Weird behaviour of nb_beamform.ipynb

petotamas / pyArgus

Antenna array signal processing library implemented in python

https://www.tamaspeto.com/pyargus

GNU General Public License v3.0

164 stars 60 forks source link

Weird behaviour of nb_beamform.ipynb #2

Closed nlhnt closed 4 years ago

nlhnt commented 4 years ago

Hello, sometimes when you try to use the pyargus.beamform.MSINR_beamform() an exception is raised during eigenvectors and eigenvalues computation.

The exception raised talks about infs or NaNs inside matrices/arrays, but it is called just after prodMatrix is calculated (line 183).

Even though Rnunu and Rss are fine, the output of their multiplication produces weird objects. This seems to be inconsistent. The example provided in docs works once, then it does not and it happens randomly.

nlhnt commented 4 years ago

I need to do further investigation, will close the issue for now.

Edit: Sorry for the commotion, seems like I have a problem with my python distribution that's supposed to be working on Windows, but it behaves badly.

After calling the same operator twice I get two different results from numpy. This is sad... I don't even know how to begin a bug report to numpy or is it Anaconda? So many things that could be responsible, even bad RAM dices.

Everything works just fine using Linux (Manjaro Stable).