__init__.py 84.5 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) 2006-2016 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
34
   all_shortest_paths
   all_predecessors
Tiago Peixoto's avatar
Tiago Peixoto committed
35
   all_paths
Tiago Peixoto's avatar
Tiago Peixoto committed
36
   pseudo_diameter
37
   similarity
38
   vertex_similarity
39
   isomorphism
40
41
   subgraph_isomorphism
   mark_subgraph
42
43
   max_cardinality_matching
   max_independent_vertex_set
44
   min_spanning_tree
45
   random_spanning_tree
46
47
48
   dominator_tree
   topological_sort
   transitive_closure
Tiago Peixoto's avatar
Tiago Peixoto committed
49
   tsp_tour
50
   sequential_vertex_coloring
51
52
   label_components
   label_biconnected_components
53
   label_largest_component
54
   label_out_component
Tiago Peixoto's avatar
Tiago Peixoto committed
55
   kcore_decomposition
56
   is_bipartite
Tiago Peixoto's avatar
Tiago Peixoto committed
57
   is_DAG
58
   is_planar
59
   make_maximal_planar
Tiago Peixoto's avatar
Tiago Peixoto committed
60
   edge_reciprocity
61
62
63

Contents
++++++++
64

65
66
"""

67
68
from __future__ import division, absolute_import, print_function

Tiago Peixoto's avatar
Tiago Peixoto committed
69
from .. dl_import import dl_import
70
dl_import("from . import libgraph_tool_topology")
71

72
from .. import _prop, Vector_int32_t, _check_prop_writable, \
73
     _check_prop_scalar, _check_prop_vector, Graph, PropertyMap, GraphView,\
74
     libcore, _get_rng, _degree, perfect_prop_hash, _limit_args
75
from .. stats import label_self_loops
76
import random, sys, numpy, collections
77

78
__all__ = ["isomorphism", "subgraph_isomorphism", "mark_subgraph",
79
           "max_cardinality_matching", "max_independent_vertex_set",
80
           "min_spanning_tree", "random_spanning_tree", "dominator_tree",
Tiago Peixoto's avatar
Tiago Peixoto committed
81
           "topological_sort", "transitive_closure", "tsp_tour",
82
83
           "sequential_vertex_coloring", "label_components",
           "label_largest_component", "label_biconnected_components",
Tiago Peixoto's avatar
Tiago Peixoto committed
84
           "label_out_component", "kcore_decomposition", "shortest_distance",
Tiago Peixoto's avatar
Tiago Peixoto committed
85
           "shortest_path", "all_shortest_paths", "all_predecessors",
Tiago Peixoto's avatar
Tiago Peixoto committed
86
           "all_paths", "pseudo_diameter", "is_bipartite", "is_DAG",
87
88
           "is_planar", "make_maximal_planar", "similarity", "vertex_similarity",
           "edge_reciprocity"]
89
90
91
92
93
94
95
96
97

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
98
        Second graph to be compared.
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
    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
    --------
125
126
127
128
129
130
131
    .. testcode::
       :hide:

       import numpy.random
       numpy.random.seed(42)
       gt.seed_rng(42)

132
133
134
135
    >>> g = gt.random_graph(100, lambda: (3,3))
    >>> u = g.copy()
    >>> gt.similarity(u, g)
    1.0
Tiago Peixoto's avatar
Tiago Peixoto committed
136
    >>> gt.random_rewire(u)
137
    24
138
    >>> gt.similarity(u, g)
139
    0.04666666666666667
140
141
142
143
144
145
146
    """

    if label1 is None:
        label1 = g1.vertex_index
    if label2 is None:
        label2 = g2.vertex_index
    if label1.value_type() != label2.value_type():
147
148
149
150
        try:
            label2 = label2.copy(label1.value_type())
        except ValueError:
            label1 = label1.copy(label2.value_type())
Tiago Peixoto's avatar
Tiago Peixoto committed
151
    if label1.is_writable() or label2.is_writable():
152
153
154
        s = libgraph_tool_topology.\
               similarity(g1._Graph__graph, g2._Graph__graph,
                          _prop("v", g1, label1), _prop("v", g2, label2))
Tiago Peixoto's avatar
Tiago Peixoto committed
155
156
157
158
    else:
        s = libgraph_tool_topology.\
               similarity_fast(g1._Graph__graph, g2._Graph__graph,
                               _prop("v", g1, label1), _prop("v", g2, label2))
159
160
161
162
163
    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
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
@_limit_args({"sim_type": ["dice", "jaccard", "inv-log-weight"]})
def vertex_similarity(g, sim_type="jaccard", vertex_pairs=None, self_loops=True,
                      sim_map=None):
    r"""Return the similarity between pairs of vertices.

    Parameters
    ----------
    g : :class:`~graph_tool.Graph`
        The graph to be used.
    sim_type : ``str`` (optional, default: ``"jaccard"``)
        Type of similarity to use. This must be one of ``"dice"``, ``"jaccard"``
        or ``"inv-log-weight"``.
    vertex_pairs : iterable of pairs of integers (optional, default: ``None``)
        Pairs of vertices to compute the similarity. If omitted, all pairs will
        be considered.
    self_loops : bool (optional, default: ``True``)
        If ``True``, vertices will be considered adjacent to themselves for the
        purpose of the similarity computation.
    sim_map : :class:`~graph_tool.PropertyMap` (optional, default: ``None``)
        If provided, and ``vertex_pairs == None``, the vertex similarities will
        be stored in this vector-valued property. Otherwise, a new one will be
        created.

    Returns
    -------
    similarities : :class:`numpy.ndarray` or :class:`~graph_tool.PropertyMap`
        If ``vertex_pairs`` was supplied, this will be a :class:`numpy.ndarray`
        with the corresponding similarities, otherwise it will be a
        vector-valued vertex :class:`~graph_tool.PropertyMap`, with the
        similarities to all other vertices.

    Notes
    -----
    According to ``sim_type``, this function computes the following similarities:

    ``sim_type == "dice"``

       The Sørensen–Dice similarity [sorensen-dice]_ is twice the number of
203
204
       common neighbours between two vertices divided by the sum of their
       degrees.
205
206
207
208

    ``sim_type == "jaccard"``

       The Jaccard similarity [jaccard]_ is the number of common neighbours
209
       between two vertices divided by the size of the set of all neighbours to
210
211
212
213
214
215
216
       both vertices.

    ``sim_type == "inv-log-weight"``

       The inverse log weighted similarity [adamic-friends-2003]_ is the sum of
       the weights of common neighbours between two vertices, where the weights
       are computed as :math:`1/\log(k)`, with :math:`k` being the degree of the
217
       vertex.
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


    For directed graphs, only out-neighbours are considered in the above
    algorthms (for "inv-log-weight", the in-degrees are used to compute the
    weights). To use the in-neighbours instead, a :class:`~graph_tool.GraphView`
    should be used to reverse the graph, e.g. ``vertex_similarity(GraphView(g,
    reversed=True))``.

    The algorithm runs with complexity :math:`O(\left<k\right>N^2)` if
    ``vertex_pairs == None``, otherwise with :math:`O(\left<k\right>P)` where
    :math:`P` is the length of ``vertex_pairs``.

    If enabled during compilation, this algorithm runs in parallel.

    Examples
    --------
    .. testcode::
       :hide:

       import matplotlib

    >>> g = gt.collection.data["polbooks"]
    >>> s = gt.vertex_similarity(g, "jaccard")
    >>> color = g.new_vp("double")
    >>> color.a = s[0].a
    >>> gt.graph_draw(g, pos=g.vp.pos, vertex_text=g.vertex_index,
    ...               vertex_color=color, vertex_fill_color=color,
    ...               vcmap=matplotlib.cm.inferno,
    ...               output="polbooks-jaccard.pdf")
    <...>

    .. testcode::
       :hide:

       gt.graph_draw(g, pos=g.vp.pos, vertex_text=g.vertex_index,
                     vertex_color=color, vertex_fill_color=color,
                     vcmap=matplotlib.cm.inferno,
                     output="polbooks-jaccard.png")

    .. figure:: polbooks-jaccard.*
258
       :align: center
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

       Jaccard similarities to vertex ``0`` in a political books network.

    References
    ----------
    .. [sorensen-dice] https://en.wikipedia.org/wiki/S%C3%B8rensen%E2%80%93Dice_coefficient
    .. [jaccard] https://en.wikipedia.org/wiki/Jaccard_index
    .. [adamic-friends-2003] Lada A. Adamic and Eytan Adar, "Friends and neighbors
       on the Web", Social Networks Volume 25, Issue 3, Pages 211–230 (2003)
       :doi:`10.1016/S0378-8733(03)00009-1`
    .. [liben-nowell-link-prediction-2007] David Liben-Nowell and Jon Kleinberg,
       "The link-prediction problem for social networks", Journal of the
       American Society for Information Science and Technology, Volume 58, Issue
       7, pages 1019–1031 (2007), :doi:`10.1002/asi.20591`
    """

    if vertex_pairs is None:
        if sim_map is None:
            s = g.new_vp("vector<double>")
        else:
            s = sim_map
        if sim_type == "dice":
            libgraph_tool_topology.dice_similarity(g._Graph__graph,
                                                   _prop("v", g, s),
                                                   self_loops)
        elif sim_type == "jaccard":
            libgraph_tool_topology.jaccard_similarity(g._Graph__graph,
                                                      _prop("v", g, s),
                                                      self_loops)
        elif sim_type == "inv-log-weight":
            libgraph_tool_topology.inv_log_weight_similarity(g._Graph__graph,
                                                             _prop("v", g, s))
    else:
        vertex_pairs = numpy.asarray(vertex_pairs, dtype="int64")
        s = numpy.zeros(vertex_pairs.shape[0], dtype="double")
        if sim_type == "dice":
            libgraph_tool_topology.dice_similarity_pairs(g._Graph__graph,
                                                         vertex_pairs,
                                                         s, self_loops)
        elif sim_type == "jaccard":
            libgraph_tool_topology.jaccard_similarity_pairs(g._Graph__graph,
                                                            vertex_pairs,
                                                            s, self_loops)
        elif sim_type == "inv-log-weight":
            libgraph_tool_topology.\
                inv_log_weight_similarity_pairs(g._Graph__graph, vertex_pairs,
                                                s)
    return s

Tiago Peixoto's avatar
Tiago Peixoto committed
308

309
def isomorphism(g1, g2, vertex_inv1=None, vertex_inv2=None, isomap=False):
310
311
    r"""Check whether two graphs are isomorphic.

312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
    Parameters
    ----------
    g1 : :class:`~graph_tool.Graph`
        First graph.
    g2 : :class:`~graph_tool.Graph`
        Second graph.
    vertex_inv1 : :class:`~graph_tool.PropertyMap` (optional, default: `None`)
        Vertex invariant of the first graph. Only vertices with with the same
        invariants are considered in the isomorphism.
    vertex_inv2 : :class:`~graph_tool.PropertyMap` (optional, default: `None`)
        Vertex invariant of the second graph. Only vertices with with the same
        invariants are considered in the isomorphism.
    isomap : ``bool`` (optional, default: ``False``)
        If ``True``, a vertex :class:`~graph_tool.PropertyMap` with the
        isomorphism mapping is returned as well.

    Returns
    -------
    is_isomorphism : ``bool``
        ``True`` if both graphs are isomorphic, otherwise ``False``.
    isomap : :class:`~graph_tool.PropertyMap`
         Isomorphism mapping corresponding to a property map belonging to the
         first graph which maps its vertices to their corresponding vertices of
         the second graph.
336
337
338

    Examples
    --------
339
340
341
342
343
344
345
    .. testcode::
       :hide:

       import numpy.random
       numpy.random.seed(42)
       gt.seed_rng(42)

346
347
348
349
350
351
352
353
354
    >>> 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

355
    """
356
    imap = g1.new_vertex_property("int32_t")
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
    if vertex_inv1 is None:
        vertex_inv1 = g1.degree_property_map("total").copy("int64_t")
    else:
        vertex_inv1 = vertex_inv1.copy("int64_t")
        d = g1.degree_property_map("total")
        vertex_inv1.fa += (vertex_inv1.fa.max() + 1) * d.a
    if vertex_inv2 is None:
        vertex_inv2 = g2.degree_property_map("total").copy("int64_t")
    else:
        vertex_inv2 = vertex_inv2.copy("int64_t")
        d = g2.degree_property_map("total")
        vertex_inv2.fa += (vertex_inv2.fa.max() + 1) * d.a

    inv_max = max(vertex_inv1.fa.max(),vertex_inv2.fa.max()) + 1

    l1 = label_self_loops(g1, mark_only=True)
    if l1.fa.max() > 0:
        g1 = GraphView(g1, efilt=1 - l1.fa)

    l2 = label_self_loops(g2, mark_only=True)
    if l2.fa.max() > 0:
        g2 = GraphView(g2, efilt=1 - l2.fa)

380
    iso = libgraph_tool_topology.\
381
           check_isomorphism(g1._Graph__graph, g2._Graph__graph,
382
383
384
                             _prop("v", g1, vertex_inv1),
                             _prop("v", g2, vertex_inv2),
                             inv_max,
Tiago Peixoto's avatar
Tiago Peixoto committed
385
                             _prop("v", g1, imap))
386
387
388
389
390
    if isomap:
        return iso, imap
    else:
        return iso

Tiago Peixoto's avatar
Tiago Peixoto committed
391

392
def subgraph_isomorphism(sub, g, max_n=0, vertex_label=None, edge_label=None,
393
                         induced=False, subgraph=True, generator=False):
394
    r"""Obtain all subgraph isomorphisms of `sub` in `g` (or at most `max_n` subgraphs, if `max_n > 0`).
395

396

Tiago Peixoto's avatar
Tiago Peixoto committed
397
398
399
400
401
402
    Parameters
    ----------
    sub : :class:`~graph_tool.Graph`
        Subgraph for which to be searched.
    g : :class:`~graph_tool.Graph`
        Graph in which the search is performed.
403
    max_n : int (optional, default: ``0``)
Tiago Peixoto's avatar
Tiago Peixoto committed
404
405
        Maximum number of matches to find. If `max_n == 0`, all matches are
        found.
406
    vertex_label : pair of :class:`~graph_tool.PropertyMap` (optional, default: ``None``)
407
        If provided, this should be a pair of :class:`~graph_tool.PropertyMap`
408
409
410
411
        objects, belonging to ``sub`` and ``g`` (in this order), which specify
        vertex labels which should match, in addition to the topological
        isomorphism.
    edge_label : pair of :class:`~graph_tool.PropertyMap` (optional, default: ``None``)
412
        If provided, this should be a pair of :class:`~graph_tool.PropertyMap`
413
414
415
416
417
418
419
        objects, belonging to ``sub`` and ``g`` (in this order), which specify
        edge labels which should match, in addition to the topological
        isomorphism.
    induced : bool (optional, default: ``False``)
        If ``True``, only node-induced subgraphs are found.
    subgraph : bool (optional, default: ``True``)
        If ``False``, all non-subgraph isomorphisms between `sub` and `g` are
420
        found.
421
422
423
424
    generator : bool (optional, default: ``False``)
        If ``True``, a generator will be returned, instead of a list. This is
        useful if the number of isomorphisms is too large to store in memory. If
        ``generator == True``, the option ``max_n`` is ignored.
Tiago Peixoto's avatar
Tiago Peixoto committed
425
426
427

    Returns
    -------
428
429
430
431
    vertex_maps : list (or generator) of :class:`~graph_tool.PropertyMap` objects
        List (or generator) containing vertex property map objects which
        indicate different isomorphism mappings. The property maps vertices in
        `sub` to the corresponding vertex index in `g`.
Tiago Peixoto's avatar
Tiago Peixoto committed
432
433
434

    Notes
    -----
435
436
437
438
439
    The implementation is based on the VF2 algorithm, introduced by Cordella et al.
    [cordella-improved-2001]_ [cordella-subgraph-2004]_. The spatial complexity
    is of order :math:`O(V)`, where :math:`V` is the (maximum) number of vertices
    of the two graphs. Time complexity is :math:`O(V^2)` in the best case and
    :math:`O(V!\times V)` in the worst case.
440
441
442

    Examples
    --------
443
    >>> from numpy.random import poisson
444
445
446
    >>> g = gt.complete_graph(30)
    >>> sub = gt.complete_graph(10)
    >>> vm = gt.subgraph_isomorphism(sub, g, max_n=100)
447
    >>> print(len(vm))
448
    100
449
    >>> for i in range(len(vm)):
450
451
    ...   g.set_vertex_filter(None)
    ...   g.set_edge_filter(None)
452
    ...   vmask, emask = gt.mark_subgraph(g, sub, vm[i])
453
454
    ...   g.set_vertex_filter(vmask)
    ...   g.set_edge_filter(emask)
455
    ...   assert gt.isomorphism(g, sub)
456
457
458
459
    >>> 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
460
461
462
    >>> ewidth.a *= 2
    >>> gt.graph_draw(g, vertex_fill_color=vmask, edge_color=emask,
    ...               edge_pen_width=ewidth, output_size=(200, 200),
463
    ...               output="subgraph-iso-embed.pdf")
464
    <...>
Tiago Peixoto's avatar
Tiago Peixoto committed
465
    >>> gt.graph_draw(sub, output_size=(200, 200), output="subgraph-iso.pdf")
466
467
    <...>

Tiago Peixoto's avatar
Tiago Peixoto committed
468
469
470
471
472
473
474
475
    .. testcode::
       :hide:

       gt.graph_draw(g, vertex_fill_color=vmask, edge_color=emask,
                     edge_pen_width=ewidth, output_size=(200, 200),
                     output="subgraph-iso-embed.png")
       gt.graph_draw(sub, output_size=(200, 200), output="subgraph-iso.png")

Tiago Peixoto's avatar
Tiago Peixoto committed
476
477
    .. image:: subgraph-iso.*
    .. image:: subgraph-iso-embed.*
478

479

Tiago Peixoto's avatar
Tiago Peixoto committed
480
    **Left:** Subgraph searched, **Right:** One isomorphic subgraph found in main graph.
481
482
483

    References
    ----------
484
485
486
487
488
489
    .. [cordella-improved-2001] L. P. Cordella, P. Foggia, C. Sansone, and M. Vento,
       "An improved algorithm for matching large graphs.", 3rd IAPR-TC15 Workshop
       on Graph-based Representations in Pattern Recognition, pp. 149-159, Cuen, 2001.
       http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.101.5342
    .. [cordella-subgraph-2004] L. P. Cordella, P. Foggia, C. Sansone, and M. Vento,
       "A (Sub)Graph Isomorphism Algorithm for Matching Large Graphs.",
Tiago Peixoto's avatar
Tiago Peixoto committed
490
       IEEE Trans. Pattern Anal. Mach. Intell., vol. 26, no. 10, pp. 1367-1372, 2004.
491
492
       :doi:`10.1109/TPAMI.2004.75`
    .. [boost-subgraph-iso] http://www.boost.org/libs/graph/doc/vf2_sub_graph_iso.html
493
    .. [subgraph-isormophism-wikipedia] http://en.wikipedia.org/wiki/Subgraph_isomorphism_problem
494
495

    """
496
497
    if sub.num_vertices() == 0:
        raise ValueError("Cannot search for an empty subgraph.")
498
499
500
501
    if vertex_label is None:
        vertex_label = (None, None)
    elif vertex_label[0].value_type() != vertex_label[1].value_type():
        raise ValueError("Both vertex label property maps must be of the same type!")
502
503
    elif vertex_label[0].value_type() != "int64_t":
        vertex_label = perfect_prop_hash(vertex_label, htype="int64_t")
504

505
506
507
508
    if edge_label is None:
        edge_label = (None, None)
    elif edge_label[0].value_type() != edge_label[1].value_type():
        raise ValueError("Both edge label property maps must be of the same type!")
509
510
    elif edge_label[0].value_type() != "int64_t":
        edge_label = perfect_prop_hash(edge_label, htype="int64_t")
511

512
513
514
515
516
517
518
519
520
    vmaps = libgraph_tool_topology.\
            subgraph_isomorphism(sub._Graph__graph, g._Graph__graph,
                                 _prop("v", sub, vertex_label[0]),
                                 _prop("v", g, vertex_label[1]),
                                 _prop("e", sub, edge_label[0]),
                                 _prop("e", g, edge_label[1]),
                                 max_n, induced, not subgraph,
                                 generator)
    if generator:
521
        return (PropertyMap(vmap, sub, "v") for vmap in vmaps)
522
    else:
523
        return [PropertyMap(vmap, sub, "v") for vmap in vmaps]
524

Tiago Peixoto's avatar
Tiago Peixoto committed
525

526
def mark_subgraph(g, sub, vmap, vmask=None, emask=None):
527
528
529
530
531
532
533
534
535
    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
536
    `sub`.
537
    """
538
    if vmask is None:
539
        vmask = g.new_vertex_property("bool")
540
    if emask is None:
541
542
543
544
545
546
547
548
        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
549
550
        us = set([g.vertex(vmap[x]) for x in v.out_neighbours()])

551
        for ew in w.out_edges():
552
553
554
            if ew.target() in us:
                emask[ew] = True

555
    return vmask, emask
556

Tiago Peixoto's avatar
Tiago Peixoto committed
557

558
def min_spanning_tree(g, weights=None, root=None, tree_map=None):
559
560
561
562
563
564
565
    """
    Return the minimum spanning tree of a given graph.

    Parameters
    ----------
    g : :class:`~graph_tool.Graph`
        Graph to be used.
566
    weights : :class:`~graph_tool.PropertyMap` (optional, default: `None`)
567
568
        The edge weights. If provided, the minimum spanning tree will minimize
        the edge weights.
569
    root : :class:`~graph_tool.Vertex` (optional, default: `None`)
570
        Root of the minimum spanning tree. If this is provided, Prim's algorithm
571
        is used. Otherwise, Kruskal's algorithm is used.
572
    tree_map : :class:`~graph_tool.PropertyMap` (optional, default: `None`)
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
        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
    --------
588
589
590
591
592
593
594
595
    .. testcode::
       :hide:

       import numpy.random
       numpy.random.seed(42)
       gt.seed_rng(42)

    >>> from numpy.random import random
596
597
598
    >>> 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
599
    ...    weight[e] = linalg.norm(pos[e.target()].a - pos[e.source()].a)
600
    >>> tree = gt.min_spanning_tree(g, weights=weight)
601
    >>> gt.graph_draw(g, pos=pos, output="triang_orig.pdf")
602
    <...>
603
604
    >>> u = gt.GraphView(g, efilt=tree)
    >>> gt.graph_draw(u, pos=pos, output="triang_min_span_tree.pdf")
605
606
    <...>

Tiago Peixoto's avatar
Tiago Peixoto committed
607
608
609
610
    .. testcode::
       :hide:

       gt.graph_draw(g, pos=pos, output="triang_orig.png")
611
       gt.graph_draw(u, pos=pos, output="triang_min_span_tree.png")
612

613
    .. image:: triang_orig.*
Tiago Peixoto's avatar
Tiago Peixoto committed
614
        :width: 400px
615
    .. image:: triang_min_span_tree.*
Tiago Peixoto's avatar
Tiago Peixoto committed
616
        :width: 400px
617
618

    *Left:* Original graph, *Right:* The minimum spanning tree.
619
620
621
622
623

    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
624
625
       American Mathematical Society, volume 7, pages 48-50, 1956.
       :doi:`10.1090/S0002-9939-1956-0078686-7`
626
627
628
629
630
    .. [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
    """
631
    if tree_map is None:
632
633
634
635
        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.")

636
637
638
639
640
641
642
643
644
645
646
    u = GraphView(g, directed=False)
    if root is None:
        libgraph_tool_topology.\
               get_kruskal_spanning_tree(u._Graph__graph,
                                         _prop("e", g, weights),
                                         _prop("e", g, tree_map))
    else:
        libgraph_tool_topology.\
               get_prim_spanning_tree(u._Graph__graph, int(root),
                                      _prop("e", g, weights),
                                      _prop("e", g, tree_map))
647
    return tree_map
648

Tiago Peixoto's avatar
Tiago Peixoto committed
649

650
def random_spanning_tree(g, weights=None, root=None, tree_map=None):
651
    r"""Return a random spanning tree of a given graph, which can be directed or
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
    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
    -----
674
675

    The running time for this algorithm is :math:`O(\tau)`, with :math:`\tau`
676
677
678
679
    being the mean hitting time of a random walk on the graph. In the worse case,
    we have :math:`\tau \sim O(V^3)`, with :math:`V` being the number of
    vertices in the graph. However, in much more typical cases (e.g. sparse
    random graphs) the running time is simply :math:`O(V)`.
680
681
682

    Examples
    --------
683
684
685
686
687
688
689
690
    .. testcode::
       :hide:

       import numpy.random
       numpy.random.seed(42)
       gt.seed_rng(42)

    >>> from numpy.random import random
691
    >>> g, pos = gt.triangulation(random((400, 2)), type="delaunay")
692
693
694
695
    >>> 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)
696
    >>> tree2 = gt.random_spanning_tree(g, weights=weight)
697
698
    >>> gt.graph_draw(g, pos=pos, output="rtriang_orig.pdf")
    <...>
699
700
701
702
703
    >>> u = gt.GraphView(g, efilt=tree)
    >>> gt.graph_draw(u, pos=pos, output="triang_random_span_tree.pdf")
    <...>
    >>> u2 = gt.GraphView(g, efilt=tree2)
    >>> gt.graph_draw(u2, pos=pos, output="triang_random_span_tree2.pdf")
704
705
    <...>

Tiago Peixoto's avatar
Tiago Peixoto committed
706
707
708
709
    .. testcode::
       :hide:

       gt.graph_draw(g, pos=pos, output="rtriang_orig.png")
710
711
       gt.graph_draw(u, pos=pos, output="triang_random_span_tree.png")
       gt.graph_draw(u2, pos=pos, output="triang_random_span_tree2.png")
712
713

    .. image:: rtriang_orig.*
714
        :width: 300px
715
    .. image:: triang_random_span_tree.*
716
717
718
        :width: 300px
    .. image:: triang_random_span_tree2.*
        :width: 300px
719

720
721
    *Left:* Original graph, *Middle:* A random spanning tree, *Right:* Another
    random spanning tree
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742

    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)
743
744
745
    u = GraphView(g, vfilt=l)
    if u.num_vertices() != g.num_vertices():
        raise ValueError("There must be a path from all vertices to the root vertex: %d" % int(root) )
746
747
748
749

    libgraph_tool_topology.\
        random_spanning_tree(g._Graph__graph, int(root),
                             _prop("e", g, weights),
750
                             _prop("e", g, tree_map), _get_rng())
751
752
753
    return tree_map


Tiago Peixoto's avatar
Tiago Peixoto committed
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
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
    --------
781
782
783
784
785
786
787
    .. testcode::
       :hide:

       import numpy.random
       numpy.random.seed(42)
       gt.seed_rng(42)

Tiago Peixoto's avatar
Tiago Peixoto committed
788
789
790
    >>> g = gt.random_graph(100, lambda: (2, 2))
    >>> tree = gt.min_spanning_tree(g)
    >>> g.set_edge_filter(tree)
791
    >>> root = [v for v in g.vertices() if v.in_degree() == 0]
Tiago Peixoto's avatar
Tiago Peixoto committed
792
    >>> dom = gt.dominator_tree(g, root[0])
793
    >>> print(dom.a)
794
795
796
    [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 1 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
797
798
799

    References
    ----------
800
    .. [dominator-bgl] http://www.boost.org/libs/graph/doc/lengauer_tarjan_dominator.htm
Tiago Peixoto's avatar
Tiago Peixoto committed
801
802

    """
803
    if dom_map is None:
Tiago Peixoto's avatar
Tiago Peixoto committed
804
805
806
        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" +
807
808
                         " int32_t.")
    if not g.is_directed():
Tiago Peixoto's avatar
Tiago Peixoto committed
809
        raise ValueError("dominator tree requires a directed graph.")
810
    libgraph_tool_topology.\
Tiago Peixoto's avatar
Tiago Peixoto committed
811
812
813
               dominator_tree(g._Graph__graph, int(root),
                              _prop("v", g, dom_map))
    return dom_map
814

Tiago Peixoto's avatar
Tiago Peixoto committed
815

816
def topological_sort(g):
Tiago Peixoto's avatar
Tiago Peixoto committed
817
818
819
820
821
822
823
    """
    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
824
    such that if edge (u,v) appears in the graph, then u comes before v in the
Tiago Peixoto's avatar
Tiago Peixoto committed
825
826
827
828
829
830
    ordering. The graph must be a directed acyclic graph (DAG).

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

    Examples
    --------
831
832
833
834
835
836
837
    .. testcode::
       :hide:

       import numpy.random
       numpy.random.seed(42)
       gt.seed_rng(42)

Tiago Peixoto's avatar
Tiago Peixoto committed
838
839
840
841
    >>> g = gt.random_graph(30, lambda: (3, 3))
    >>> tree = gt.min_spanning_tree(g)
    >>> g.set_edge_filter(tree)
    >>> sort = gt.topological_sort(g)
842
    >>> print(sort)
Tiago Peixoto's avatar
Tiago Peixoto committed
843
844
    [29 28 27 26 23 24 22 21 20 18 17 16 15 14 11 10  9  6  5  4 19 12 13  3  2
     25  1  0  7  8]
Tiago Peixoto's avatar
Tiago Peixoto committed
845
846
847

    References
    ----------
848
    .. [topological-boost] http://www.boost.org/libs/graph/doc/topological_sort.html
Tiago Peixoto's avatar
Tiago Peixoto committed
849
850
851
852
    .. [topological-wiki] http://en.wikipedia.org/wiki/Topological_sorting

    """

853
    topological_order = Vector_int32_t()
Tiago Peixoto's avatar
Tiago Peixoto committed
854
855
856
857
    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).");
858
    return topological_order.a[::-1].copy()
859

Tiago Peixoto's avatar
Tiago Peixoto committed
860

861
def transitive_closure(g):
Tiago Peixoto's avatar
Tiago Peixoto committed
862
863
864
865
866
867
868
869
870
871
872
873
874
    """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
    --------
875
876
877
878
879
880
881
    .. testcode::
       :hide:

       import numpy.random
       numpy.random.seed(42)
       gt.seed_rng(42)

Tiago Peixoto's avatar
Tiago Peixoto committed
882
883
884
885
886
    >>> g = gt.random_graph(30, lambda: (3, 3))
    >>> tc = gt.transitive_closure(g)

    References
    ----------
887
    .. [transitive-boost] http://www.boost.org/libs/graph/doc/transitive_closure.html
Tiago Peixoto's avatar
Tiago Peixoto committed
888
889
890
891
    .. [transitive-wiki] http://en.wikipedia.org/wiki/Transitive_closure

    """

892
893
894
895
896
897
898
    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
899

900
def label_components(g, vprop=None, directed=None, attractors=False):
901
    """
902
    Label the components to which each vertex in the graph belongs. If the
903
904
    graph is directed, it finds the strongly connected components.

905
906
907
    A property map with the component labels is returned, together with an
    histogram of component labels.

908
909
    Parameters
    ----------
910
    g : :class:`~graph_tool.Graph`
911
        Graph to be used.
912
    vprop : :class:`~graph_tool.PropertyMap` (optional, default: ``None``)
913
914
        Vertex property to store the component labels. If none is supplied, one
        is created.
915
    directed : bool (optional, default: ``None``)
916
917
        Treat graph as directed or not, independently of its actual
        directionality.
918
919
920
921
    attractors : bool (optional, default: ``False``)
        If ``True``, and the graph is directed, an additional array with Boolean
        values is returned, specifying if the strongly connected components are
        attractors or not.
922
923
924

    Returns
    -------
925
    comp : :class:`~graph_tool.PropertyMap`
926
        Vertex property map with component labels.
927
928
    hist : :class:`~numpy.ndarray`
        Histogram of component labels.
929
930
931
932
    is_attractor : :class:`~numpy.ndarray`
        A Boolean array specifying if the strongly connected components are
        attractors or not. This returned only if ``attractors == True``, and the
        graph is directed.
933
934
935
936
937
938

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

939
    The algorithm runs in :math:`O(V + E)` time.
940
941
942

    Examples
    --------
943
944
945
946
947
948
    .. testcode::
       :hide:

       numpy.random.seed(43)
       gt.seed_rng(43)

949
950
    >>> g = gt.random_graph(100, lambda: (poisson(2), poisson(2)))
    >>> comp, hist, is_attractor = gt.label_components(g, attractors=True)
951
    >>> print(comp.a)
952
953
954
955
    [13 13 13 13 14 12 13 15 16 13 17 19 13 13 13 20 13 13 13 10 13 13 22 13 13
      4 13 13  2 23 13 13 24 13 13 26 27 13 13 13 13  0 13 13  3 13 13 13 28  1
      6 13 13 13 13  5 13 13 13 13 13 13 13  9 13 11 13 29 13 13 13 13 18 13 30
     31 13 13 32 13 33 34 35 13 13 21 13 25  8 36 13 13 13 13 13 37 13 13  7 13]
956
    >>> print(hist)
957
958
    [ 1  1  1  1  1  1  1  1  1  1  1  1  1 63  1  1  1  1  1  1  1  1  1  1  1
      1  1  1  1  1  1  1  1  1  1  1  1  1]
959
    >>> print(is_attractor)
960
961
962
963
    [ True False  True  True  True False False  True False  True  True  True
      True False  True False False False False False False False False False
     False False False False False False False False False  True False  True
     False False]
964
965
    """

966
    if vprop is None:
967
968
969
970
971
        vprop = g.new_vertex_property("int32_t")

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

972
973
    if directed is not None:
        g = GraphView(g, directed=directed)
974

975
976
    hist = libgraph_tool_topology.\
               label_components(g._Graph__graph, _prop("v", g, vprop))
977
978
979
980
981
982
983
984
985

    if attractors and g.is_directed() and directed != False:
        is_attractor = numpy.ones(len(hist), dtype="bool")
        libgraph_tool_topology.\
               label_attractors(g._Graph__graph, _prop("v", g, vprop),
                                is_attractor)
        return vprop, hist, is_attractor
    else:
        return vprop, hist
986
987
988
989


def label_largest_component(g, directed=None):
    """
990
991
    Label the largest component in the graph. If the graph is directed, then the
    largest strongly connected component is labelled.
992
993
994
995
996
997
998
999
1000

    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
For faster browsing, not all history is shown. View entire blame