def setUp(self):
sim.setup()
self.p1 = sim.Population(7, sim.IF_cond_exp())
self.p2 = sim.Population(4, sim.IF_cond_exp())
self.p3 = sim.Population(5, sim.IF_curr_alpha())
self.syn_rnd = sim.StaticSynapse(weight=0.123, delay=0.5)
self.syn_a2a = sim.StaticSynapse(weight=0.456, delay=0.4)
self.random_connect = sim.FixedNumberPostConnector(n=2)
self.all2all = sim.AllToAllConnector()
def inhibitory_connect(layers, source, dest1, dest2, dest3, weight):
sim.Projection(layers[source].population, layers[dest1].population,
sim.OneToOneConnector(),
sim.StaticSynapse(weight=weight))
sim.Projection(layers[source].population, layers[dest2].population,
sim.OneToOneConnector(),
sim.StaticSynapse(weight=weight))
sim.Projection(layers[source].population, layers[dest3].population,
sim.OneToOneConnector(),
sim.StaticSynapse(weight=weight))
def create_C2_layers(S2_layers: Dict[float, Sequence[Layer]],
s2_prototype_cells: int) -> List[sim.Population]:
"""
Creates the populations of the C2 layer, one for each S2 prototype cell,
containing only a single cell which max-pools the spikes of all layers of a
prototype.
Parameters:
`S2_layers`: A dictionary containing for each scale a list of S2
layers, one for each prototype cell
`s2_prototype_cells`: The number of S2 prototype cells
Returns:
A list of populations of size one, one population for each prototype
cell
"""
no_inh_w = 17.15 # synapse weight without S2 inhibitions
with_inh_w = 4 * no_inh_w # synapse weight with S2 inhibitions
C2_populations = [sim.Population(1, sim.IF_curr_exp(),
label=str(prot))\
for prot in range(s2_prototype_cells)]
total_connections = sum(
map(lambda ll: ll[0].shape[0] * ll[0].shape[1], S2_layers.values()))
for s2ll in S2_layers.values():
for prot in range(s2_prototype_cells):
sim.Projection(
s2ll[prot].population, C2_populations[prot],
sim.AllToAllConnector(),
sim.StaticSynapse(weight=with_inh_w / total_connections))
return C2_populations
def getSynapseType(lat_conn_strength_params,
use_stdp,
tau_plus=20.0,
tau_minus=20.0,
A_plus=0.01,
A_minus=0.012,
w_min=0,
w_max=1.0,
static_weight=0.02):
"""
For getting the required synapse type, fixed or STDP
Arguments: lat_conn_strength_params, parameters for the uniform distribution. The connections evolve during training anyway.
use_stdp, flag
the other Arguments are defaults.
Return: the synapse_type to use
"""
stdp_weight_distn = pynn.random.RandomDistribution(
'uniform', lat_conn_strength_params)
stdp = pynn.STDPMechanism(
weight=stdp_weight_distn,
timing_dependence=pynn.SpikePairRule(tau_plus=tau_plus,
tau_minus=tau_minus,
A_plus=A_plus,
A_minus=A_minus),
weight_dependence=pynn.AdditiveWeightDependence(w_min=w_min,
w_max=w_max))
synapse_to_use = stdp if use_stdp else pynn.StaticSynapse(
weight=static_weight)
return synapse_to_use
def connect_layers(input_layer,
output_layer,
weights,
i_s,
j_s,
i_e,
j_e,
k_out,
stdp=False,
initial_weight=0,
label_dicts=None):
"""
Connects a neuron of an output layer to the corresponding square of an input
layer. This is a helper function of connect_layer_to_layer()
Returns:
The created projection between the input and output layers
"""
m = input_layer.shape[1]
view_elements = []
i = i_s
while i < i_e:
j = j_s
while j < j_e:
view_elements.append(m * i + j)
j += 1
i += 1
if stdp:
w_max = initial_weight * 15
stdp_shared = sim.native_synapse_type('stdp_synapse')\
(Wmax=w_max * 1000, mu_plus=0.0, mu_minus=1.0)
proj = sim.Projection(input_layer.population[view_elements],
output_layer.population[[k_out]],
sim.AllToAllConnector(), stdp_shared)
ol = int(output_layer.population.label)
il = input_layer.population.label
out_neuron = output_layer.population[k_out]
if label_dicts == None:
for i in range(len(view_elements)):
label = '{}_{}_{}'.format(ol, il, i)
in_neuron = input_layer.population[view_elements[i]]
conn = nest.GetConnections(source=[in_neuron],
target=[out_neuron])
nest.SetStatus(conn, {'label': label, 'weight': weights[i][0]})
else:
for i in range(len(view_elements)):
label = '{}_{}_{}'.format(ol, il, i)
if not label in label_dicts[ol]:
label_dicts[ol][label] = ([], [])
in_neuron = input_layer.population[view_elements[i]]
label_dicts[ol][label][0].append(in_neuron)
label_dicts[ol][label][1].append(out_neuron)
else:
proj = sim.Projection(input_layer.population[view_elements],
output_layer.population[[k_out]],
sim.AllToAllConnector(),
sim.StaticSynapse(weight=weights))
return proj
def test_ticket244():
nest = pyNN.nest
nest.setup(threads=4)
p1 = nest.Population(4, nest.IF_curr_exp())
p1.record('spikes')
poisson_generator = nest.Population(3, nest.SpikeSourcePoisson(rate=1000.0))
conn = nest.OneToOneConnector()
syn = nest.StaticSynapse(weight=1.0)
nest.Projection(poisson_generator, p1.sample(3), conn, syn, receptor_type="excitatory")
nest.run(15)
p1.get_data()
def test_ticket240():
nest = pyNN.nest
nest.setup(threads=4)
parameters = {'Tau_m': 17.0}
p1 = nest.Population(4, nest.IF_curr_exp())
p2 = nest.Population(5, nest.native_cell_type("ht_neuron")(**parameters))
conn = nest.AllToAllConnector()
syn = nest.StaticSynapse(weight=1.0)
prj = nest.Projection(p1, p2, conn, syn, receptor_type='AMPA') # This should be a nonstandard receptor type but I don't know of one to use.
connections = prj.get(('weight',), format='list')
assert len(connections) > 0
def test_single_presynaptic_and_single_postsynaptic_neuron(self):
prj = sim.Projection(self.p4, self.p4, sim.AllToAllConnector(),
synapse_type=sim.StaticSynapse(weight=0.123))
assert prj.shape == (1, 1)
weight = 0.456
prj.set(weight=weight)
self.assertEqual(prj.get("weight", format="array")[0][0], weight)
weight_array = numpy.ones(prj.shape) * weight
prj.set(weight=weight_array)
self.assertTrue((weight_array == prj.get("weight", format="array")).all())
def create_corner_layer_for(input_layers):
shape = input_layers[0].shape
total_output_neurons = np.prod(shape)
output_population = sim.Population(total_output_neurons,
sim.IF_curr_exp(),
label='corner')
for layer in input_layers:
sim.Projection(layer.population, output_population,
sim.OneToOneConnector(),
sim.StaticSynapse(weight=1., delay=0.5))
return Layer(output_population, shape)
def test_record_native_model():
if not have_nest:
raise SkipTest
nest = pyNN.nest
from pyNN.random import RandomDistribution
init_logging(logfile=None, debug=True)
nest.setup()
parameters = {'tau_m': 17.0}
n_cells = 10
p1 = nest.Population(n_cells,
nest.native_cell_type("ht_neuron")(**parameters))
p1.initialize(V_m=-70.0, Theta=-50.0)
p1.set(theta_eq=-51.5)
#assert_arrays_equal(p1.get('theta_eq'), -51.5*numpy.ones((10,)))
assert_equal(p1.get('theta_eq'), -51.5)
print(p1.get('tau_m'))
p1.set(tau_m=RandomDistribution('uniform', low=15.0, high=20.0))
print(p1.get('tau_m'))
current_source = nest.StepCurrentSource(
times=[50.0, 110.0, 150.0, 210.0],
amplitudes=[0.01, 0.02, -0.02, 0.01])
p1.inject(current_source)
p2 = nest.Population(
1,
nest.native_cell_type("poisson_generator")(rate=200.0))
print("Setting up recording")
p2.record('spikes')
p1.record('V_m')
connector = nest.AllToAllConnector()
syn = nest.StaticSynapse(weight=0.001)
prj_ampa = nest.Projection(p2, p1, connector, syn, receptor_type='AMPA')
tstop = 250.0
nest.run(tstop)
vm = p1.get_data().segments[0].analogsignals[0]
n_points = int(tstop / nest.get_time_step()) + 1
assert_equal(vm.shape, (n_points, n_cells))
assert vm.max() > 0.0 # should have some spikes
def create_local_inhibition(layers_dict):
"""
Creates local inhibitory connections from a neuron to its neighbors in an
area of a fixed distance. The latency of its neighboring neurons decreases
linearly with the distance from the spike from 15% to 5%, as described in
Masquelier's paper. Here we assumed that a weight of -10 inhibits the
neuron completely and took that as a starting point.
"""
for size, layers in layers_dict.items():
print('Create local inhibition for size', size)
for layer in layers:
sim.Projection(layer.population,
layer.population,
sim.DistanceDependentProbabilityConnector(
'd < 5', allow_self_connections=False),
sim.StaticSynapse(weight='.25 * d - 1.75'),
space=space.Space(axes='xy'))
def test():
if not HAVE_H5PY and HAVE_NEST:
raise SkipTest
sim.setup()
p1 = sim.Population(10,
sim.IF_cond_exp(v_rest=-65,
tau_m=lambda i: 10 + 0.1 * i,
cm=RD('normal', (0.5, 0.05))),
label="population_one")
p2 = sim.Population(20,
sim.IF_curr_alpha(v_rest=-64,
tau_m=lambda i: 11 + 0.1 * i),
label="population_two")
prj = sim.Projection(p1,
p2,
sim.FixedProbabilityConnector(p_connect=0.5),
synapse_type=sim.StaticSynapse(weight=RD(
'uniform', [0.0, 0.1]),
delay=0.5),
receptor_type='excitatory')
net = Network(p1, p2, prj)
export_to_sonata(net, "tmp_serialization_test", overwrite=True)
net2 = import_from_sonata("tmp_serialization_test/circuit_config.json",
sim)
for orig_population in net.populations:
imp_population = net2.get_component(orig_population.label)
assert orig_population.size == imp_population.size
for name in orig_population.celltype.default_parameters:
assert_array_almost_equal(orig_population.get(name),
imp_population.get(name), 12)
w1 = prj.get('weight', format='array')
prj2 = net2.get_component(asciify(prj.label).decode('utf-8') + "-0")
w2 = prj2.get('weight', format='array')
assert_array_almost_equal(w1, w2, 12)
def set_synapses(self, conn_type, sim_params, E_neurons, I_neurons, N_inh,
conn_types, verbose):
syn = {}
proj = {}
verbose = True
if verbose: print('Building %s synapses..' % conn_types[conn_type])
weight = sim_params['w_{}'.format(conn_types[conn_type])]
delay = sim_params['s_{}'.format(conn_types[conn_type])]
syn[conn_types[conn_type]] = sim.StaticSynapse(delay=delay)
if conn_types[
conn_type][:3] == 'exc': #string slicing, this co is TO exc
pre_neurons = E_neurons
receptor_type = 'excitatory'
else:
pre_neurons = I_neurons #TO inh
receptor_type = 'inhibitory'
if conn_types[conn_type][-3:] == 'exc': #FROM exc
post_neurons = E_neurons
else:
post_neurons = I_neurons #FROM inh
sparseness = sim_params['c_{}'.format(conn_types[conn_type])]
proj[conn_types[conn_type]] = sim.Projection(
pre_neurons,
post_neurons,
connector=sim.FixedProbabilityConnector(sparseness,
rng=sim.NumpyRNG(seed=42)),
synapse_type=syn[conn_types[conn_type]],
receptor_type=receptor_type)
bw = sim_params['b_{}'.format(conn_types[conn_type])]
angle_pre = 1. * np.arange(proj[conn_types[conn_type]].pre.size)
angle_post = 1. * np.arange(proj[conn_types[conn_type]].post.size)
w_ij = self.tuning_function(angle_pre[:, np.newaxis],
angle_post[np.newaxis, :], bw,
N_inh) * weight
proj[conn_types[conn_type]].set(weight=w_ij)
return proj
label="inhib_pop")
spike_source_1 = sim.Population(
1, sim.SpikeSourceArray(spike_times=spike_time_for_id_1))
spike_source_2 = sim.Population(
1, sim.SpikeSourceArray(spike_times=spike_time_for_id_2))
spike_source_3 = sim.Population(
1, sim.SpikeSourceArray(spike_times=spike_time_for_id_3))
spike_source_4 = sim.Population(
1, sim.SpikeSourceArray(spike_times=spike_time_for_id_4))
spike_source_9 = sim.Population(
1, sim.SpikeSourceArray(spike_times=spike_time_for_id_9))
spike_2_conn_for_1 = sim.Projection(spike_source_1, cann_pop[1:2],
sim.OneToOneConnector(),
sim.StaticSynapse(weight=0.002, delay=0.1))
spike_2_conn_for_2 = sim.Projection(spike_source_2, cann_pop[2:3],
sim.OneToOneConnector(),
sim.StaticSynapse(weight=0.002, delay=0.1))
spike_2_conn_for_3 = sim.Projection(spike_source_3, cann_pop[3:4],
sim.OneToOneConnector(),
sim.StaticSynapse(weight=0.002, delay=0.1))
spike_2_conn_for_4 = sim.Projection(spike_source_4, cann_pop[4:5],
sim.OneToOneConnector(),
sim.StaticSynapse(weight=0.002, delay=0.1))
spike_2_conn_for_9 = sim.Projection(spike_source_9, cann_pop[9:10],
sim.OneToOneConnector(),
sim.StaticSynapse(weight=0.002, delay=0.1))
cann_2_cann = sim.Projection(
cann_pop, cann_pop,
###### Neurons #######
input_neurons = p.Population(param.input_nr, p.SpikeSourcePoisson())
reservoir_exc = p.Population(param.res_exc_nr,
p.IF_curr_exp, {},
label="reservoir_exc")
reservoir_inh = p.Population(param.res_inh_nr,
p.IF_curr_exp, {},
label="reservoir_inh")
######################
###### Synapses #######
stat_syn_input = p.StaticSynapse(weight=dist_input, delay=1)
stat_syn_exc = p.StaticSynapse(weight=dist_exc, delay=1)
stat_syn_inh = p.StaticSynapse(weight=dist_inh, delay=1)
#stat_syn_input = p.StaticSynapse(weight =50.0, delay=1)
######################
###### Connections #######
exc_conn = p.FixedProbabilityConnector(param.res_pconn, rng=param.rng)
inh_conn = p.FixedProbabilityConnector(param.res_pconn, rng=param.rng)
inp_conn = p.FixedProbabilityConnector(0.5, rng=param.rng)
rout_conn = p.AllToAllConnector()
connections = {}
neuron1 = sim.create(sim.HH_cond_exp(**cellparams))
neuron1.record(["v"])
neuron1.initialize(v=cellparams["e_rev_leak"])
neuron2 = sim.create(sim.HH_cond_exp(**cellparams))
neuron2.record(["v"])
neuron2.initialize(v=cellparams["e_rev_leak"])
spike_times = [2.0]
spike_source = sim.Population(1, sim.SpikeSourceArray(spike_times=spike_times))
sim.Projection(spike_source,
neuron1,
sim.OneToOneConnector(),
sim.StaticSynapse(weight=0.076, delay=0.1),
receptor_type='excitatory'),
sim.Projection(spike_source,
neuron2,
sim.OneToOneConnector(),
sim.StaticSynapse(weight=0.0755, delay=0.1),
receptor_type='excitatory'),
sim.run(tEnd)
data1 = neuron1.get_data()
data2 = neuron2.get_data()
signal_names = [s.name for s in data1.segments[0].analogsignalarrays]
fig = plt.figure(figsize=(cm2inch(12.4), cm2inch(7)))
###### Neurons #######
input_neurons = p.Population(param.input_nr, p.SpikeSourcePoisson())
readout_neurons = p.Population(param.readout_nr,
p.IF_curr_exp, {},
label="readout")
reservoir = p.Population(param.reservoir_nr,
p.IF_curr_exp, {},
label="reservoir")
######################
###### Synapses #######
stat_syn_res = p.StaticSynapse(weight=5.0, delay=1)
stat_syn_input = p.StaticSynapse(weight=50.0, delay=1)
stat_syn_rout = p.StaticSynapse(weight=0.0, delay=1)
######################
###### Connections #######
res_conn = p.FixedProbabilityConnector(param.res_pconn, rng=param.rng)
inp_conn = p.AllToAllConnector()
rout_conn = p.AllToAllConnector()
connections = {}
connections['r2r'] = p.Projection(reservoir,
reservoir,
label='Excitatory layer')
cortical_neurons_inh = simulator.Population(Ncell_inh**2,
inhibitory_cell,
structure=inhibitory_structure,
label='Inhibitory layer')
#############################
# Add background noise
#############################
correlated = False
noise_rate = 5800 # Hz
g_noise = 0.00089 * 0 # Microsiemens (0.89 Nanosiemens)
noise_delay = 1 # ms
noise_model = simulator.SpikeSourcePoisson(rate=noise_rate)
noise_syn = simulator.StaticSynapse(weight=g_noise, delay=noise_delay)
if correlated:
# If correlated is True all the cortical neurons receive noise from the same cell.
noise_population = simulator.Population(1,
noise_model,
label='Background Noise')
background_noise_to_exc = simulator.Projection(
noise_population, cortical_neurons_exc, simulator.AllToAllConnector(),
noise_syn)
background_noise_to_inh = simulator.Projection(
noise_population, cortical_neurons_inh, simulator.AllToAllConnector(),
noise_syn)
else:
# If correlated is False, all cortical neurons receive independent noise
noise_population_exc = simulator.Population(
p.setup(timestep=param.dt)
dist_input = RandomDistribution('uniform', (1, 10), rng=param.rng)
dist_reservoir = RandomDistribution('uniform', (1, 10), rng=param.rng)
#print(distribution.next(5))
###### Neurons #######
input_neurons = p.Population(param.input_nr, p.SpikeSourcePoisson())
readout_neurons = p.Population(param.readout_nr, p.IF_curr_exp, {}, label="readout")
reservoir = p.Population(param.reservoir_nr,p.IF_curr_exp, {}, label="reservoir")
######################
###### Synapses #######
stat_syn_res = p.StaticSynapse(weight =dist_reservoir, delay=1)
stat_syn_input = p.StaticSynapse(weight =dist_input, delay=1)
stat_syn_rout = p.StaticSynapse(weight =0.0, delay=1)
######################
###### Connections #######
res_conn = p.FixedProbabilityConnector(param.res_pconn, rng=param.rng)
inp_conn = p.AllToAllConnector()
rout_conn = p.AllToAllConnector()
connections = {}
connections['r2r'] = p.Projection(reservoir, reservoir, res_conn,
synapse_type=stat_syn_res, receptor_type='excitatory')
if dist[i][j] > n / 2:
dist[i][j] = n - dist[i][j]
weights[i][j] = round(weight_to_spike * \
(1 / (sig * np.sqrt(2 * np.pi)) * \
(np.exp(-np.power(dist[i][j] - mu, 2.) \
/ (2 * np.power(sig, 2.))))), 2)
cann_connector.append((i, j, weights[i][j])) #, delay_cann2cann))
print("Weight matrix:\n", weights)
spike_times = [1000., 2000.]
spike_source = sim.Population(1, sim.SpikeSourceArray(spike_times=spike_times))
conn = sim.Projection(spike_source, cann_pop[spiky:spiky + 1],
sim.AllToAllConnector(),
sim.StaticSynapse(weight=0.002, delay=0.1))
n2n_conn = sim.Projection(
cann_pop, cann_pop,
sim.FromListConnector(cann_connector, column_names=["weight"]),
sim.StaticSynapse(weight=0.0001, delay=100))
spike_source.record('spikes')
cann_pop.record(('v', 'spikes'))
sim.run(5000.0)
from pyNN.utility.plotting import Figure, Panel
data1 = cann_pop.get_data().segments[0]
vm = data1.filter(name="v")[0]
readout_neurons = p.Population(param.readout_nr,
p.IF_curr_exp, {},
label="readout")
reservoir_exc = p.Population(param.res_exc_nr,
p.IF_curr_exp, {},
label="reservoir_exc")
reservoir_inh = p.Population(param.res_inh_nr,
p.IF_curr_exp, {},
label="reservoir_inh")
######################
###### Synapses #######
stat_syn_exc = p.StaticSynapse(weight=5.0, delay=1)
stat_syn_inh = p.StaticSynapse(weight=20.0, delay=1)
stat_syn_input = p.StaticSynapse(weight=50.0, delay=1)
######################
###### Connections #######
exc_conn = p.FixedProbabilityConnector(param.res_pconn, rng=param.rng)
inh_conn = p.FixedProbabilityConnector(param.res_pconn, rng=param.rng)
inp_conn = p.AllToAllConnector()
rout_conn = p.AllToAllConnector()
connections = {}
connections['e2e'] = p.Projection(reservoir_exc,
reservoir_exc,
sim.Izhikevich(**parameters_e),
initial_values=initialize_e,
label="Ex")
population_i = sim.Population(n_neurons_i,
sim.Izhikevich(**parameters_i),
initial_values=initialize_i,
label="In")
# Assembly entire population
population = sim.Assembly(population_e, population_i)
# Create synapses
syn_weight_e = max_syn_weight_e * np.random.rand(n_neurons_e, n_neurons)
synapses_e = sim.Projection(population_e, population,
sim.AllToAllConnector(allow_self_connections=False),
sim.StaticSynapse(weight=syn_weight_e,
delay=1.0),
receptor_type='excitatory')
syn_weight_i = -max_syn_weight_i * np.random.rand(n_neurons_i, n_neurons)
synapses_i = sim.Projection(population_i, population,
sim.AllToAllConnector(allow_self_connections=False),
sim.StaticSynapse(weight=syn_weight_i,
delay=1.0),
receptor_type='inhibitory')
# Create noise Gaussian generator
gauss_gen_e = sim.NoisyCurrentSource(**gauss_gen_e_params)
population_e.inject(gauss_gen_e)
gauss_gen_i = sim.NoisyCurrentSource(**gauss_gen_i_params)
population_i.inject(gauss_gen_i)
# Set spike times to be recorded
#V value initialization
sim.initialize(Neurons_RS, v=-65.0, gsyn_exc=0,
gsyn_inh=0) # v:mV, gsyn_exc:nS, gsyn_inh:nS
sim.initialize(Neurons_FS, v=-65.0, gsyn_exc=0, gsyn_inh=0)
## RECURRENT CONNECTIONS
#The two populations of neurons are randomly connected
# (internally and mutually) with a connectivity probability of 5%
# exc_exc
exc_exc = sim.Projection(Neurons_RS,
Neurons_RS,
sim.FixedProbabilityConnector(
0.05, allow_self_connections=False),
receptor_type='excitatory',
synapse_type=sim.StaticSynapse(weight=0.001))
# exc_inh
exc_inh = sim.Projection(Neurons_RS,
Neurons_FS,
sim.FixedProbabilityConnector(
0.05, allow_self_connections=False),
receptor_type='excitatory',
synapse_type=sim.StaticSynapse(weight=0.001))
# inh_exc
inh_exc = sim.Projection(Neurons_FS,
Neurons_RS,
sim.FixedProbabilityConnector(
0.05, allow_self_connections=False),
receptor_type='inhibitory',
synapse_type=sim.StaticSynapse(weight=0.005))
# inh_inh
'v_thresh': -60.0, # (mV)
'cm': 0.5
}
sim.setup()
neuron1 = sim.Population(n, sim.IF_cond_alpha(**cell_params), label="neuron1")
neuron2 = sim.Population(n, sim.IF_cond_alpha(**cell_params), label="neuron2")
spike_times = [1000., 2000.]
spike_source = sim.Population(1, sim.SpikeSourceArray(spike_times=spike_times))
conn = sim.Projection(spike_source, neuron1[spiky:spiky + 1],
sim.AllToAllConnector(),
sim.StaticSynapse(weight=0.002, delay=1.))
n2n_conn = sim.Projection(neuron1, neuron2, sim.OneToOneConnector(),
sim.StaticSynapse(weight=0.002, delay=1.))
spike_source.record('spikes')
neuron1.record(('v', 'spikes'))
neuron2.record(('v', 'spikes'))
sim.run(5000.0)
#print neuron1.get_spike_counts()
from pyNN.utility.plotting import Figure, Panel
data1 = neuron1.get_data().segments[0]
data2 = neuron1.get_data().segments[0]
vm = data1.filter(name="v")[0]
def test_callback(data_input):
global message
message = data_input.actual.positions
msg_list = list(message)
#msg_list[0] = int(message[0].encode('hex'),16)
#for i in
#msg_list = int(message.encode('hex'),16)
#print('============= Received image data.',message)
rospy.loginfo('=====received data %r', msg_list[0])
timer = Timer()
dt = 0.1
p.setup(timestep=dt) # 0.1ms
pop_1 = p.Population(1, p.IF_curr_exp, {}, label="pop_1")
#input = p.Population(1, p.SpikeSourceArray, {'spike_times': [[0,3,6]]}, label='input')
input = p.Population(1, p.SpikeSourcePoisson,
{'rate': (msg_list[0] + 1.6) * 100})
stat_syn = p.StaticSynapse(weight=50.0, delay=1)
input_proj = p.Projection(input,
pop_1,
p.OneToOneConnector(),
synapse_type=stat_syn,
receptor_type='excitatory')
pop_1.record(['v', 'spikes'])
p.run(10)
pop_1_data = pop_1.get_data()
spikes = pop_1_data.segments[0].spiketrains[0]
mean_rate = int(gaussian_convolution(spikes, dt))
rospy.loginfo('=====mean_rate %r', mean_rate) # mean_rate = 64
rate_command = mean_rate
# rate coding of the spike train
'''
pub = rospy.Publisher('/cmd_vel_mux/input/teleop', Twist, queue_size=10)
# construct the output command
command = Twist()
command.linear.x = rate_command*0.02
command.angular.z = rate_command/50000.
pub.publish(command)
'''
pub = rospy.Publisher('/arm_controller/follow_joint_trajectory/goal',
FollowJointTrajectoryActionGoal,
queue_size=10)
command = FollowJointTrajectoryActionGoal()
command.header.stamp = rospy.Time.now()
command.goal.trajectory.joint_names = ['elbow']
point = JointTrajectoryPoint()
point.positions = [rate_command / 10]
point.time_from_start = rospy.Duration(1)
command.goal.trajectory.points.append(point)
pub.publish(command)
rospy.loginfo('=====send command %r', command.goal.trajectory.points[0])
fig_settings = {
'lines.linewidth': 0.5,
'axes.linewidth': 0.5,
'axes.labelsize': 'small',
'legend.fontsize': 'small',
'font.size': 8
}
plt.rcParams.update(fig_settings)
fig1 = plt.figure(1, figsize=(6, 8))
def plot_spiketrains(segment):
for spiketrain in segment.spiketrains:
y = np.ones_like(spiketrain) * spiketrain.annotations['source_id']
plt.plot(spiketrain, y, '.')
plt.ylabel(segment.name)
plt.setp(plt.gca().get_xticklabels(), visible=False)
def plot_signal(signal, index, colour='b'):
label = "Neuron %d" % signal.annotations['source_ids'][index]
plt.plot(signal.times, signal[:, index], colour, label=label)
plt.ylabel("%s (%s)" %
(signal.name, signal.units._dimensionality.string))
plt.setp(plt.gca().get_xticklabels(), visible=False)
plt.legend()
print("now plotting the network---------------")
rospy.loginfo('--------now plotting---------------')
n_panels = sum(a.shape[1]
for a in pop_1_data.segments[0].analogsignalarrays) + 2
plt.subplot(n_panels, 1, 1)
plot_spiketrains(pop_1_data.segments[0])
panel = 3
for array in pop_1_data.segments[0].analogsignalarrays:
for i in range(array.shape[1]):
plt.subplot(n_panels, 1, panel)
plot_signal(array, i, colour='bg'[panel % 2])
panel += 1
plt.xlabel("time (%s)" % array.times.units._dimensionality.string)
plt.setp(plt.gca().get_xticklabels(), visible=True) #
I_vec, u_vec, s_vec = LIF_CUBA(spikes=spikes_e)
# %% PyNN CUBA
# Setup
sim.setup(timestep=0.1, min_delay=0.1, max_delay=10.0)
IF_sim = sim.Population(1, sim.IF_curr_exp(), label="IF_curr_exp")
IF_sim.record('v')
spike_times = np.arange(50,100,10)
spike_input = sim.Population(1,sim.SpikeSourceArray(spike_times=spike_times),label="Input spikes")
# Connections
w = 1
connections = sim.Projection(spike_input, IF_sim,
connector=sim.AllToAllConnector(),
synapse_type=sim.StaticSynapse(weight=w,delay=0.1),
receptor_type="excitatory")
# Running simulation in MS
IF_sim.record('v')
sim.run(100.0)
# Data
v_data = IF_sim.get_data()
data = IF_sim.get_data().segments[0]
v_cuba = data.filter(name="v")[0]
# Plotting
plt.plot(v_cuba)
# End
def __init__(self, sim_params=None, cell_params=None, verbose=True):
'''
Parameters : Stimulus, Population, Synapses, Recording, Running
'''
self.verbose = verbose
self.sim_params = sim_params
self.cell_params = cell_params
sim.setup() #spike_precision='on_grid')#timestep = .1)
N_inh = int(sim_params['nb_neurons'] *
sim_params['p']) #total pop * proportion of inhib
self.spike_source = sim.Population(
N_inh,
sim.SpikeSourcePoisson(rate=sim_params['input_rate'],
duration=sim_params['simtime'] / 2))
#orientation stimulus, see bottom section of notebook
angle = 1. * np.arange(N_inh)
rates = self.tuning_function(angle,
sim_params['angle_input'] / 180. * N_inh,
sim_params['b_input'], N_inh)
rates /= rates.mean()
rates *= sim_params['input_rate']
for i, cell in enumerate(self.spike_source):
cell.set_parameters(rate=rates[i])
#neuron model selection
if sim_params['neuron_model'] == 'IF_cond_alpha':
model = sim.IF_cond_alpha #LIF with nice dynamics
else:
model = sim.IF_cond_exp #LIF with exp dynamics
#populations
E_neurons = sim.Population(
N_inh,
model(**cell_params),
initial_values={
'v':
rnd('uniform',
(sim_params['v_init_min'], sim_params['v_init_max']))
},
label="Excitateurs")
I_neurons = sim.Population(
int(sim_params['nb_neurons'] - N_inh),
model(**cell_params),
initial_values={
'v':
rnd('uniform',
(sim_params['v_init_min'], sim_params['v_init_max']))
},
label="Inhibiteurs")
#input to excitatories
input_exc = sim.Projection(
self.spike_source, E_neurons, sim.OneToOneConnector(),
sim.StaticSynapse(weight=sim_params['w_input_exc'],
delay=sim_params['s_input_exc']))
#loop through connections type and use associated params, can be a bit slow
conn_types = ['exc_inh', 'inh_exc', 'exc_exc',
'inh_inh'] #connection types
'''
self.proj = self.set_synapses(conn_types = conn_types, sim_params =sim_params,
E_neurons = E_neurons, I_neurons = I_neurons,
N_inh = N_inh)
'''
#Multi threading support NE MARCHE PAS LAISSER LE NJOBS EN 1
self.proj = Parallel(n_jobs=1, backend='multiprocessing')(
delayed(self.set_synapses)(conn_type,
sim_params=sim_params,
E_neurons=E_neurons,
I_neurons=I_neurons,
N_inh=N_inh,
conn_types=conn_types,
verbose=verbose)
for conn_type in range(len(conn_types)))
if verbose: print('Done building synapses !')
#record
self.spike_source.record('spikes')
E_neurons.record('spikes')
I_neurons.record('spikes')
#run
if verbose: print('Running simulation..')
sim.run(sim_params['simtime'])
if verbose: print('Done running !')
#get the spikes
self.E_spikes = E_neurons #.get_data().segments[0]
self.I_spikes = I_neurons #.get_data().segments[0]
self.P_spikes = self.spike_source #.get_data().segments[0]
def create_S2_layers(C1_layers: Dict[float, Sequence[Layer]], feature_size,
s2_prototype_cells, refrac_s2=.1, stdp=True,
inhibition=True)\
-> Dict[float, List[Layer]]:
"""
Creates all prototype S2 layers for all sizes.
Parameters:
`layers_dict`: A dictionary containing for each size a list of C1
layers, for each feature one
`feature_size`:
`s2_prototype_cells`:
`refrac_s2`:
`stdp`:
Returns:
A dictionary containing for each size a list of different S2
layers, for each prototype one.
"""
f_s = feature_size
initial_weight = 25 / (f_s * f_s)
weight_rng = rnd.RandomDistribution('normal',
mu=initial_weight,
sigma=initial_weight / 20)
i_offset_rng = rnd.RandomDistribution('normal', mu=.5, sigma=.45)
weights = list(
map(lambda x: weight_rng.next() * 1000, range(4 * f_s * f_s)))
S2_layers = {}
i_offsets = list(
map(lambda x: i_offset_rng.next(), range(s2_prototype_cells)))
ndicts = list(map(lambda x: {}, range(s2_prototype_cells)))
ondicts = list(map(lambda x: {}, range(s2_prototype_cells)))
omdicts = list(map(lambda x: {}, range(s2_prototype_cells)))
for size, layers in C1_layers.items():
n, m = how_many_squares_in_shape(layers[0].shape, (f_s, f_s), f_s)
if stdp:
l_i_offsets = [list(map(lambda x: rnd.RandomDistribution('normal',
mu=i_offsets[i], sigma=.25).next(), range(n * m)))\
for i in range(s2_prototype_cells)]
else:
l_i_offsets = np.zeros((s2_prototype_cells, n * m))
print('S2 Shape', n, m)
layer_list = list(
map(
lambda i: Layer(
sim.Population(n * m,
sim.IF_curr_exp(i_offset=l_i_offsets[i],
tau_refrac=refrac_s2),
structure=space.Grid2D(aspect_ratio=m / n),
label=str(i)), (n, m)),
range(s2_prototype_cells)))
for S2_layer in layer_list:
for C1_layer in layers:
S2_layer.projections[C1_layer.population.label] =\
connect_layer_to_layer(C1_layer, S2_layer, (f_s, f_s), f_s,
[[w] for w in weights[:f_s * f_s]],
stdp=stdp,
initial_weight=initial_weight,
ndicts=ndicts, ondicts=ondicts,
omdicts=omdicts)
S2_layers[size] = layer_list
# Set the labels of the shared connections
if stdp:
t = time.clock()
print('Set shared labels')
for s2_label_dicts in [ndicts, ondicts, omdicts]:
for i in range(s2_prototype_cells):
w_iter = weights.__iter__()
for label, (source, target) in s2_label_dicts[i].items():
conns = nest.GetConnections(source=source, target=target)
nest.SetStatus(conns, {
'label': label,
'weight': w_iter.__next__()
})
print('Setting labels took', time.clock() - t)
if inhibition:
# Create inhibitory connections between the S2 cells
# First between the neurons of the same layer...
inh_weight = -10
inh_delay = .1
print('Create S2 self inhibitory connections')
for layer_list in S2_layers.values():
for layer in layer_list:
sim.Projection(
layer.population, layer.population,
sim.AllToAllConnector(allow_self_connections=False),
sim.StaticSynapse(weight=inh_weight, delay=inh_delay))
# ...and between the layers
print('Create S2 cross-scale inhibitory connections')
for i in range(s2_prototype_cells):
for layer_list1 in S2_layers.values():
for layer_list2 in S2_layers.values():
if layer_list1[i] != layer_list2[i]:
sim.Projection(
layer_list1[i].population,
layer_list2[i].population, sim.AllToAllConnector(),
sim.StaticSynapse(weight=inh_weight,
delay=inh_delay))
if stdp:
# Create the inhibition between different prototype layers
print('Create S2 cross-prototype inhibitory connections')
for layer_list in S2_layers.values():
for layer1 in layer_list:
for layer2 in layer_list:
if layer1 != layer2:
sim.Projection(
layer1.population, layer2.population,
sim.OneToOneConnector(),
sim.StaticSynapse(weight=inh_weight - 1,
delay=inh_delay))
return S2_layers
sim.setup(simparams['dt'])
inp = sim.Population(
1,
sim.SpikeSourcePoisson(duration=simparams['duration'],
rate=simparams['input_rate']))
inp.label = 'input cell'
outp = sim.Population(1, sim.IF_curr_exp, cellparams=cellparams)
outp.label = 'output cell'
inp.record('spikes')
outp.record(['v', 'spikes'])
synapse = sim.StaticSynapse(weight=1, delay=simparams['delay'])
connector = sim.OneToOneConnector()
connection = sim.Projection(inp, outp, connector, synapse)
def report_time(t):
print("Time: {}".format(t))
return t + simparams['dt']
par = 'i_offset'
for p in [0.01]:
outp.set(**{par: p})
cellparams[par] = p
outp.initialize(v=cellparams['v_rest'])
sim.run(simparams['duration'], callbacks=[report_time])
# ============= Validation ============= #
print('Constructing new network with the learned weights')
sim.setup(threads=args.threads, min_delay=.1)
# Create the validation network and connect the C2 neurons to it
validation_spiketrains = [[s for s in st] for st in validation_pair[1]]
C2_populations, compound_C2_population =\
create_C2_populations(validation_spiketrains)
classifier_neurons = [sim.Population(1, sim.IF_curr_exp())\
for cat in range(categories)]
for category in range(categories):
# for category in range(1):
sim.Projection(compound_C2_population, classifier_neurons[category],
sim.AllToAllConnector(),
sim.StaticSynapse(weight=classifier_weights[category]))
# Record the spikes for visualization purposes
compound_C2_population.record('spikes')
for pop in classifier_neurons:
pop.record(['spikes', 'v'])
# Let the simulation run to "fill" the layer pipeline with spikes
sim.run(40)
for pop in C2_populations:
pop.get_data(clear=True)
predicted_labels = []
# Simulate and classify the images
for i in range(validation_image_count):