Dear Nest Users,
I hope all of you be in health during these times.
I want to create a balanced network as it’s the first step of my master thesis. The model I want to use is “iaf_cond_exp” which is a must for later purposes. There’s an example in Pynest folder of a balanced network, unless it’s using “iaf_psc_alpha” and it doesn’t fit my goals. When I try to change the model, and run the program, the network doesn’t get active and there’s nothing to record. I divided my question into two part: - Does anyone have a balanced network with “iaf_cond_exp” neurons and all of its necessary parameters to run? - In general, how do people find or calculate their network parameters to fit the neuron model they use and don’t get lost in the massive number of parameters. Note: I also tried to use the example of “brunel_alpha_evolution_strategies.py” example which is a genetic algorithm to find the best parameter, although it finds the parameters after 50 generation, I use those parameters afterwards. It just doesn’t work!
Kind regards, Nosratullah Mohammadi
Dear Nosratullah,
I am surprised you find the ``pynest/examples/brunel_alpha_nest.py`` example to be not working; it is working fine for me. I would be happy to troubleshoot this, could you tell us a bit more about what you observe/don't observe?
It should not be too difficult to modify this example, changing the alpha synaptic kernels for exponential kernels. You will need to look at how the synaptic time constant and amplitude, and rate of the external Poisson generators need to be updated for consistency.
Finding parameters is, in general, one of the more difficult parts of computational neuroscience. It very much helps if you can find a publication that contains a set of parameter values known to work (in your case, based on cond-exp synapses). Then, you can modify the NEST script with these parameters. I am not an expert on evolutionary algorithms for parameter optimisation; this is a whole field of research in and of itself, and in the early stages of neural network development, I would recommend to use your own skill and knowledge to get the parameters in a reasonable operating regime. Again, having a publication that lists a set of known-good parameters is invaluable as a starting point. Perhaps the following paper contains some useful leads: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6048296/
Please let us know if you run into any issues with NEST Simulator during your further development.
With kind regards, Charl
On Wed, May 13, 2020, at 12:13, nosratullah mohammadi wrote:
Dear Nest Users,
I hope all of you be in health during these times.
I want to create a balanced network as it’s the first step of my master thesis. The model I want to use is “iaf_cond_exp” which is a must for later purposes. There’s an example in Pynest folder of a balanced network, unless it’s using “iaf_psc_alpha” and it doesn’t fit my goals. When I try to change the model, and run the program, the network doesn’t get active and there’s nothing to record. I divided my question into two part:
- Does anyone have a balanced network with “iaf_cond_exp” neurons and
all of its necessary parameters to run?
- In general, how do people find or calculate their network parameters
to fit the neuron model they use and don’t get lost in the massive number of parameters. Note: I also tried to use the example of “brunel_alpha_evolution_strategies.py” example which is a genetic algorithm to find the best parameter, although it finds the parameters after 50 generation, I use those parameters afterwards. It just doesn’t work!
Kind regards, Nosratullah Mohammadi
NEST Users mailing list -- users@nest-simulator.org To unsubscribe send an email to users-leave@nest-simulator.org
Dear Charl,
Thanks for your advices. I started from initial parameter’s of the NEST and succeeded to get record from the EI network with ‘iaf_exp_cond’. After working the network with initial parameter’s and using the Pynest example parameters for the number of connections I changed the synaptic time constant and so far everything works fine. From the Pynest folder, I only chose: p_rate, CI, CE and left all other parameters to their initial values. Also thanks for the paper.
Kind regards, Nosratullah Mohammadi
On 25 May 2020 AD, at 16:19, Charl Linssen nest-users@turingbirds.com wrote:
Dear Nosratullah,
I am surprised you find the ``pynest/examples/brunel_alpha_nest.py`` example to be not working; it is working fine for me. I would be happy to troubleshoot this, could you tell us a bit more about what you observe/don't observe?
It should not be too difficult to modify this example, changing the alpha synaptic kernels for exponential kernels. You will need to look at how the synaptic time constant and amplitude, and rate of the external Poisson generators need to be updated for consistency.
Finding parameters is, in general, one of the more difficult parts of computational neuroscience. It very much helps if you can find a publication that contains a set of parameter values known to work (in your case, based on cond-exp synapses). Then, you can modify the NEST script with these parameters. I am not an expert on evolutionary algorithms for parameter optimisation; this is a whole field of research in and of itself, and in the early stages of neural network development, I would recommend to use your own skill and knowledge to get the parameters in a reasonable operating regime. Again, having a publication that lists a set of known-good parameters is invaluable as a starting point. Perhaps the following paper contains some useful leads: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6048296/
Please let us know if you run into any issues with NEST Simulator during your further development.
With kind regards, Charl
On Wed, May 13, 2020, at 12:13, nosratullah mohammadi wrote:
Dear Nest Users,
I hope all of you be in health during these times.
I want to create a balanced network as it’s the first step of my master thesis. The model I want to use is “iaf_cond_exp” which is a must for later purposes. There’s an example in Pynest folder of a balanced network, unless it’s using “iaf_psc_alpha” and it doesn’t fit my goals. When I try to change the model, and run the program, the network doesn’t get active and there’s nothing to record. I divided my question into two part:
- Does anyone have a balanced network with “iaf_cond_exp” neurons and
all of its necessary parameters to run?
- In general, how do people find or calculate their network parameters
to fit the neuron model they use and don’t get lost in the massive number of parameters. Note: I also tried to use the example of “brunel_alpha_evolution_strategies.py” example which is a genetic algorithm to find the best parameter, although it finds the parameters after 50 generation, I use those parameters afterwards. It just doesn’t work!
Kind regards, Nosratullah Mohammadi
NEST Users mailing list -- users@nest-simulator.org To unsubscribe send an email to users-leave@nest-simulator.org
NEST Users mailing list -- users@nest-simulator.org To unsubscribe send an email to users-leave@nest-simulator.org
-
Charl Linssen -
nosratullah mohammadi