Dear Andreas,
Thank you for your request. Bjoern's swift answer is correct: Each spike
source is either excitatory or inhibitory.
Here is a word on the technical background: The input spike train is fed
into the system through a so-called synapse driver which operates an
entire row of the synapse array. Then, the pre-synaptic row connects to
a post-synaptic neuron via one column of the array, i.e., the chip
really implements a "weight matrix".
The reason for the curious behaviour is that a synapse driver can only
be configured to be either excitatory or inhibitory. So the behaviour is
expected. Yet, I agree that the software should raise an exception since
this is an invalid configuration.
What you could do for now is: Duplicate your input spike trains pop_src
to have an exc. and an inh. source with identical spike times. Then
connect those as desired to your population.
I hope this answer is helpful. Please don't hesitate to contact me if
you have further questions.
Best regards,
Johannes
On 03/26/2016 07:41 PM, Deiseroth, Bjoern wrote:
> Hey Andreas,
>
> Neurons run in single-target-out mode due to the hardware gating. So your pop_src's neuron spikes are either excitatory, OR inhibitory.
>
> Greets
> bjoern
>
> -----Original Message-----
> From: Spikey neuromorphic system - User [mailto:[log in to unmask]] On Behalf Of Andreas Stoeckel
> Sent: Samstag, 26. März 2016 00:02
> To: [log in to unmask]
> Subject: Neurons cannot target both excitatory and inhibitory synapses
>
> * PGP Signed by an unknown key
>
> Dear Spikey developers,
>
> to get a better feeling for the behaviour of the Spikey system I ran
> an experiment which measured the neuron "tuning curves", i.e. their
> mapping from input to output fire rate.
>
> To do so I've generated a test spike train which is sent from a
> population of SpikeSourceArrays to a population of neurons. Half of
> the neurons (every second neuron) also receives a bias input from a
> constantly spiking SpikeSourceArray, and should receive the test spike
> train via inhibitory synapses [2].
>
> The curious behaviour I observed is the following: as soon as a
> SpikeSourceArray is targeted both at excitatory and inhibitory
> synapses (of different target neurons), only inhibitory connections
> are made.
>
> I wrote a small PyNN test script which reproduces this behaviour, you
> can find it at [1].
>
> Is this a known restriction? If not, I'll gladly open a bug ticket.
>
> Cheers,
> Andreas
>
> [1] https://gist.github.com/astoeckel/6b252f3ce955f6393586
> [2]
> https://github.com/hbp-sanncs/cypress/blob/408bdc34437ef901386424380ba4b76617b5a77f/examples/measure_tuning_curves.cpp
>
>
>
>
> ########################################################################
>
> To unsubscribe from the KIP-SPIKEY-USERS list, click the following link:
> http://listserv.uni-heidelberg.de/cgi-bin/wa?TICKET=NzM2MDc5IGJqb2Vybi5kZWlzZXJvdGhAU0FQLkNPTSBLSVAtU1BJS0VZLVVTRVJTIGx2fnCXvdrc&c=SIGNOFF
>
> * Unknown Key
> * 0x213223DD
>
> ########################################################################
>
> To unsubscribe from the KIP-SPIKEY-USERS list, click the following link:
> http://listserv.uni-heidelberg.de/cgi-bin/wa?TICKET=NzM2MDc5IGJpbGxAS0lQLlVOSS1IRUlERUxCRVJHLkRFIEtJUC1TUElLRVktVVNFUlMgIE5GapWjcojh&c=SIGNOFF
>
########################################################################
To unsubscribe from the KIP-SPIKEY-USERS list, click the following link:
https://listserv.uni-heidelberg.de/cgi-bin/wa?SUBED1=KIP-SPIKEY-USERS
|
