def batch_rz_opt(batch_params):
batch_params = batch_params /2
batch_real = None
batch_imag = None
for i in range(batch_params.size(1)):
params_slice = batch_params[:,i:i+1]
params_real = tcat([ones(1,2)*cos(param) for param in params_slice], dim=0)
params_imag = tcat([tcat([-sin(param).view(1,1),sin(param).view(1,1)], dim=1) for param in params_slice], dim=0)
batch_real_ = params_real if batch_real is None else batch_k_rz(params_real,batch_real) - batch_k_rz(params_imag,batch_imag)
batch_imag_ = params_imag if batch_imag is None else batch_k_rz(params_real,batch_imag) + batch_k_rz(params_imag,batch_real)
batch_real = batch_real_
batch_imag = batch_imag_
hm_params = batch_real.size(1)
batch_real = tstack([tstack([batch_real[:,rowcol]]*hm_params,dim=1) for rowcol in range(hm_params)],dim=2)
batch_imag = tstack([tstack([batch_imag[:,rowcol]]*hm_params,dim=1) for rowcol in range(hm_params)],dim=2)
return ComplexTensor.__graph_copy__(None,batch_real,batch_imag)
def ry_a0(params_slice, qbit_id):
batch_size = params_slice.size(0)
sector_size = 2**qbit_id
group_size = sector_size*2
params_slice = params_slice.view(batch_size,1,1)
group = cos(params_slice) * tstack([eye(group_size,group_size)]*batch_size,dim=0)
sector_leftdown = sin(params_slice) * tstack([eye(sector_size,sector_size)]*batch_size,dim=0)
sector_rightup = -sector_leftdown
group[:,sector_size:,:sector_size] = sector_leftdown
group[:,:sector_size,sector_size:] = sector_rightup
hm_repeats = config.statevec_size//group_size
matrix = zeros(batch_size,config.statevec_size,config.statevec_size)
for i in range(hm_repeats):
f = i*group_size
t = (i+1)*group_size
matrix[:,f:t,f:t] = group
return matrix
def batch_rz(batch_params):
batch_params = batch_params /2
batch_real = None
batch_imag = None
for i in range(batch_params.size(1)):
params_slice = batch_params[:,i:i+1]
params_real = tcat([tcat([cos(param).view(1,1)]*2, dim=1) for param in params_slice], dim=0)
params_imag = tcat([tcat([-sin(param).view(1,1),sin(param).view(1,1)], dim=1) for param in params_slice], dim=0)
batch_real_ = params_real if batch_real is None else batch_k_rz(params_real,batch_real) - batch_k_rz(params_imag,batch_imag)
batch_imag_ = params_imag if batch_imag is None else batch_k_rz(params_real,batch_imag) + batch_k_rz(params_imag,batch_real)
batch_real = batch_real_
batch_imag = batch_imag_
batch_real = tstack([eye(batch_real.size(1),batch_real.size(1))]*batch_real.size(0),dim=0) \
* batch_real.view(batch_real.size(0),batch_real.size(1),1)
batch_imag = tstack([eye(batch_imag.size(1),batch_imag.size(1))]*batch_imag.size(0),dim=0) \
* batch_imag.view(batch_real.size(0),batch_real.size(1),1)
return ComplexTensor.__graph_copy__(None,batch_real,batch_imag)
def rotation_2d(points, angles):
rot_sin = torch.sin(angles)
rot_cos = torch.cos(angles)
rot_mat_T = torch.stack(
[tstack([rot_cos, rot_sin]),
tstack([-rot_sin, rot_cos])])
return torch.einsum('aij,jka->aik', (points, rot_mat_T))
def rotation_2d(points, angles):
"""rotation 2d points based on origin point clockwise when angle positive.
Args:
points (float array, shape=[N, point_size, 2]): points to be rotated.
angles (float array, shape=[N]): rotation angle.
Returns:
float array: same shape as points
"""
rot_sin = torch.sin(angles)
rot_cos = torch.cos(angles)
rot_mat_T = torch.stack([tstack([rot_cos, -rot_sin]), tstack([rot_sin, rot_cos])])
return torch.einsum("aij,jka->aik", (points, rot_mat_T))
def batch_rx(batch_params):
batch_params = batch_params /2
batch_real = None
batch_imag = None
for i in range(batch_params.size(1)):
params_slice = batch_params[:,i:i+1]
params_real = tstack([tcat([tcat([cos(param).view(1,1),tensor(.0).view(1,1)],dim=1),tcat([tensor(.0).view(1,1),cos(param).view(1,1)],dim=1)],dim=0) for param in params_slice],dim=0)
params_imag = tstack([tcat([tcat([tensor(.0).view(1,1),-sin(param).view(1,1)],dim=1),tcat([-sin(param).view(1,1),tensor(.0).view(1,1)],dim=1)],dim=0) for param in params_slice],dim=0)
batch_real_ = params_real if batch_real is None else batch_k_rx(params_real,batch_real) - batch_k_rx(params_imag,batch_imag)
batch_imag_ = params_imag if batch_imag is None else batch_k_rx(params_real,batch_imag) + batch_k_rx(params_imag,batch_real)
batch_real = batch_real_
batch_imag = batch_imag_
return ComplexTensor.__graph_copy__(None, batch_real, batch_imag)
def batch_u3(batch_params):
batch_real = None
batch_imag = None
for i in range(batch_params.size(1) // 3):
params_slice = batch_params[:,i*3:(i+1)*3]
params_real = tstack([tcat([tcat([cos(p0/2).view(1,1), -(cos(p2)*sin(p0/2)).view(1,1)], dim=1),
tcat([(cos(p1)*sin(p0/2)).view(1,1), (cos(p1+p2)*cos(p0/2)).view(1,1)], dim=1)], dim=0) for p0,p1,p2 in
params_slice], dim=0)
params_imag = tstack([tcat([tcat([tensor(.0).view(1,1), -(sin(p2)*sin(p0/2)).view(1,1)], dim=1),
tcat([(sin(p1)*sin(p0/2)).view(1,1), (sin(p1+p2)*cos(p0/2)).view(1,1)], dim=1)], dim=0) for p0,p1,p2 in
params_slice], dim=0)
batch_real_ = params_real if batch_real is None else batch_k_rx(params_real, batch_real) - batch_k_rx(params_imag, batch_imag)
batch_imag_ = params_imag if batch_imag is None else batch_k_rx(params_real, batch_imag) + batch_k_rx(params_imag, batch_real)
batch_real = batch_real_
batch_imag = batch_imag_
return ComplexTensor.__graph_copy__(None, batch_real, batch_imag)
def batch_ry_opt(batch_params):
batch_params = batch_params /2
batch_real = None
for i in range(batch_params.size(1)):
params_slice = batch_params[:,i:i+1]
params_real = tstack([tcat([tcat([cos(param).view(1,1), -sin(param).view(1,1)],dim=1),
tcat([sin(param).view(1,1), cos(param).view(1,1)],dim=1)], dim=0) for param in
params_slice], dim=0)
batch_real = params_real if batch_real is None else batch_k_rx(params_real, batch_real)
return batch_real
def batch_ry(batch_params):
batch_params = batch_params /2
batch_real = None
for i in range(batch_params.size(1)):
params_slice = batch_params[:,i:i+1]
params_real = tstack([tcat([tcat([cos(param).view(1,1), -sin(param).view(1,1)],dim=1),
tcat([sin(param).view(1,1), cos(param).view(1,1)],dim=1)], dim=0) for param in
params_slice], dim=0)
batch_real = params_real if batch_real is None else batch_k_rx(params_real, batch_real)
return ComplexTensor.__graph_copy__(None, batch_real, zeros(batch_real.size()))
def batch_a0(batch_params):
batch_params = batch_params /2
cx_guide, circuit = circuit_a0()
batch_real = tstack([eye(config.statevec_size,config.statevec_size)]*batch_params.size(0),dim=0)
param_ctr = 0
for _ in range(config.circuit_layers):
for element in circuit:
if element[0] == 'RY':
batch_real = ry_a0(batch_params[:,param_ctr],element[1]) @ batch_real
param_ctr += 1
elif element[0] == 'CX':
batch_real = cx_guide[element[1]][element[2]] @ batch_real
return batch_real
def rotation_3d_in_axis(points, angles, axis=0):
# points: [N, point_size, 3]
# angles: [N]
rot_sin = torch.sin(angles)
rot_cos = torch.cos(angles)
ones = torch.ones_like(rot_cos)
zeros = torch.zeros_like(rot_cos)
if axis == 1:
rot_mat_T = tstack([
tstack([rot_cos, zeros, -rot_sin]),
tstack([zeros, ones, zeros]),
tstack([rot_sin, zeros, rot_cos])
])
elif axis == 2 or axis == -1:
rot_mat_T = tstack([
tstack([rot_cos, -rot_sin, zeros]),
tstack([rot_sin, rot_cos, zeros]),
tstack([zeros, zeros, ones])
])
elif axis == 0:
rot_mat_T = tstack([
tstack([zeros, rot_cos, -rot_sin]),
tstack([zeros, rot_sin, rot_cos]),
tstack([ones, zeros, zeros])
])
else:
raise ValueError("axis should in range")
return torch.einsum('aij,jka->aik', (points, rot_mat_T))