Increase Code Coverage

[JHartzer]

I recommend making use of pytest-cov or any other coverage tool to calculate and track overall code coverage. I've included an initial high-level pass at the current state of testing below:

Module	statements	missing	excluded	coverage
pyswarming/init.py	11	0	0	100%
pyswarming/auto_differentiation.py	160	91	0	43%
pyswarming/behaviors.py	160	19	0	88%
pyswarming/swarm.py	74	64	0	14%
tests/init.py	0	0	0	100%
tests/test_behaviors.py	312	0	0	100%
Total	717	174	0	76%

It is apparent that most of the main behavior functions are being tested, but the main swarm class is not. I believe adding tests for the swarm class initialization and simulation would improve this work. Simply adding the two examples in a test function would likely be sufficient.

[mrsonandrade]

@JHartzer, thank you for the recommendation! I did not know pytest-cov. I have worked on increasing the code coverage of the behaviors class and I was able to increase it from 88% to 98%. However, I could not handle that much the swarm class due to the plt.show() inside simulate(). So, it increased from 14% to 39% only. I would appreciate it if you have any suggestions on how to overcome that.

[JHartzer]

Excellent! Typically you can get around the plotting test coverage limitation by separating the simulation from the plotting routines. For example, if simulate() produced an array of state data, and a separate animate() function ingested those states and produced the animation. That way, you could test the simulation portion of the code and ignore the plotting.

[mrsonandrade]

Great! Thanks for the suggestion! Instead of separating and adding one more step to the new users, I have created a mode argument for simulate(), so, it is possible to just get the states, get the animation object, or produce the animation. Now the code coverage increased from 39% to 82% 😃