Problem with handling of units of electrode parameters in Brian

[appukuttan-shailesh]

Consider the following snippet of code:

c = DCSource(amplitude=0.5)
print c.amplitude
c.amplitude += 0.0
print c.amplitude

The output from Brian is:

5e-10
5e-19

Note the difference. This arises owing to the conversion from nA to A for the Brian implementation. During the increment (with 0.0 here), the value is re-converted from nA to A, resulting in the extra 1e-9.

When running the same on NEURON, things are safe as there is no conversion involved:

0.5
0.5

Things seem to be in order with the NEST implementation (which also involves conversion of units):

<larray: base_value=500.0 shape=None dtype=<type 'float'>, operations=[(<built-in function div>, 1000.0)]>
<larray: base_value=500.0 shape=None dtype=<type 'float'>, operations=[(<built-in function div>, 1000.0)]>

@apdavison : It would also be useful (if feasible) to ensure that printing the parameters across the various simulator backends results in values with same units.