def to_nest(self, send_only=None, weights=True):
'''
Send the network to NEST.
.. seealso::
:func:`~nngt.simulation.make_nest_network` for parameters
'''
from nngt.simulation import make_nest_network
if nngt._config['with_nest']:
return make_nest_network(self,
send_only=send_only,
weights=weights)
else:
raise RuntimeError("NEST is not present.")
def randomize_neural_states(network, instructions, make_nest=False):
'''
Randomize the neural states according to the instructions.
Parameters
----------
network : :class:`~nngt.Network` subclass instance
Network that will be simulated.
instructions : dict
Variables to initialize. Allowed keys are "V_m" and "w". Values are
3-tuples of type ``("distrib_name", double, double)``.
make_nest : bool, optional (default: False)
If ``True`` and network has not been converted to NEST, automatically
generate the network, else raises an exception.
Example
-------
python::
instructions = {
"V_m": ("uniform", -80., -60.),
"w": ("normal", 50., 5.)
}
'''
allowed = ("V_m", "w")
# check whether network is in NEST
if network._nest_gid is None:
if make_nest:
make_nest_network(network)
else:
raise AttributeError('`network` has not converted to NEST yet.')
num_neurons = network.node_nb()
for key, val in instructions.items():
if key in allowed:
state = _generate_random(num_neurons, val)
nest.SetStatus(list(network.nest_gid), key, state)
else:
raise RuntimeError('Only "V_m" and "w" can be randomized.')
def test_delays(self, graph, instructions, **kwargs):
'''
Test entirely run only if NEST is present on the computer.
Check that delay distribution generated in NNGT, then in NEST, is
conform to what was instructed.
'''
di_distrib = instructions["weights"]
distrib = di_distrib["distribution"]
delays = graph.set_delays(distribution=distrib, parameters=di_distrib)
ref_result = _results_theo(instructions)
computed_result = _results_exp(delays, instructions)
self.assertTrue(np.allclose(ref_result,computed_result,self.tolerance))
if nngt.config['with_nest']:
# @todo
from nngt.simulation import make_nest_network
subnet, gids = make_nest_network(graph)
graph = nngt.SpatialNetwork(pop, weight_prop={"distrib":"gaussian", "distrib_prop":{"avg_distrib": 60.}})
#~ graph = nngt.SpatialNetwork(pop, weight_prop={"distrib":"lognormal"})
nngt.generation.connect_neural_types(graph, 1, -1, "erdos_renyi", {"density": 0.035})
#~ nngt.generation.connect_neural_types(graph, 1, 1, "newman_watts", {"coord_nb":10, "proba_shortcut": 0.1})
#~ nngt.generation.connect_neural_types(graph, 1, 1, "random_scale_free", {"in_exp": 2.1, "out_exp": 2.9, "density": 0.065})
nngt.generation.connect_neural_types(graph, 1, 1, "erdos_renyi", {"density": 0.077})
nngt.generation.connect_neural_types(graph, -1, 1, "erdos_renyi", {"density": 0.2})
nngt.generation.connect_neural_types(graph, -1, -1, "erdos_renyi", {"density": 0.04})
#-----------------------------------------------------------------------------#
# NEST objects
#------------------------
#
subnet, gids = make_nest_network(graph)
recorders, record = nngt.simulation.monitor_nodes(gids, ["spike_detector", "multimeter"], [["spikes"], ["V_m"]], network=graph)
nngt.simulation.set_noise(gids, 70., 80.)
nngt.simulation.set_poisson_input(gids[570:870], 44000.)
#-----------------------------------------------------------------------------#
# Names
#------------------------
#
simtime = 1000.
nest.Simulate(simtime)
