def setup(timestep=DEFAULT_TIMESTEP, min_delay=DEFAULT_MIN_DELAY,
max_delay=DEFAULT_MAX_DELAY, **extra_params):
"""
Should be called at the very beginning of a script.
`extra_params` contains any keyword arguments that are required by a given
simulator but not by others.
NEST-specific extra_params:
`spike_precision`:
should be "on_grid" (default) or "off_grid"
`verbosity`:
INSERT DESCRIPTION OF POSSIBLE VALUES
`recording_precision`:
number of decimal places (OR SIGNIFICANT FIGURES?) in recorded data
`threads`:
number of threads to use
`grng_seed`:
one seed for the global random number generator of NEST
`rng_seeds`:
a list of seeds, one for each thread on each MPI process
`rng_seeds_seed`:
a single seed that will be used to generate random values for `rng_seeds`
"""
common.setup(timestep, min_delay, max_delay, **extra_params)
simulator.state.clear()
for key in ("verbosity", "spike_precision", "recording_precision",
"threads"):
if key in extra_params:
setattr(simulator.state, key, extra_params[key])
# set kernel RNG seeds
simulator.state.num_threads = extra_params.get('threads') or 1
if 'grng_seed' in extra_params:
simulator.state.grng_seed = extra_params['grng_seed']
if 'rng_seeds' in extra_params:
simulator.state.rng_seeds = extra_params['rng_seeds']
else:
rng = NumpyRNG(extra_params.get('rng_seeds_seed', 42))
n = simulator.state.num_processes * simulator.state.threads
simulator.state.rng_seeds = rng.next(n, 'uniform_int', {'low': 0, 'high': 100000}).tolist()
# set resolution
simulator.state.dt = timestep
# Set min_delay and max_delay for all synapse models
simulator.state.set_delays(min_delay, max_delay)
nest.SetDefaults('spike_generator', {'precise_times': True})
return rank()
def fixed_number_pre_with_replacement_heterogeneous_parameters(sim):
sim.setup()
p1 = sim.Population(5, sim.IF_cond_exp())
p2 = sim.Population(7, sim.IF_cond_exp())
connector1 = sim.FixedNumberPreConnector(n=3, with_replacement=True, rng=NumpyRNG())
synapse_type2 = sim.TsodyksMarkramSynapse(weight=lambda d: d, delay=0.5, U=lambda d: 0.02 * d + 0.1)
#synapse_type2 = sim.TsodyksMarkramSynapse(weight=0.001, delay=0.5, U=lambda d: 2*d+0.1)
prj2 = sim.Projection(p1, p2, connector1, synapse_type2)
print("Projection 2")
x = prj2.get(['weight', 'delay', 'U'], format='list', gather=False)
from pprint import pprint
pprint(x)
i, j, w, d, u = numpy.array(x).T
assert_arrays_equal(w, abs(i - j))
assert_arrays_equal(d, 0.5 * numpy.ones(p2.size * connector1.n))
assert_arrays_equal(u, 0.02 * abs(i - j) + 0.1)
sim.end()
def initialize():
global sim
global options
global extra
global rngseed
global parallel_safe
global rng
global n_ext
global n_exc
global n_inh
sim, options = get_simulator(
("--plot-figure", "Plot the connections to a file."))
init_logging(None, debug=True)
# === General parameters =================================================
threads = 1
rngseed = 98765
parallel_safe = True
rng = NumpyRNG(seed=rngseed, parallel_safe=parallel_safe)
# === general network parameters (except connections) ====================
n_ext = 60 # number of external stimuli
n_exc = 60 # number of excitatory cells
n_inh = 60 # number of inhibitory cells
# === Options ============================================================
extra = {
'loglevel': 2,
'useSystemSim': True,
'maxNeuronLoss': 0.,
'maxSynapseLoss': 0.4,
'hardwareNeuronSize': 8,
'threads': threads,
'filename': "connections.xml",
'label': 'VA'
}
if sim.__name__ == "pyNN.hardware.brainscales":
extra['hardware'] = sim.hardwareSetup['small']
if options.simulator == "neuroml":
extra["file"] = "connections.xml"
def scenario4(sim):
"""
Network with spatial structure
"""
init_logging(logfile=None, debug=True)
sim.setup()
rng = NumpyRNG(seed=76454, parallel_safe=False)
input_layout = RandomStructure(boundary=Cuboid(width=500.0, height=500.0, depth=100.0),
origin=(0, 0, 0), rng=rng)
inputs = sim.Population(100, sim.SpikeSourcePoisson(rate=RandomDistribution('uniform', [3.0, 7.0], rng=rng)),
structure=input_layout, label="inputs")
output_layout = Grid3D(aspect_ratioXY=1.0, aspect_ratioXZ=5.0, dx=10.0, dy=10.0, dz=10.0,
x0=0.0, y0=0.0, z0=200.0)
outputs = sim.Population(200, sim.EIF_cond_exp_isfa_ista(),
initial_values = {'v': RandomDistribution('normal', [-65.0, 5.0], rng=rng),
'w': RandomDistribution('normal', [0.0, 1.0], rng=rng)},
structure=output_layout, # 10x10x2 grid
label="outputs")
logger.debug("Output population positions:\n %s", outputs.positions)
DDPC = sim.DistanceDependentProbabilityConnector
input_connectivity = DDPC("0.5*exp(-d/100.0)", rng=rng)
recurrent_connectivity = DDPC("sin(pi*d/250.0)**2", rng=rng)
depressing = sim.TsodyksMarkramSynapse(weight=RandomDistribution('normal', (0.1, 0.02), rng=rng),
delay="0.5 + d/100.0",
U=0.5, tau_rec=800.0, tau_facil=0.0)
facilitating = sim.TsodyksMarkramSynapse(weight=0.05,
delay="0.2 + d/100.0",
U=0.04, tau_rec=100.0,
tau_facil=1000.0)
input_connections = sim.Projection(inputs, outputs, input_connectivity,
receptor_type='excitatory',
synapse_type=depressing,
space=Space(axes='xy'),
label="input connections")
recurrent_connections = sim.Projection(outputs, outputs, recurrent_connectivity,
receptor_type='inhibitory',
synapse_type=facilitating,
space=Space(periodic_boundaries=((-100.0, 100.0), (-100.0, 100.0), None)), # should add "calculate_boundaries" method to Structure classes
label="recurrent connections")
outputs.record('spikes')
outputs.sample(10, rng=rng).record('v')
sim.run(1000.0)
data = outputs.get_data()
sim.end()
return data
def do_run(split, seed=None):
p.setup(1.0)
if split:
p.set_number_of_neurons_per_core(p.SpikeSourcePoisson, 27)
p.set_number_of_neurons_per_core(p.IF_curr_exp, 22)
inp = p.Population(100, p.SpikeSourcePoisson(rate=100),
label="input", additional_parameters={"seed": seed})
pop = p.Population(100, p.IF_curr_exp, {}, label="pop")
p.Projection(inp, pop, p.OneToOneConnector(),
synapse_type=p.StaticSynapse(weight=5))
pop.record("spikes")
inp.record("spikes")
p.run(100)
inp.set(rate=10)
# pop.set("cm", 0.25)
pop.set(tau_syn_E=1)
p.run(100)
pop_spikes1 = pop.spinnaker_get_data('spikes')
inp_spikes1 = inp.spinnaker_get_data('spikes')
p.reset()
inp.set(rate=0)
pop.set(i_offset=1.0)
vs = p.RandomDistribution(
"uniform", [-65.0, -55.0], rng=NumpyRNG(seed=seed))
pop.initialize(v=vs)
p.run(100)
pop_spikes2 = pop.spinnaker_get_data('spikes')
inp_spikes2 = inp.spinnaker_get_data('spikes')
p.end()
return (pop_spikes1, inp_spikes1, pop_spikes2, inp_spikes2)
def __init__(self, mean, stdev, dt=None, start=0.0, stop=None, rng=None):
"""Construct the current source.
Required arguments:
mean -- mean current amplitude in nA
stdev -- standard deviation of the current amplitude in nA
Optional arguments:
dt -- interval between updates of the current amplitude. Must be
a multiple of the simulation time step. If not specified,
the simulation time step will be used.
start -- onset of the current injection in ms. If not specified, the
current will begin at the start of the simulation.
stop -- end of the current injection in ms. If not specified, the
current will continue until the end of the simulation.
rng -- an RNG object from the `pyNN.random` module. For speed,
this should be a `NativeRNG` instance (uses the simulator's
internal random number generator). For reproducibility
across simulators, use one of the other RNG types. If not
specified, a NumpyRNG is used.
"""
self.rng = rng or NumpyRNG()
self.dt = dt or state.dt
if dt:
assert self.dt % dt == 0
self.start = start
self.stop = stop
self.mean = mean
self.stdev = stdev
if isinstance(rng, NativeRNG):
self._device = nest.Create('noise_generator')
nest.SetStatus(
self._device, {
'mean': mean * 1000.0,
'std': stdev * 1000.0,
'start': self.delay_correction(start),
'dt': self.dt
})
if stop:
nest.SetStatus(self._device,
{'stop': self.delay_correction(stop)})
else:
raise NotImplementedError(
"Only using a NativeRNG is currently supported.")
def unif(n_spikes, T, rng=NumpyRNG(), rounding=False, t_min=1.0, min_isi=10.0):
"""
Generate uniformally distributed spikes between [t_min, T),
with minimum inter-spike separation and optional rounding
pyNN.nest is generally unstable with rounding for input spikes
pyNN.nest errors if lowest spike value is exactly equal to dt
Input spikes between 0.0 and dt are not integrated over
Parameters
----------
n_spikes : int
T : float
Time interval (ms)
t_min : float
Lower bound on generated time value (ms)
min_isi : float
Minimum inter-spike separation : n_spikes*MIN_ISI << T (default 10 ms)
Returns
-------
spike_times : pyNN.parameters.Sequence (float64)
"""
spike_times = np.empty([0], dtype=float)
while spike_times.size < n_spikes:
timing = rng.uniform(t_min, T)
# Ensure minimum separation w.r.t. existing spikes
if (spike_times.size > 0
and np.min(np.abs(timing - spike_times)) < min_isi):
continue
else:
spike_times = np.append(spike_times, timing)
spike_times.sort()
if rounding:
return Sequence(np.floor(spike_times))
else:
return Sequence(spike_times)
def fixed_number_post_with_replacement(sim):
sim.setup()
p1 = sim.Population(5, sim.IF_cond_exp())
p2 = sim.Population(7, sim.IF_cond_exp())
synapse_type1 = sim.StaticSynapse(weight=0.5, delay=0.5)
connector1 = sim.FixedNumberPostConnector(n=9, with_replacement=True, rng=NumpyRNG())
prj1 = sim.Projection(p1, p2, connector1, synapse_type1)
print("Projection #1\n", connection_plot(prj1))
delays = prj1.get('delay', format='list', gather=False)
assert_equal(len(delays), connector1.n * p1.size)
weights = prj1.get('weight', format='array', gather=False)
for row in weights:
row[np.isnan(row)] = 0
assert_equal(row.sum(), 4.5)
weights2 = prj1.get('weight', format='array', gather=False, multiple_synapses='min')
for row in weights2:
n_nan = np.isnan(row).sum()
row[np.isnan(row)] = 0
assert_equal(row.sum(), (row.size - n_nan)*0.5)
def _generate_random_values(
self, values, n_connections, pre_vertex_slice, post_vertex_slice):
"""
:param ~pyNN.random.NumpyRNG values:
:param int n_connections:
:param ~pacman.model.graphs.common.Slice pre_vertex_slice:
:param ~pacman.model.graphs.common.Slice post_vertex_slice:
:rtype: ~numpy.ndarray
"""
key = (id(pre_vertex_slice), id(post_vertex_slice), id(values))
seed = self.__param_seeds.get(key, None)
if seed is None:
seed = int(values.rng.next() * 0x7FFFFFFF)
self.__param_seeds[key] = seed
new_rng = NumpyRNG(seed)
copy_rd = RandomDistribution(
values.name, parameters_pos=None, rng=new_rng,
**values.parameters)
if n_connections == 1:
return numpy.array([copy_rd.next(1)], dtype="float64")
return copy_rd.next(n_connections)
def __init__(self, connector, pre_population, post_population,
prepop_is_view, postpop_is_view, rng,
synapse_dynamics, synapse_type, is_virtual_machine,
weights=None, delays=None):
"""
:param AbstractConnector connector:
The connector connected to the synapse
:param pre_population: The population sending spikes to the synapse
:type pre_population: ~spynnaker.pyNN.models.populations.Population or
~spynnaker.pyNN.models.populations.PopulationView
:param post_population: The population hosting the synapse
:type post_population: ~spynnaker.pyNN.models.populations.Population
or ~spynnaker.pyNN.models.populations.PopulationView
:param bool prepop_is_view: Whether the ``pre_population`` is a view
:param bool postpop_is_view: Whether the ``post_population`` is a view
:param rng: Seeded random number generator
:type rng: ~pyNN.random.NumpyRNG or None
:param AbstractSynapseDynamics synapse_dynamics:
The dynamic behaviour of the synapse
:param int synapse_type: The type of the synapse
:param bool is_virtual_machine: Whether the machine is virtual
:param weights: The synaptic weights
:type weights: float or list(float) or ~numpy.ndarray(float) or None
:param delays: The total synaptic delays
:type delays: float or list(float) or ~numpy.ndarray(float) or None
"""
self.__connector = connector
self.__pre_population = pre_population
self.__post_population = post_population
self.__prepop_is_view = prepop_is_view
self.__postpop_is_view = postpop_is_view
self.__rng = (rng or NumpyRNG())
self.__synapse_dynamics = synapse_dynamics
self.__synapse_type = synapse_type
self.__weights = weights
self.__delays = delays
self.__is_virtual_machine = is_virtual_machine
# Make a list of holders to be updated
self.__pre_run_connection_holders = list()
class param:
seed = 8658764 # Seed for reproduction of random number
rng = NumpyRNG() # Use seed to reproduce
input_nr = 9 # Number of input neurons
readout_nr = 2 # Number of readout neurons
reservoir_nr = 50 # Number of reservour neurons
simulation_time = 19.0 # Simulation time for each input
dt = 1 # Timestep in simulation
res_pconn = 0.1 # sparse connection probability for reservoir
images_train_nr = 9 # Number of training images to train with,
# Must be a factor of 3
images_test_nr = 9 # Number of test images
images_train = generate_labeledImages(images_train_nr)
images_test = generate_labeledImages(images_test_nr)
# If network uses excitatory and inhibatory neurons
res_exc_nr = int(math.ceil(reservoir_nr *
0.8)) # Number of excitatory neurons
res_inh_nr = int(math.floor(reservoir_nr *
0.2)) # Number of inhibitory neurons
print('exc:', res_exc_nr)
def test_initial_value(self):
sim.setup(timestep=1.0)
pop = sim.Population(5, sim.IF_curr_exp(), label="pop_1")
self.assertEqual([-65, -65, -65, -65, -65], pop.get_initial_value("v"))
view = PopulationView(pop, [1, 3], label="Odds")
view2 = PopulationView(pop, [1, 2], label="OneTwo")
view_iv = view.initial_values
self.assertEqual(1, len(view_iv))
self.assertEqual([-65, -65], view_iv["v"])
view.initialize(v=-60)
self.assertEqual([-65, -60, -65, -60, -65], pop.get_initial_value("v"))
self.assertEqual([-60, -60], view.initial_values["v"])
self.assertEqual([-60, -65], view2.initial_values["v"])
rand_distr = RandomDistribution("uniform",
parameters_pos=[-65.0, -55.0],
rng=NumpyRNG(seed=85524))
view.initialize(v=rand_distr)
self.assertEqual([-64.43349869042906, -63.663421790102184],
view.initial_values["v"])
view.initialize(v=lambda i: -65 + i / 10.0)
self.assertEqual([-64.9, -64.7], view.initial_values["v"])
sim.end()
def obtain_synapses(wiring_plan):
rng = NumpyRNG(seed=64754)
delay_distr = RandomDistribution('normal', [2, 1e-1], rng=rng)
weight_distr = RandomDistribution('normal', [45, 1e-1], rng=rng)
flat_iter = [ (i,j,k,xaxis) for i,j in enumerate(filtered) for k,xaxis in enumerate(j) ]
index_exc = list(set( source for (source,j,target,xaxis) in flat_iter if xaxis==1 or xaxis == 2 ))
index_inh = list(set( source for (source,j,target,xaxis) in flat_iter if xaxis==-1 or xaxis == -2 ))
EElist = []
IIlist = []
EIlist = []
IElist = []
for (source,j,target,xaxis) in flat_iter:
delay = delay_distr.next()
weight = 1.0 # will be updated later.
if xaxis==1 or xaxis == 2:
if target in index_inh:
EIlist.append((source,target,delay,weight))
else:
EElist.append((source,target,delay,weight))
if xaxis==-1 or xaxis == -2:
if target in index_exc:
IElist.append((source,target,delay,weight))
else:
IIlist.append((source,target,delay,weight))
conn_ee = sim.FromListConnector(EElist)
conn_ie = sim.FromListConnector(IElist)
conn_ei = sim.FromListConnector(EIlist)
conn_ii = sim.FromListConnector(IIlist)
return (conn_ee, conn_ie, conn_ei, conn_ii,index_exc,index_inh)
dfEE.loc[0].keys()
dfm = dfEE.as_matrix()
rcls = dfm[:, :1] # real cell labels.
rcls = rcls[1:]
rcls = {k: v
for k, v in enumerate(rcls)} # real cell labels, cast to dictionary
import pickle
with open('cell_names.p', 'wb') as f:
pickle.dump(rcls, f)
import pandas as pd
pd.DataFrame(rcls).to_csv('cell_names.csv', index=False)
filtered = dfm[:, 3:]
filtered = filtered[1:]
rng = NumpyRNG(seed=64754)
delay_distr = RandomDistribution('normal', [45, 1e-1], rng=rng)
index_exc = [i for i, d in enumerate(dfm) if '+' in d[0]]
index_inh = [i for i, d in enumerate(dfm) if '-' in d[0]]
EElist = []
IIlist = []
EIlist = []
IElist = []
for i, j in enumerate(filtered):
for k, xaxis in enumerate(j):
if xaxis == 1 or xaxis == 2:
source = i
target = k
def setupNetwork():
node = pynn.setup(timestep=0.1,
min_delay=1.0,
max_delay=1.0,
debug=True,
quit_on_end=False)
print "Process with rank %d running on %s" % (node, socket.gethostname())
rng = NumpyRNG(seed=seed, parallel_safe=True)
print "[%d] Creating populations" % node
# 1) create excitatory populations
l0_exc_population = pynn.Population(num['l0_exc_neurons'],
native_cell_type('aeif_cond_exp'),
cell_params_adex,
label="exc0")
l0_exc_population.record()
l0_exc_population.record_v()
l1_exc_population = pynn.Population(num['l1_exc_neurons'],
pynn.IF_cond_exp,
cell_params,
label="exc1")
l1_exc_population.record()
# 2) create inhibitory population
l0_inh_population = pynn.Population(num['l0_inh_neurons'],
pynn.IF_cond_exp,
cell_params,
label="inh0")
l0_inh_population.record()
l1_inh_population = pynn.Population(num['l1_inh_neurons'],
pynn.IF_cond_exp,
cell_params,
label="inh1")
l1_inh_population.record()
# 3) exc0 -> inh0
inh_connector = pynn.FixedProbabilityConnector(p_exc0_inh0,
weights=w_exc0_inh0)
l0_exc_inh_projection = pynn.Projection(l0_exc_population,
l0_inh_population, inh_connector)
# 4) exc1 -> inh1
inh_connector = pynn.FixedProbabilityConnector(p_exc1_inh1,
weights=w_exc1_inh1)
l1_exc_inh_projection = pynn.Projection(l1_exc_population,
l1_inh_population, inh_connector)
# 5) exc0 -> exc0
exc_connector = pynn.AllToAllConnector(weights=0.0)
l0_exc_exc_projection = pynn.Projection(l0_exc_population,
l0_exc_population, exc_connector)
exc0_exc0_weights = l0_exc_exc_projection.getWeights()
exc0_exc0_weights = connect_gauss(num['l0_exc_neurons'],
num['l0_exc_neurons'], sigma_exc0_exc0,
w_exc0_exc0_max,
num['l0_exc_maxneighbors'],
exc0_exc0_weights, True)
l0_exc_exc_projection.setWeights(exc0_exc0_weights)
# 6) exc1 -> exc1
exc_connector = pynn.AllToAllConnector(weights=0.0)
l1_exc_exc_projection = pynn.Projection(l1_exc_population,
l1_exc_population, exc_connector)
exc1_exc1_weights = l1_exc_exc_projection.getWeights()
exc1_exc1_weights = connect_gauss(num['l1_exc_neurons'],
num['l1_exc_neurons'], sigma_exc1_exc1,
w_exc1_exc1_max,
num['l1_exc_maxneighbors'],
exc1_exc1_weights, True)
l1_exc_exc_projection.setWeights(exc1_exc1_weights)
# 7) inh0 -> exc0
connector = pynn.FixedProbabilityConnector(p_inh0_exc0,
weights=w_inh0_exc0)
l0_inh_exc_projection = pynn.Projection(l0_inh_population,
l0_exc_population,
connector,
target="inhibitory")
# 8) inh1 -> exc1
connector = pynn.FixedProbabilityConnector(p_inh1_exc1,
weights=w_inh1_exc1)
l1_inh_exc_projection = pynn.Projection(l1_inh_population,
l1_exc_population,
connector,
target="inhibitory")
# 9) create input population
input_population = pynn.Population(num['inputs'],
pynn.SpikeSourcePoisson,
{'rate': input_rate},
label="input")
input_population.record()
# 10) input -> exc0
stdp_model = pynn.STDPMechanism(
timing_dependence=pynn.SpikePairRule(tau_plus=10.0, tau_minus=15.0),
weight_dependence=pynn.AdditiveWeightDependence(w_min=0,
w_max=w_inp_exc0_max,
A_plus=0.012,
A_minus=0.012))
connector = pynn.AllToAllConnector(weights=0.0)
input_projection = pynn.Projection(
input_population,
l0_exc_population,
connector,
rng=rng,
synapse_dynamics=pynn.SynapseDynamics(slow=stdp_model))
input_weights = input_projection.getWeights()
print "[%d] Creating input projections" % node
input_weights = connect_gauss(num['inputs'], num['l0_exc_neurons'],
sigma_inp_exc0, w_inp_exc0_peak,
num['inputs_maxneighbors'], input_weights,
False)
input_projection.setWeights(input_weights)
# 11) exc0 -> exc1
stdp_model = pynn.STDPMechanism(
timing_dependence=pynn.SpikePairRule(tau_plus=20.0, tau_minus=20.0),
weight_dependence=pynn.AdditiveWeightDependence(w_min=0,
w_max=w_exc0_exc1_max,
A_plus=0.012,
A_minus=0.012))
connector = pynn.AllToAllConnector(weights=0.0)
l1_projection = pynn.Projection(
l0_exc_population,
l1_exc_population,
connector,
rng=rng,
synapse_dynamics=pynn.SynapseDynamics(slow=stdp_model))
exc0_exc1_weights = l1_projection.getWeights()
print "[%d] Creating input projections" % node
exc0_exc1_weights = connect_gauss(num['l0_exc_neurons'],
num['l1_exc_neurons'], sigma_exc0_exc1,
w_exc0_exc1_peak,
num['l0_l1_maxneighbors'],
exc0_exc1_weights, False)
l1_projection.setWeights(exc0_exc1_weights)
return node, l0_exc_population, l1_exc_population, l0_inh_population, l1_inh_population, input_population, input_projection, l1_projection
import pyNN.brian as sim
from pyNN.random import RandomDistribution, NumpyRNG
Exc_in = 32
Inh_in = 32
noSpikes = 20 # number of spikes per chanel per simulation run
stimSpikes = RandomDistribution(
'uniform', low=0, high=500.0, rng=NumpyRNG(seed=72386)
).next(
[Exc_in + Inh_in, noSpikes]
) # generate a time uniform distributed signal with Exc_in + Inh_in chanels and noSpikes for each chanel
# print stimSpikes
for i in range(Exc_in):
if i == 0:
Excinp = sim.Population(
1, sim.SpikeSourceArray(spike_times=stimSpikes[i, :]))
else:
spike_source = sim.Population(
1, sim.SpikeSourceArray(spike_times=stimSpikes[i, :]))
Excinp = Excinp + spike_source
for i in range(Inh_in):
if i == 0:
Inhinp = sim.Population(
1, sim.SpikeSourceArray(spike_times=stimSpikes[i + Exc_in, :]))
else:
spike_source = sim.Population(
1, sim.SpikeSourceArray(spike_times=stimSpikes[i + Exc_in, :]))
Inhinp = Inhinp + spike_source
celltype(**cell_params),
label="Excitatory_Cells")
inh_cells = sim.Population(n_inh,
celltype(**cell_params),
label="Inhibitory_Cells")
if options.benchmark == "COBA":
ext_stim = sim.Population(20,
sim.SpikeSourcePoisson(rate=rate,
duration=stim_dur),
label="expoisson")
rconn = 0.01
ext_conn = sim.FixedProbabilityConnector(rconn)
ext_syn = sim.StaticSynapse(weight=0.1)
print "%s Initialising membrane potential to random values..." % node_id
rng = NumpyRNG(seed=rngseed, parallel_safe=parallel_safe)
uniformDistr = RandomDistribution('uniform', [v_reset, v_thresh], rng=rng)
exc_cells.initialize(v=uniformDistr)
inh_cells.initialize(v=uniformDistr)
print "%s Connecting populations..." % node_id
progress_bar = ProgressBar(width=20)
connector = sim.FixedProbabilityConnector(pconn,
rng=rng,
callback=progress_bar)
exc_syn = sim.StaticSynapse(weight=w_exc, delay=delay)
inh_syn = sim.StaticSynapse(weight=w_inh, delay=delay)
connections = {}
connections['e2e'] = sim.Projection(exc_cells,
exc_cells,
import spynnaker8 as sim
import spynnaker8.spynnaker_plotting as splot
import pyNN.utility.plotting as plot
import matplotlib.pyplot as plt
from pyNN.random import RandomDistribution, NumpyRNG
# Remove random effect for testing
# Set to None to randomize
rng = NumpyRNG(seed=1)
# Choose the number of neurons to be simulated in the network.
n_neurons = 100
n_exc = int(round(n_neurons * 0.8))
n_inh = int(round(n_neurons * 0.2))
simtime = 5000
# Set up the simulation to use 0.1ms timesteps.
sim.setup(timestep=0.1)
# Create an excitatory population with 80% of the neurons and
# an inhibitory population with 20% of the neurons.
pop_exc = sim.Population(n_exc, sim.IF_curr_exp(), label="Excitatory")
pop_inh = sim.Population(n_inh, sim.IF_curr_exp(), label="Inhibitory")
# Create excitatory poisson stimulation population with 80%
# of the neurons and
# an inhibitory poisson stimulation population with 20% of the neurons,
# both with a rate of 1000Hz
# TODO RATE?
if rng is None:
seed = None
def do_run(plot):
p.setup(timestep=1.0)
cell_params_lif = {
'cm': 0.25,
'i_offset': 0.0,
'tau_m': 20.0,
'tau_refrac': 2.0,
'tau_syn_E': 5.0,
'tau_syn_I': 5.0,
'v_reset': -70.0,
'v_rest': -65.0,
'v_thresh': -50.0
}
def create_grid(n, label, dx=1.0, dy=1.0):
grid_structure = p.Grid2D(dx=dx, dy=dy, x0=0.0, y0=0.0)
return p.Population(n * n,
p.IF_curr_exp(**cell_params_lif),
structure=grid_structure,
label=label)
# Parameters
n = 5
weight_to_spike = 2.0
delay = 2
runtime = 1000
p.set_number_of_neurons_per_core(p.IF_curr_exp, 100)
# Network population
small_world = create_grid(n, 'small_world')
# SpikeInjector
injectionConnection = [(0, 0)]
spikeArray = {'spike_times': [[0]]}
inj_pop = p.Population(1,
p.SpikeSourceArray(**spikeArray),
label='inputSpikes_1')
# Injector projection
p.Projection(inj_pop, small_world,
p.FromListConnector(injectionConnection),
p.StaticSynapse(weight=weight_to_spike, delay=delay))
# Connectors
degree = 2.0
rewiring = 0.4
rng = NumpyRNG(seed=1)
small_world_connector = p.SmallWorldConnector(degree, rewiring, rng=rng)
# Projection for small world grid
sw_pro = p.Projection(small_world, small_world, small_world_connector,
p.StaticSynapse(weight=2.0, delay=5))
small_world.record(['v', 'spikes'])
p.run(runtime)
v = small_world.get_data('v')
spikes = small_world.get_data('spikes')
weights = sw_pro.get('weight', 'list')
if plot:
# pylint: disable=no-member
Figure(
# raster plot of the presynaptic neuron spike times
Panel(spikes.segments[0].spiketrains,
yticks=True,
markersize=0.2,
xlim=(0, runtime),
xticks=True),
# membrane potential of the postsynaptic neuron
Panel(v.segments[0].filter(name='v')[0],
ylabel="Membrane potential (mV)",
data_labels=[small_world.label],
yticks=True,
xlim=(0, runtime),
xticks=True),
title="Simple small world connector",
annotations="Simulated with {}".format(p.name()))
plt.show()
p.end()
return v, spikes, weights
def setupNetwork():
node = pynn.setup(timestep=0.1,
min_delay=1.0,
max_delay=1.0,
debug=True,
quit_on_end=False)
print "Process with rank %d running on %s" % (node, socket.gethostname())
rng = NumpyRNG(seed=seed, parallel_safe=True)
print "[%d] Creating populations" % node
# 1) create excitatory populations
l0_exc_population = pynn.Population(num['l0_exc_neurons'],
pynn.IF_cond_exp,
cell_params,
label="exc0")
l0_exc_population.record()
#l0_exc_population.record_v()
l1_exc_population = pynn.Population(num['l1_exc_neurons'],
pynn.IF_cond_exp,
cell_params,
label="exc1")
l1_exc_population.record()
# 2) create inhibitory population
l0_inh_population = pynn.Population(num['l0_inh_neurons'],
pynn.IF_cond_exp,
cell_params,
label="inh0")
l0_inh_population.record()
l1_inh_population = pynn.Population(num['l1_inh_neurons'],
pynn.IF_cond_exp,
cell_params,
label="inh1")
l1_inh_population.record()
# 3) exc0 -> inh0
inh_connector = pynn.FixedProbabilityConnector(p_exc0_inh0,
weights=w_exc0_inh0)
l0_exc_inh_projection = pynn.Projection(l0_exc_population,
l0_inh_population, inh_connector)
# 4) exc1 -> inh1
inh_connector = pynn.FixedProbabilityConnector(p_exc1_inh1,
weights=w_exc1_inh1)
l1_exc_inh_projection = pynn.Projection(l1_exc_population,
l1_inh_population, inh_connector)
# 5) exc0 -> exc0
exc_connector = pynn.FixedProbabilityConnector(
p_exc0_exc0, weights=w_exc0_exc0, allow_self_connections=False)
l0_exc_exc_projection = pynn.Projection(l0_exc_population,
l0_exc_population, exc_connector)
# 6) exc1 -> exc1
#l1_exc_exc_projection = pynn.Projection(l1_exc_population,l1_exc_population,exc_connector)
#for i in range(num['nodes']):
# exc_inh_projections.append(Projection(exc_populations[i],inh_population,inh_connector))
# for j in range(i-num['neighbours'],i+num['neighbours']+1):
# if j != i:
# exc_connector = OneToOneConnector(weights=1.0/abs(j-i))
# if j<0:
# j+=num['nodes']
# if j> num['nodes']-1:
# j-=num['nodes']
# exc_exc_projections.append(Projection(exc_populations[i],exc_populations[j],exc_connector))
# 7) inh0 -> exc0
connector = pynn.FixedProbabilityConnector(p_inh0_exc0,
weights=w_inh0_exc0)
l0_inh_exc_projection = pynn.Projection(l0_inh_population,
l0_exc_population,
connector,
target="inhibitory")
# 8) inh1 -> exc1
connector = pynn.FixedProbabilityConnector(p_inh1_exc1,
weights=w_inh1_exc1)
l1_inh_exc_projection = pynn.Projection(l1_inh_population,
l1_exc_population,
connector,
target="inhibitory")
# 9) create input population
input_population = pynn.Population(num['inputs'],
pynn.SpikeSourcePoisson,
{'rate': input_rate},
label="input")
input_population.record()
# 10) input -> exc0
stdp_model = pynn.STDPMechanism(
timing_dependence=pynn.SpikePairRule(tau_plus=20.0, tau_minus=20.0),
weight_dependence=pynn.AdditiveWeightDependence(w_min=0,
w_max=w_inp_exc0_max,
A_plus=0.01,
A_minus=0.012))
connector = pynn.AllToAllConnector(weights=pynn.RandomDistribution(
distribution='uniform', parameters=[0.00, w_inp_exc0_max], rng=rng))
#connector = pynn.FixedProbabilityConnector(p_inp_exc0,weights=w_inp_exc0)
input_projection = pynn.Projection(
input_population,
l0_exc_population,
connector,
rng=rng,
#synapse_dynamics=pynn.SynapseDynamics(slow=stdp_model)
)
# 11) exc0 -> exc1
stdp_model = pynn.STDPMechanism(
timing_dependence=pynn.SpikePairRule(tau_plus=20.0, tau_minus=20.0),
weight_dependence=pynn.AdditiveWeightDependence(w_min=0,
w_max=w_exc0_exc1_max,
A_plus=0.01,
A_minus=0.012))
connector = pynn.AllToAllConnector(weights=pynn.RandomDistribution(
distribution='uniform', parameters=[0.00, w_exc0_exc1_max], rng=rng))
#connector = pynn.FixedProbabilityConnector(0.05, weights=0.01)
l1_projection = pynn.Projection(
l0_exc_population,
l1_exc_population,
connector,
rng=rng,
#synapse_dynamics=pynn.SynapseDynamics(slow=stdp_model)
)
return node, l0_exc_population, l1_exc_population, l0_inh_population, l1_inh_population, input_population, input_projection, l1_projection
def __init__(self, boundary, origin=(0.0, 0.0, 0.0), rng=None):
assert isinstance(boundary, Shape)
assert len(origin) == 3
self.boundary = boundary
self.origin = origin
self.rng = rng or NumpyRNG()
def setup(self, sim):
# Create matrix of synaptic weights
self.w = create_weight_matrix()
model = getattr(sim, 'IF_curr_exp')
script_rng = NumpyRNG(seed=6508015, parallel_safe=parallel_safe)
distr = RandomDistribution('normal', [V0_mean, V0_sd], rng=script_rng)
# Create cortical populations
self.pops = {}
for layer in layers:
self.pops[layer] = {}
for pop in pops:
self.pops[layer][pop] = sim.Population(
int(N_full[layer][pop] * N_scaling),
model,
cellparams=neuron_params)
self.pops[layer][pop].initialize(v=distr)
# Store whether population is inhibitory or excitatory
self.pops[layer][pop].annotate(type=pop)
this_pop = self.pops[layer][pop]
# Spike recording
if record_fraction:
num_spikes = int(round(this_pop.size * frac_record_spikes))
else:
num_spikes = n_record
this_pop[0:num_spikes].record('spikes')
# Membrane potential recording
if record_v:
if record_fraction:
num_v = int(round(this_pop.size * frac_record_v))
else:
num_v = n_record_v
this_pop[0:num_v].record('v')
# Create thalamic population
if thalamic_input:
self.thalamic_population = sim.Population(
thal_params['n_thal'], sim.SpikeSourcePoisson, {
'rate': thal_params['rate'],
'start': thal_params['start'],
'duration': thal_params['duration']
})
# Compute DC input before scaling
if input_type == 'DC':
self.DC_amp = {}
for target_layer in layers:
self.DC_amp[target_layer] = {}
for target_pop in pops:
self.DC_amp[target_layer][target_pop] = bg_rate * \
K_ext[target_layer][target_pop] * w_mean * neuron_params['tau_syn_E'] / \
1000.
else:
self.DC_amp = {
'L23': {
'E': 0.,
'I': 0.
},
'L4': {
'E': 0.,
'I': 0.
},
'L5': {
'E': 0.,
'I': 0.
},
'L6': {
'E': 0.,
'I': 0.
}
}
# Scale and connect
# In-degrees of the full-scale model
K_full = scaling.get_indegrees()
if K_scaling != 1:
self.w, self.w_ext, self.K_ext, self.DC_amp = scaling.adjust_w_and_ext_to_K(
K_full, K_scaling, self.w, self.DC_amp)
else:
self.w_ext = w_ext
if sim.rank() == 0:
print('w: %s' % self.w)
for target_layer in layers:
for target_pop in pops:
target_index = structure[target_layer][target_pop]
this_pop = self.pops[target_layer][target_pop]
# External inputs
if input_type == 'DC' or K_scaling != 1:
this_pop.set(
i_offset=self.DC_amp[target_layer][target_pop])
if input_type == 'poisson':
poisson_generator = sim.Population(
this_pop.size, sim.SpikeSourcePoisson, {
'rate':
bg_rate * self.K_ext[target_layer][target_pop]
})
conn = sim.OneToOneConnector()
syn = sim.StaticSynapse(weight=self.w_ext)
sim.Projection(poisson_generator,
this_pop,
conn,
syn,
receptor_type='excitatory')
if thalamic_input:
# Thalamic inputs
if sim.rank() == 0:
print('creating thalamic connections to %s%s' %
(target_layer, target_pop))
C_thal = thal_params['C'][target_layer][target_pop]
n_target = N_full[target_layer][target_pop]
K_thal = round(
np.log(1 - C_thal) / np.log(
(n_target * thal_params['n_thal'] - 1.) /
(n_target * thal_params['n_thal']))) / n_target
FixedTotalNumberConnect(sim, self.thalamic_population,
this_pop, K_thal, w_ext,
w_rel * w_ext, d_mean['E'],
d_sd['E'])
# Recurrent inputs
for source_layer in layers:
for source_pop in pops:
source_index = structure[source_layer][source_pop]
if sim.rank() == 0:
print('creating connections from %s%s to %s%s' %
(source_layer, source_pop, target_layer,
target_pop))
weight = self.w[target_index][source_index]
if source_pop == 'E' and source_layer == 'L4' and target_layer == 'L23' and target_pop == 'E':
w_sd = weight * w_rel_234
else:
w_sd = abs(weight * w_rel)
FixedTotalNumberConnect(
sim, self.pops[source_layer][source_pop],
self.pops[target_layer][target_pop],
K_full[target_index][source_index] * K_scaling,
weight, w_sd, d_mean[source_pop], d_sd[source_pop])
def sim_runner(wgf):
wg = wgf
import pyNN.neuron as sim
nproc = sim.num_processes()
node = sim.rank()
print(nproc)
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
import matplotlib as mpl
mpl.rcParams.update({'font.size':16})
#import mpi4py
#threads = sim.rank()
threads = 1
rngseed = 98765
parallel_safe = False
#extra = {'threads' : threads}
import os
import pandas as pd
import sys
import numpy as np
from pyNN.neuron import STDPMechanism
import copy
from pyNN.random import RandomDistribution, NumpyRNG
import pyNN.neuron as neuron
from pyNN.neuron import h
from pyNN.neuron import StandardCellType, ParameterSpace
from pyNN.random import RandomDistribution, NumpyRNG
from pyNN.neuron import STDPMechanism, SpikePairRule, AdditiveWeightDependence, FromListConnector, TsodyksMarkramSynapse
from pyNN.neuron import Projection, OneToOneConnector
from numpy import arange
import pyNN
from pyNN.utility import get_simulator, init_logging, normalized_filename
import random
import socket
#from neuronunit.optimization import get_neab
import networkx as nx
sim = pyNN.neuron
# Get some hippocampus connectivity data, based on a conversation with
# academic researchers on GH:
# https://github.com/Hippocampome-Org/GraphTheory/issues?q=is%3Aissue+is%3Aclosed
# scrape hippocamome connectivity data, that I intend to use to program neuromorphic hardware.
# conditionally get files if they don't exist.
path_xl = '_hybrid_connectivity_matrix_20171103_092033.xlsx'
if not os.path.exists(path_xl):
os.system('wget https://github.com/Hippocampome-Org/GraphTheory/files/1657258/_hybrid_connectivity_matrix_20171103_092033.xlsx')
xl = pd.ExcelFile(path_xl)
dfEE = xl.parse()
dfEE.loc[0].keys()
dfm = dfEE.as_matrix()
rcls = dfm[:,:1] # real cell labels.
rcls = rcls[1:]
rcls = { k:v for k,v in enumerate(rcls) } # real cell labels, cast to dictionary
import pickle
with open('cell_names.p','wb') as f:
pickle.dump(rcls,f)
import pandas as pd
pd.DataFrame(rcls).to_csv('cell_names.csv', index=False)
filtered = dfm[:,3:]
filtered = filtered[1:]
rng = NumpyRNG(seed=64754)
delay_distr = RandomDistribution('normal', [2, 1e-1], rng=rng)
weight_distr = RandomDistribution('normal', [45, 1e-1], rng=rng)
sanity_e = []
sanity_i = []
EElist = []
IIlist = []
EIlist = []
IElist = []
for i,j in enumerate(filtered):
for k,xaxis in enumerate(j):
if xaxis == 1 or xaxis == 2:
source = i
sanity_e.append(i)
target = k
if xaxis ==-1 or xaxis == -2:
sanity_i.append(i)
source = i
target = k
index_exc = list(set(sanity_e))
index_inh = list(set(sanity_i))
import pickle
with open('cell_indexs.p','wb') as f:
returned_list = [index_exc, index_inh]
pickle.dump(returned_list,f)
import numpy
a = numpy.asarray(index_exc)
numpy.savetxt('pickles/'+str(k)+'excitatory_nunber_labels.csv', a, delimiter=",")
import numpy
a = numpy.asarray(index_inh)
numpy.savetxt('pickles/'+str(k)+'inhibitory_nunber_labels.csv', a, delimiter=",")
for i,j in enumerate(filtered):
for k,xaxis in enumerate(j):
if xaxis==1 or xaxis == 2:
source = i
sanity_e.append(i)
target = k
delay = delay_distr.next()
weight = 1.0
if target in index_inh:
EIlist.append((source,target,delay,weight))
else:
EElist.append((source,target,delay,weight))
if xaxis==-1 or xaxis == -2:
sanity_i.append(i)
source = i
target = k
delay = delay_distr.next()
weight = 1.0
if target in index_exc:
IElist.append((source,target,delay,weight))
else:
IIlist.append((source,target,delay,weight))
internal_conn_ee = sim.FromListConnector(EElist)
ee = internal_conn_ee.conn_list
ee_srcs = ee[:,0]
ee_tgs = ee[:,1]
internal_conn_ie = sim.FromListConnector(IElist)
ie = internal_conn_ie.conn_list
ie_srcs = set([ int(e[0]) for e in ie ])
ie_tgs = set([ int(e[1]) for e in ie ])
internal_conn_ei = sim.FromListConnector(EIlist)
ei = internal_conn_ei.conn_list
ei_srcs = set([ int(e[0]) for e in ei ])
ei_tgs = set([ int(e[1]) for e in ei ])
internal_conn_ii = sim.FromListConnector(IIlist)
ii = internal_conn_ii.conn_list
ii_srcs = set([ int(e[0]) for e in ii ])
ii_tgs = set([ int(e[1]) for e in ii ])
for e in internal_conn_ee.conn_list:
assert e[0] in ee_srcs
assert e[1] in ee_tgs
for i in internal_conn_ii.conn_list:
assert i[0] in ii_srcs
assert i[1] in ii_tgs
ml = len(filtered[1])+1
pre_exc = []
post_exc = []
pre_inh = []
post_inh = []
rng = NumpyRNG(seed=64754)
delay_distr = RandomDistribution('normal', [2, 1e-1], rng=rng)
plot_EE = np.zeros(shape=(ml,ml), dtype=bool)
plot_II = np.zeros(shape=(ml,ml), dtype=bool)
plot_EI = np.zeros(shape=(ml,ml), dtype=bool)
plot_IE = np.zeros(shape=(ml,ml), dtype=bool)
for i in EElist:
plot_EE[i[0],i[1]] = int(0)
#plot_ss[i[0],i[1]] = int(1)
if i[0]!=i[1]: # exclude self connections
plot_EE[i[0],i[1]] = int(1)
pre_exc.append(i[0])
post_exc.append(i[1])
assert len(pre_exc) == len(post_exc)
for i in IIlist:
plot_II[i[0],i[1]] = int(0)
if i[0]!=i[1]:
plot_II[i[0],i[1]] = int(1)
pre_inh.append(i[0])
post_inh.append(i[1])
for i in IElist:
plot_IE[i[0],i[1]] = int(0)
if i[0]!=i[1]: # exclude self connections
plot_IE[i[0],i[1]] = int(1)
pre_inh.append(i[0])
post_inh.append(i[1])
for i in EIlist:
plot_EI[i[0],i[1]] = int(0)
if i[0]!=i[1]:
plot_EI[i[0],i[1]] = int(1)
pre_exc.append(i[0])
post_exc.append(i[1])
plot_excit = plot_EI + plot_EE
plot_inhib = plot_IE + plot_II
assert len(pre_inh) == len(post_inh)
num_exc = [ i for i,e in enumerate(plot_excit) if sum(e) > 0 ]
num_inh = [ y for y,i in enumerate(plot_inhib) if sum(i) > 0 ]
# the network is dominated by inhibitory neurons, which is unusual for modellers.
assert num_inh > num_exc
assert np.sum(plot_inhib) > np.sum(plot_excit)
assert len(num_exc) < ml
assert len(num_inh) < ml
# # Plot all the Projection pairs as a connection matrix (Excitatory and Inhibitory Connections)
import pickle
with open('graph_inhib.p','wb') as f:
pickle.dump(plot_inhib,f, protocol=2)
import pickle
with open('graph_excit.p','wb') as f:
pickle.dump(plot_excit,f, protocol=2)
#with open('cell_names.p','wb') as f:
# pickle.dump(rcls,f)
import pandas as pd
pd.DataFrame(plot_EE).to_csv('ee.csv', index=False)
import pandas as pd
pd.DataFrame(plot_IE).to_csv('ie.csv', index=False)
import pandas as pd
pd.DataFrame(plot_II).to_csv('ii.csv', index=False)
import pandas as pd
pd.DataFrame(plot_EI).to_csv('ei.csv', index=False)
from scipy.sparse import coo_matrix
m = np.matrix(filtered[1:])
bool_matrix = np.add(plot_excit,plot_inhib)
with open('bool_matrix.p','wb') as f:
pickle.dump(bool_matrix,f, protocol=2)
if not isinstance(m, coo_matrix):
m = coo_matrix(m)
Gexc_ud = nx.Graph(plot_excit)
avg_clustering = nx.average_clustering(Gexc_ud)#, nodes=None, weight=None, count_zeros=True)[source]
rc = nx.rich_club_coefficient(Gexc_ud,normalized=False)
print('This graph structure as rich as: ',rc[0])
gexc = nx.DiGraph(plot_excit)
gexcc = nx.betweenness_centrality(gexc)
top_exc = sorted(([ (v,k) for k, v in dict(gexcc).items() ]), reverse=True)
in_degree = gexc.in_degree()
top_in = sorted(([ (v,k) for k, v in in_degree.items() ]))
in_hub = top_in[-1][1]
out_degree = gexc.out_degree()
top_out = sorted(([ (v,k) for k, v in out_degree.items() ]))
out_hub = top_out[-1][1]
mean_out = np.mean(list(out_degree.values()))
mean_in = np.mean(list(in_degree.values()))
mean_conns = int(mean_in + mean_out/2)
k = 2 # number of neighbouig nodes to wire.
p = 0.25 # probability of instead wiring to a random long range destination.
ne = len(plot_excit)# size of small world network
small_world_ring_excit = nx.watts_strogatz_graph(ne,mean_conns,0.25)
k = 2 # number of neighbouring nodes to wire.
p = 0.25 # probability of instead wiring to a random long range destination.
ni = len(plot_inhib)# size of small world network
small_world_ring_inhib = nx.watts_strogatz_graph(ni,mean_conns,0.25)
nproc = sim.num_processes()
nproc = 8
host_name = socket.gethostname()
node_id = sim.setup(timestep=0.01, min_delay=1.0)#, **extra)
print("Host #%d is on %s" % (node_id + 1, host_name))
rng = NumpyRNG(seed=64754)
#pop_size = len(num_exc)+len(num_inh)
#num_exc = [ i for i,e in enumerate(plot_excit) if sum(e) > 0 ]
#num_inh = [ y for y,i in enumerate(plot_inhib) if sum(i) > 0 ]
#pop_exc = sim.Population(len(num_exc), sim.Izhikevich(a=0.02, b=0.2, c=-65, d=8, i_offset=0))
#pop_inh = sim.Population(len(num_inh), sim.Izhikevich(a=0.02, b=0.25, c=-65, d=2, i_offset=0))
#index_exc = list(set(sanity_e))
#index_inh = list(set(sanity_i))
all_cells = sim.Population(len(index_exc)+len(index_inh), sim.Izhikevich(a=0.02, b=0.2, c=-65, d=8, i_offset=0))
#all_cells = None
#all_cells = pop_exc + pop_inh
pop_exc = sim.PopulationView(all_cells,index_exc)
pop_inh = sim.PopulationView(all_cells,index_inh)
#print(pop_exc)
#print(dir(pop_exc))
for pe in pop_exc:
print(pe)
#import pdb
pe = all_cells[pe]
#pdb.set_trace()
#pe = all_cells[i]
r = random.uniform(0.0, 1.0)
pe.set_parameters(a=0.02, b=0.2, c=-65+15*r, d=8-r**2, i_offset=0)
#pop_exc.append(pe)
#pop_exc = sim.Population(pop_exc)
for pi in index_inh:
pi = all_cells[pi]
#print(pi)
#pi = all_cells[i]
r = random.uniform(0.0, 1.0)
pi.set_parameters(a=0.02+0.08*r, b=0.25-0.05*r, c=-65, d= 2, i_offset=0)
#pop_inh.append(pi)
#pop_inh = sim.Population(pop_inh)
'''
for pe in pop_exc:
r = random.uniform(0.0, 1.0)
pe.set_parameters(a=0.02, b=0.2, c=-65+15*r, d=8-r**2, i_offset=0)
for pi in pop_inh:
r = random.uniform(0.0, 1.0)
pi.set_parameters(a=0.02+0.08*r, b=0.25-0.05*r, c=-65, d= 2, i_offset=0)
'''
NEXC = len(num_exc)
NINH = len(num_inh)
exc_syn = sim.StaticSynapse(weight = wg, delay=delay_distr)
assert np.any(internal_conn_ee.conn_list[:,0]) < ee_srcs.size
prj_exc_exc = sim.Projection(all_cells, all_cells, internal_conn_ee, exc_syn, receptor_type='excitatory')
prj_exc_inh = sim.Projection(all_cells, all_cells, internal_conn_ei, exc_syn, receptor_type='excitatory')
inh_syn = sim.StaticSynapse(weight = wg, delay=delay_distr)
delay_distr = RandomDistribution('normal', [1, 100e-3], rng=rng)
prj_inh_inh = sim.Projection(all_cells, all_cells, internal_conn_ii, inh_syn, receptor_type='inhibitory')
prj_inh_exc = sim.Projection(all_cells, all_cells, internal_conn_ie, inh_syn, receptor_type='inhibitory')
inh_distr = RandomDistribution('normal', [1, 2.1e-3], rng=rng)
def prj_change(prj,wg):
prj.setWeights(wg)
prj_change(prj_exc_exc,wg)
prj_change(prj_exc_inh,wg)
prj_change(prj_inh_exc,wg)
prj_change(prj_inh_inh,wg)
def prj_check(prj):
for w in prj.weightHistogram():
for i in w:
print(i)
prj_check(prj_exc_exc)
prj_check(prj_exc_inh)
prj_check(prj_inh_exc)
prj_check(prj_inh_inh)
#print(rheobase['value'])
#print(float(rheobase['value']),1.25/1000.0)
'''Old values that worked
noise = sim.NoisyCurrentSource(mean=0.85/1000.0, stdev=5.00/1000.0, start=0.0, stop=2000.0, dt=1.0)
pop_exc.inject(noise)
#1000.0 pA
noise = sim.NoisyCurrentSource(mean=1.740/1000.0, stdev=5.00/1000.0, start=0.0, stop=2000.0, dt=1.0)
pop_inh.inject(noise)
#1750.0 pA
'''
noise = sim.NoisyCurrentSource(mean=0.74/1000.0, stdev=4.00/1000.0, start=0.0, stop=2000.0, dt=1.0)
pop_exc.inject(noise)
#1000.0 pA
noise = sim.NoisyCurrentSource(mean=1.440/1000.0, stdev=4.00/1000.0, start=0.0, stop=2000.0, dt=1.0)
pop_inh.inject(noise)
##
# Setup and run a simulation. Note there is no current injection into the neuron.
# All cells in the network are in a quiescent state, so its not a surprise that xthere are no spikes
##
sim = pyNN.neuron
arange = np.arange
import re
all_cells.record(['v','spikes']) # , 'u'])
all_cells.initialize(v=-65.0, u=-14.0)
# === Run the simulation =====================================================
tstop = 2000.0
sim.run(tstop)
data = None
data = all_cells.get_data().segments[0]
#print(len(data.analogsignals[0].times))
with open('pickles/qi'+str(wg)+'.p', 'wb') as f:
pickle.dump(data,f)
# make data none or else it will grow in a loop
all_cells = None
data = None
noise = None
'e_rev_I': e_rev_I,
'v_reset': v_reset,
'tau_refrac': tau_refrac,
'i_offset': i_offset
}
cell_type = sim.IF_cond_exp(**neuronParameters)
input = sim.Population(20, sim.SpikeSourcePoisson(rate=50.0))
output = sim.Population(25, cell_type)
rand_distr = RandomDistribution('uniform', (v_reset, v_thresh), rng=NumpyRNG(seed=85524))
output.initialize(v=rand_distr)
stdp = sim.STDPMechanism(weight_dependence=sim.AdditiveWeightDependence(w_min=0.0, w_max=0.1),
timing_dependence=sim.Vogels2011Rule(eta=0.0, rho=1e-3),
weight=0.005, delay=0.5)
fpc = sim.FixedProbabilityConnector(0.02, rng=NumpyRNG(seed=854))
v_rest = list()
v_thresh = list()
for atom in range(0, nNeurons):
cm.append(0.25)
i_off.append(0.0)
tau_m.append(10.0)
tau_re.append(2.0)
tau_syn_e.append(0.5)
tau_syn_i.append(0.5)
v_reset.append(-65.0)
v_rest.append(-65.0)
v_thresh.append(-64.4)
gbar_na_distr = RandomDistribution('normal', (20.0, 2.0),
rng=NumpyRNG(seed=85524))
cell_params_lif = {
'cm': cm, # nF
'i_offset': i_off,
'tau_m': tau_m,
'tau_refrac': tau_re,
'tau_syn_E': tau_syn_e,
'tau_syn_I': tau_syn_i,
'v_reset': v_reset,
'v_rest': v_rest,
'v_thresh': v_thresh
}
populations = list()
projections = list()
timer = Timer()
seed = 764756387
tstop = 1000.0 # ms
input_rate = 100.0 # Hz
cell_params = {'tau_refrac': 2.0, # ms
'v_thresh': -50.0, # mV
'tau_syn_E': 2.0, # ms
'tau_syn_I': 2.0} # ms
n_record = 5
node = setup(timestep=0.025, min_delay=1.0, max_delay=10.0, debug=True, quit_on_end=False)
print("Process with rank %d running on %s" % (node, socket.gethostname()))
rng = NumpyRNG(seed=seed, parallel_safe=True)
print("[%d] Creating populations" % node)
n_spikes = int(2 * tstop * input_rate / 1000.0)
spike_times = numpy.add.accumulate(rng.next(n_spikes, 'exponential',
{'beta': 1000.0 / input_rate}, mask_local=False))
input_population = Population(100, SpikeSourceArray(spike_times=spike_times), label="input")
output_population = Population(10, IF_curr_exp(**cell_params), label="output")
print("[%d] input_population cells: %s" % (node, input_population.local_cells))
print("[%d] output_population cells: %s" % (node, output_population.local_cells))
print("[%d] Connecting populations" % node)
timer.start()
connector = CSAConnector(csa.random(0.5))
syn = StaticSynapse(weight=0.1)
def do_run(nNeurons):
p.setup(timestep=1.0, min_delay=1.0, max_delay=32.0)
p.set_number_of_neurons_per_core("IF_curr_exp", 100)
cm = list()
i_off = list()
tau_m = list()
tau_re = list()
tau_syn_e = list()
tau_syn_i = list()
v_reset = list()
v_rest = list()
v_thresh = list()
cell_params_lif = {'cm': cm, 'i_offset': i_off, 'tau_m': tau_m,
'tau_refrac': tau_re, 'tau_syn_E': tau_syn_e,
'tau_syn_I': tau_syn_i, 'v_reset': v_reset,
'v_rest': v_rest, 'v_thresh': v_thresh}
for atom in range(0, nNeurons):
cm.append(0.25)
i_off.append(0.0)
tau_m.append(10.0)
tau_re.append(2.0)
tau_syn_e.append(0.5)
tau_syn_i.append(0.5)
v_reset.append(-65.0)
v_rest.append(-65.0)
v_thresh.append(-64.4)
gbar_na_distr = RandomDistribution('normal', (20.0, 2.0),
rng=NumpyRNG(seed=85524))
cell_params_lif = {'cm': cm, 'i_offset': i_off, 'tau_m': tau_m,
'tau_refrac': tau_re, 'tau_syn_E': tau_syn_e,
'tau_syn_I': tau_syn_i, 'v_reset': v_reset,
'v_rest': v_rest, 'v_thresh': v_thresh}
populations = list()
projections = list()
weight_to_spike = 2
delay = 1
connections = list()
for i in range(0, nNeurons):
singleConnection = (i, ((i + 1) % nNeurons), weight_to_spike, delay)
connections.append(singleConnection)
injectionConnection = [(0, 0, weight_to_spike, delay)]
spikeArray = {'spike_times': [[0]]}
populations.append(p.Population(nNeurons, p.IF_curr_exp, cell_params_lif,
label='pop_1'))
populations.append(p.Population(1, p.SpikeSourceArray, spikeArray,
label='inputSpikes_1'))
populations[0].set({'cm': 0.25})
populations[0].set('cm', cm)
populations[0].set({'tau_m': tau_m, 'v_thresh': v_thresh})
populations[0].set('i_offset', gbar_na_distr)
populations[0].set('i_offset', i_off)
projections.append(p.Projection(populations[0], populations[0],
p.FromListConnector(connections)))
projections.append(p.Projection(populations[1], populations[0],
p.FromListConnector(injectionConnection)))
populations[0].record_v()
populations[0].record_gsyn()
populations[0].record()
p.run(100)
v = populations[0].get_v(compatible_output=True)
gsyn = populations[0].get_gsyn(compatible_output=True)
spikes = populations[0].getSpikes(compatible_output=True)
p.end()
return (v, gsyn, spikes)
def network(para, simulator):
if simulator == "-nest":
sim = import_module("pyNN.nest")
elif simulator == "-neuron":
sim = import_module("pyNN.neuron")
elif simulator == "-brian":
sim = import_module("pyNN.brian")
# initialize pyNN simulation
sim.setup(timestep=para["dt"])
# Parameters for excitatory neurons
E_parameters = {
"tau_m": para["taumE"],
"cm": para["taumE"] / para["R"],
"v_rest": para["Vr"],
"v_reset": para["Vreset"],
"v_thresh": para["Vt"],
"tau_refrac": para["tref"],
"tau_syn_E": para["tau_syn_e"],
"tau_syn_I": para["tau_syn_e"],
"i_offset": (para["VextE"] / para["R"])
}
# Paramters for inhibitory neurons
I_parameters = {
"tau_m": para["taumI"],
"cm": para["taumI"] / para["R"],
"v_rest": para["Vr"],
"v_reset": para["Vreset"],
"v_thresh": para["Vt"],
"tau_refrac": para["tref"],
"tau_syn_E": para["tau_syn_e"],
"tau_syn_I": para["tau_syn_e"],
"i_offset": (para["VextI"] / para["R"])
}
############ All Excitatory and inhibitory neurons ########################
# number of excitatory neurons in one network
NE_net = int(para['N'] * para['Ne'])
# Total number of excitatory neurons
NE = int(para['NAreas'] * NE_net)
# number of excitatory neurons in one network
NI_net = int(para['N'] * (1 - para['Ne']))
# Total number of inhibitory neurons
NI = int((para['NAreas'] * NI_net))
popE = sim.Population(NE, sim.IF_curr_alpha, E_parameters, label="popE")
popI = sim.Population(NI, sim.IF_curr_alpha, I_parameters, label="popI")
################################ Noise ####################################
# Noise on excitatory neurons
stdNoiseE = (para["sigma"] / para["R"]) * (para["taumE"]**
0.5) / (para["dt"]**0.5)
for i in range(NE):
noise = sim.NoisyCurrentSource(mean=0,
stdev=stdNoiseE,
start=0.0,
stop=para["duration"])
popE[i].inject(noise)
# Noise on inhibitory neurons
stdNoiseI = (para["sigma"] / para["R"]) * (para["taumI"]**
0.5) / (para["dt"]**0.5)
for i in range(NI):
noise = sim.NoisyCurrentSource(mean=0,
stdev=stdNoiseI,
start=0.0,
stop=para["duration"])
popI[i].inject(noise)
# paramters for initial conditions
kernelseed = 5456532
rng = NumpyRNG(kernelseed, parallel_safe=True)
uniformDistr = RandomDistribution('uniform',
low=para["Vr"],
high=para["Vt"],
rng=rng)
sim.initialize(popE, v=uniformDistr)
sim.initialize(popI, v=uniformDistr)
# Separate population in population views
popEList = []
popIList = []
# Store projections
EE = []
IE = []
EI = []
II = []
EElongRange = []
IElongRange = []
for i in range(para['NAreas']):
# store sub populations in lists
popEList.append(popE[(i * NE_net):((i + 1) * NE_net)])
popIList.append(popI[(i * NI_net):((i + 1) * NI_net)])
#### Synapses
# Weights for recurrent connections
wEE_alpha = (((1 + para['alpha'] * para["hier"][i]) * para["wEE"]) /
para['coeffE'])[0] #[nA]
wIE_alpha = (((1 + para['alpha'] * para["hier"][i]) * para["wIE"]) /
para['coeffI'])[0] #[nA]
wEI_alpha = (para["wEI"] / para['coeffE']) * -1 #[nA]
wII_alpha = (para["wII"] / para['coeffI']) * -1 #[nA]
# Connections
EE_connections = sim.Projection(
popEList[i], popEList[i],
sim.FixedProbabilityConnector(p_connect=para["probIntra"]),
sim.StaticSynapse(weight=wEE_alpha, delay=para["dlocal"]))
IE_connections = sim.Projection(
popEList[i], popIList[i],
sim.FixedProbabilityConnector(p_connect=para["probIntra"]),
sim.StaticSynapse(weight=wIE_alpha, delay=para["dlocal"]))
EI_connections = sim.Projection(
popIList[i], popEList[i],
sim.FixedProbabilityConnector(p_connect=para["probIntra"]),
sim.StaticSynapse(weight=wEI_alpha, delay=para["dlocal"]))
II_connections = sim.Projection(
popIList[i], popIList[i],
sim.FixedProbabilityConnector(p_connect=para["probIntra"]),
sim.StaticSynapse(weight=wII_alpha, delay=para["dlocal"]))
# Store projections in lists
EE.append(EE_connections)
IE.append(IE_connections)
EI.append(EI_connections)
II.append(II_connections)
# Long Range connections
for i in range(para['NAreas']):
for j in range(para['NAreas']):
if i != j:
# Weights
wEE_alphaLR = (
(1 + para['alpha'] * para["hier"][j]) * para['muEE'] *
para["conn"][j, i])[0] / para['coeffE']
wIE_alphaLR = (
(1 + para['alpha'] * para["hier"][j]) * para['muIE'] *
para["conn"][j, i])[0] / para['coeffI']
# Delay
# Mean for delay distribution
meanlr = para["delayMat"][j, i]
# Standard deviation for delay distribution
varlr = para['lrvar'] * meanlr
dLR = RandomDistribution('normal', [meanlr, varlr],
rng=NumpyRNG(seed=4242))
# Connections
EE_connectionsLR = sim.Projection(
popEList[i], popEList[j],
sim.FixedProbabilityConnector(p_connect=para["probInter"]),
sim.StaticSynapse(weight=wEE_alphaLR, delay=dLR))
IE_connectionsLR = sim.Projection(
popEList[i], popIList[j],
sim.FixedProbabilityConnector(p_connect=para["probInter"]),
sim.StaticSynapse(weight=wIE_alphaLR, delay=dLR))
# Store projections in list
EElongRange.append(EE_connectionsLR)
IElongRange.append(IE_connectionsLR)
# Stimulus
amplitude = para['currval'] / para['R'] #[mV]
pulse = sim.DCSource(amplitude=amplitude,
start=300.0,
stop=300.0 + para['currdur'])
pulse.inject_into(popEList[0])
# Record data
popE.record('spikes')
popI.record('spikes')
# Run
sim.run(para['duration'])
# Store spikes
spikesE_in = popE.get_data()
spikesI_in = popI.get_data()
# Generate array with spike data for popE
spkpopE = spikeData(popE)
spkpopI = spikeData(popI)
return spkpopE, spkpopI
print "%d Initialising the simulator with %d threads..." % (
rank, extra['threads'])
else:
print "%d Initialising the simulator with single thread..." % (rank)
# Small function to display information only on node 1
def nprint(s):
if (rank == 0):
print s
timer.start() # start timer on construction
print "%d Setting up random number generator" % rank
rng = NumpyRNG(kernelseed, parallel_safe=True)
print "%d Creating excitatory population with %d neurons." % (rank, NE)
celltype = IF_curr_alpha(**cell_params)
E_net = Population(NE, celltype, label="E_net")
print "%d Creating inhibitory population with %d neurons." % (rank, NI)
I_net = Population(NI, celltype, label="I_net")
print "%d Initialising membrane potential to random values between %g mV and %g mV." % (
rank, U0, theta)
uniformDistr = RandomDistribution('uniform', [U0, theta], rng)
E_net.initialize(v=uniformDistr)
I_net.initialize(v=uniformDistr)
print "%d Creating excitatory Poisson generator with rate %g spikes/s." % (
def do_run(seed=None):
random.seed(seed)
# SpiNNaker setup
sim.setup(timestep=1.0, min_delay=1.0, max_delay=10.0)
# +-------------------------------------------------------------------+
# | General Parameters |
# +-------------------------------------------------------------------+
# Population parameters
model = sim.IF_curr_exp
cell_params = {
'cm': 0.25,
'i_offset': 0.0,
'tau_m': 10.0,
'tau_refrac': 2.0,
'tau_syn_E': 2.5,
'tau_syn_I': 2.5,
'v_reset': -70.0,
'v_rest': -65.0,
'v_thresh': -55.4
}
# Other simulation parameters
e_rate = 200
in_rate = 350
n_stim_test = 5
n_stim_pairing = 10
dur_stim = 20
pop_size = 40
ISI = 150.
start_test_pre_pairing = 200.
start_pairing = 1500.
start_test_post_pairing = 700.
simtime = start_pairing + start_test_post_pairing + \
ISI*(n_stim_pairing + n_stim_test) + 550. # let's make it 5000
# Initialisations of the different types of populations
IAddPre = []
IAddPost = []
# +-------------------------------------------------------------------+
# | Creation of neuron populations |
# +-------------------------------------------------------------------+
# Neuron populations
pre_pop = sim.Population(pop_size, model(**cell_params))
post_pop = sim.Population(pop_size, model(**cell_params))
# Test of the effect of activity of the pre_pop population on the post_pop
# population prior to the "pairing" protocol : only pre_pop is stimulated
for i in range(n_stim_test):
IAddPre.append(
sim.Population(
pop_size,
sim.SpikeSourcePoisson(rate=in_rate,
start=start_test_pre_pairing + ISI *
(i),
duration=dur_stim,
seed=random.randint(0, 100000000))))
# Pairing protocol : pre_pop and post_pop are stimulated with a 10 ms
# difference
for i in range(n_stim_pairing):
IAddPre.append(
sim.Population(
pop_size,
sim.SpikeSourcePoisson(rate=in_rate,
start=start_pairing + ISI * (i),
duration=dur_stim,
seed=random.randint(0, 100000000))))
IAddPost.append(
sim.Population(
pop_size,
sim.SpikeSourcePoisson(rate=in_rate,
start=start_pairing + ISI * (i) + 10.,
duration=dur_stim,
seed=random.randint(0, 100000000))))
# Test post pairing : only pre_pop is stimulated
# (and should trigger activity in Post)
for i in range(n_stim_test):
start = start_pairing + ISI * n_stim_pairing + \
start_test_post_pairing + ISI * i
IAddPre.append(
sim.Population(
pop_size,
sim.SpikeSourcePoisson(rate=in_rate,
start=start,
duration=dur_stim,
seed=random.randint(0, 100000000))))
# Noise inputs
INoisePre = sim.Population(pop_size,
sim.SpikeSourcePoisson(rate=e_rate,
start=0,
duration=simtime,
seed=random.randint(
0, 100000000)),
label="expoisson")
INoisePost = sim.Population(pop_size,
sim.SpikeSourcePoisson(rate=e_rate,
start=0,
duration=simtime,
seed=random.randint(
0, 100000000)),
label="expoisson")
# +-------------------------------------------------------------------+
# | Creation of connections |
# +-------------------------------------------------------------------+
# Connection parameters
JEE = 3.
# Connection type between noise poisson generator and
# excitatory populations
ee_connector = sim.OneToOneConnector()
# Noise projections
sim.Projection(INoisePre,
pre_pop,
ee_connector,
receptor_type='excitatory',
synapse_type=sim.StaticSynapse(weight=JEE * 0.05))
sim.Projection(INoisePost,
post_pop,
ee_connector,
receptor_type='excitatory',
synapse_type=sim.StaticSynapse(weight=JEE * 0.05))
# Additional Inputs projections
for i in range(len(IAddPre)):
sim.Projection(IAddPre[i],
pre_pop,
ee_connector,
receptor_type='excitatory',
synapse_type=sim.StaticSynapse(weight=JEE * 0.05))
for i in range(len(IAddPost)):
sim.Projection(IAddPost[i],
post_pop,
ee_connector,
receptor_type='excitatory',
synapse_type=sim.StaticSynapse(weight=JEE * 0.05))
# Plastic Connections between pre_pop and post_pop
stdp_model = sim.STDPMechanism(
timing_dependence=sim.SpikePairRule(tau_plus=20.,
tau_minus=50.0,
A_plus=0.02,
A_minus=0.02),
weight_dependence=sim.AdditiveWeightDependence(w_min=0, w_max=0.9))
rng = NumpyRNG(seed=seed, parallel_safe=True)
plastic_projection = \
sim.Projection(pre_pop, post_pop, sim.FixedProbabilityConnector(
p_connect=0.5, rng=rng), synapse_type=stdp_model)
# +-------------------------------------------------------------------+
# | Simulation and results |
# +-------------------------------------------------------------------+
# Record spikes and neurons' potentials
pre_pop.record(['v', 'spikes'])
post_pop.record(['v', 'spikes'])
# Run simulation
sim.run(simtime)
weights = plastic_projection.get('weight', 'list')
pre_spikes = neo_convertor.convert_spikes(pre_pop.get_data('spikes'))
post_spikes = neo_convertor.convert_spikes(post_pop.get_data('spikes'))
# End simulation on SpiNNaker
sim.end()
return (pre_spikes, post_spikes, weights)