def Atb(projections, geo, angles, krylov="matched"):
if projections.dtype != np.float32:
raise TypeError("Input data should be float32, not " +
str(projections.dtype))
if not np.isreal(projections).all():
raise ValueError("Complex types not compatible for back projection.")
geox = copy.deepcopy(geo)
geox.check_geo(angles)
"""
Here we cast all values in geo to single point precision float. This way we
know what behaviour to expect from pytigre to Cuda and can change
single parameters accordingly.
"""
geox.cast_to_single()
#geox.checknans()
if projections.shape != (geox.angles.shape[0], geox.nDetector[0],
geox.nDetector[1]):
raise ValueError("Expected data shape: " +
str((geox.angles.shape[0], geox.nDetector[0],
geox.nDetector[1])) + " not compatible with: " +
str(projections.shape))
return _Atb_ext(projections, geox, geox.angles, krylov, geox.mode)
def Atb(projections, geo, angles, krylov="matched"):
if projections.dtype != np.float32:
raise TypeError("Input data should be float32, not " +
str(projections.dtype))
if not np.isreal(projections).all():
raise ValueError("Complex types not compatible for back projection.")
geox = copy.deepcopy(geo)
geox.check_geo(angles)
if projections.shape != (geox.angles.shape[0], geox.nDetector[0],
geox.nDetector[1]):
raise ValueError("Expected data shape: " +
str((geox.angles.shape[0], geox.nDetector[0],
geox.nDetector[1])) + " not compatible with: " +
str(projections.shape))
return _Atb_ext(projections, geox, geox.angles, krylov, geox.mode)