BUG: `net.add_poisson_drive()` throws error depending on `rate_constant` type

[ntolley]

Weird bug, if you set cell_specific=False and use a rate_constant of type float, you get the following error at simulation time

net = jones_2009_model()
weights_ampa_noise = {'L2_basket': 0.01, 'L2_pyramidal': 0.002,
                      'L5_pyramidal': 0.02}
net.add_poisson_drive('noise_global', rate_constant=2.0, location='distal',
                       weights_ampa=weights_ampa_noise, space_constant=100,
                       n_drive_cells=1, cell_specific=False)
simulate_dipole(net, tstop=100)

Now if you set rate_constant=2, an integer instead of a float, the simulation runs fine...

@rythorpe maybe you have an idea? I remember you improved the terminology for n_drive_cells API but this might have existed previously since a uniform poisson drive is a pretty uncommon use case

[jasmainak]

It sounds like a bug with type checks and type casting.

On Tue 4 Jul 2023 at 08:06, Nicholas Tolley @.***> wrote:

Weird bug, if you set cell_specific=False and use a rate_constant of type float, you get the following error at simulation time

net = jones_2009_model()weights_ampa_noise = {'L2_basket': 0.01, 'L2_pyramidal': 0.002, 'L5_pyramidal': 0.02}net.add_poisson_drive('noise_global', rate_constant=2.0, location='distal', weights_ampa=weights_ampa_noise, space_constant=100, n_drive_cells=1, cell_specific=False)simulate_dipole(net, tstop=100)

[image: image] https://user-images.githubusercontent.com/55253912/250895599-6235c5bf-adf1-4348-808c-17261ce7fdb6.png

Now if you set rate_constant=2, an integer instead of a float, the simulation runs fine...

@rythorpe https://github.com/rythorpe maybe you have an idea? I remember you improved the terminology for n_drive_cells API but this might have existed previously since a uniform poisson drive is a pretty uncommon use case

— Reply to this email directly, view it on GitHub https://github.com/jonescompneurolab/hnn-core/issues/662, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADY6FIQT25VW42MUBWESJBLXOQBMLANCNFSM6AAAAAAZ5URWLE . You are receiving this because you are subscribed to this thread.Message ID: @.***>

-- Sent from my iPhone

[rythorpe]

Yeah, I'm guessing the rate_constant variable somewhere gets typecasted differently based on it's target population if it gets entered by the user as type int vs float vs dict. Probably here.

Then there looks like an inconsistency in _drive_cell_event_times() where when the target_type is set to 'any' (i.e., for a non-specific drive that targets all cells), the function assumes the value of drive['dynamics']['rate_constant'] is of type int or float (which it's not, because the float was typecasted into a dict of identical floats in Network.add_poisson_drive()).

I'd suggest just removing the typecasting in Network.add_poisson_drive().

[rythorpe]

Can we close this @dylansdaniels, or is there still stuff left to do?

[dylansdaniels]

This is good to close! I'll do so now 😁