Chris Havlin

oh, hmm, maybe just adding the keyword isn't a great idea: https://stackoverflow.com/questions/54717862/how-do-i-know-if-the-usedforsecurity-flag-is-supported-by-hashlib-md5 , but the answer there would be easy enough to do instead (using `hashlib.new('md5', usedforsecurity=False)` instead of `hashlib.md5(usedforsecurity=False)`...

Looks like a factor of 2 difference: make the substitution q = 2x in equation 5 of the paper you link and it matches the yt code that you link...

> q_ij = sqrt(x2+y2)/h but not x = q/2 `x` in my comment and in the `sph_kernel_cubic` function is an arbitrary (perhaps poorly chosen) variable name, not a coordinate. I...

#4964 should fix the error when the bbox is None. For the changing coordinate values, they are indeed being clipped here: https://github.com/yt-project/yt/blob/39a7563e64e8ec1e8d12ea3af2cc6da5354419d1/yt/frontends/tipsy/io.py#L81-L88 but I'm not sure why...

I'd guess it was likely added to avoid a particle falling identically on the bounding box of the dataset (maybe to address some particle selection issue?). I think you're only...

> As for the bounding box argument: I just copied it over from the arguments used in the image test. oh interesting! I'll go take a look at that to...

> Maybe Tipsy wasn't formally run with a bounding box? Ya! I think that's right -- there's actually a comment in the `TipsyDataset.__init__` that reads ``` # Because Tipsy outputs...

Ya, I agree with you -- raising an error or at least a warning would be ideal. I'll hit slack to see if anyone more familiar with Tipsy can weigh...

finally got around to trying on my mac (which already had a miniconda installation) and I can't reproduce the failure. Tried a couple different numpy versions, went back a bit...

One quick comment not related to the details of the implementation: > This PR will add an option (to be set as default) I could be wrong, but this feels...