Robert Kern

Yeah, it's more natural in MATLAB than in Python because they have different conventions for dealing with different argument structures. I prefer keeping to Python's standard semantics when we can....

Add `cdf_interval(lower, upper)` and be done with it. Leave `cdf()` alone.

I am not aware that "we" as a project have a policy on the matter. We definitely don't want opaque archives of large amounts of code (but then, it's the...

I have confirmed that the file is as described and restored the link.

If I had to guess, we're probably running into `long` integer overflow on Windows where it's 32-bit while on Linux it's 64-bit. I don't have a Windows environment handy to...

Yeah, that's likely the issue, then. If you do want the `-1000`s to count that way, using `arr.astype(np.float32)` would also be a workaround.

Something like the following will probably reproduce the platform difference (if not quite so dramatically as to flip signs). ```python z = np.full((512, 512, 284), -1000, dtype=np.int16) z[256-16:256+16, 256-16:256+16, 142-16:142+16]...

RMS is frequently used in signal processing on complex arrays and uses the `abs()` in there. C.f. [MATLAB's implementation](https://www.mathworks.com/help/matlab/ref/rms.html). That is the most useful definition for complex inputs, IMO, and...

In any case where I want something that I call "RMS" on complex values, I need the `abs()`. There might be an abs-less case that's of use somewhere, but I...

For that matter, I'm not really sure that generalized means are abs-less either. Certainly, the formulae you will see usually have no abses, but they also tend to define their...