def test_pid_dep3():
"""
Test that idep and ibroja differ on reduced or.
"""
d = bivariates['reduced or']
pid1 = PID_BROJA(d)
pid2 = PID_dep(d)
assert pid1 != pid2
def test_pid_broja3():
"""
Test that iproj and ibroja are the same on reduced or.
"""
d = bivariates['reduced or']
pid1 = PID_BROJA(d)
pid2 = PID_Proj(d)
assert pid1 == pid2
def test_pid_broja4():
"""
Test that xor cat is incomplete, but nonnegative and consistent.
"""
d = trivariates['xor cat']
pid = PID_BROJA(d)
assert not pid.complete
assert pid.nonnegative
assert pid.consistent
def test_pid_broja1():
"""
Test ibroja on a generic distribution.
"""
d = bivariates['diff']
pid = PID_BROJA(d, ((0, ), (1, )), (2, ))
assert pid[((0, ), (1, ))] == pytest.approx(0.12255624891826589, abs=1e-4)
assert pid[((0, ), )] == pytest.approx(0.18872187554086706, abs=1e-4)
assert pid[((1, ), )] == pytest.approx(0.18872187554086706, abs=1e-4)
assert pid[((0, 1), )] == pytest.approx(0.0, abs=1e-4)
def test_pid_broja2():
"""
Test ibroja on another generic distribution.
"""
d = trivariates['anddup']
pid = PID_BROJA(d, [[0], [1], [2]], [3])
for atom in pid._lattice:
if atom == ((0, ), (1, ), (2, )):
assert pid[atom] == pytest.approx(0.31127812445913305, abs=1e-4)
elif atom == ((0, 1), (1, 2)):
assert pid[atom] == pytest.approx(0.5, abs=1e-4)
else:
assert pid[atom] == pytest.approx(0.0, abs=1e-4)
dist1 = Distribution(outcomes, pmf)
print(" ")
print("Probability distribution")
print(" ")
print(dist1)
print(" ")
# Compute five PIDs and print the output
print("Partial Information Decompositions")
print(" ")
print("{0}{1} is Shd, {0} is UnqB, {1} is UnqA, {0:1} is Syn")
print(" ")
rcyBROJA = PID_BROJA(dist1, [[0], [1]], [2])
print(" ")
print(" The I_broja PID is plotted in Fig. 6 ")
print(rcyBROJA)
print(" ")
rcyCCS = PID_CCS(dist1, [[0], [1]], [2])
print(rcyCCS)
rcyDEP = PID_dep(dist1, [[0], [1]], [2])
print(rcyDEP)
rcyWB = PID_WB(dist1, [[0], [1]], [2])
output = np.zeros((16, 26))
for i in range(16):
print(i)
freq = mat[i, ]
aa = freq / np.sum(freq)
dat = np.reshape(aa, (4, 4, 3))
dist1 = Distribution.from_ndarray(dat)
p = PID_BROJA(dist1, [[0], [1]], [2])
output[i, 0:4] = [
p.get_partial(n)
for n in sorted(p._lattice, key=dit.pid.lattice.sort_key(p._lattice))
]
p = PID_WB(dist1, [[0], [1]], [2])
output[i, 4:8] = [
p.get_partial(n)
for n in sorted(p._lattice, key=dit.pid.lattice.sort_key(p._lattice))
]
p = PID_Proj(dist1, [[0], [1]], [2])