ContinuousSpace: Fix get_distance calculation on toroidal boundary condition

[rht]

cc: @simonchx, @clmerlos, @JamesArruda, @njvrzm

This fixes the distance calculation back to before this commit change: 600c62b#diff-84ac902399d236cac5d91bacaad69b2bL547, additionally, uses numpy for the arithmetic as suggested by @JamesArruda's implementation. This version of distance calculation passes the most number of test cases.

This fixes #410.

[coveralls]

Coverage increased (+0.05%) to 82.177% when pulling d85599b on rht:torus_distance into 275674d on projectmesa:master.

[simonchx]

@rht Thanks for the fix! It works well.

[Corvince]

Thanks @rht. But what do you mean by

This version of distance calculation passes the most number of test cases.

Are there any test cases where the calculation fails?

[rht]

Are there any test cases where the calculation fails?

There are not. In retrospect, I should have added this point in my description though.

[TaylorMutch]

For clarification (maybe I don't understand), how does this PR differ from #425 ?

[rht]

#425 fails the test at [1]

pos_6 = (-30, -29)
pos_7 = (21, -5)
assert self.space.get_distance(pos_6, pos_7) ==  np.sqrt(49 ** 2 + 24 ** 2)

[1] #410 (comment)

[coveralls]

Coverage increased (+0.04%) to 82.955% when pulling e50af5c on rht:torus_distance into 67f14ab on projectmesa:master.

[coveralls]

Coverage increased (+0.04%) to 82.955% when pulling f047ae4 on rht:torus_distance into 67f14ab on projectmesa:master.

[coveralls]

Coverage increased (+0.04%) to 82.955% when pulling a3014a1 on rht:torus_distance into 67f14ab on projectmesa:master.

[rht]

I added that test case assert self.space.get_distance(pos_6, pos_7) == np.sqrt(49 ** 2 + 24 ** 2).

[Corvince]

I think this can finally be merged, thanks everyone for the contributions!