Fix mass absolute nan

[plodocus]

Fixes #286

Pull Request Checklist

Below is a simple checklist but please do not hesitate to ask for assistance!

• Fork, clone, and checkout the newest version of the code
• Create a new branch
• Make necessary code changes
• Install black (i.e., python -m pip install black or conda install -c conda-forge black)
• Install flake8 (i.e., python -m pip install flake8 or conda install -c conda-forge flake8)
• Install pytest-cov (i.e., python -m pip install pytest-cov or conda install -c conda-forge pytest-cov)
• Run black . in the root stumpy directory
• Run flake8 . in the root stumpy directory
• Run ./setup.sh && ./test.sh in the root stumpy directory
• Reference a Github issue (and create one if one doesn't already exist)

[seanlaw]

@DanBenHa Thank you for the PR! I couldn't find much information on when negative values may occur except for one article that simply set those values to zero before taking the square root. I am wondering if there are situations where setting the negative values to zero is wrong?

Otherwise, everything else looks good and this should be ready to merge.

[plodocus]

I am wondering if there are situations where setting the negative values to zero is wrong?

Good question. I don't know. But that would imply calculating a complex value for the distance of two real-valued vectors. Would that make sense?

[codecov-io]

Codecov Report

Merging #288 (593098f) into master (4adf60a) will increase coverage by 0.00%.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##           master     #288   +/-   ##
=======================================
  Coverage   99.82%   99.82%           
=======================================
  Files          19       19           
  Lines        1732     1734    +2     
=======================================
+ Hits         1729     1731    +2     
  Misses          3        3           

Impacted Files	Coverage Δ
stumpy/core.py	100.00% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 4adf60a...593098f. Read the comment docs.

[seanlaw]

Good question. I don't know. But that would imply calculating a complex value for the distance of two real-valued vectors. Would that make sense?

Since negative values imply some sort of deficiency in the numerical precision, I think what I'm really asking is if there might be situations where the amplitude of the negative value may be indicative of something that is a lot worst (while a small, near zero amplitude is due to numerical precision). Just thinking out loud here though :)

[plodocus]

Since negative values imply some sort of deficiency in the numerical precision, I think what I'm really asking is if there might be situations where the amplitude of the negative value may be indicative of something that is a lot worst (while a small, near zero amplitude is due to numerical precision). Just thinking out loud here though :)

I see. I guess it could be indicative of problems in the computation of Q_squared, T_squared and QT, so primarily in the functions rolling_is_finite, sliding_dot_product and rolling_window.

[seanlaw]

Okay, I spent most of my morning looking for an ideal solution and, while there were many discussions, there were no obvious solutions

[plodocus]

Okay, I spent most of my morning looking for an ideal solution and, while there were many discussions, there were no obvious solutions

Haha, reminds me of some of my mornings ;)

Thanks for merging and the interesting question raised.