Dear Nest community,
I have a network of LIF neurons ('iaf_cond_alpha') and I would like to implement STDP synapses, rather than the
standard 'static_synapse'. I would like to be able to set a subset of the network to STDP synapses (e.g. all excitatory ones)
and leave the rest static. This is how have implemented that currently (see code). I am using a for loop because I want to
see how STDP changes the activity over time (over many iterations), which leaves me with the following two questions:
*1. Is this a correct way to implement STDP in nest? *
*2. Is there a way to reset the network such that the connections are not lost? *ResetNetwork() and ResetKernel() both seem to
destroy all the network connections. While I would like to keep them, so that the network becomes less and less randomly
connected per simulation.
Thank you!
Best,
Daphne
*CODE*
# define network connectivity conn_dict = {'rule': 'pairwise_bernoulli', 'p': 0.1} # define dicts for static synapses static_ex_params = {'model':'static_synapse','weight': 6.0, 'delay': 1.5} static_in_params = {'model':'static_synapse','weight': -96.0, 'delay': 1.5}
if self.STDP == 'ALL': # all synapses should be stdp ones, make connections between the two populations: # from exc neurons to all neurons nest.Connect(neurons_all[:self.NE], neurons_all, conn_dict, self.syn_params_ex) # from interneurons to all neurons nest.Connect(neurons_all[self.NE:], neurons_all, conn_dict, self.syn_params_in)
elif self.STDP == 'EXC': # keep the inhibitory synapses static # connections from exc neurons to all neurons nest.Connect(neurons_all[:self.NE], neurons_all, conn_dict, self.syn_params_ex) # connections from interneurons to all neurons nest.Connect(neurons_all[self.NE:], neurons_all, conn_dict, static_in_params)
where
synapse_params_in = { 'model':'stdp_synapse', 'lambda': 0.01, 'alpha': 1.0, 'delay': 1.5, 'weight':-95.0 }
Dear Daphne,
What is the for loop you are referring to?
In the Connect() calls you show, nothing seems incorrect, but this doesn't shed light on the part where you need to reset the simulation. Could you please attach that part of the code?
By the way, the autapses option is True by default, so each neuron in your setup, if I understand correctly, has some probability to get connected to itself. I am just reminding because to me such behaviour was a surprise the first time I stumbled upon it.
Best wishes, Alex Serenko
чт, 16 апр. 2020 г. в 19:00, Daphne Cornelisse daphn3cor@gmail.com:
Dear Nest community,
I have a network of LIF neurons ('iaf_cond_alpha') and I would like to implement STDP synapses, rather than the
standard 'static_synapse'. I would like to be able to set a subset of the network to STDP synapses (e.g. all excitatory ones)
and leave the rest static. This is how have implemented that currently (see code). I am using a for loop because I want to
see how STDP changes the activity over time (over many iterations), which leaves me with the following two questions:
*1. Is this a correct way to implement STDP in nest? *
*2. Is there a way to reset the network such that the connections are not lost? *ResetNetwork() and ResetKernel() both seem to
destroy all the network connections. While I would like to keep them, so that the network becomes less and less randomly
connected per simulation.
Thank you!
Best,
Daphne
*CODE*
# define network connectivity conn_dict = {'rule': 'pairwise_bernoulli', 'p': 0.1} # define dicts for static synapses static_ex_params = {'model':'static_synapse','weight': 6.0, 'delay': 1.5} static_in_params = {'model':'static_synapse','weight': -96.0, 'delay': 1.5} if self.STDP == 'ALL': # all synapses should be stdp ones, make connections between the two populations: # from exc neurons to all neurons nest.Connect(neurons_all[:self.NE], neurons_all, conn_dict, self.syn_params_ex) # from interneurons to all neurons nest.Connect(neurons_all[self.NE:], neurons_all, conn_dict, self.syn_params_in) elif self.STDP == 'EXC': # keep the inhibitory synapses static # connections from exc neurons to all neurons nest.Connect(neurons_all[:self.NE], neurons_all, conn_dict, self.syn_params_ex) # connections from interneurons to all neurons nest.Connect(neurons_all[self.NE:], neurons_all, conn_dict, static_in_params)where
synapse_params_in = { 'model':'stdp_synapse', 'lambda': 0.01, 'alpha': 1.0, 'delay': 1.5, 'weight':-95.0 }
NEST Users mailing list -- users@nest-simulator.org To unsubscribe send an email to users-leave@nest-simulator.org
Dear Daphne,
Please excuse my late reply.
I have a network of LIF neurons ('iaf_cond_alpha') and I would like to implement STDP synapses, rather than the
standard 'static_synapse'. I would like to be able to set a subset of the network to STDP synapses (e.g. all excitatory ones)
and leave the rest static. This is how have implemented that currently (see code). I am using a for loop because I want to
see how STDP changes the activity over time (over many iterations), which leaves me with the following two questions:
1. Is this a correct way to implement STDP in nest?
You code looks correct. To use STDP for excitatory-excitatory connections, simply replace static_synapse with stdp_synapse. See the documentation for stdp_synapse for additional parameters you can set to control plasticity, which allows you also to choose between different STDP variants.
See hpc_benchmark.py for an example of a network where only E->E connections are plastic.
2. Is there a way to reset the network such that the connections are not lost? ResetNetwork() and ResetKernel() both seem to
destroy all the network connections. While I would like to keep them, so that the network becomes less and less randomly
connected per simulation.
No, resetting the network dynamics without deleting the connections is not possible. We considered this thoroughly some years ago, but doing this in a perfectly robust fashion turned out to be very challenging and we decided against supporting this.
Best, Hans Ekkehard
Thank you!
Best,
Daphne
CODE
# define network connectivity conn_dict = {'rule': 'pairwise_bernoulli', 'p': 0.1} # define dicts for static synapses static_ex_params = {'model':'static_synapse','weight': 6.0, 'delay': 1.5} static_in_params = {'model':'static_synapse','weight': -96.0, 'delay': 1.5}
if self.STDP == 'ALL': # all synapses should be stdp ones, make connections between the two populations: # from exc neurons to all neurons nest.Connect(neurons_all[:self.NE], neurons_all, conn_dict, self.syn_params_ex) # from interneurons to all neurons nest.Connect(neurons_all[self.NE:], neurons_all, conn_dict, self.syn_params_in)
elif self.STDP == 'EXC': # keep the inhibitory synapses static # connections from exc neurons to all neurons nest.Connect(neurons_all[:self.NE], neurons_all, conn_dict, self.syn_params_ex) # connections from interneurons to all neurons nest.Connect(neurons_all[self.NE:], neurons_all, conn_dict, static_in_params)
where
synapse_params_in = { 'model':'stdp_synapse', 'lambda': 0.01, 'alpha': 1.0, 'delay': 1.5, 'weight':-95.0 }
_______________________________________________ NEST Users mailing list -- users@nest-simulator.orgmailto:users@nest-simulator.org To unsubscribe send an email to users-leave@nest-simulator.orgmailto:users-leave@nest-simulator.org
--
Prof. Dr. Hans Ekkehard Plesser Head, Data Science Section
Faculty of Science and Technology Norwegian University of Life Sciences PO Box 5003, 1432 Aas, Norway
Phone +47 6723 1560 Email hans.ekkehard.plesser@nmbu.no Home http://arken.nmbu.no/~plesser
-
Alexey Serenko -
Daphne Cornelisse -
Hans Ekkehard Plesser