Nothing wrong with those intrinsics, I use them sometimes, I just couldn't get through all the instructions because there's so many. Mostly I just wanted to give people a taste of the most common ones and what kinds of operations exist. Also should be a good start-off point for learning the other intrinsics that I didn't cover.

Don't follow what you're asking. But there's two examples shown, one where you're optimising for a desired intensity distribution (and you don't care about the phase), and another example where you care about both amplitude and phase in a certain region, but don't care what happens outside that region. In both cases, you're reinforcing the aspect and/or region of the field you want, and letting some other aspect change (such as the phase) to let the aspect you want get optimised. If you end up in a situation where you've thrown away everything in one of the planes (i.e. both amplitude and phase everywhere in the plane), then you've done something wrong. There's always some bit you change, and some bit that you let stay whatever it already is.

Not aware of any fundamental reason, but I think it's just to keep with convention of how you normally write numbers from least significant digit to most significant digit right-to-left. Think of it as one big 256-digit binary number, just like you'd write a decimal number '1234' meaning one thousand, two hundred and thirty-four, with the least significant digit is on the right, and then you work to the left with each digit being a higher-power of 10. Same idea, except this is binary. Or in C/C++, if you were writing a binary number it'd be 0b1101010101, with the least significant bits on the right. The confusion can be if you're thinking of it like an array in C/C++, where you'd define it left-to-right, A = {0,1,2,3} where the first element is on the left. Don't get me wrong, you can also think of the AVX blocks as little arrays, it's just the way they're normally illustrated is such that the significance of the bits in the whole block increase from right-to-left, like one big number (even if you address them as blocks of individual numbers). Then no matter how you slice it, every chunk would just read like it's own binary number right-to-left.

@@joelacarpenter Thank you so much for explaining!!

Version with the song had copyright flag, so students in some countries wouldn't be able to see it.

It looks like the phase is changing on the 'smiley face' pattern, but that's just floating-point noise because all those background pixels are just zeros (so it doesn't have meaningful phase). So what you're seeing is some floating-point noise from the FFT in the dark background (very very small, but non-zero values), being replaced with strictly zero when I enforce the amplitude of the target. Then all those meaningless noise phases just get wiped to zero because of the amplitude. If you look at the phase of the positions where the smiley face actually exists and has a proper amplitude, those areas aren't changing phase.

Yep, if you look in the supplemental material for our 'Laguerre-Gaussian mode sorter' paper, there's Matlab code provided...doi.org/10.14264/uql.2019.81

​@@joelacarpenter Thank you so much, the paper was really helpful, and the code is very well-documented. For context, I am a Physics undergraduate student that is trying to design a MPLC device that works with HG beams. But here I'm optimizing the patterns of the phase masks to target a certain transformation in the HG mode basis, and not to target specific N output modes given N input modes, so it's not exactly inverse design. For now, I am trying to work with 1D HG beams to later generalize it to 2D. I'd really appreciate if you could suggest any resources that I could check to learn new topics that could help me with my project.

The textbook 'Fundamentals of Photonics' has a chapter on beam optics, and a chapter on Fourier optics that's useful for getting acquainted when you're starting out in this area.

Short answer no, but look at my tutorial on off-axis digital holography for how you can measure phase reliably if you've got a phase reference arm. Gerchberg-Saxton essentially 'guesses' the phase, so you can't rely on it to give a unique correct solution. OAM beams are particularly ambiguous because both the + and - topological charge beams look identical in the near and far-field in terms of intensity, so there's no way to tell the sign of the topological charge from a CCD image in the near field and far-field. The other way to measure OAM beams is with devices like SLM which display correlation filters, but the easiest/best approach is off-axis digital holography, if you've got access to a reference beam.

Thanks for the answer ❤️

I made my own script for that. It's basically just the hsv colormap for the phase, multiplied by the normalised amplitude of the thing you're plotting. i.e. a hsv colormap for the phase, which fades to black depending on the amplitude information.

yep, it was a multimode coupler used as a crude mode demultiplexer

You often put an extra grating over the top to steer the pattern away from the zero-order position. The zero-order (on-axis) position is often where imperfections end up. So there's often undiffracted light, stray light, unwanted reflections etc. that hang around in that area.