How to use the torchkbnufft.KbInterpForw function in torchkbnufft
To help you get started, we’ve selected a few torchkbnufft examples, based on popular ways it is used in public projects.
Secure your code as it's written. Use Snyk Code to scan source code in minutes - no build needed - and fix issues immediately.
mmuckley / torchkbnufft / tests / test_pytorch_grad_adj_matching.py View on Github 
im_size = params_2d['im_size']
grid_size = params_2d['grid_size']
numpoints = params_2d['numpoints']
x = np.random.normal(size=(batch_size, 1) + grid_size) + \
1j*np.random.normal(size=(batch_size, 1) + grid_size)
x = torch.tensor(np.stack((np.real(x), np.imag(x)), axis=2))
y = params_2d['y']
ktraj = params_2d['ktraj']
for device in device_list:
x = x.detach().to(dtype=dtype, device=device)
y = y.detach().to(dtype=dtype, device=device)
ktraj = ktraj.detach().to(dtype=dtype, device=device)
kbinterp_ob = KbInterpForw(
im_size=im_size,
grid_size=grid_size,
numpoints=numpoints
).to(dtype=dtype, device=device)
adjkbinterp_ob = KbInterpBack(
im_size=im_size,
grid_size=grid_size,
numpoints=numpoints
).to(dtype=dtype, device=device)
x.requires_grad = True
y = kbinterp_ob.forward(x, ktraj)
((y ** 2) / 2).sum().backward()
x_grad = x.grad.clone().detach()def test_3d_init_inputs():
# all object initializations have assertions
# this should result in an error if any dimensions don't match
# test 3d scalar inputs
im_sz = (10, 256, 256)
smap = torch.randn(*((1,) + im_sz))
grid_sz = (10, 512, 512)
n_shift = (5, 128, 128)
numpoints = 6
table_oversamp = 2**10
kbwidth = 2.34
order = 0
norm = 'None'
ob = KbInterpForw(
im_size=im_sz, grid_size=grid_sz, n_shift=n_shift, numpoints=numpoints,
table_oversamp=table_oversamp, kbwidth=kbwidth, order=order)
ob = KbInterpBack(
im_size=im_sz, grid_size=grid_sz, n_shift=n_shift, numpoints=numpoints,
table_oversamp=table_oversamp, kbwidth=kbwidth, order=order)
ob = KbNufft(
im_size=im_sz, grid_size=grid_sz, n_shift=n_shift, numpoints=numpoints,
table_oversamp=table_oversamp, kbwidth=kbwidth, order=order, norm=norm)
ob = AdjKbNufft(
im_size=im_sz, grid_size=grid_sz, n_shift=n_shift, numpoints=numpoints,
table_oversamp=table_oversamp, kbwidth=kbwidth, order=order, norm=norm)
ob = MriSenseNufft(
smap=smap, im_size=im_sz, grid_size=grid_sz, n_shift=n_shift, numpoints=numpoints,
table_oversamp=table_oversamp, kbwidth=kbwidth, order=order, norm=norm)im_size = params_3d['im_size']
grid_size = params_3d['grid_size']
numpoints = params_3d['numpoints']
x = np.random.normal(size=(batch_size, 1) + grid_size) + \
1j*np.random.normal(size=(batch_size, 1) + grid_size)
x = torch.tensor(np.stack((np.real(x), np.imag(x)), axis=2))
y = params_3d['y']
ktraj = params_3d['ktraj']
for device in device_list:
x = x.detach().to(dtype=dtype, device=device)
y = y.detach().to(dtype=dtype, device=device)
ktraj = ktraj.detach().to(dtype=dtype, device=device)
kbinterp_ob = KbInterpForw(
im_size=im_size,
grid_size=grid_size,
numpoints=numpoints
).to(dtype=dtype, device=device)
adjkbinterp_ob = KbInterpBack(
im_size=im_size,
grid_size=grid_size,
numpoints=numpoints
).to(dtype=dtype, device=device)
real_mat, imag_mat = precomp_sparse_mats(ktraj, kbinterp_ob)
interp_mats = {
'real_interp_mats': real_mat,
'imag_interp_mats': imag_mat
}im_size = params_2d['im_size']
grid_size = params_2d['grid_size']
numpoints = params_2d['numpoints']
x = np.random.normal(size=(batch_size, 1) + grid_size) + \
1j*np.random.normal(size=(batch_size, 1) + grid_size)
x = torch.tensor(np.stack((np.real(x), np.imag(x)), axis=2))
y = params_2d['y']
ktraj = params_2d['ktraj']
for device in device_list:
x = x.detach().to(dtype=dtype, device=device)
y = y.detach().to(dtype=dtype, device=device)
ktraj = ktraj.detach().to(dtype=dtype, device=device)
kbinterp_ob = KbInterpForw(
im_size=im_size,
grid_size=grid_size,
numpoints=numpoints
).to(dtype=dtype, device=device)
adjkbinterp_ob = KbInterpBack(
im_size=im_size,
grid_size=grid_size,
numpoints=numpoints
).to(dtype=dtype, device=device)
real_mat, imag_mat = precomp_sparse_mats(ktraj, kbinterp_ob)
interp_mats = {
'real_interp_mats': real_mat,
'imag_interp_mats': imag_mat
}im_size = params_2d['im_size']
grid_size = params_2d['grid_size']
numpoints = params_2d['numpoints']
x = np.random.normal(size=(batch_size, 1) + grid_size) + \
1j*np.random.normal(size=(batch_size, 1) + grid_size)
x = torch.tensor(np.stack((np.real(x), np.imag(x)), axis=2))
y = params_2d['y']
ktraj = params_2d['ktraj']
for device in device_list:
x = x.detach().to(dtype=dtype, device=device)
y = y.detach().to(dtype=dtype, device=device)
ktraj = ktraj.detach().to(dtype=dtype, device=device)
kbinterp_ob = KbInterpForw(
im_size=im_size,
grid_size=grid_size,
numpoints=numpoints
).to(dtype=dtype, device=device)
adjkbinterp_ob = KbInterpBack(
im_size=im_size,
grid_size=grid_size,
numpoints=numpoints
).to(dtype=dtype, device=device)
x_forw = kbinterp_ob(x, ktraj)
y_back = adjkbinterp_ob(y, ktraj)
inprod1 = inner_product(y, x_forw, dim=2)
inprod2 = inner_product(y_back, x, dim=2)im_size = params_2d['im_size']
grid_size = params_2d['grid_size']
numpoints = params_2d['numpoints']
x = np.random.normal(size=(batch_size, 1) + grid_size) + \
1j*np.random.normal(size=(batch_size, 1) + grid_size)
x = torch.tensor(np.stack((np.real(x), np.imag(x)), axis=2))
y = params_2d['y']
ktraj = params_2d['ktraj']
for device in device_list:
x = x.detach().to(dtype=dtype, device=device)
y = y.detach().to(dtype=dtype, device=device)
ktraj = ktraj.detach().to(dtype=dtype, device=device)
kbinterp_ob = KbInterpForw(
im_size=im_size,
grid_size=grid_size,
numpoints=numpoints
).to(dtype=dtype, device=device)
adjkbinterp_ob = KbInterpBack(
im_size=im_size,
grid_size=grid_size,
numpoints=numpoints
).to(dtype=dtype, device=device)
real_mat, imag_mat = precomp_sparse_mats(ktraj, kbinterp_ob)
interp_mats = {
'real_interp_mats': real_mat,
'imag_interp_mats': imag_mat
}im_size = params_3d['im_size']
grid_size = params_3d['grid_size']
numpoints = params_3d['numpoints']
x = np.random.normal(size=(batch_size, 1) + grid_size) + \
1j*np.random.normal(size=(batch_size, 1) + grid_size)
x = torch.tensor(np.stack((np.real(x), np.imag(x)), axis=2))
y = params_3d['y']
ktraj = params_3d['ktraj']
for device in device_list:
x = x.detach().to(dtype=dtype, device=device)
y = y.detach().to(dtype=dtype, device=device)
ktraj = ktraj.detach().to(dtype=dtype, device=device)
kbinterp_ob = KbInterpForw(
im_size=im_size,
grid_size=grid_size,
numpoints=numpoints
).to(dtype=dtype, device=device)
adjkbinterp_ob = KbInterpBack(
im_size=im_size,
grid_size=grid_size,
numpoints=numpoints
).to(dtype=dtype, device=device)
real_mat, imag_mat = precomp_sparse_mats(ktraj, kbinterp_ob)
interp_mats = {
'real_interp_mats': real_mat,
'imag_interp_mats': imag_mat
}im_size = params_2d['im_size']
grid_size = params_2d['grid_size']
numpoints = params_2d['numpoints']
x = np.random.normal(size=(batch_size, 1) + grid_size) + \
1j*np.random.normal(size=(batch_size, 1) + grid_size)
x = torch.tensor(np.stack((np.real(x), np.imag(x)), axis=2))
y = params_2d['y']
ktraj = params_2d['ktraj']
for device in device_list:
x = x.detach().to(dtype=dtype, device=device)
y = y.detach().to(dtype=dtype, device=device)
ktraj = ktraj.detach().to(dtype=dtype, device=device)
kbinterp_ob = KbInterpForw(
im_size=im_size,
grid_size=grid_size,
numpoints=numpoints
).to(dtype=dtype, device=device)
adjkbinterp_ob = KbInterpBack(
im_size=im_size,
grid_size=grid_size,
numpoints=numpoints
).to(dtype=dtype, device=device)
real_mat, imag_mat = precomp_sparse_mats(ktraj, kbinterp_ob)
interp_mats = {
'real_interp_mats': real_mat,
'imag_interp_mats': imag_mat
}im_size = params_2d['im_size']
grid_size = params_2d['grid_size']
numpoints = params_2d['numpoints']
x = np.random.normal(size=(batch_size, 1) + grid_size) + \
1j*np.random.normal(size=(batch_size, 1) + grid_size)
x = torch.tensor(np.stack((np.real(x), np.imag(x)), axis=2))
y = params_2d['y']
ktraj = params_2d['ktraj']
for device in device_list:
x = x.detach().to(dtype=dtype, device=device)
y = y.detach().to(dtype=dtype, device=device)
ktraj = ktraj.detach().to(dtype=dtype, device=device)
kbinterp_ob = KbInterpForw(
im_size=im_size,
grid_size=grid_size,
numpoints=numpoints
).to(dtype=dtype, device=device)
adjkbinterp_ob = KbInterpBack(
im_size=im_size,
grid_size=grid_size,
numpoints=numpoints
).to(dtype=dtype, device=device)
y.requires_grad = True
x = adjkbinterp_ob.forward(y, ktraj)
((x ** 2) / 2).sum().backward()
y_grad = y.grad.clone().detach()im_size = params_3d['im_size']
grid_size = params_3d['grid_size']
numpoints = params_3d['numpoints']
x = np.random.normal(size=(batch_size, 1) + grid_size) + \
1j*np.random.normal(size=(batch_size, 1) + grid_size)
x = torch.tensor(np.stack((np.real(x), np.imag(x)), axis=2))
y = params_3d['y']
ktraj = params_3d['ktraj']
for device in device_list:
x = x.detach().to(dtype=dtype, device=device)
y = y.detach().to(dtype=dtype, device=device)
ktraj = ktraj.detach().to(dtype=dtype, device=device)
kbinterp_ob = KbInterpForw(
im_size=im_size,
grid_size=grid_size,
numpoints=numpoints
).to(dtype=dtype, device=device)
adjkbinterp_ob = KbInterpBack(
im_size=im_size,
grid_size=grid_size,
numpoints=numpoints
).to(dtype=dtype, device=device)
x_forw = kbinterp_ob(x, ktraj)
y_back = adjkbinterp_ob(y, ktraj)
inprod1 = inner_product(y, x_forw, dim=2)
inprod2 = inner_product(y_back, x, dim=2)torchkbnufft
A high-level, easy-to-deploy non-uniform Fast Fourier Transform in PyTorch.
Package Health Score
72 / 100Popular torchkbnufft functions
- torchkbnufft.AdjKbNufft
- torchkbnufft.AdjMriSenseNufft
- torchkbnufft.functional.kbinterp.AdjKbInterpFunction.apply
- torchkbnufft.functional.kbinterp.KbInterpFunction.apply
- torchkbnufft.functional.kbnufft.AdjKbNufftFunction
- torchkbnufft.functional.kbnufft.KbNufftFunction
- torchkbnufft.kbinterp.KbInterpModule
- torchkbnufft.KbInterpBack
- torchkbnufft.KbInterpForw
- torchkbnufft.kbmodule.KbModule
- torchkbnufft.KbNufft
- torchkbnufft.kbnufft.KbNufftModule
- torchkbnufft.math.complex_mult
- torchkbnufft.math.conj_complex_mult
- torchkbnufft.math.inner_product
- torchkbnufft.MriSenseNufft
- torchkbnufft.mrisensenufft.SenseNufftModule
- torchkbnufft.nufft.sparse_interp_mat.precomp_sparse_mats