Since cosine and sine are symmetric in both the x- and y- direction, we can reduce the number of calls even further. This shouldn't be too difficult, I just need some time to think about how to do so elegantly.
There is no restrictions on num_angular_voxels, num_azimuthal_voxels other than its > 0, so odd numbers also need to be considered.
Currently, this is still far from being the performance bottleneck, so I'm keeping in mind Amdahl's law for now.
Since cosine and sine are symmetric in both the x- and y- direction, we can reduce the number of calls even further. This shouldn't be too difficult, I just need some time to think about how to do so elegantly.
There is no restrictions on num_angular_voxels, num_azimuthal_voxels other than its > 0, so odd numbers also need to be considered.
Currently, this is still far from being the performance bottleneck, so I'm keeping in mind Amdahl's law for now.