def setup(self):
a = np.arange(4)
vec = self.declare_input('a', val=a)
# Summation of all the entries of a vector
self.register_output('einsum_summ1',
ot.einsum_new_api(vec, operation=[(33, )]))
def setup(self):
a = np.arange(4)
vec = self.declare_input('a', val=a)
self.register_output(
'einsum_summ1_sparse_derivs',
ot.einsum_new_api(vec, operation=[(33, )],
partial_format='sparse'))
def setup(self):
shape2 = (5, 4)
b = np.arange(20).reshape(shape2)
mat = self.declare_input('b', val=b)
# reorder of a matrix
self.register_output(
'einsum_reorder1',
ot.einsum_new_api(mat, operation=[(46, 99), (99, 46)]))
def setup(self):
a = np.arange(4)
vec = self.declare_input('a', val=a)
# Inner Product of 2 vectors
self.register_output(
'einsum_inner1',
ot.einsum_new_api(vec, vec, operation=[(0, ), (0, )]))
def setup(self):
a = np.arange(4)
vec = self.declare_input('a', val=a)
# Special operation: sum all the entries of the first and second
# vector to a single scalar
self.register_output(
'einsum_special2',
ot.einsum_new_api(vec, vec, operation=[(1, ), (2, )]))
def setup(self):
a = np.arange(4)
vec = self.declare_input('a', val=a)
self.register_output(
'einsum_outer1',
ot.einsum_new_api(
vec,
vec,
operation=[('rows', ), ('cols', ), ('rows', 'cols')],
))
def setup(self):
shape2 = (5, 4)
b = np.arange(20).reshape(shape2)
mat = self.declare_input('b', val=b)
self.register_output(
'einsum_reorder1_sparse_derivs',
ot.einsum_new_api(mat,
operation=[(46, 99), (99, 46)],
partial_format='sparse'))
def setup(self):
# Shape of Tensor
shape3 = (2, 4, 3)
c = np.arange(24).reshape(shape3)
# Declaring tensor
tens = self.declare_input('c', val=c)
self.register_output(
'einsum_summ2_sparse_derivs',
ot.einsum_new_api(tens,
operation=[(33, 66, 99)],
partial_format='sparse'))
def setup(self):
# Shape of Tensor
shape3 = (2, 4, 3)
c = np.arange(24).reshape(shape3)
# Declaring tensor
tens = self.declare_input('c', val=c)
# Summation of all the entries of a tensor
self.register_output(
'einsum_summ2', ot.einsum_new_api(
tens,
operation=[(33, 66, 99)],
))
def setup(self):
a = np.arange(4)
vec = self.declare_input('a', val=a)
# Special operation: summation of all the entries of first
# vector and scalar multiply the second vector with the computed
# sum
self.register_output(
'einsum_special1',
ot.einsum_new_api(
vec,
vec,
operation=[(1, ), (2, ), (2, )],
))
def setup(self):
# Shape of Tensor
shape3 = (2, 4, 3)
c = np.arange(24).reshape(shape3)
# Declaring tensor
tens = self.declare_input('c', val=c)
# reorder of a tensor
self.register_output(
'einsum_reorder2',
ot.einsum_new_api(
tens,
operation=[(33, 66, 99), (99, 66, 33)],
))
def setup(self):
a = np.arange(4)
vec = self.declare_input('a', val=a)
# Shape of Tensor
shape3 = (2, 4, 3)
c = np.arange(24).reshape(shape3)
# Declaring tensor
tens = self.declare_input('c', val=c)
# Outer Product of a tensor and a vector
self.register_output(
'einsum_outer2',
ot.einsum_new_api(
tens,
vec,
operation=[(0, 1, 30), (2, ), (0, 1, 30, 2)],
))
def setup(self):
a = np.arange(4)
vec = self.declare_input('a', val=a)
# Shape of Tensor
shape3 = (2, 4, 3)
c = np.arange(24).reshape(shape3)
# Declaring tensor
tens = self.declare_input('c', val=c)
self.register_output(
'einsum_outer2_sparse_derivs',
ot.einsum_new_api(tens,
vec,
operation=[
(0, 1, 30),
(2, ),
(0, 1, 30, 2),
],
partial_format='sparse'))
