Rayfront.opd analysis functions

[Jashcraf]

We are storing the optical path difference of the rays in both CODE V and Zemax but need a way to actually make the data meaningful. As of now it's a raveled array of data, but diffraction calculations are done on square grids!

So, we need

• method of decomposing the WFE into polynomials (Zernikes are easy)
• method of differentiating ray OPD with OPD data
• relevant analysis of OPD

Hyukmo was interested in this utility so we can work with him on that.

[Jashcraf]

@hyukmo I do think priority #1 is coming up with some code to do zernike decomposition in poke

do you think you can write some code in python to generate a zernike polynomial given some index?

edit: taking a given x,y coordinate as inputs would be great too

[hyukmo]

@Jashcraf I can try, captain. But it would take some time....

[Jashcraf]

No worries just need the contribution before SPIE haha

we can either write our own, or make poke dependent on another package. If possible i think it'd be nice to do our own for low order polynomials (say Z1-36 or whatever). I can also help out with it and we can use this issue to workshop the code.

[trentjbrendel]

Jaren, how much do we need to generate code from scratch? Would it be okay if I prompted ChatGPT for some boilerplate code?

[Jashcraf]

le mao I have a preference against it so that contributors can use this as a learning experience, also I have no clue how to cite ChatGPT or it's products.

but zernike decomposition is just a problem of generating the polynomials, which wikipedia tells you how to do, and computing the dot product of them with your data.

[Jashcraf]

Turns out it might be better to use np.linalg.lstq to solve for the coefficient vector!

[Jashcraf]

Emory added the ability to do Zernike decomposition, so now we can do that with OPD data. Might be worth the assertion that OPD has consistent units in poke.raytrace.