Alexander Clausen

> The error message is shown as an error in the web GUI and the user can continue browsing Rough plan: This means that `detect_params` needs to be able to...

> Can I work on this issue? I found it interesting to work on but don't know a lot about this Hi, thank you for expressing your interest! Feel free...

I can try :) So the issue is that our helper `count_nonzero(a)` gives a different result than `count_nonzero(a.todense())`. This happens if `a` is a sparse array with a different fill...

> I was wondering if this has ever been discussed or thought about ? I have indeed thought about this before, the last time when working on [the WDD UDF](https://github.com/Ptychography-4-0/ptychography/pull/72)...

> Here is a benchmark for `np.fft`, `cp.fft` and `pyfftw.interfaces.scipy_fftpack` : :+1: > ```python > for stack_size in stack_sizes: > for sig_shape in sig_shapes.keys(): > warmup_data = xp.random.uniform(size=(stack_size, *sig_shape)).astype(xp.float32) >...

> Hi @matbryan52 , @sk1p, so yeah it is very interesting if we can do full stacks of frames compression. I am wondering, why we could not directly implement broadcasting...

> Here is the same graph but running fft2 through the UDF interface on frames / stacks of frames. [...] Thanks for the updated graph! I didn't fully digest it...

As a workaround on Linux or Mac OS, you can run the following in the shell before running the tests: `ulimit -n 4096`. On systems where this is not globally...

This should actually be fixed at least since #1561- I'm currently running `pytest tests/ -m "slow and not dist" -n 12` again and will close the issue if that succeeds.

We could also think about updating the code generated by the notebook template feature of the GUI by @twentyse7en to use UDFs where it makes sense.