.\
164
           check_isomorphism(g1._Graph__graph, g2._Graph__graph,
Tiago Peixoto's avatar
Tiago Peixoto committed
165
                             _prop("v", g1, imap))
166
167
168
169
170
    if isomap:
        return iso, imap
    else:
        return iso

Tiago Peixoto's avatar
Tiago Peixoto committed
171

172
def subgraph_isomorphism(sub, g, max_n=0, random=False):
173
    r"""
174
175
    Obtain all subgraph isomorphisms of `sub` in `g` (or at most `max_n`
    subgraphs, if `max_n > 0`).
176

177

Tiago Peixoto's avatar
Tiago Peixoto committed
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
    Parameters
    ----------
    sub : :class:`~graph_tool.Graph`
        Subgraph for which to be searched.
    g : :class:`~graph_tool.Graph`
        Graph in which the search is performed.
    max_n : int (optional, default: 0)
        Maximum number of matches to find. If `max_n == 0`, all matches are
        found.
    random : bool (optional, default: False)
        If `True`, the vertices of `g` are indexed in random order before
        the search.

    Returns
    -------
    vertex_maps : list of :class:`~graph_tool.PropertyMap` objects
        List containing vertex property map objects which indicate different
        isomorphism mappings. The property maps vertices in `sub` to the
        corresponding vertex index in `g`.
    edge_maps : list of :class:`~graph_tool.PropertyMap` objects
        List containing edge property map objects which indicate different
        isomorphism mappings. The property maps edges in `sub` to the
        corresponding edge index in `g`.

    Notes
    -----
    The algorithm used is described in [ullmann-algorithm-1976]_. It has a
    worse-case complexity of :math:`O(N_g^{N_{sub}})`, but for random graphs it
    typically has a complexity of :math:`O(N_g^\gamma)` with :math:`\gamma`
    depending sub-linearly on the size of `sub`.
208
209
210
211
212
213
214
215

    Examples
    --------
    >>> from numpy.random import seed, poisson
    >>> seed(42)
    >>> g = gt.random_graph(30, lambda: (poisson(6),poisson(6)))
    >>> sub = gt.random_graph(10, lambda: (poisson(1.8), poisson(1.9)))
    >>> vm, em = gt.subgraph_isomorphism(sub, g)
216
    >>> print(len(vm))
217
    102
218
    >>> for i in range(len(vm)):
219
220
221
222
223
224
225
226
227
228
    ...   g.set_vertex_filter(None)
    ...   g.set_edge_filter(None)
    ...   vmask, emask = gt.mark_subgraph(g, sub, vm[i], em[i])
    ...   g.set_vertex_filter(vmask)
    ...   g.set_edge_filter(emask)
    ...   assert(gt.isomorphism(g, sub))
    >>> g.set_vertex_filter(None)
    >>> g.set_edge_filter(None)
    >>> ewidth = g.copy_property(emask, value_type="double")
    >>> ewidth.a += 0.5
Tiago Peixoto's avatar
Tiago Peixoto committed
229
230
231
    >>> ewidth.a *= 2
    >>> gt.graph_draw(g, vertex_fill_color=vmask, edge_color=emask,
    ...               edge_pen_width=ewidth, output_size=(200, 200),
232
    ...               output="subgraph-iso-embed.pdf")
233
    <...>
Tiago Peixoto's avatar
Tiago Peixoto committed
234
    >>> gt.graph_draw(sub, output_size=(200, 200), output="subgraph-iso.pdf")
235
236
    <...>

Tiago Peixoto's avatar
Tiago Peixoto committed
237
238
    .. image:: subgraph-iso.*
    .. image:: subgraph-iso-embed.*
239

240

Tiago Peixoto's avatar
Tiago Peixoto committed
241
    **Left:** Subgraph searched, **Right:** One isomorphic subgraph found in main graph.
242
243
244

    References
    ----------
245
    .. [ullmann-algorithm-1976] Ullmann, J. R., "An algorithm for subgraph
Tiago Peixoto's avatar
Tiago Peixoto committed
246
       isomorphism", Journal of the ACM 23 (1): 31–42, 1976, :doi:`10.1145/321921.321925`
247
    .. [subgraph-isormophism-wikipedia] http://en.wikipedia.org/wiki/Subgraph_isomorphism_problem
248
249
250
251

    """
    # vertex and edge labels disabled for the time being, until GCC is capable
    # of compiling all the variants using reasonable amounts of memory
Tiago Peixoto's avatar
Tiago Peixoto committed
252
253
    vlabels=(None, None)
    elabels=(None, None)
254
255
    vmaps = []
    emaps = []
256
    if random:
257
        seed = numpy.random.randint(0, sys.maxsize)
258
259
    else:
        seed = 42
260
261
262
263
264
265
    libgraph_tool_topology.\
           subgraph_isomorphism(sub._Graph__graph, g._Graph__graph,
                                _prop("v", sub, vlabels[0]),
                                _prop("v", g, vlabels[1]),
                                _prop("e", sub, elabels[0]),
                                _prop("e", g, elabels[1]),
266
                                vmaps, emaps, max_n, seed)
267
    for i in range(len(vmaps)):
268
269
270
271
        vmaps[i] = PropertyMap(vmaps[i], sub, "v")
        emaps[i] = PropertyMap(emaps[i], sub, "e")
    return vmaps, emaps

Tiago Peixoto's avatar
Tiago Peixoto committed
272

273
274
275
276
277
278
279
280
281
282
def mark_subgraph(g, sub, vmap, emap, vmask=None, emask=None):
    r"""
    Mark a given subgraph `sub` on the graph `g`.

    The mapping must be provided by the `vmap` and `emap` parameters,
    which map vertices/edges of `sub` to indexes of the corresponding
    vertices/edges in `g`.

    This returns a vertex and an edge property map, with value type 'bool',
    indicating whether or not a vertex/edge in `g` corresponds to the subgraph
283
    `sub`.
284
    """
285
    if vmask is None:
286
        vmask = g.new_vertex_property("bool")
287
    if emask is None:
288
289
290
291
292
293
294
295
296
297
298
299
300
301
        emask = g.new_edge_property("bool")

    vmask.a = False
    emask.a = False

    for v in sub.vertices():
        w = g.vertex(vmap[v])
        vmask[w] = True
        for ew in w.out_edges():
            for ev in v.out_edges():
                if emap[ev] == g.edge_index[ew]:
                    emask[ew] = True
                    break
    return vmask, emask
302

Tiago Peixoto's avatar
Tiago Peixoto committed
303

304
def min_spanning_tree(g, weights=None, root=None, tree_map=None):
305
306
307
308
309
310
311
    """
    Return the minimum spanning tree of a given graph.

    Parameters
    ----------
    g : :class:`~graph_tool.Graph`
        Graph to be used.
312
    weights : :class:`~graph_tool.PropertyMap` (optional, default: `None`)
313
314
        The edge weights. If provided, the minimum spanning tree will minimize
        the edge weights.
315
    root : :class:`~graph_tool.Vertex` (optional, default: `None`)
316
        Root of the minimum spanning tree. If this is provided, Prim's algorithm
317
        is used. Otherwise, Kruskal's algorithm is used.
318
    tree_map : :class:`~graph_tool.PropertyMap` (optional, default: `None`)
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
        If provided, the edge tree map will be written in this property map.

    Returns
    -------
    tree_map : :class:`~graph_tool.PropertyMap`
        Edge property map with mark the tree edges: 1 for tree edge, 0
        otherwise.

    Notes
    -----
    The algorithm runs with :math:`O(E\log E)` complexity, or :math:`O(E\log V)`
    if `root` is specified.

    Examples
    --------
Tiago Peixoto's avatar
Tiago Peixoto committed
334
    >>> from numpy.random import seed, random
335
    >>> seed(42)
336
337
338
    >>> g, pos = gt.triangulation(random((400, 2)) * 10, type="delaunay")
    >>> weight = g.new_edge_property("double")
    >>> for e in g.edges():
Tiago Peixoto's avatar
Tiago Peixoto committed
339
    ...    weight[e] = linalg.norm(pos[e.target()].a - pos[e.source()].a)
340
    >>> tree = gt.min_spanning_tree(g, weights=weight)
341
    >>> gt.graph_draw(g, pos=pos, output="triang_orig.pdf")
342
343
    <...>
    >>> g.set_edge_filter(tree)
344
    >>> gt.graph_draw(g, pos=pos, output="triang_min_span_tree.pdf")
345
346
347
    <...>


348
    .. image:: triang_orig.*
Tiago Peixoto's avatar
Tiago Peixoto committed
349
        :width: 400px
350
    .. image:: triang_min_span_tree.*
Tiago Peixoto's avatar
Tiago Peixoto committed
351
        :width: 400px
352
353

    *Left:* Original graph, *Right:* The minimum spanning tree.
354
355
356
357
358

    References
    ----------
    .. [kruskal-shortest-1956] J. B. Kruskal.  "On the shortest spanning subtree
       of a graph and the traveling salesman problem",  In Proceedings of the
Tiago Peixoto's avatar
Tiago Peixoto committed
359
360
       American Mathematical Society, volume 7, pages 48-50, 1956.
       :doi:`10.1090/S0002-9939-1956-0078686-7`
361
362
363
364
365
    .. [prim-shortest-1957] R. Prim.  "Shortest connection networks and some
       generalizations",  Bell System Technical Journal, 36:1389-1401, 1957.
    .. [boost-mst] http://www.boost.org/libs/graph/doc/graph_theory_review.html#sec:minimum-spanning-tree
    .. [mst-wiki] http://en.wikipedia.org/wiki/Minimum_spanning_tree
    """
366
    if tree_map is None:
367
368
369
370
        tree_map = g.new_edge_property("bool")
    if tree_map.value_type() != "bool":
        raise ValueError("edge property 'tree_map' must be of value type bool.")

371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
    try:
        g.stash_filter(directed=True)
        g.set_directed(False)
        if root is None:
            libgraph_tool_topology.\
                   get_kruskal_spanning_tree(g._Graph__graph,
                                             _prop("e", g, weights),
                                             _prop("e", g, tree_map))
        else:
            libgraph_tool_topology.\
                   get_prim_spanning_tree(g._Graph__graph, int(root),
                                          _prop("e", g, weights),
                                          _prop("e", g, tree_map))
    finally:
        g.pop_filter(directed=True)
386
    return tree_map
387

Tiago Peixoto's avatar
Tiago Peixoto committed
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
def random_spanning_tree(g, weights=None, root=None, tree_map=None):
    """
    Return a random spanning tree of a given graph, which can be directed or
    undirected.

    Parameters
    ----------
    g : :class:`~graph_tool.Graph`
        Graph to be used.
    weights : :class:`~graph_tool.PropertyMap` (optional, default: `None`)
        The edge weights. If provided, the probability of a particular spanning
        tree being selected is the product of its edge weights.
    root : :class:`~graph_tool.Vertex` (optional, default: `None`)
        Root of the spanning tree. If not provided, it will be selected randomly.
    tree_map : :class:`~graph_tool.PropertyMap` (optional, default: `None`)
        If provided, the edge tree map will be written in this property map.

    Returns
    -------
    tree_map : :class:`~graph_tool.PropertyMap`
        Edge property map with mark the tree edges: 1 for tree edge, 0
        otherwise.

    Notes
    -----
    The typical running time for random graphs is :math:`O(N\log N)`.

    Examples
    --------
    >>> from numpy.random import seed, random
    >>> seed(42)
    >>> g, pos = gt.triangulation(random((400, 2)) * 10, type="delaunay")
    >>> weight = g.new_edge_property("double")
    >>> for e in g.edges():
    ...    weight[e] = linalg.norm(pos[e.target()].a - pos[e.source()].a)
    >>> tree = gt.random_spanning_tree(g, weights=weight)
    >>> gt.graph_draw(g, pos=pos, output="rtriang_orig.pdf")
    <...>
    >>> g.set_edge_filter(tree)
Tiago Peixoto's avatar
Tiago Peixoto committed
428
    >>> gt.graph_draw(g, pos=pos, output="triang_random_span_tree.pdf")
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
    <...>


    .. image:: rtriang_orig.*
        :width: 400px
    .. image:: triang_random_span_tree.*
        :width: 400px

    *Left:* Original graph, *Right:* A random spanning tree.

    References
    ----------

    .. [wilson-generating-1996] David Bruce Wilson, "Generating random spanning
       trees more quickly than the cover time", Proceedings of the twenty-eighth
       annual ACM symposium on Theory of computing, Pages 296-303, ACM New York,
       1996, :doi:`10.1145/237814.237880`
    .. [boost-rst] http://www.boost.org/libs/graph/doc/random_spanning_tree.html
    """
    if tree_map is None:
        tree_map = g.new_edge_property("bool")
    if tree_map.value_type() != "bool":
        raise ValueError("edge property 'tree_map' must be of value type bool.")

    if root is None:
        root = g.vertex(numpy.random.randint(0, g.num_vertices()),
                        use_index=False)

    # we need to restrict ourselves to the in-component of root
    l = label_out_component(GraphView(g, reversed=True), root)
    g = GraphView(g, vfilt=l)

    seed = numpy.random.randint(0, sys.maxsize)
    libgraph_tool_topology.\
        random_spanning_tree(g._Graph__graph, int(root),
                             _prop("e", g, weights),
                             _prop("e", g, tree_map), seed)
    return tree_map


Tiago Peixoto's avatar
Tiago Peixoto committed
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
def dominator_tree(g, root, dom_map=None):
    """Return a vertex property map the dominator vertices for each vertex.

    Parameters
    ----------
    g : :class:`~graph_tool.Graph`
        Graph to be used.
    root : :class:`~graph_tool.Vertex`
        The root vertex.
    dom_map : :class:`~graph_tool.PropertyMap` (optional, default: None)
        If provided, the dominator map will be written in this property map.

    Returns
    -------
    dom_map : :class:`~graph_tool.PropertyMap`
        The dominator map. It contains for each vertex, the index of its
        dominator vertex.

    Notes
    -----
    A vertex u dominates a vertex v, if every path of directed graph from the
    entry to v must go through u.

    The algorithm runs with :math:`O((V+E)\log (V+E))` complexity.

    Examples
    --------
    >>> from numpy.random import seed
    >>> seed(42)
    >>> g = gt.random_graph(100, lambda: (2, 2))
    >>> tree = gt.min_spanning_tree(g)
    >>> g.set_edge_filter(tree)
501
    >>> root = [v for v in g.vertices() if v.in_degree() == 0]
Tiago Peixoto's avatar
Tiago Peixoto committed
502
    >>> dom = gt.dominator_tree(g, root[0])
503
    >>> print(dom.a)
Tiago Peixoto's avatar
Tiago Peixoto committed
504
505
506
    [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 5 0 0 0 0 0 0 0 0 0 0
     0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
     0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
Tiago Peixoto's avatar
Tiago Peixoto committed
507
508
509

    References
    ----------
510
    .. [dominator-bgl] http://www.boost.org/libs/graph/doc/lengauer_tarjan_dominator.htm
Tiago Peixoto's avatar
Tiago Peixoto committed
511
512

    """
513
    if dom_map is None:
Tiago Peixoto's avatar
Tiago Peixoto committed
514
515
516
        dom_map = g.new_vertex_property("int32_t")
    if dom_map.value_type() != "int32_t":
        raise ValueError("vertex property 'dom_map' must be of value type" +
517
518
                         " int32_t.")
    if not g.is_directed():
Tiago Peixoto's avatar
Tiago Peixoto committed
519
        raise ValueError("dominator tree requires a directed graph.")
520
    libgraph_tool_topology.\
Tiago Peixoto's avatar
Tiago Peixoto committed
521
522
523
               dominator_tree(g._Graph__graph, int(root),
                              _prop("v", g, dom_map))
    return dom_map
524

Tiago Peixoto's avatar
Tiago Peixoto committed
525

526
def topological_sort(g):
Tiago Peixoto's avatar
Tiago Peixoto committed
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
    """
    Return the topological sort of the given graph. It is returned as an array
    of vertex indexes, in the sort order.

    Notes
    -----
    The topological sort algorithm creates a linear ordering of the vertices
    such that if edge (u,v) appears in the graph, then v comes before u in the
    ordering. The graph must be a directed acyclic graph (DAG).

    The time complexity is :math:`O(V + E)`.

    Examples
    --------
    >>> from numpy.random import seed
    >>> seed(42)
    >>> g = gt.random_graph(30, lambda: (3, 3))
    >>> tree = gt.min_spanning_tree(g)
    >>> g.set_edge_filter(tree)
    >>> sort = gt.topological_sort(g)
547
    >>> print(sort)
Tiago Peixoto's avatar
Tiago Peixoto committed
548
549
    [ 3 20  9 29 15  0 10 23  1  2 21  7  4 12 11  5 26 27  6  8 13 14 22 16 17
     28 18 19 24 25]
Tiago Peixoto's avatar
Tiago Peixoto committed
550
551
552

    References
    ----------
553
    .. [topological-boost] http://www.boost.org/libs/graph/doc/topological_sort.html
Tiago Peixoto's avatar
Tiago Peixoto committed
554
555
556
557
    .. [topological-wiki] http://en.wikipedia.org/wiki/Topological_sorting

    """

558
    topological_order = Vector_int32_t()
Tiago Peixoto's avatar
Tiago Peixoto committed
559
560
561
562
563
    is_DAG = libgraph_tool_topology.\
        topological_sort(g._Graph__graph, topological_order)
    if not is_DAG:
        raise ValueError("Graph is not a directed acylic graph (DAG).");
    return topological_order.a.copy()
564

Tiago Peixoto's avatar
Tiago Peixoto committed
565

566
def transitive_closure(g):
Tiago Peixoto's avatar
Tiago Peixoto committed
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
    """Return the transitive closure graph of g.

    Notes
    -----
    The transitive closure of a graph G = (V,E) is a graph G* = (V,E*) such that
    E* contains an edge (u,v) if and only if G contains a path (of at least one
    edge) from u to v. The transitive_closure() function transforms the input
    graph g into the transitive closure graph tc.

    The time complexity (worst-case) is :math:`O(VE)`.

    Examples
    --------
    >>> from numpy.random import seed
    >>> seed(42)
    >>> g = gt.random_graph(30, lambda: (3, 3))
    >>> tc = gt.transitive_closure(g)

    References
    ----------
587
    .. [transitive-boost] http://www.boost.org/libs/graph/doc/transitive_closure.html
Tiago Peixoto's avatar
Tiago Peixoto committed
588
589
590
591
    .. [transitive-wiki] http://en.wikipedia.org/wiki/Transitive_closure

    """

592
593
594
595
596
597
598
    if not g.is_directed():
        raise ValueError("graph must be directed for transitive closure.")
    tg = Graph()
    libgraph_tool_topology.transitive_closure(g._Graph__graph,
                                              tg._Graph__graph)
    return tg

Tiago Peixoto's avatar
Tiago Peixoto committed
599

600
601
def label_components(g, vprop=None, directed=None):
    """
602
    Label the components to which each vertex in the graph belongs. If the
603
604
    graph is directed, it finds the strongly connected components.

605
606
607
    A property map with the component labels is returned, together with an
    histogram of component labels.

608
609
    Parameters
    ----------
610
    g : :class:`~graph_tool.Graph`
611
        Graph to be used.
612
    vprop : :class:`~graph_tool.PropertyMap` (optional, default: None)
613
614
615
616
617
618
619
620
        Vertex property to store the component labels. If none is supplied, one
        is created.
    directed : bool (optional, default:None)
        Treat graph as directed or not, independently of its actual
        directionality.

    Returns
    -------
621
    comp : :class:`~graph_tool.PropertyMap`
622
        Vertex property map with component labels.
623
624
    hist : :class:`~numpy.ndarray`
        Histogram of component labels.
625
626
627
628
629
630

    Notes
    -----
    The components are arbitrarily labeled from 0 to N-1, where N is the total
    number of components.

631
    The algorithm runs in :math:`O(V + E)` time.
632
633
634

    Examples
    --------
635
636
637
    >>> from numpy.random import seed
    >>> seed(43)
    >>> g = gt.random_graph(100, lambda: (1, 1))
638
    >>> comp, hist = gt.label_components(g)
639
    >>> print(comp.a)
Tiago Peixoto's avatar
Tiago Peixoto committed
640
641
642
    [0 0 0 1 0 2 0 0 0 0 2 0 0 0 2 1 0 2 0 1 2 0 1 0 0 1 0 2 0 2 1 0 2 0 0 0 0
     0 0 1 0 0 2 2 2 0 0 0 0 0 0 2 0 0 1 1 0 0 2 0 1 0 0 0 2 0 0 2 2 1 2 1 0 0
     2 0 0 1 2 1 2 2 0 0 0 0 0 2 0 0 0 1 1 0 0 0 1 1 2 2]
643
    >>> print(hist)
Tiago Peixoto's avatar
Tiago Peixoto committed
644
    [58 18 24]
645
646
    """

647
    if vprop is None:
648
649
650
651
652
        vprop = g.new_vertex_property("int32_t")

    _check_prop_writable(vprop, name="vprop")
    _check_prop_scalar(vprop, name="vprop")

653
654
    if directed is not None:
        g = GraphView(g, directed=directed)
655

656
657
658
659
660
661
662
    hist = libgraph_tool_topology.\
               label_components(g._Graph__graph, _prop("v", g, vprop))
    return vprop, hist


def label_largest_component(g, directed=None):
    """
663
664
    Label the largest component in the graph. If the graph is directed, then the
    largest strongly connected component is labelled.
665
666
667
668
669
670
671
672
673
674
675
676
677
678

    A property map with a boolean label is returned.

    Parameters
    ----------
    g : :class:`~graph_tool.Graph`
        Graph to be used.
    directed : bool (optional, default:None)
        Treat graph as directed or not, independently of its actual
        directionality.

    Returns
    -------
    comp : :class:`~graph_tool.PropertyMap`
679
         Boolean vertex property map which labels the largest component.
680
681
682
683
684
685
686
687
688
689
690

    Notes
    -----
    The algorithm runs in :math:`O(V + E)` time.

    Examples
    --------
    >>> from numpy.random import seed, poisson
    >>> seed(43)
    >>> g = gt.random_graph(100, lambda: poisson(1), directed=False)
    >>> l = gt.label_largest_component(g)
691
    >>> print(l.a)
Tiago Peixoto's avatar
Tiago Peixoto committed
692
693
694
    [1 0 0 0 0 0 0 0 1 1 0 1 0 0 0 0 0 0 1 0 0 1 1 1 1 0 0 0 1 0 0 0 0 0 0 0 1
     1 1 0 0 0 0 1 0 1 1 0 0 0 1 1 0 0 1 1 0 0 0 0 0 0 1 0 1 0 0 0 0 0 0 0 0 0
     0 0 0 1 1 0 1 1 0 0 0 0 0 1 1 0 1 0 1 0 1 0 0 0 0 0]
695
    >>> u = gt.GraphView(g, vfilt=l)   # extract the largest component as a graph
696
    >>> print(u.num_vertices())
Tiago Peixoto's avatar
Tiago Peixoto committed
697
    31
698
699
700
701
    """

    label = g.new_vertex_property("bool")
    c, h = label_components(g, directed=directed)
702
703
704
705
706
    vfilt, inv = g.get_vertex_filter()
    if vfilt is None:
        label.a = c.a == h.argmax()
    else:
        label.a = (c.a == h.argmax()) & (vfilt.a ^ inv)
707
    return label
708

Tiago Peixoto's avatar
Tiago Peixoto committed
709

710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
def label_out_component(g, root):
    """
    Label the out-component (or simply the component for undirected graphs) of a
    root vertex.

    Parameters
    ----------
    g : :class:`~graph_tool.Graph`
        Graph to be used.
    root : :class:`~graph_tool.Vertex`
        The root vertex.

    Returns
    -------
    comp : :class:`~graph_tool.PropertyMap`
         Boolean vertex property map which labels the out-component.

    Notes
    -----
    The algorithm runs in :math:`O(V + E)` time.

    Examples
    --------
    >>> from numpy.random import seed, poisson
    >>> seed(43)
    >>> g = gt.random_graph(100, lambda: poisson(1), directed=False)
    >>> l = gt.label_out_component(g, g.vertex(0))
    >>> print(l.a)
    [1 0 0 0 0 0 0 0 1 1 0 1 0 0 0 0 0 0 1 0 0 1 1 1 1 0 0 0 1 0 0 0 0 0 0 0 1
     1 1 0 0 0 0 1 0 1 1 0 0 0 1 1 0 0 1 1 0 0 0 0 0 0 1 0 1 0 0 0 0 0 0 0 0 0
     0 0 0 1 1 0 1 1 0 0 0 0 0 1 1 0 1 0 1 0 1 0 0 0 0 0]

    The in-component can be obtained by reversing the graph.

Tiago Peixoto's avatar
Tiago Peixoto committed
744
745
    >>> l = gt.label_out_component(gt.GraphView(g, reversed=True, directed=True),
    ...                            g.vertex(0))
746
747
748
749
750
751
752
753
754
755
756
757
758
    >>> print(l.a)
    [1 0 0 0 0 0 0 0 1 1 0 1 0 0 0 0 0 0 1 0 0 1 1 1 1 0 0 0 1 0 0 0 0 0 0 0 1
     1 1 0 0 0 0 1 0 1 1 0 0 0 1 1 0 0 1 1 0 0 0 0 0 0 1 0 1 0 0 0 0 0 0 0 0 0
     0 0 0 1 1 0 1 1 0 0 0 0 0 1 1 0 1 0 1 0 1 0 0 0 0 0]
    """

    label = g.new_vertex_property("bool")
    libgraph_tool_topology.\
             label_out_component(g._Graph__graph, int(root),
                                 _prop("v", g, label))
    return label


759
def label_biconnected_components(g, eprop=None, vprop=None):
760
761
762
763
    """
    Label the edges of biconnected components, and the vertices which are
    articulation points.

764
765
766
767
    An edge property map with the component labels is returned, together a
    boolean vertex map marking the articulation points, and an histogram of
    component labels.

768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
    Parameters
    ----------
    g : :class:`~graph_tool.Graph`
        Graph to be used.

    eprop : :class:`~graph_tool.PropertyMap` (optional, default: None)
        Edge property to label the biconnected components.

    vprop : :class:`~graph_tool.PropertyMap` (optional, default: None)
        Vertex property to mark the articulation points. If none is supplied,
        one is created.


    Returns
    -------
    bicomp : :class:`~graph_tool.PropertyMap`
        Edge property map with the biconnected component labels.
    articulation : :class:`~graph_tool.PropertyMap`
        Boolean vertex property map which has value 1 for each vertex which is
        an articulation point, and zero otherwise.
    nc : int
        Number of biconnected components.

    Notes
    -----

    A connected graph is biconnected if the removal of any single vertex (and
    all edges incident on that vertex) can not disconnect the graph. More
    generally, the biconnected components of a graph are the maximal subsets of
    vertices such that the removal of a vertex from a particular component will
    not disconnect the component. Unlike connected components, vertices may
    belong to multiple biconnected components: those vertices that belong to
    more than one biconnected component are called "articulation points" or,
    equivalently, "cut vertices". Articulation points are vertices whose removal
    would increase the number of connected components in the graph. Thus, a
    graph without articulation points is biconnected. Vertices can be present in
    multiple biconnected components, but each edge can only be contained in a
    single biconnected component.

    The algorithm runs in :math:`O(V + E)` time.

    Examples
    --------
    >>> from numpy.random import seed
Tiago Peixoto's avatar
Tiago Peixoto committed
812
    >>> seed(43)
813
    >>> g = gt.random_graph(100, lambda: 2, directed=False)
814
    >>> comp, art, hist = gt.label_biconnected_components(g)
815
    >>> print(comp.a)
Tiago Peixoto's avatar
Tiago Peixoto committed
816
817
818
    [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0
     0 0 1 0 0 1 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 1 0 0 0 0 1
     0 0 0 0 0 0 0 1 0 1 0 0 0 0 0 0 0 0 1 0 1 0 0 0 0 0]
819
    >>> print(art.a)
820
821
822
    [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
     0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
     0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
823
    >>> print(hist)
Tiago Peixoto's avatar
Tiago Peixoto committed
824
    [87 13]
825
    """
826

827
    if vprop is None:
828
        vprop = g.new_vertex_property("bool")
829
    if eprop is None:
830
831
832
833
834
835
836
        eprop = g.new_edge_property("int32_t")

    _check_prop_writable(vprop, name="vprop")
    _check_prop_scalar(vprop, name="vprop")
    _check_prop_writable(eprop, name="eprop")
    _check_prop_scalar(eprop, name="eprop")

837
838
    g = GraphView(g, directed=False)
    hist = libgraph_tool_topology.\
839
840
             label_biconnected_components(g._Graph__graph, _prop("e", g, eprop),
                                          _prop("v", g, vprop))
841
    return eprop, vprop, hist
842

Tiago Peixoto's avatar
Tiago Peixoto committed
843

844
def shortest_distance(g, source=None, weights=None, max_dist=None,
845
846
                      directed=None, dense=False, dist_map=None,
                      pred_map=False):
847
848
849
850
851
852
853
854
855
    """
    Calculate the distance of all vertices from a given source, or the all pairs
    shortest paths, if the source is not specified.

    Parameters
    ----------
    g : :class:`~graph_tool.Graph`
        Graph to be used.
    source : :class:`~graph_tool.Vertex` (optional, default: None)
856
        Source vertex of the search. If unspecified, the all pairs shortest
857
858
859
860
861
862
        distances are computed.
    weights : :class:`~graph_tool.PropertyMap` (optional, default: None)
        The edge weights. If provided, the minimum spanning tree will minimize
        the edge weights.
    max_dist : scalar value (optional, default: None)
        If specified, this limits the maximum distance of the vertices
863
        are searched. This parameter has no effect if source is None.
864
865
866
867
    directed : bool (optional, default:None)
        Treat graph as directed or not, independently of its actual
        directionality.
    dense : bool (optional, default: False)
868
869
        If true, and source is None, the Floyd-Warshall algorithm is used,
        otherwise the Johnson algorithm is used. If source is not None, this option
870
871
872
873
        has no effect.
    dist_map : :class:`~graph_tool.PropertyMap` (optional, default: None)
        Vertex property to store the distances. If none is supplied, one
        is created.
874
875
876
    pred_map : bool (optional, default: False)
        If true, a vertex property map with the predecessors is returned.
        Ignored if source=None.
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902

    Returns
    -------
    dist_map : :class:`~graph_tool.PropertyMap`
        Vertex property map with the distances from source. If source is 'None',
        it will have a vector value type, with the distances to every vertex.

    Notes
    -----

    If a source is given, the distances are calculated with a breadth-first
    search (BFS) or Dijkstra's algorithm [dijkstra]_, if weights are given. If
    source is not given, the distances are calculated with Johnson's algorithm
    [johnson-apsp]_. If dense=True, the Floyd-Warshall algorithm
    [floyd-warshall-apsp]_ is used instead.

    If source is specified, the algorithm runs in :math:`O(V + E)` time, or
    :math:`O(V \log V)` if weights are given. If source is not specified, it
    runs in :math:`O(VE\log V)` time, or :math:`O(V^3)` if dense == True.

    Examples
    --------
    >>> from numpy.random import seed, poisson
    >>> seed(42)
    >>> g = gt.random_graph(100, lambda: (poisson(3), poisson(3)))
    >>> dist = gt.shortest_distance(g, source=g.vertex(0))
903
    >>> print(dist.a)
Tiago Peixoto's avatar
Tiago Peixoto committed
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
    [         0          3          6          4 2147483647          3
              4          3          4          2          3          4
              3          4          2          4          2          5
              4          4 2147483647          4 2147483647          6
              4          7          5 2147483647          3          4
              2          3          5          5          4          5
              1          5          6          1 2147483647          8
              4          2          1          5          5          6
              7          4          5          3          4          4
              5          3          3          5          4          5
              4          3          5          4          2 2147483647
              6          5          4          5          1 2147483647
              5          5          4          2          5          4
              6          3          5          3          4 2147483647
              4          4          7          4          3          5
              5          2          7          3          4          4
              4          3          4          4]
921
    >>> dist = gt.shortest_distance(g)
922
    >>> print(dist[g.vertex(0)].a)
Tiago Peixoto's avatar
Tiago Peixoto committed
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
    [         0          3          6          4 2147483647          3
              4          3          4          2          3          4
              3          4          2          4          2          5
              4          4 2147483647          4 2147483647          6
              4          7          5 2147483647          3          4
              2          3          5          5          4          5
              1          5          6          1 2147483647          8
              4          2          1          5          5          6
              7          4          5          3          4          4
              5          3          3          5          4          5
              4          3          5          4          2 2147483647
              6          5          4          5          1 2147483647
              5          5          4          2          5          4
              6          3          5          3          4 2147483647
              4          4          7          4          3          5
              5          2          7          3          4          4
              4          3          4          4]
940
941
942
943
944

    References
    ----------
    .. [bfs] Edward Moore, "The shortest path through a maze", International
       Symposium on the Theory of Switching (1959), Harvard University
Tiago Peixoto's avatar
Tiago Peixoto committed
945
946
       Press;
    .. [bfs-boost] http://www.boost.org/libs/graph/doc/breadth_first_search.html
947
948
    .. [dijkstra] E. Dijkstra, "A note on two problems in connexion with
       graphs." Numerische Mathematik, 1:269-271, 1959.
Tiago Peixoto's avatar
Tiago Peixoto committed
949
    .. [dijkstra-boost] http://www.boost.org/libs/graph/doc/dijkstra_shortest_paths.html
950
951
952
953
    .. [johnson-apsp] http://www.boost.org/libs/graph/doc/johnson_all_pairs_shortest.html
    .. [floyd-warshall-apsp] http://www.boost.org/libs/graph/doc/floyd_warshall_shortest.html
    """

954
    if weights is None:
955
956
957
958
        dist_type = 'int32_t'
    else:
        dist_type = weights.value_type()

959
960
    if dist_map is None:
        if source is not None:
961
962
963
964
965
            dist_map = g.new_vertex_property(dist_type)
        else:
            dist_map = g.new_vertex_property("vector<%s>" % dist_type)

    _check_prop_writable(dist_map, name="dist_map")
966
    if source is not None:
967
968
969
970
        _check_prop_scalar(dist_map, name="dist_map")
    else:
        _check_prop_vector(dist_map, name="dist_map")

971
    if max_dist is None:
972
973
        max_dist = 0

974
    if directed is not None:
975
976
977
978
        g.stash_filter(directed=True)
        g.set_directed(directed)

    try:
979
        if source is not None:
980
            pmap = g.copy_property(g.vertex_index, value_type="int64_t")
981
982
983
            libgraph_tool_topology.get_dists(g._Graph__graph, int(source),
                                             _prop("v", g, dist_map),
                                             _prop("e", g, weights),
984
                                             _prop("v", g, pmap),
985
986
987
988
989
990
991
                                             float(max_dist))
        else:
            libgraph_tool_topology.get_all_dists(g._Graph__graph,
                                                 _prop("v", g, dist_map),
                                                 _prop("e", g, weights), dense)

    finally:
992
        if directed is not None:
993
            g.pop_filter(directed=True)
994
    if source is not None and pred_map:
995
996
997
998
        return dist_map, pmap
    else:
        return dist_map

Tiago Peixoto's avatar
Tiago Peixoto committed
999

1000
def shortest_path(g, source, target, weights=None, pred_map=None):