1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
|
import importlib
import os
import sparse
import networkx as nx
from utils import benchmark
import numpy as np
ITERS = 3
if __name__ == "__main__":
print("Counting Triangles Example:\n")
G = nx.gnp_random_graph(n=200, p=0.2)
# ======= Finch =======
os.environ[sparse._ENV_VAR_NAME] = "Finch"
importlib.reload(sparse)
a_sps = nx.to_scipy_sparse_array(G)
a = sparse.asarray(a_sps)
@sparse.compiled()
def count_triangles_finch(a):
return sparse.sum(a @ a * a) / sparse.asarray(6)
# Compile & Benchmark
result_finch = benchmark(count_triangles_finch, args=[a], info="Finch", iters=ITERS)
# ======= SciPy =======
def count_triangles_scipy(a):
return (a @ a * a).sum() / 6
a = nx.to_scipy_sparse_array(G)
# Compile & Benchmark
result_scipy = benchmark(count_triangles_scipy, args=[a], info="SciPy", iters=ITERS)
# ======= NetworkX =======
def count_triangles_networkx(a):
return sum(nx.triangles(a).values()) / 3
a = G
# Compile & Benchmark
result_networkx = benchmark(count_triangles_networkx, args=[a], info="NetworkX", iters=ITERS)
np.testing.assert_equal(result_finch.todense(), result_scipy)
np.testing.assert_equal(result_finch.todense(), result_networkx)
assert result_networkx == result_scipy
|
