Skip to content

Commit

Permalink
Merge pull request #1196 from fabratu/20240322_fix_gen_powerlaw
Browse files Browse the repository at this point in the history 
Fix PowerlawSequence generator (Python)
  • Loading branch information
@fabratu
fabratu committed Mar 27, 2024
2 parents faf5e01 + 3cad44a commit 3c7d0d2
Show file tree
Hide file tree
Showing 2 changed files with 53 additions and 38 deletions.
9 changes: 5 additions & 4 deletions networkit/generators.pyx
View file
Expand Up @@ -653,10 +653,11 @@ cdef class PowerlawDegreeSequence:
def __cinit__(self, minDeg, count maxDeg = 0, double gamma = -2):
if isinstance(minDeg, Graph):
self._this = new _PowerlawDegreeSequence((<Graph>minDeg)._this)
try:
self._this = new _PowerlawDegreeSequence(<vector[double]?>minDeg)
except TypeError:
self._this = new _PowerlawDegreeSequence((<count?>minDeg), maxDeg, gamma)
else:
try:
self._this = new _PowerlawDegreeSequence(<vector[double]?>minDeg)
except TypeError:
self._this = new _PowerlawDegreeSequence((<count?>minDeg), maxDeg, gamma)

def __dealloc__(self):
del self._this
Expand Down
82 changes: 48 additions & 34 deletions networkit/test/test_generators.py
View file
Expand Up @@ -2,11 +2,12 @@
import unittest
import networkit as nk


class TestGraph(unittest.TestCase):

def setUp(self):
self.G = nk.readGraph("input/looptest1.gml", nk.Format.GML)

def testClusteredRandomGraphGenerator(self):
n, k = 100, 10
pIntra, pInter = 0.1, 0.05
Expand All @@ -16,22 +17,22 @@ def testClusteredRandomGraphGenerator(self):
nCommunities = len(set(gen.getCommunities().getVector()))
self.assertGreater(nCommunities, 0)
self.assertLessEqual(nCommunities, k)

def testBarabasiAlbertGenerator(self):
g = nk.generators.BarabasiAlbertGenerator(2, 50, 1)
g.fit(self.G)
g.fit(self.G)
self.assertEqual(g.generate().numberOfNodes(), 50)
self.assertEqual(g.generate().numberOfEdges(), 98)

def testBTERReplicator(self):
g = nk.generators.BTERReplicator(self.G)
g.setPaths("./BTER.data")
g.fit(self.G)

def testChungLuGenerator(self):
g = nk.generators.ChungLuGenerator([1.3, 1.5, 1.7])
g.fit(self.G)
self.assertEqual(g.generate().numberOfNodes(), 3)
self.assertEqual(g.generate().numberOfNodes(), 3)

def testDorogovtsevMendesGenerator(self):
g = nk.generators.DorogovtsevMendesGenerator(100)
Expand All @@ -45,7 +46,7 @@ def testDynamicHyperbolicGenerator(self):
res = g.getGraph()
coords = g.getCoordinates()
self.assertEqual(res.numberOfNodes(), 100)
self.assertEqual(len(coords), 100)
self.assertEqual(len(coords), 100)

def testDynamicPubWebGenerator(self):
n = 100
Expand All @@ -59,9 +60,9 @@ def testDynamicPubWebGenerator(self):

# check that all points are within the unit square and that they are non-empt
self.assertTrue(all(0.0 <= pt[0] <= 1 for pt in coords))
self.assertTrue(any(0.0 < pt[0] < 1 for pt in coords))
self.assertTrue(any(0.0 < pt[0] < 1 for pt in coords))
self.assertTrue(all(0.0 <= pt[1] <= 1 for pt in coords))
self.assertTrue(any(0.0 < pt[1] < 1 for pt in coords))
self.assertTrue(any(0.0 < pt[1] < 1 for pt in coords))

gen.generate(10)

Expand All @@ -72,16 +73,18 @@ def testDynamicPubWebGenerator(self):
def testDynamicForestFireGenerator(self):
dyn = nk.generators.DynamicForestFireGenerator(0.5, directed=True)
G = dyn.generate(10)
self.assertGreaterEqual(len(G), 20)
self.assertGreaterEqual(len(G), 20)

def testEdgeSwitchingMarkovChainGenerator(self):
n = 100
while True:
degseq = nk.generators.PowerlawDegreeSequence(2, n).run().getDegreeSequence(n)
degseq = (
nk.generators.PowerlawDegreeSequence(2, n).run().getDegreeSequence(n)
)
gen = nk.generators.EdgeSwitchingMarkovChainGenerator(degseq, False, 10)
if not gen.isRealizable():
continue

G = gen.generate()
self.assertEqual(G.numberOfNodes(), n)
for u in range(G.numberOfNodes()):
Expand All @@ -95,13 +98,13 @@ def testErdosRenyiGenerator(self):
self.assertEqual(g.generate().numberOfNodes(), 10)

def testHavelHakimiGenerator(self):
g = nk.generators.HavelHakimiGenerator([2,2,2,2])
self.assertTrue(g.isRealizable())
g = nk.generators.HavelHakimiGenerator([2, 2, 2, 2])
self.assertTrue(g.isRealizable())
self.assertTrue(g.getRealizable())

def testHavelHakimiGeneratorFit(self):
g = nk.generators.HavelHakimiGenerator([2,2,2,2])
g.fit(self.G)
g = nk.generators.HavelHakimiGenerator([2, 2, 2, 2])
g.fit(self.G)
self.assertEqual(g.generate().numberOfNodes(), 4)
self.assertEqual(g.generate().numberOfEdges(), 4)

Expand All @@ -114,25 +117,25 @@ def testHyperbolicGenerator(self):

def testLFRGenerator(self):
g = nk.generators.LFRGenerator(5)
g.setDegreeSequence([0,0,0,0,0])
g.setDegreeSequence([0, 0, 0, 0, 0])
g.setMu(1.5)
g.setCommunitySizeSequence([1,1,1,1,1])
g.setCommunitySizeSequence([1, 1, 1, 1, 1])
self.assertEqual(g.generate().numberOfNodes(), 5)
g.run()
self.assertEqual(g.getGraph().numberOfEdges(), 0)
self.assertEqual(g.getPartition().numberOfElements(), 5)

def testLFRGeneratorFit(self):
g = nk.generators.LFRGenerator(5)
g.setDegreeSequence([0,0,0,0,0])
g.setDegreeSequence([0, 0, 0, 0, 0])
g.setMu(1.5)
g.setCommunitySizeSequence([1,1,1,1,1])
g.setCommunitySizeSequence([1, 1, 1, 1, 1])
g.fit(self.G)
self.assertEqual(g.generate().numberOfNodes(), 5)
self.assertEqual(g.generate().numberOfEdges(), 0)
self.assertEqual(g.generate().numberOfEdges(), 0)

def testLFRGeneratorPowerlawDegreeSequence(self):
#Example with generatingPowerlaw functions
# Example with generatingPowerlaw functions
g2 = nk.generators.LFRGenerator(100)
g2.generatePowerlawDegreeSequence(10, 20, -2)
g2.generatePowerlawCommunitySizeSequence(10, 50, -1)
Expand All @@ -155,9 +158,9 @@ def testLFRGeneratorPowerlawDegreeSequenceFitVanilla(self):
g2.generatePowerlawDegreeSequence(5, 6, -2)
g2.generatePowerlawCommunitySizeSequence(5, 6, -1)
g2.setMu(0.5)
g2.fit(self.G, vanilla=True, plfit=False)
g2.fit(self.G, vanilla=True, plfit=False)
self.assertEqual(g2.generate().numberOfNodes(), 10)

def testPowerLawDegreeSequence(self):
n = 100
PowGen = nk.generators.PowerlawDegreeSequence(2, n)
Expand All @@ -172,6 +175,16 @@ def testPowerLawDegreeSequence(self):
# Algorithm is not deterministic
self.assertIsInstance(PowGen.getDegree(), int)

def testPowerLawDegreeSequenceFromGraph(self):
PowGen = nk.generators.PowerlawDegreeSequence(self.G)
PowGen.run()

degG = nk.centrality.DegreeCentrality(self.G).run().scores()
minDegG = min(degG)
maxDegG = max(degG)
self.assertEqual(PowGen.getMinimumDegree(), minDegG)
self.assertEqual(PowGen.getMaximumDegree(), maxDegG)

def testPubWebGenerator(self):
n = 100
gen = nk.generators.PubWebGenerator(n, 8, 0.1, 4)
Expand All @@ -184,9 +197,9 @@ def testPubWebGenerator(self):

# check that all points are within the unit square and that they are non-empt
self.assertTrue(all(0.0 <= pt[0] <= 1 for pt in coords))
self.assertTrue(any(0.0 < pt[0] < 1 for pt in coords))
self.assertTrue(any(0.0 < pt[0] < 1 for pt in coords))
self.assertTrue(all(0.0 <= pt[1] <= 1 for pt in coords))
self.assertTrue(any(0.0 < pt[1] < 1 for pt in coords))
self.assertTrue(any(0.0 < pt[1] < 1 for pt in coords))

def testRmatGenerator(self):
g = nk.generators.RmatGenerator(2, 1, 0.25, 0.25, 0.25, 0.25)
Expand All @@ -199,19 +212,20 @@ def testRmatGeneratorFit(self):
g = nk.generators.RmatGenerator(2, 1, 0.25, 0.25, 0.25, 0.25)
g.setPaths("./rMat.data")
inputGraph = nk.Graph(3)
inputGraph.addEdge(0,1)
inputGraph.addEdge(1,2)
g.fit(inputGraph, kronfit=False)
inputGraph.addEdge(0, 1)
inputGraph.addEdge(1, 2)
g.fit(inputGraph, kronfit=False)
self.assertEqual(g.generate().numberOfNodes(), 4)
self.assertEqual(g.generate().numberOfEdges(), 4)
self.assertEqual(g.generate().numberOfEdges(), 4)

def testWattsStrogatzGenerator(self):
n = 12
nbrs = 5
G = nk.generators.WattsStrogatzGenerator(n, nbrs, 0.5).generate()

self.assertEqual(G.numberOfNodes(), n)
self.assertEqual(G.numberOfEdges(), n*nbrs)
self.assertEqual(G.numberOfEdges(), n * nbrs)


if __name__ == "__main__":
unittest.main()

0 comments on commit 3c7d0d2

Please sign in to comment.