def test_tensor_product_2(k1, k2):
k = TensorProduct(x=k1, y=k2)
mirror = eval(repr(k)) # representation meaningness test
assert(mirror.theta == k.theta)
for i1, j1 in [(0, 0), (0, 1.5), (-1, 1), (-1.0, 0)]:
for i2, j2 in [(0, 0), (0, 1.5), (-1, 1), (-1.0, 0)]:
''' default and corner cases '''
assert(k(dict(x=i1, y=i2), dict(x=j1, y=j2))
== k1(i1, j1) * k2(i2, j2))
assert(k(dict(x=i1, y=i2), dict(x=j1, y=j2))
== mirror(dict(x=i1, y=i2), dict(x=j1, y=j2)))
for _ in range(10000):
i1 = random.paretovariate(0.1)
j1 = random.paretovariate(0.1)
i2 = random.paretovariate(0.1)
j2 = random.paretovariate(0.1)
''' check by definition '''
assert(k(dict(x=i1, y=j1), dict(x=i2, y=j2))
== k1(i1, i2) * k2(j1, j2))
assert(k(dict(x=i1, y=i2), dict(x=j1, y=j2))
== mirror(dict(x=i1, y=i2), dict(x=j1, y=j2)))
''' hyperparameter retrieval '''
assert(k1.theta in k.theta)
assert(k2.theta in k.theta)
k.theta = k.theta
''' representation generation '''
assert(len(str(k).split('⊗')) == 2)
assert(str(k1) in str(k))
assert(str(k2) in str(k))
assert(repr(k1) in repr(k))
assert(repr(k2) in repr(k))
''' C++ code generation '''
assert(k.dtype.isalignedstruct)
assert(isinstance(k.gencode('x', 'y'), str))
def test_tensor_product_3(k1, k2, k3):
k = TensorProduct(x=k1, y=k2, z=k3)
mirror = eval(repr(k)) # representation meaningness test
assert(mirror.theta == k.theta)
''' default and corner cases only '''
for x1, y1, z1 in [(0, 0, 0), (0, 1, -1), (-1, 1, 0.5), (0, -42., 1)]:
for x2, y2, z2 in [(0, 0, 0), (0, 1, -1), (-1, 1, 0.5), (0, -42., 1)]:
''' default and corner cases '''
assert(k(dict(x=x1, y=y1, z=z1), dict(x=x2, y=y2, z=z2))
== (k1(x1, x2)
* k2(y1, y2)
* k3(z1, z2)))
assert(k(dict(x=x1, y=y1, z=z1), dict(x=x2, y=y2, z=z2))
== mirror(dict(x=x1, y=y1, z=z1), dict(x=x2, y=y2, z=z2)))
''' hyperparameter retrieval '''
assert(k1.theta in k.theta)
assert(k2.theta in k.theta)
assert(k3.theta in k.theta)
k.theta = k.theta
''' representation generation '''
assert(len(str(k).split('⊗')) == 3)
assert(str(k1) in str(k))
assert(str(k2) in str(k))
assert(str(k3) in str(k))
assert(repr(k1) in repr(k))
assert(repr(k2) in repr(k))
assert(repr(k3) in repr(k))
''' C++ code generation '''
assert(k.dtype.isalignedstruct)
assert(isinstance(k.gencode('x', 'y'), str))
def _makekernel(self, **kwargs):
self.kernel = MarginalizedGraphKernel(
TensorProduct(element=KroneckerDelta(self.element_prior, 1.0)),
TensorProduct(length=SquareExponential(self.edge_length_scale)),
q=self.stopping_probability,
p=self.starting_probability,
**kwargs)
# {1.0, 1} --[1.0]-- {2.0, 1}
# \ /
# [0.5] [2.0]
# \ /
# {1.0, 2}
g3 = nx.Graph()
g3.add_node(0, radius=1.0, category=1)
g3.add_node(1, radius=2.0, category=1)
g3.add_node(2, radius=1.0, category=2)
g3.add_edge(0, 1, w=1.0)
g3.add_edge(0, 2, w=0.5)
g3.add_edge(1, 2, w=2.0)
# define node and edge kernelets
knode = TensorProduct(radius=SquareExponential(0.5),
category=KroneckerDelta(0.5))
kedge = Constant(1.0)
# compose the marginalized graph kernel and compute pairwise similarity
mlgk = MarginalizedGraphKernel(knode, kedge, q=0.05)
R = mlgk([Graph.from_networkx(g, weight='w') for g in [g1, g2, g3]])
# normalize the similarity matrix
d = np.diag(R)**-0.5
K = np.diag(d).dot(R).dot(np.diag(d))
print(K)
Graph.from_networkx(unlabeled_graph1),
Graph.from_networkx(unlabeled_graph2)
],
'knode':
Constant(1.0),
'kedge':
Constant(1.0),
'q': [0.01, 0.05, 0.1, 0.5]
},
'labeled': {
'graphs': [
Graph.from_networkx(labeled_graph1),
Graph.from_networkx(labeled_graph2)
],
'knode':
TensorProduct(hybridization=KroneckerDelta(0.3, 1.0),
charge=SquareExponential(1.0)),
'kedge':
TensorProduct(order=KroneckerDelta(0.3, 1.0),
length=SquareExponential(0.05)),
'q': [0.01, 0.05, 0.1, 0.5]
},
'weighted': {
'graphs': [
Graph.from_networkx(weighted_graph1, weight='w'),
Graph.from_networkx(weighted_graph2, weight='w')
],
'knode':
TensorProduct(hybridization=KroneckerDelta(0.3, 1.0),
charge=SquareExponential(1.0)),
'kedge':
TensorProduct(order=KroneckerDelta(0.3, 1.0),
'CC', # ethane
'CCO', # acetic acid
'CCN', # ethylamine
'C=C', # ethene
'CC=C', # propene
'CC=CC', # 2-n-butene
]
# convert to molecular graphs
# nodes(atoms) has 'aromatic', 'charge', 'element', 'hcount' attributes
# edges(bonds) has the 'order' attribute
graphs = [Graph.from_smiles(smi) for smi in smiles_list]
# define node and edge kernelets
knode = TensorProduct(aromatic=KroneckerDelta(0.8, 1.0),
charge=SquareExponential(1.0),
element=KroneckerDelta(0.5, 1.0),
hcount=SquareExponential(1.0))
kedge = TensorProduct(order=KroneckerDelta(0.5, 1.0))
# compose the marginalized graph kernel and compute pairwise similarity
kernel = MarginalizedGraphKernel(knode, kedge, q=0.05)
R = kernel(graphs)
# normalize the similarity matrix and then print
d = np.diag(R)**-0.5
K = np.diag(d).dot(R).dot(np.diag(d))
print(pd.DataFrame(K, columns=smiles_list, index=smiles_list))
# \ /
# \ /
# {2.0}[0.5] 1.0}[2.0]
# \ /
# \ /
# {1.0, 2}
g3 = nx.Graph()
g3.add_node(0, radius=1.0, category=1)
g3.add_node(1, radius=2.0, category=1)
g3.add_node(2, radius=1.0, category=2)
g3.add_edge(0, 1, w=1.0, length=1.5)
g3.add_edge(0, 2, w=0.5, length=2.0)
g3.add_edge(1, 2, w=2.0, length=1.0)
# define node and edge kernelets
knode = TensorProduct(radius=SquareExponential(1.0),
category=KroneckerDelta(0.5))
kedge = TensorProduct(length=SquareExponential(1.0))
# compose the marginalized graph kernel and compute pairwise similarity
mlgk = MarginalizedGraphKernel(knode, kedge, q=0.05)
R = mlgk([Graph.from_networkx(g, weight='w') for g in [g1, g2, g3]])
# normalize the similarity matrix
d = np.diag(R)**-0.5
K = np.diag(d).dot(R).dot(np.diag(d))
print(K)