Access Dipole array

[wagdy88]

Hello,
Is there a way to access the Dipole's array? I'd like to access - and operate on- the arrays for aggregates, L2, and L5.

Thanks
Mohamed

[jasmainak]

Did you look at the documentation ? ;-)

You can do:

dipole.data['L2']
dipole.data['L5']
dipole.data['agg']

[jasmainak]

A little hard to debug with the information you provided to me. The Dipole object has an attribute called scale_applied which should be 1. by default by can be larger if you assume that the HNN network represents a larger network of cells than 270 :) Can you share the commands you used?

[wagdy88]

Thanks @jasmainak. I want to keep the network as is i.e. 270 cells. Here is what I have been trying to do:
We are trying to quantify the gamma signals using PSD. When we plotted the original dipole, it seemed that the signals are shifted up around the zero. So, our thought was to correct that by subtracting the mean dipole from the original dipole and then re-plot it.

Here are the commands I used:

dpl_dataL2 = dpl.data['L2']
dpl_dataL2_avg = dpl.data['L2'].mean()
dpl_dataL2_modified = dpl_dataL2 - dpl_dataL2_avg

plot_dipole(dpl_dataL2_modified)
and that's when I get the message " 'numpy.float64' object has no attribute 'scale_applied'"

[jasmainak]

this is easy to fix ;-) You passed an array instead of Dipole object to plot_dipole. This merits a check in our function. You should do:

dpl_new = Dipole(dpl.times, dpl_dataL2_modified)
dpl_new.plot()

[wagdy88]

Thanks! I'll try this :-)

On the other hand, if the issue seems doable for a new contributor, I'd be happy to work on it

[jasmainak]

I think it's totally doable. You will mostly have to look around in the code how we do type checks and copy that.

[jasmainak]

There is also a private function _baseline_renormalize in the code but it seems that at some point we stopped using it ... @rythorpe @ntolley @cjayb any clue, opinion?

[rythorpe]

I think we stopped using it in #264 because it really only made sense for a very specific use-case (ERPs), and even then, it didn't do a good job. Note that _baseline_renormalize is still used in the tests that compare simulated waveforms between here and legacy HNN.

[rythorpe]

Correction: we still call _baseline_renormalize by default in _gather_trial_data().