I want to initialize a network with some random connectivity, following
some standard connectivity rules, like is achieved by standard use of the
nest.Connect function.Then I want to make several runs of this network,
with the same connectivity. I learned however that this is not a
straightforward thing to do repeated runs of the same network in NEST.
Although I saw different solutions online, it is not clear that there is a
'standard' solution for how to achieve this goal of just initializing the
network with the same connectivity several times, which seems fairly
simple. This link
<https://nest-simulator.readthedocs.io/en/stable/auto_examples/store_restore_network.html>
here, for example, shows how to store and restore a network, but it seems
like overshoot for my purpose (the described problem is a network with
plasticity). I also thought of creating a network, extract the
connectivity, and then impose this connectivity in new networks, following
what is explained here
<https://nest-simulator.readthedocs.io/en/v3.4/synapses/connection_management.html>
in the 'Connecting sparse matrices with array indexing' section, but it's a
bit tedious to implement (I started implementing it) and error prone.
The alternative I'm thinking of now is to just re create the network on
each pass of my loop with standard NEST functions, but giving the same
random seed to NEST before making the connectivity each time. After the
connectivity is done, I'd pass a new random seed, sampled from numpy's
random number generator, and so each simulation would be a different repeat.
I was wondering whether this very simple approach I propose would work, or
whether I'm missing something.
Thanks
Show replies by date