Added Q functions and demos

[wcmelby]

No description provided.

[codecov-commenter]

Welcome to Codecov 🎉

Once you merge this PR into your default branch, you're all set! Codecov will compare coverage reports and display results in all future pull requests.

Thanks for integrating Codecov - We've got you covered ☂️

[Jashcraf]

Well done 👏 👏 👏 just have a few minor comments before merging, you should be able to see them as comments attached to each of the file changes.

[Jashcraf]

Since this notebook doesn’t use any functions from Katsu yet, I’m thinking we should wait before including it in the website. It’s great to know that there are other packages that enable motion control though! Maybe we can either add Pylab as an optional dependency in the future.

[wcmelby]

Sounds good. I'll see if I can keep it as a draft in the branch but take it out of the pull request. I could also add more about how to integrate the motion commands in a DRRP data-taking script which might be more relevant to katsu

[Jashcraf]

Good idea - if it’s too difficult we can also leave it in and then just take it out of the index.rst that loads it into the website.

We can also save the integrated motion control / data reduction for another PR in the future :^)

[Jashcraf]

Love this, just a couple of comments that I’lll include in the replies

[Jashcraf]

In the beginning, where you say “(for more information, see FullMuellerExample.ipynb)”, can you replace this with the following:

(for more information, see [The Full Mueller Polarimetry Demo](FullMuellerExample.ipynb))

If I wrote this correctly, it should hyperlink to the Jupyter notebook

[Jashcraf]

However, there is a way to get the same result using a slightly more robust method that uses measurements of the Q Stokes parameter.

Has the relative robustness been verified in the lab? If so, we should consider a demo where you demonstrate it because that’s awesome 👀 . If not, maybe we should call this an “emerging technique” instead and cite your SPIE proceedings.

[Jashcraf]

As opposed to measuring a single linear polarization, this measurement takes information from a greater sampling of polarization states and is therefore more thorough. 

Same comment as before, has “more thorough” been verified? If not, maybe just a comment on how we (you) are investigating its potential for greater robustness.

[Jashcraf]

the last linear polarizer in the polarization state analyzer (PSA) can be replaced with a Wollaston prism to allow for Q sampling.

“Q sampling” phrasing could be strengthened, maybe “direct measurement of Q” or something similar?

[Jashcraf]

Instead of a list of zeros that you have here:

# Now feed this data into a data reduction function to see how well the matrix correlates
M = q_calibrated_full_mueller_polarimetry(theta, 0, 0, 0, 0, 0, 2, 2, M_rand)

It might be more illustrative to show what each variable is in the demo, ex:

# Initialize polarizer angles
psg_polarizer_angle = 0
psg_qwp_angle = 0
psa_qwp_angle = 0
psa_polarizer_angle = 0
psg_retardance = 0
psa_retardance = 0
I_vertical = 2
I_horizontal = 2

# Now feed this data into a data reduction function to see how well the matrix correlates
M = q_calibrated_full_mueller_polarimetry(theta,
                                          psg_polarizer_angle,
                                          psg_qwp_angle,
                                          psa_qwp_angle,
                                          psg_retardance,
                                          psg_retardance,
                                          I_vertical,
                                          I_horizontal,
                                          Mrand)

[Jashcraf]

Last thing is to maybe add some kind of tie-up comment at the end. Like how we are using Katsu as a platform for developing new polarimetric data reduction techniques like this one you came up with!

[wcmelby]

I’m a little confused by the equation. Specifically the [0, 1, 0, 0] on the left. Is this a dot product or a matrix multiplication?

Also, I think you can remove the * in between the matrices. The multiplication is implicit for matrix and matrix-vector multiplications.

Sorry, it should be matrix/vector multiplication. The [0, 1, 0, 0] simply isolates the Q parameter, but maybe it's simpler to leave the whole Stokes vector

[Jashcraf]

Maybe writing it as two separate equations? Replace the stokes vector [0, 1, 0, 0] with a linear analyzer oriented along the horizontal in one equation, and along the vertical in another. Then you can write something like:

$$Q = P_{H} - P_{V}$$

And move on to say that you build the data reduction matrix with Q

[Jashcraf]

@wcmelby THE DEMO LOOKS SO GOOD

It’s not clear from the demo what the following variables are, could you add some elaboration? Presumably these are power measurements, but I’m not sure why they are 2x1 arrays.

I_vertical = np.array([1, 2]) # some numpy array, can be anything for now
I_horizontal = np.array([2, 3]) # some numpy array, can be anything for now

Also, in the following

By using the optional input for M_rand, this function simulates what the data would look like for that matrix and uses that data for the matrix reconstruction.

I think it might be a bit clearer to say “setting the keyword argument M_in to M_rand, this function simulates what…” or something along those lines.

After that, I think we are good to go

[Jashcraf]

ooh I think these files come from a git add ., please remove this one

[Jashcraf]

ooh I think these files come from a git add ., please remove this one

[Jashcraf]

ooh I think these files come from a git add ., please remove this one

[Jashcraf]

ooh I think these files come from a git add ., please remove this one

[wcmelby]

Should I delete them or just take it out of the pull request?

[Jashcraf]

Definitely take em out of the pull request. I think if you delete them they just end up re-generating when you run jupyter notebooks haha.

[Jashcraf]

Oh man this looks pretty good, especially good job on preserving the numpy docstring style 👍

Can you move the RMS_Calculuator and propagated_error functions to katsu_math.py and then import them here? Also, you can remove the polar decomposition functions and just import them from mueller.py, we already have them there!

[wcmelby]

The polar decomposition functions I imported are the same as yours except for one line where I normalize the matrices. I remember having an issue I think when calculating the retardance if these weren't normalized. I moved these two functions to mueller.py for now

[Jashcraf]

Ah I see. Maybe the path forward is to add a normalization option. Something like

def decompose_depolarizer(*args, **kwargs, normalize=False)

    if normalize:
        return M / M[…, 0, 0]
    else:
        return M

Thoughts?

[wcmelby]

Good idea, I'll add that

[Jashcraf]

Same comment here re: tests in a future PR, but otherwise this looks great!

[Jashcraf]

Good stuff! Just had one last comment on the demo, after that I think we can go ahead and approve :)

[Jashcraf]

@wcmelby THE DEMO LOOKS SO GOOD

It’s not clear from the demo what the following variables are, could you add some elaboration? Presumably these are power measurements, but I’m not sure why they are 2x1 arrays.

I_vertical = np.array([1, 2]) # some numpy array, can be anything for now
I_horizontal = np.array([2, 3]) # some numpy array, can be anything for now

Also, in the following

By using the optional input for M_rand, this function simulates what the data would look like for that matrix and uses that data for the matrix reconstruction.

I think it might be a bit clearer to say “setting the keyword argument M_in to M_rand, this function simulates what…” or something along those lines.

After that, I think we are good to go

[Jashcraf]

Great! We should consider adding tests in a future PR, but this is good for now.

[Jashcraf]

Same comment here re: tests in a future PR, but otherwise this looks great!

[Jashcraf]

It took me a minute to see the change @becca9808 made but good catch 👏 I didn’t notice that one

[wcmelby]

Thanks for the feedback, the edits are in! @Jashcraf are you able to selectively accept changes from the pull request or is it easier if I take out the Thorlabs demo now? Also I'm not sure how to revert changes to the pycache files.

[Jashcraf]

Ah no worries I think I can do it on my end

[Jashcraf]

LGTM!