Thanks Pooja.
My problem was loading a nestml generated neural model into PyNN. In particular, the problem is loading into PyNN and then executing the simulation in nest. (It seems one of the benefits of nestML
is that it generates code for SpiNNaker, and I hope to try that eventually.)
From python I
import pyNN.nest as sim
from pyNN.nest import native_cell_type
igSpike_neuron = native_cell_type('ignore_and_spike')
cells=sim.Population(2,igSpike_neuron,{})
sim.run(100.0)
#the model gives two emited spike
sim.setup(timestep=1.0)
cells=sim.Population(2,igSpike_neuron,{})
#the error output culminating with….
#error File "/home/christian3/CAs/jun24/nest-simulator-3.7/nest-install/lib/python3.12/site-packages/nest/ll_api.py", line 103, in catching_sli_run
# raise exceptionCls(commandname, message)
#nest.lib.hl_api_exceptions.NESTErrors.UnknownComponent: UnknownComponent in SLI function GetDefaults_l: /ignore_and_spike is not a known component.
So, it loads the model properly, and runs it. However, when I do a setup, it no longer recognises ignore_and_spike. (Almost all of my existing scripts use setup.)
The name of the model wasn’t my problem. The name of model is easy to modify as I did. (I am sure your answer about that is correct, though I haven’t tried it.)
-Chris
From: Babu, Pooja <p.babu@fz-juelich.de>
Sent: 19 June 2024 12:29
To: NEST User Mailing List <users@nest-simulator.org>
Subject: [NEST Users] Re: nestml syntax problem
Hello Chris,
If you want to generate the neuron models which has the same name in NEST, you can use the “suffix” option in the “generate_nest_target” function. This adds the suffix to the model name during code generation.
generate_nest_target(input_path=“test_model.nestml",
nest.Install(“nestmlmodule”)
n = nest.Create(“test_model_nestml”)
We do not currently provide pyNN support for NESTML models. If you manage to run NESTML models in pyNN, do let us know.
Pooja
I seem to be having problems with pyNN accessing the new model. I should note that the first problem is that all of the 7.0.2 nesml models are already in the standard nest. So, it’s not clear, which version one is using. So, I switched
to the ignore_and_fire model, but changed the name to ignore_and_spike. A simple change of file name and model name was sufficient. I got that running in pyNest. (It’s a really simple model to work with.)
I can also get it running in pyNN, but I can’t use the Setup. If I do, it changes something and I no longer have access to ignore_and_spike model. I’m using pyNN 0.12.3.
From: Chris Huyck
Sent: 13 June 2024 16:34
To: NEST User Mailing List <users@nest-simulator.org>
Subject: RE: [NEST Users] Re: nestml syntax problem
This was a pretty straight forward solution. I used the hill_tononi.nestml file below, and nestml ran and compiled. The nest.Install works, but I can’t actually run the model. I’ll try to sort that out.
The keywords “model”, “OnCondition” and “OnReceieve” are all the new changes to the NESTML language which is not yet released.
We will fix the documentation to point to the "release version" to avoid further confusion.
Firstly, thanks for the quick response.
The latest release of NESTML (v7.0.2) uses the “neuron” or “synapse” at the beginning of the model definition. We have unified this in our latest changes to the language where they are replaced by “model”, however, this is not yet released.
It looks like you have conflicting versions of NESTML installed. I would suggest you uninstall all the old NESTML instances and install it again via pip.
Apologies for the confusion. We are in the process of doing a pre-release of the new language changes soon.
Personal I‘d like to recommend you to use the newest version of Nestml and to adapt your old neuron model to the new syntax.
For doing it you have to compare a nestml model of any neuron in the current nestml version to the old one.
Normally you have to put a word “end” at the end of each segment in model
BR Binh
Gesendet mit der mobilen Mail App
Am 13.06.24 um 10:59 schrieb Chris Huyck
I was using an old version of nest (3.5) and nestml. Someone sent me a nestml model, and that worked. I wanted to make my own model, but the tutorials had a different format. In particular, it seems like the first line of code in the
tutorials start with model, and the old version started with neuron of synapse.
So, I thought it was time to upgrade, so installed nest 3.7, and pip-ed nestml. (I also installed PyNN as that's the way I typically work.) When I imported generate_target (through python 3.12), I got the typical PyGSL warning; I think
that's because I'm using python 3.12 not 3.11, and I don't think it matters in this case.
Then I ran the generate_target and I got an error mismatched input 'model' expecting {newline, 'neuron', 'synapse}
I expect there's something like a yacc table that specifies the syntax, and somehow I've got the old syntax.
I've tried it with several version of nestml 7.0.2, 7.0.0, 6.0.0 and 5.0.0. I think 5.0.0 had the old syntax. I've tried several models from the library, and the Izhikevich tutorial.
Any thoughts about how to fix it?
_______________________________________________
NEST Users mailing list -- users@nest-simulator.org
To unsubscribe send an email to users-leave@nest-simulator.org