__init__.py 47.1 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
``graph_tool.generation`` - Random graph generation
23
---------------------------------------------------
24
25
26
27
28
29
30
31
32
33
34
35

Summary
+++++++

.. autosummary::
   :nosignatures:

   random_graph
   random_rewire
   predecessor_tree
   line_graph
   graph_union
36
   triangulation
37
38
   lattice
   geometric_graph
39
   price_network
40
41
42

Contents
++++++++
43
44
"""

45
46
from __future__ import division, absolute_import, print_function

Tiago Peixoto's avatar
Tiago Peixoto committed
47
from .. dl_import import dl_import
48
dl_import("from . import libgraph_tool_generation")
49

50
from .. import Graph, GraphView, _check_prop_scalar, _prop, _limit_args, _gt_type
Tiago Peixoto's avatar
Tiago Peixoto committed
51
from .. stats import label_parallel_edges, label_self_loops
52
53
import inspect
import types
54
import sys, numpy, numpy.random
55

Tiago Peixoto's avatar
Tiago Peixoto committed
56
__all__ = ["random_graph", "random_rewire", "predecessor_tree", "line_graph",
57
58
           "graph_union", "triangulation", "lattice", "geometric_graph",
           "price_network"]
59

Tiago Peixoto's avatar
Tiago Peixoto committed
60

61
def random_graph(N, deg_sampler, deg_corr=None, cache_probs=True, directed=True,
62
                 parallel_edges=False, self_loops=False, blockmodel=None,
63
                 block_type="int", degree_block=False,
64
                 random=True, mix_time=10, verbose=False):
Tiago Peixoto's avatar
Tiago Peixoto committed
65
66
67
68
69
70
71
72
73
74
75
76
77
    r"""
    Generate a random graph, with a given degree distribution and correlation.

    Parameters
    ----------
    N : int
        Number of vertices in the graph.
    deg_sampler : function
        A degree sampler function which is called without arguments, and returns
        a tuple of ints representing the in and out-degree of a given vertex (or
        a single int for undirected graphs, representing the out-degree). This
        function is called once per vertex, but may be called more times, if the
        degree sequence cannot be used to build a graph.
78

79
80
81
82
83
        Optionally, you can also pass a function which receives one or two
        arguments: If ``blockmodel == None``, the single argument passed will
        be the index of the vertex which will receive the degree.
        If ``blockmodel != None``, the first value passed will be the vertex
        index, and the second will be the block value of the vertex.
84
    deg_corr : function (optional, default: ``None``)
Tiago Peixoto's avatar
Tiago Peixoto committed
85
        A function which gives the degree correlation of the graph. It should be
Tiago Peixoto's avatar
Tiago Peixoto committed
86
87
88
89
90
        callable with two parameters: the in,out-degree pair of the source
        vertex an edge, and the in,out-degree pair of the target of the same
        edge (for undirected graphs, both parameters are single values). The
        function should return a number proportional to the probability of such
        an edge existing in the generated graph.
91
92
93

        If ``blockmodel != None``, the value passed to the function will be the
        block value of the respective vertices, not the in/out-degree pairs.
94
95
96
97
98
99
100
    cache_probs : bool (optional, default: ``True``)
        If ``True``, the probabilities returned by the ``deg_corr`` parameter
        will be cached internally. This is crucial for good performance, since
        in this case the supplied python function is called only a few times,
        and not at every attempted edge rewire move. However, in the case were
        the different parameter combinations to the probability function is very
        large, the memory requirements to keep the cache may be very large.
101
    directed : bool (optional, default: ``True``)
Tiago Peixoto's avatar
Tiago Peixoto committed
102
        Whether the generated graph should be directed.
103
104
105
106
107
108
109
110
111
112
113
114
115
    parallel_edges : bool (optional, default: ``False``)
        If ``True``, parallel edges are allowed.
    self_loops : bool (optional, default: ``False``)
        If ``True``, self-loops are allowed.
    blockmodel : list or :class:`~numpy.ndarray` or function (optional, default: ``None``)
        If supplied, the graph will be sampled from a blockmodel ensemble. If
        the value is a list or a :class:`~numpy.ndarray`, it must have
        ``len(block_model) == N``, and the values will define to which block
        each vertex belongs.

        If this value is a function, it will be used to sample the block
        types. It must be callable either with no arguments or with a single
        argument which will be the vertex index. In either case it must return
116
117
118
119
120
121
122
123
        a type compatible with the ``block_type`` parameter.
    block_type : string (optional, default: ``"int"``)
        Value type of block labels. Valid only if ``blockmodel != None``.
    degree_block : bool (optional, default: ``False``)
        If ``True``, the degree of each vertex will be appended to block labels
        when constructing the blockmodel, such that the resulting block type
        will be a pair :math:`(r, k)`, where :math:`r` is the original block
        label.
124
125
126
127
    random : bool (optional, default: ``True``)
        If ``True``, the returned graph is randomized. Otherwise a deterministic
        placement of the edges will be used.
    mix_time : int (optional, default: ``10``)
128
129
130
        Number of edge sweeps to perform in order to mix the graph. This value
        is ignored if ``parallel_edges == self_loops == True`` and
        ``strat != "probabilistic"``.
131
132
    verbose : bool (optional, default: ``False``)
        If ``True``, verbose information is displayed.
Tiago Peixoto's avatar
Tiago Peixoto committed
133
134
135

    Returns
    -------
136
    random_graph : :class:`~graph_tool.Graph`
Tiago Peixoto's avatar
Tiago Peixoto committed
137
        The generated graph.
138
139
140
    blocks : :class:`~graph_tool.PropertyMap`
        A vertex property map with the block values. This is only returned if
        ``blockmodel != None``.
Tiago Peixoto's avatar
Tiago Peixoto committed
141
142
143
144
145
146
147

    See Also
    --------
    random_rewire: in place graph shuffling

    Notes
    -----
Tiago Peixoto's avatar
Tiago Peixoto committed
148
149
150
    The algorithm makes sure the degree sequence is graphical (i.e. realizable)
    and keeps re-sampling the degrees if is not. With a valid degree sequence,
    the edges are placed deterministically, and later the graph is shuffled with
151
152
    the :func:`~graph_tool.generation.random_rewire` function, with the
    ``mix_time`` parameter passed as ``n_iter``.
Tiago Peixoto's avatar
Tiago Peixoto committed
153

154
    The complexity is :math:`O(V + E)` if parallel edges are allowed, and
155
    :math:`O(V + E \times\text{mix-time})` if parallel edges are not allowed.
156
157
158
159
160
161
162
163
164
165
166
167
168


    .. note ::

        If ``parallel_edges == False`` this algorithm only guarantees that the
        returned graph will be a random sample from the desired ensemble if
        ``mix_time`` is sufficiently large. The algorithm implements an
        efficient Markov chain based on edge swaps, with a mixing time which
        depends on the degree distribution and correlations desired. If degree
        correlations are provided, the mixing time tends to be larger.

        If ``strat == "probabilistic"``, the Markov chain still needs to be
        mixed, even if parallel edges and self-loops are allowed. In this case
169
170
171
        the Markov chain is implemented using the Metropolis-Hastings
        [metropolis-equations-1953]_ [hastings-monte-carlo-1970]_
        acceptance/rejection algorithm.
Tiago Peixoto's avatar
Tiago Peixoto committed
172
173
174
175
176

    Examples
    --------
    >>> from numpy.random import randint, random, seed, poisson
    >>> from pylab import *
177
    >>> seed(43)
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

    This is a degree sampler which uses rejection sampling to sample from the
    distribution :math:`P(k)\propto 1/k`, up to a maximum.

    >>> def sample_k(max):
    ...     accept = False
    ...     while not accept:
    ...         k = randint(1,max+1)
    ...         accept = random() < 1.0/k
    ...     return k
    ...

    The following generates a random undirected graph with degree distribution
    :math:`P(k)\propto 1/k` (with k_max=40) and an *assortative* degree
    correlation of the form:

    .. math::

        P(i,k) \propto \frac{1}{1+|i-k|}

    >>> g = gt.random_graph(1000, lambda: sample_k(40),
199
200
    ...                     lambda i, k: 1.0 / (1 + abs(i - k)), directed=False,
    ...                     mix_time=100)
Tiago Peixoto's avatar
Tiago Peixoto committed
201
    >>> gt.scalar_assortativity(g, "out")
Tiago Peixoto's avatar
Tiago Peixoto committed
202
    (0.6435658697163692, 0.010420519538259333)
Tiago Peixoto's avatar
Tiago Peixoto committed
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217

    The following samples an in,out-degree pair from the joint distribution:

    .. math::

        p(j,k) = \frac{1}{2}\frac{e^{-m_1}m_1^j}{j!}\frac{e^{-m_1}m_1^k}{k!} +
                 \frac{1}{2}\frac{e^{-m_2}m_2^j}{j!}\frac{e^{-m_2}m_2^k}{k!}

    with :math:`m_1 = 4` and :math:`m_2 = 20`.

    >>> def deg_sample():
    ...    if random() > 0.5:
    ...        return poisson(4), poisson(4)
    ...    else:
    ...        return poisson(20), poisson(20)
218
    ... 
Tiago Peixoto's avatar
Tiago Peixoto committed
219
220
221
222
223
224
225

    The following generates a random directed graph with this distribution, and
    plots the combined degree correlation.

    >>> g = gt.random_graph(20000, deg_sample)
    >>>
    >>> hist = gt.combined_corr_hist(g, "in", "out")
226
227
    >>>
    >>> clf()
228
    >>> imshow(hist[0].T, interpolation="nearest", origin="lower")
Tiago Peixoto's avatar
Tiago Peixoto committed
229
230
231
    <...>
    >>> colorbar()
    <...>
232
    >>> xlabel("in-degree")
Tiago Peixoto's avatar
Tiago Peixoto committed
233
    <...>
234
    >>> ylabel("out-degree")
Tiago Peixoto's avatar
Tiago Peixoto committed
235
    <...>
236
    >>> savefig("combined-deg-hist.pdf")
Tiago Peixoto's avatar
Tiago Peixoto committed
237

238
    .. figure:: combined-deg-hist.*
Tiago Peixoto's avatar
Tiago Peixoto committed
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
        :align: center

        Combined degree histogram.

    A correlated directed graph can be build as follows. Consider the following
    degree correlation:

    .. math::

         P(j',k'|j,k)=\frac{e^{-k}k^{j'}}{j'!}
         \frac{e^{-(20-j)}(20-j)^{k'}}{k'!}

    i.e., the in->out correlation is "disassortative", the out->in correlation
    is "assortative", and everything else is uncorrelated.
    We will use a flat degree distribution in the range [1,20).

    >>> p = scipy.stats.poisson
    >>> g = gt.random_graph(20000, lambda: (sample_k(19), sample_k(19)),
257
258
259
    ...                     lambda a,b: (p.pmf(a[0], b[1]) *
    ...                                  p.pmf(a[1], 20 - b[0])),
    ...                     mix_time=100)
Tiago Peixoto's avatar
Tiago Peixoto committed
260
261
262

    Lets plot the average degree correlations to check.

263
    >>> clf()
264
265
    >>> axes([0.1,0.15,0.63,0.8])
    <...>
Tiago Peixoto's avatar
Tiago Peixoto committed
266
    >>> corr = gt.avg_neighbour_corr(g, "in", "in")
267
    >>> errorbar(corr[2][:-1], corr[0], yerr=corr[1], fmt="o-",
268
    ...         label=r"$\left<\text{in}\right>$ vs in")
269
    <...>
Tiago Peixoto's avatar
Tiago Peixoto committed
270
    >>> corr = gt.avg_neighbour_corr(g, "in", "out")
271
    >>> errorbar(corr[2][:-1], corr[0], yerr=corr[1], fmt="o-",
272
    ...         label=r"$\left<\text{out}\right>$ vs in")
273
    <...>
Tiago Peixoto's avatar
Tiago Peixoto committed
274
    >>> corr = gt.avg_neighbour_corr(g, "out", "in")
275
    >>> errorbar(corr[2][:-1], corr[0], yerr=corr[1], fmt="o-",
276
    ...          label=r"$\left<\text{in}\right>$ vs out")
277
    <...>
Tiago Peixoto's avatar
Tiago Peixoto committed
278
    >>> corr = gt.avg_neighbour_corr(g, "out", "out")
279
    >>> errorbar(corr[2][:-1], corr[0], yerr=corr[1], fmt="o-",
280
    ...          label=r"$\left<\text{out}\right>$ vs out")
Tiago Peixoto's avatar
Tiago Peixoto committed
281
    <...>
282
283
284
    >>> legend(bbox_to_anchor=(1.01, 0.5), loc="center left", borderaxespad=0.)
    <...>
    >>> xlabel("Source degree")
Tiago Peixoto's avatar
Tiago Peixoto committed
285
    <...>
286
    >>> ylabel("Average target degree")
Tiago Peixoto's avatar
Tiago Peixoto committed
287
    <...>
288
    >>> savefig("deg-corr-dir.pdf")
Tiago Peixoto's avatar
Tiago Peixoto committed
289

290
    .. figure:: deg-corr-dir.*
Tiago Peixoto's avatar
Tiago Peixoto committed
291
292
293
        :align: center

        Average nearest neighbour correlations.
294
295
296
297
298


    **Blockmodels**


299
300
301
    The following example shows how a stochastic blockmodel
    [holland-stochastic-1983]_ [karrer-stochastic-2011]_ can be generated. We
    will consider a system of 10 blocks, which form communities. The connection
302
303
304
305
306
307
308
309
310
311
    probability will be given by

    >>> def corr(a, b):
    ...    if a == b:
    ...        return 0.999
    ...    else:
    ...        return 0.001

    The blockmodel can be generated as follows.

Tiago Peixoto's avatar
Tiago Peixoto committed
312
    >>> g, bm = gt.random_graph(5000, lambda: poisson(10), directed=False,
313
314
    ...                         blockmodel=lambda: randint(10), deg_corr=corr,
    ...                         mix_time=500)
315
    >>> gt.graph_draw(g, vertex_fill_color=bm, output="blockmodel.pdf")
316
317
    <...>

318
    .. figure:: blockmodel.*
319
320
321
322
323
324
325
326
327
328
329
330
331
332
        :align: center

        Simple blockmodel with 10 blocks.


    References
    ----------
    .. [metropolis-equations-1953]  Metropolis, N.; Rosenbluth, A.W.;
       Rosenbluth, M.N.; Teller, A.H.; Teller, E. "Equations of State
       Calculations by Fast Computing Machines". Journal of Chemical Physics 21
       (6): 1087–1092 (1953). :doi:`10.1063/1.1699114`
    .. [hastings-monte-carlo-1970] Hastings, W.K. "Monte Carlo Sampling Methods
       Using Markov Chains and Their Applications". Biometrika 57 (1): 97–109 (1970).
       :doi:`10.1093/biomet/57.1.97`
333
334
335
336
337
338
    .. [holland-stochastic-1983] Paul W. Holland, Kathryn Blackmond Laskey, and
       Samuel Leinhardt, "Stochastic blockmodels: First steps," Social Networks
       5, no. 2: 109-13 (1983) :doi:`10.1016/0378-8733(83)90021-7`
    .. [karrer-stochastic-2011] Brian Karrer and M. E. J. Newman, "Stochastic
       blockmodels and community structure in networks," Physical Review E 83,
       no. 1: 016107 (2011) :doi:`10.1103/PhysRevE.83.016107` :arxiv:`1008.3926`
Tiago Peixoto's avatar
Tiago Peixoto committed
339
    """
340

341
    seed = numpy.random.randint(0, sys.maxsize)
342
343
344
345
346
    g = Graph()
    if deg_corr == None:
        uncorrelated = True
    else:
        uncorrelated = False
347
348
349

    if (type(blockmodel) is types.FunctionType or
        type(blockmodel) is types.LambdaType):
350
351
        btype = block_type
        bm = []
352
        if len(inspect.getargspec(blockmodel)[0]) == 0:
353
            for i in range(N):
354
                bm.append(blockmodel())
355
        else:
356
            for i in range(N):
357
358
                bm.append(blockmodel(i))
        blockmodel = bm
Tiago Peixoto's avatar
Tiago Peixoto committed
359
    elif blockmodel is not None:
360
        btype = _gt_type(blockmodel[0])
361
362
363

    if len(inspect.getargspec(deg_sampler)[0]) > 0:
        if blockmodel is not None:
364
            sampler = lambda i: deg_sampler(i, blockmodel[i])
365
        else:
Tiago Peixoto's avatar
Tiago Peixoto committed
366
            sampler = deg_sampler
367
368
369
370
    else:
        sampler = lambda i: deg_sampler()

    libgraph_tool_generation.gen_graph(g._Graph__graph, N, sampler,
371
372
373
                                       uncorrelated, not parallel_edges,
                                       not self_loops, not directed,
                                       seed, verbose, True)
374
375
    g.set_directed(directed)

376
377
378
379
380
381
382
383
384
385
386
387
    if degree_block:
        if btype in ["object", "string"] or "vector" in btype:
            btype = "object"
        elif btype in ["int", "int32_t", "bool"]:
            btype = "vector<int32_t>"
        elif btype in ["long", "int64_t"]:
            btype = "vector<int64_t>"
        elif btype in ["double"]:
            btype = "vector<double>"
        elif btype in ["long double"]:
            btype = "vector<long double>"

388
389
390
391
    if blockmodel is not None:
        bm = g.new_vertex_property(btype)
        if btype in ["object", "string"] or "vector" in btype:
            for v in g.vertices():
392
393
394
395
396
397
398
399
                if not degree_block:
                    bm[v] = blockmodel[int(v)]
                else:
                    if g.is_directed():
                        bm[v] = (blockmodel[int(v)], v.in_degree(),
                                 v.out_degree())
                    else:
                        bm[v] = (blockmodel[int(v)], v.out_degree())
400
401
402
403
404
405
406
407
408
        else:
            try:
                bm.a = blockmodel
            except ValueError:
                bm = g.new_vertex_property("object")
                for v in g.vertices():
                    bm[v] = blockmodel[int(v)]
    else:
        bm = None
409

410
    if parallel_edges and self_loops and deg_corr is None:
411
        mix_time = 1
Tiago Peixoto's avatar
Tiago Peixoto committed
412
    if random:
413
414
        if deg_corr is not None:
            random_rewire(g, strat="probabilistic", n_iter=mix_time,
Tiago Peixoto's avatar
Tiago Peixoto committed
415
                          parallel_edges=parallel_edges, deg_corr=deg_corr,
416
417
                          cache_probs=cache_probs, self_loops=self_loops,
                          blockmodel=bm, verbose=verbose)
418
419
420
421
        else:
            random_rewire(g, parallel_edges=parallel_edges, n_iter=mix_time,
                          self_loops=self_loops, verbose=verbose)

422
423
424
425
    if bm is None:
        return g
    else:
        return g, bm
426

Tiago Peixoto's avatar
Tiago Peixoto committed
427

428
429
@_limit_args({"strat": ["erdos", "correlated", "uncorrelated",
                        "probabilistic"]})
430
431
def random_rewire(g, strat="uncorrelated", n_iter=1, edge_sweep=True,
                  parallel_edges=False, self_loops=False, deg_corr=None,
432
433
                  cache_probs=True, blockmodel=None, ret_fail=False,
                  verbose=False):
434
    r"""
435
436
437
438
439
440
441
442
    Shuffle the graph in-place.

    If ``strat != "erdos"``, the degrees (either in or out) of each vertex are
    always the same, but otherwise the edges are randomly placed. If
    ``strat == "correlated"``, the degree correlations are also maintained: The
    new source and target of each edge both have the same in and out-degree. If
    ``strat == "probabilistic"``, then edges are rewired according to the degree
    correlation given by the parameter ``deg_corr``.
443
444
445

    Parameters
    ----------
446
    g : :class:`~graph_tool.Graph`
447
        Graph to be shuffled. The graph will be modified.
448
449
450
451
    strat : string (optional, default: ``"uncorrelated"``)
        If ``strat == "erdos"``, the resulting graph will be entirely random. If
        ``strat == "uncorrelated"`` only the degrees of the vertices will be
        maintained, nothing else. If ``strat == "correlated"``, additionally the
452
        new source and target of each edge both have the same in and out-degree.
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
        If ``strat == "probabilistic"``, than edges are rewired according to the
        degree correlation given by the parameter ``deg_corr``.
    n_iter : int (optional, default: ``1``)
        Number of iterations. If ``edge_sweep == True``, each iteration
        corresponds to an entire "sweep" over all edges. Otherwise this
        corresponds to the total number of edges which are randomly chosen for a
        swap attempt (which may repeat).
    edge_sweep : bool (optional, default: ``True``)
        If ``True``, each iteration will perform an entire "sweep" over the
        edges, where each edge is visited once in random order, and a edge swap
        is attempted.
    parallel : bool (optional, default: ``False``)
        If ``True``, parallel edges are allowed.
    self_loops : bool (optional, default: ``False``)
        If ``True``, self-loops are allowed.
    deg_corr : function (optional, default: ``None``)
469
470
471
472
473
474
        A function which gives the degree correlation of the graph. It should be
        callable with two parameters: the in,out-degree pair of the source
        vertex an edge, and the in,out-degree pair of the target of the same
        edge (for undirected graphs, both parameters are single values). The
        function should return a number proportional to the probability of such
        an edge existing in the generated graph. This parameter is ignored,
475
        unless ``strat == "probabilistic"``.
476
477
478

        If ``blockmodel != None``, the value passed to the function will be the
        block value of the respective vertices, not the in/out-degree pairs.
479
480
481
482
483
484
485
    cache_probs : bool (optional, default: ``True``)
        If ``True``, the probabilities returned by the ``deg_corr`` parameter
        will be cached internally. This is crucial for good performance, since
        in this case the supplied python function is called only a few times,
        and not at every attempted edge rewire move. However, in the case were
        the different parameter combinations to the probability function is very
        large, the memory requirements to keep the cache may be very large.
486
487
488
489
    blockmodel : :class:`~graph_tool.PropertyMap` (optional, default: ``None``)
        If supplied, the graph will be rewired to conform to a blockmodel
        ensemble. The value must be a vertex property map which defines the
        block of each vertex.
490
491
492
493
494
495
496
497
498
499
500
    ret_fail : bool (optional, default: ``False``)
        If ``True``, the number of failed edge moves (due to parallel edges or
        self-loops) is returned.
    verbose : bool (optional, default: ``False``)
        If ``True``, verbose information is displayed.


    Returns
    -------
    fail_count : int
        Number of failed edge moves (due to parallel edges or self-loops).
501
502
503
504
505
506
507

    See Also
    --------
    random_graph: random graph generation

    Notes
    -----
Tiago Peixoto's avatar
Tiago Peixoto committed
508
    This algorithm iterates through all the edges in the network and tries to
509
    swap its target or source with the target or source of another edge.
Tiago Peixoto's avatar
Tiago Peixoto committed
510
511

    .. note::
512

513
514
515
516
517
518
519
520
521
522
        If ``parallel_edges = False``, parallel edges are not placed during
        rewiring. In this case, the returned graph will be a uncorrelated sample
        from the desired ensemble only if ``n_iter`` is sufficiently large. The
        algorithm implements an efficient Markov chain based on edge swaps, with
        a mixing time which depends on the degree distribution and correlations
        desired. If degree probabilistic correlations are provided, the mixing
        time tends to be larger.

        If ``strat == "probabilistic"``, the Markov chain still needs to be
        mixed, even if parallel edges and self-loops are allowed. In this case
523
524
525
        the Markov chain is implemented using the Metropolis-Hastings
        [metropolis-equations-1953]_ [hastings-monte-carlo-1970]_
        acceptance/rejection algorithm.
526

Tiago Peixoto's avatar
Tiago Peixoto committed
527

528
    Each edge is tentatively swapped once per iteration, so the overall
529
530
    complexity is :math:`O(V + E \times \text{n-iter})`. If ``edge_sweep ==
    False``, the complexity becomes :math:`O(V + E + \text{n-iter})`.
531

532
533
534
535
536
    Examples
    --------

    Some small graphs for visualization.

537
    >>> from numpy.random import random, seed
538
    >>> from pylab import *
539
    >>> seed(43)
540
    >>> g, pos = gt.triangulation(random((1000,2)))
541
542
    >>> pos = gt.arf_layout(g)
    >>> gt.graph_draw(g, pos=pos, output="rewire_orig.pdf", output_size=(200, 200))
543
    <...>
544
    >>> gt.random_rewire(g, "correlated")
545
546
    >>> pos = gt.arf_layout(g)
    >>> gt.graph_draw(g, pos=pos, output="rewire_corr.pdf", output_size=(200, 200))
547
    <...>
548
    >>> gt.random_rewire(g)
549
550
    >>> pos = gt.arf_layout(g)
    >>> gt.graph_draw(g, pos=pos, output="rewire_uncorr.pdf", output_size=(200, 200))
551
    <...>
552
    >>> gt.random_rewire(g, "erdos")
553
554
    >>> pos = gt.arf_layout(g)
    >>> gt.graph_draw(g, pos=pos, output="rewire_erdos.pdf", output_size=(200, 200))
555
    <...>
556

557
    Some `ridiculograms <http://www.youtube.com/watch?v=YS-asmU3p_4>`_ :
558

559
560
561
562
    .. image:: rewire_orig.*
    .. image:: rewire_corr.*
    .. image:: rewire_uncorr.*
    .. image:: rewire_erdos.*
563

564
565
    **From left to right**: Original graph; Shuffled graph, with degree correlations;
    Shuffled graph, without degree correlations; Shuffled graph, with random degrees.
566

567
    We can try with larger graphs to get better statistics, as follows.
568

569
570
    >>> figure()
    <...>
571
    >>> g = gt.random_graph(30000, lambda: sample_k(20),
572
573
    ...                     lambda i, j: exp(abs(i-j)), directed=False,
    ...                     mix_time=100)
574
    >>> corr = gt.avg_neighbour_corr(g, "out", "out")
575
576
    >>> errorbar(corr[2][:-1], corr[0], yerr=corr[1], fmt="o-", label="Original")
    <...>
577
578
    >>> gt.random_rewire(g, "correlated")
    >>> corr = gt.avg_neighbour_corr(g, "out", "out")
579
580
    >>> errorbar(corr[2][:-1], corr[0], yerr=corr[1], fmt="*", label="Correlated")
    <...>
581
582
    >>> gt.random_rewire(g)
    >>> corr = gt.avg_neighbour_corr(g, "out", "out")
583
584
    >>> errorbar(corr[2][:-1], corr[0], yerr=corr[1], fmt="o-", label="Uncorrelated")
    <...>
585
586
    >>> gt.random_rewire(g, "erdos")
    >>> corr = gt.avg_neighbour_corr(g, "out", "out")
587
588
    >>> errorbar(corr[2][:-1], corr[0], yerr=corr[1], fmt="o-", label=r"Erd\H{o}s")
    <...>
589
590
591
592
593
594
    >>> xlabel("$k$")
    <...>
    >>> ylabel(r"$\left<k_{nn}\right>$")
    <...>
    >>> legend(loc="best")
    <...>
595
    >>> savefig("shuffled-stats.pdf")
596

597
    .. figure:: shuffled-stats.*
598
599
600
601
602
603
604
605
606
607
608
        :align: center

        Average degree correlations for the different shuffled and non-shuffled
        graphs. The shuffled graph with correlations displays exactly the same
        correlation as the original graph.

    Now let's do it for a directed graph. See
    :func:`~graph_tool.generation.random_graph` for more details.

    >>> p = scipy.stats.poisson
    >>> g = gt.random_graph(20000, lambda: (sample_k(19), sample_k(19)),
609
610
    ...                     lambda a, b: (p.pmf(a[0], b[1]) * p.pmf(a[1], 20 - b[0])),
    ...                     mix_time=100)
611
    >>> figure()
612
613
614
    <...>
    >>> axes([0.1,0.15,0.6,0.8])
    <...>
615
    >>> corr = gt.avg_neighbour_corr(g, "in", "out")
616
    >>> errorbar(corr[2][:-1], corr[0], yerr=corr[1], fmt="o-",
617
    ...          label=r"$\left<\text{o}\right>$ vs i")
618
    <...>
619
    >>> corr = gt.avg_neighbour_corr(g, "out", "in")
620
    >>> errorbar(corr[2][:-1], corr[0], yerr=corr[1], fmt="o-",
621
    ...          label=r"$\left<\text{i}\right>$ vs o")
622
    <...>
623
624
    >>> gt.random_rewire(g, "correlated")
    >>> corr = gt.avg_neighbour_corr(g, "in", "out")
625
    >>> errorbar(corr[2][:-1], corr[0], yerr=corr[1], fmt="o-",
626
    ...          label=r"$\left<\text{o}\right>$ vs i, corr.")
627
    <...>
628
    >>> corr = gt.avg_neighbour_corr(g, "out", "in")
629
    >>> errorbar(corr[2][:-1], corr[0], yerr=corr[1], fmt="o-",
630
    ...          label=r"$\left<\text{i}\right>$ vs o, corr.")
631
    <...>
632
633
    >>> gt.random_rewire(g, "uncorrelated")
    >>> corr = gt.avg_neighbour_corr(g, "in", "out")
634
    >>> errorbar(corr[2][:-1], corr[0], yerr=corr[1], fmt="o-",
635
    ...          label=r"$\left<\text{o}\right>$ vs i, uncorr.")
636
    <...>
637
    >>> corr = gt.avg_neighbour_corr(g, "out", "in")
638
    >>> errorbar(corr[2][:-1], corr[0], yerr=corr[1], fmt="o-",
639
    ...          label=r"$\left<\text{i}\right>$ vs o, uncorr.")
640
    <...>
641
642
643
    >>> legend(bbox_to_anchor=(1.01, 0.5), loc="center left", borderaxespad=0.)
    <...>
    >>> xlabel("Source degree")
644
    <...>
645
    >>> ylabel("Average target degree")
646
    <...>
647
    >>> savefig("shuffled-deg-corr-dir.pdf")
648

649
    .. figure:: shuffled-deg-corr-dir.*
650
651
652
653
654
655
        :align: center

        Average degree correlations for the different shuffled and non-shuffled
        directed graphs. The shuffled graph with correlations displays exactly
        the same correlation as the original graph.

656
657
658
659
660
661
662
663
664
    References
    ----------
    .. [metropolis-equations-1953]  Metropolis, N.; Rosenbluth, A.W.;
       Rosenbluth, M.N.; Teller, A.H.; Teller, E. "Equations of State
       Calculations by Fast Computing Machines". Journal of Chemical Physics 21
       (6): 1087–1092 (1953). :doi:`10.1063/1.1699114`
    .. [hastings-monte-carlo-1970] Hastings, W.K. "Monte Carlo Sampling Methods
       Using Markov Chains and Their Applications". Biometrika 57 (1): 97–109 (1970).
       :doi:`10.1093/biomet/57.1.97`
665
666
667
668
669
670
    .. [holland-stochastic-1983] Paul W. Holland, Kathryn Blackmond Laskey, and
       Samuel Leinhardt, "Stochastic blockmodels: First steps," Social Networks
       5, no. 2: 109-13 (1983) :doi:`10.1016/0378-8733(83)90021-7`
    .. [karrer-stochastic-2011] Brian Karrer and M. E. J. Newman, "Stochastic
       blockmodels and community structure in networks," Physical Review E 83,
       no. 1: 016107 (2011) :doi:`10.1103/PhysRevE.83.016107` :arxiv:`1008.3926`
671
672

    """
673
    seed = numpy.random.randint(0, sys.maxsize)
674

Tiago Peixoto's avatar
Tiago Peixoto committed
675
676
677
678
679
680
681
682
683
684
685
686
687
    if not parallel_edges:
        p = label_parallel_edges(g)
        if p.a.max() != 0:
            raise ValueError("Parallel edge detected. Can't rewire " +
                             "graph without parallel edges if it " +
                             "already contains parallel edges!")
    if not self_loops:
        l = label_self_loops(g)
        if l.a.max() != 0:
            raise ValueError("Self-loop detected. Can't rewire graph " +
                             "without self-loops if it already contains" +
                             " self-loops!")

688
    if (deg_corr is not None and not g.is_directed()) and blockmodel is None:
Tiago Peixoto's avatar
Tiago Peixoto committed
689
        corr = lambda i, j: deg_corr(i[1], j[1])
690
691
692
    else:
        corr = deg_corr

693
694
    if strat != "probabilistic":
        g = GraphView(g, reversed=False)
695
696
    elif blockmodel is not None:
        strat = "blockmodel"
697
698
699
    pcount = libgraph_tool_generation.random_rewire(g._Graph__graph, strat,
                                                    n_iter, not edge_sweep,
                                                    self_loops, parallel_edges,
700
                                                    corr, _prop("v", g, blockmodel),
701
                                                    cache_probs,
702
                                                    seed, verbose)
703
704
    if ret_fail:
        return pcount
Tiago Peixoto's avatar
Tiago Peixoto committed
705

Tiago Peixoto's avatar
Tiago Peixoto committed
706

Tiago Peixoto's avatar
Tiago Peixoto committed
707
def predecessor_tree(g, pred_map):
Tiago Peixoto's avatar
Tiago Peixoto committed
708
    """Return a graph from a list of predecessors given by the ``pred_map`` vertex property."""
Tiago Peixoto's avatar
Tiago Peixoto committed
709
710
711
712
713
714
715

    _check_prop_scalar(pred_map, "pred_map")
    pg = Graph()
    libgraph_tool_generation.predecessor_graph(g._Graph__graph,
                                               pg._Graph__graph,
                                               _prop("v", g, pred_map))
    return pg
716

Tiago Peixoto's avatar
Tiago Peixoto committed
717

718
def line_graph(g):
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
    """Return the line graph of the given graph `g`.

    Notes
    -----
    Given an undirected graph G, its line graph L(G) is a graph such that

        * each vertex of L(G) represents an edge of G; and
        * two vertices of L(G) are adjacent if and only if their corresponding
          edges share a common endpoint ("are adjacent") in G.

    For a directed graph, the second criterion becomes:

       * Two vertices representing directed edges from u to v and from w to x in
         G are connected by an edge from uv to wx in the line digraph when v =
         w.

    References
    ----------
    .. [line-wiki] http://en.wikipedia.org/wiki/Line_graph
    """
739
740
741
742
743
744
745
746
    lg = Graph(directed=g.is_directed())

    vertex_map = lg.new_vertex_property("int64_t")

    libgraph_tool_generation.line_graph(g._Graph__graph,
                                        lg._Graph__graph,
                                        _prop("v", lg, vertex_map))
    return lg, vertex_map
Tiago Peixoto's avatar
Tiago Peixoto committed
747

Tiago Peixoto's avatar
Tiago Peixoto committed
748

749
def graph_union(g1, g2, intersection=None, props=None, include=False):
750
751
752
753
754
755
756
757
758
    """Return the union of graphs g1 and g2, composed of all edges and vertices
    of g1 and g2, without overlap.

    Parameters
    ----------
    g1 : :class:`~graph_tool.Graph`
       First graph in the union.
    g2 : :class:`~graph_tool.Graph`
       Second graph in the union.
759
760
761
762
763
    intersection : :class:`~graph_tool.PropertyMap` (optional, default: ``None``)
       Vertex property map owned by `g1` which maps each of each of its vertices
       to vertex indexes belonging to `g2`. Negative values mean no mapping
       exists, and thus both vertices in `g1` and `g2` will be present in the
       union graph.
764
    props : list of tuples of :class:`~graph_tool.PropertyMap` (optional, default: ``[]``)
765
766
767
768
       Each element in this list must be a tuple of two PropertyMap objects. The
       first element must be a property of `g1`, and the second of `g2`. The
       values of the property maps are propagated into the union graph, and
       returned.
769
    include : bool (optional, default: ``False``)
770
771
772
773
774
775
776
777
778
779
       If true, graph `g2` is inserted into `g1` which is modified. If false, a
       new graph is created, and both graphs remain unmodified.

    Returns
    -------
    ug : :class:`~graph_tool.Graph`
        The union graph
    props : list of :class:`~graph_tool.PropertyMap` objects
        List of propagated properties.  This is only returned if `props` is not
        empty.
780
781
782
783
784
785
786
787
788

    Examples
    --------

    >>> from numpy.random import random, seed
    >>> seed(42)
    >>> g = gt.triangulation(random((300,2)))[0]
    >>> ug = gt.graph_union(g, g)
    >>> uug = gt.graph_union(g, ug)
789
790
    >>> pos = gt.arf_layout(g)
    >>> gt.graph_draw(g, pos=pos, output_size=(300,300), output="graph_original.pdf")
791
    <...>
792
793
    >>> pos = gt.arf_layout(ug)
    >>> gt.graph_draw(ug, pos=pos, output_size=(300,300), output="graph_union.pdf")
794
    <...>
795
796
    >>> pos = gt.arf_layout(uug)
    >>> gt.graph_draw(uug, pos=pos, output_size=(300,300), output="graph_union2.pdf")
797
798
    <...>

799
800
801
    .. image:: graph_original.*
    .. image:: graph_union.*
    .. image:: graph_union2.*
802

803
    """
Tiago Peixoto's avatar
Tiago Peixoto committed
804
805
    if props == None:
        props = []
Tiago Peixoto's avatar
Tiago Peixoto committed
806
807
    if not include:
        g1 = Graph(g1)
808
809
810
811
812
813
814
815
    if intersection is None:
        intersection = g1.new_vertex_property("int32_t")
        intersection.a = 0
    else:
        intersection = intersection.copy("int32_t")
        intersection.a[intersection.a >= 0] += 1
        intersection.a[intersection.a < 0] = 0

Tiago Peixoto's avatar
Tiago Peixoto committed
816
817
818
819
820
821
822
823
    g1.stash_filter(directed=True)
    g1.set_directed(True)
    g2.stash_filter(directed=True)
    g2.set_directed(True)
    n_props = []

    try:
        vmap, emap = libgraph_tool_generation.graph_union(g1._Graph__graph,
824
825
826
827
                                                          g2._Graph__graph,
                                                          _prop("v", g1,
                                                                intersection))
        for p1, p2 in props:
Tiago Peixoto's avatar
Tiago Peixoto committed
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
            if not include:
                p1 = g1.copy_property(p1)
            if p2.value_type() != p1.value_type():
                p2 = g2.copy_property(p2, value_type=p1.value_type())
            if p1.key_type() == 'v':
                libgraph_tool_generation.\
                      vertex_property_union(g1._Graph__graph, g2._Graph__graph,
                                            vmap, emap,
                                            _prop(p1.key_type(), g1, p1),
                                            _prop(p2.key_type(), g2, p2))
            else:
                libgraph_tool_generation.\
                      edge_property_union(g1._Graph__graph, g2._Graph__graph,
                                          vmap, emap,
                                          _prop(p1.key_type(), g1, p1),
                                          _prop(p2.key_type(), g2, p2))
            n_props.append(p1)
    finally:
        g1.pop_filter(directed=True)
        g2.pop_filter(directed=True)

    if len(n_props) > 0:
        return g1, n_props
    else:
        return g1
853

Tiago Peixoto's avatar
Tiago Peixoto committed
854
855

@_limit_args({"type": ["simple", "delaunay"]})
856
def triangulation(points, type="simple", periodic=False):
857
858
859
860
861
862
863
864
    r"""
    Generate a 2D or 3D triangulation graph from a given point set.

    Parameters
    ----------
    points : :class:`~numpy.ndarray`
        Point set for the triangulation. It may be either a N x d array, where N
        is the number of points, and d is the space dimension (either 2 or 3).
865
    type : string (optional, default: ``'simple'``)
866
        Type of triangulation. May be either 'simple' or 'delaunay'.
867
868
869
    periodic : bool (optional, default: ``False``)
        If ``True``, periodic boundary conditions will be used. This is
        parameter is valid only for type="delaunay", and is otherwise ignored.
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884

    Returns
    -------
    triangulation_graph : :class:`~graph_tool.Graph`
        The generated graph.
    pos : :class:`~graph_tool.PropertyMap`
        Vertex property map with the Cartesian coordinates.

    See Also
    --------
    random_graph: random graph generation

    Notes
    -----

Tiago Peixoto's avatar
Tiago Peixoto committed
885
    A triangulation [cgal-triang]_ is a division of the convex hull of a point
886
    set into triangles, using only that set as triangle vertices.
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905

    In simple triangulations (`type="simple"`), the insertion of a point is done
    by locating a face that contains the point, and splitting this face into
    three new faces (the order of insertion is therefore important). If the
    point falls outside the convex hull, the triangulation is restored by
    flips. Apart from the location, insertion takes a time O(1). This bound is
    only an amortized bound for points located outside the convex hull.

    Delaunay triangulations (`type="delaunay"`) have the specific empty sphere
    property, that is, the circumscribing sphere of each cell of such a
    triangulation does not contain any other vertex of the triangulation in its
    interior. These triangulations are uniquely defined except in degenerate
    cases where five points are co-spherical. Note however that the CGAL
    implementation computes a unique triangulation even in these cases.

    Examples
    --------
    >>> from numpy.random import seed, random
    >>> seed(42)
906
    >>> points = random((500, 2)) * 4
907
    >>> g, pos = gt.triangulation(points)
908
909
910
911
912
913
914
    >>> weight = g.new_edge_property("double") # Edge weights corresponding to
    ...                                        # Euclidean distances
    >>> for e in g.edges():
    ...    weight[e] = sqrt(sum((array(pos[e.source()]) -
    ...                          array(pos[e.target()]))**2))
    >>> b = gt.betweenness(g, weight=weight)
    >>> b[1].a *= 100
915
916
    >>> gt.graph_draw(g, pos=pos, output_size=(300,300), vertex_fill_color=b[0],
    ...               edge_pen_width=b[1], output="triang.pdf")
917
918
    <...>
    >>> g, pos = gt.triangulation(points, type="delaunay")
919
920
921
922
923
924
    >>> weight = g.new_edge_property("double")
    >>> for e in g.edges():
    ...    weight[e] = sqrt(sum((array(pos[e.source()]) -
    ...                          array(pos[e.target()]))**2))
    >>> b = gt.betweenness(g, weight=weight)
    >>> b[1].a *= 120
925
926
    >>> gt.graph_draw(g, pos=pos, output_size=(300,300), vertex_fill_color=b[0],
    ...               edge_pen_width=b[1], output="triang-delaunay.pdf")
927
928
929
930
    <...>

    2D triangulation of random points:

931
932
    .. image:: triang.*
    .. image:: triang-delaunay.*
933

934
935
936
    *Left:* Simple triangulation. *Right:* Delaunay triangulation. The vertex
    colors and the edge thickness correspond to the weighted betweenness
    centrality.
937
938
939

    References
    ----------
Tiago Peixoto's avatar
Tiago Peixoto committed
940
    .. [cgal-triang] http://www.cgal.org/Manual/last/doc_html/cgal_manual/Triangulation_3/Chapter_main.html
941
942
943

    """

Tiago Peixoto's avatar
Tiago Peixoto committed
944
    if points.shape[1] not in [2, 3]:
945
946
947
948
949
950
951
952
953
954
        raise ValueError("points array must have shape N x d, with d either 2 or 3.")
    # copy points to ensure continuity and correct data type
    points = numpy.array(points, dtype='float64')
    if points.shape[1] == 2:
        npoints = numpy.zeros((points.shape[0], 3))
        npoints[:,:2] = points
        points = npoints
    g = Graph(directed=False)
    pos = g.new_vertex_property("vector<double>")
    libgraph_tool_generation.triangulation(g._Graph__graph, points,
955
                                           _prop("v", g, pos), type, periodic)
956
    return g, pos
957
958
959
960
961
962
963
964
965
966


def lattice(shape, periodic=False):
    r"""
    Generate a N-dimensional square lattice.

    Parameters
    ----------
    shape : list or :class:`~numpy.ndarray`
        List of sizes in each dimension.
967
    periodic : bool (optional, default: ``False``)
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
        If ``True``, periodic boundary conditions will be used.

    Returns
    -------
    lattice_graph : :class:`~graph_tool.Graph`
        The generated graph.

    See Also
    --------
    triangulation: 2D or 3D triangulation
    random_graph: random graph generation

    Examples
    --------
    >>> g = gt.lattice([10,10])
Tiago Peixoto's avatar
Tiago Peixoto committed
983
    >>> gt.graph_draw(g, pos=gt.sfdp_layout(g, cooling_step=0.95, epsilon=1e-2),
984
    ...               output_size=(300,300), output="lattice.pdf")
985
986
    <...>
    >>> g = gt.lattice([10,20], periodic=True)
Tiago Peixoto's avatar
Tiago Peixoto committed
987
    >>> gt.graph_draw(g, pos=gt.sfdp_layout(g, cooling_step=0.95, epsilon=1e-2, multilevel=True),
988
    ...               output_size=(300,300), output="lattice_periodic.pdf")
989
990
    <...>
    >>> g = gt.lattice([10,10,10])
Tiago Peixoto's avatar
Tiago Peixoto committed
991
    >>> gt.graph_draw(g, pos=gt.sfdp_layout(g, cooling_step=0.95, epsilon=1e-2, multilevel=True),
992
    ...               output_size=(300,300), output="lattice_3d.pdf")
993
994
    <...>

995
996
997
    .. image:: lattice.*
    .. image:: lattice_periodic.*
    .. image:: lattice_3d.*
998
999
1000

    *Left:* 10x10 2D lattice. *Middle:* 10x20 2D periodic lattice (torus).
    *Right:* 10x10x10 3D lattice.