Adding jones adapter, docs, vortex retarder, jones-to-mueller conversion #111
Conversation
supported functions: focus, unfocus, angular spectrum, focus_fixed_sampling, unfocus_fixed_sampling
prysm/x/polarization.py
Outdated
| supported_propagation_funcs = ['focus','unfocus','focus_fixed_sampling','angular_spectrum'] | ||
|
|
||
|
|
||
| U = np.array([[1, 0, 0, 1], |
There was a problem hiding this comment.
Please create this where it is needed (at most with a def _U():; U = _U()) -- by pre-computing it like this, GPU support is broken, since it will always reside on the device where "np" is at import time, which is always numpy-numpy.
Please also add dtype=config.precision_complex to the array() call
There was a problem hiding this comment.
Moved into jones_to_mueller and added the dtype=config.precision_complex
prysm/x/polarization.py
Outdated
| @@ -273,31 +286,108 @@ def linear_polarizer(theta=0, shape=None): | |||
|
|
|||
| return linear_diattenuator(0, theta=theta, shape=shape) | |||
|
|
|||
| def vector_vortex_retarder(charge, shape, retardance=np.pi, theta=0): | |||
There was a problem hiding this comment.
taking shape as an input is not prysm-like; please take appropriate coordinate arrays and do not make a grid internally
There was a problem hiding this comment.
I’ll have it take in an angular coordinate theta and update the tutorial to include it
prysm/x/polarization.py
Outdated
|
|
||
| @functools.wraps(prop_func) | ||
| def wrapper(*args,**kwargs): | ||
|
|
There was a problem hiding this comment.
delete two extra white spaces
prysm/x/polarization.py
Outdated
| if name in funcs_to_change: | ||
| setattr(propagation, name, jones_adapter(func)) | ||
|
|
||
| def apply_polarization_to_field(field): |
There was a problem hiding this comment.
This is not used in the module and does not have an intuitive mapping between its name and what it does; please remove
There was a problem hiding this comment.
From the Polarized_Propagation.ipynb demo:
This was added in response to a comment on #91 regarding the demo containing A = A[…, np.newaxis, np.newaxis] to enable element-wise multiplication. Would you prefer apply_polarization_component_to_field or something similar?
There was a problem hiding this comment.
It is a design question, I think
I understand that vvr * A is a very nice syntax since it reads like the calculus
What I meant was a function ~=
def apply_polarization_optic(field, pol):
# internal logic understands 2D -> 4D broadcasts
But this is fairly verbose if it is going to be used all the time
Perhaps instead the routines that create polarization components could return wavefronts, and the math overrides on wavefronts could be extended to include the newaxis's when there is a dimensionality mismatch?
There was a problem hiding this comment.
Ah apologies I misunderstood. I can add apply_polarization_optic now and we can work on adding support for Wavefronts in a future PR?
There was a problem hiding this comment.
The change in 4dbd050 is not quite right; in the event that the field is already polarization compatible, it will make it (N,M,2,2,1,1); please add an if field.ndim == 2 check (or != 4, up to you)
There was a problem hiding this comment.
change made in b885996
|
Change made in 558aad3 |
|
merged :) |
No description provided.