def get_steady_state(net):
lab_manager.reset_lab(net)
f, initial_conditions, _ = lab_manager.set_up_lab(net)
total_time = 100.
# step 4: run the lab and gather data
time_sampled_range = [total_time]
data = lab_manager.run_lab(f, initial_conditions, time_sampled_range)
return data[-1, :]
def get_data(delta_time):
"""Return the integration solution."""
# step 2: design an experiment
T0 = 250.
TIME_DELAY = delta_time
ex.delay_pulses_on_layer_0_and_1(net,
t0s=[T0, T0 + TIME_DELAY],
i_max=55.,
w=1.0)
# step 3: ask our lab manager to set up the lab for the experiment
f, initial_conditions, neuron_inds = lm.set_up_lab(net)
# step 4: run the lab and gather data
data = lm.run_lab(f,
initial_conditions,
time_sampled_range,
integrator='dopri5')
return data
def stdp_expt(net, dt, ics): # please put a 1-1 network
def sigmoid(x):
return 1. / (1. + np.exp(-x))
lab_manager.reset_lab(net)
dts = [dt]
experiments.delay_pulses_on_layer_0_and_1(net, [50., 50 + dt])
# step 3: ask our lab manager to set up the lab for the experiment.
f, _, _ = lab_manager.set_up_lab(net)
ii = 0
for edge in net.edges:
n1, n2 = edge
syn = net[n1][n2]["synapse"]
ii = syn.ii
# step 4: run the lab and gather data
total_time = 100.
time_sampled_range = np.arange(0., total_time, 0.1)
data = lab_manager.run_lab(f, initial_conditions, time_sampled_range)
#print(ii)
dw = sigmoid(data[-1, ii]) - sigmoid(data[0, ii])
return dw, data
def get_stdp_profile(Neuron, Synapse, initial_conditions):
Neuron = get_InsulatedNeuron(Neuron)
# run stdp experiment
stdp_time_range = np.arange(-40, 40, 2)
stdp_data = np.empty_like(stdp_time_range)
for i, dt in enumerate(stdp_time_range):
stdp_data[i], _ = stdp_expt(net, dt, initial_conditions)
plt.figure()
plt.plot(stdp_time_range, stdp_data, marker="o")
plt.grid()
# verify
get_stdp_profile(Neuron, Synapse, initial_conditions)
# A experiment with pulse train
Neuron, Synapse = nm.HHNeuronWithCaJL, nm.PlasticNMDASynapseWithCaJL
neuron_nums = [1, 1]
net = networks.get_multilayer_fc(Neuron, Synapse, neuron_nums)
total_time = 1000.
time_sampled_range = np.arange(0., total_time, 0.1)
f, _, _ = lab_manager.set_up_lab(net)
t0s = np.arange(0, 1000, 50)
experiments.pulse_train_on_layer(net, 0, t0s, i_max=50.)
data = lab_manager.run_lab(f, initial_conditions, time_sampled_range)
lab_manager.show_all_neuron_in_layer(time_sampled_range, data, net, 0)
lab_manager.show_all_neuron_in_layer(time_sampled_range, data, net, 1)
lab_manager.show_all_synaspe_onto_layer(time_sampled_range, data, net, 1)
#AL = net.create_AL_man(nm.LN, nm.PN_2, nm.Synapse_gaba_LN, nm.Synapse_nAch_PN_2)
#Set up the experiment
num_layers = 2
#The value of the input current is 400 pA
val = 250 #nA
#val = 400
#These are the neuron indicies within each layer which receive the current
neuron_inds = [[0, 1], [1, 3]]
current_vals = [[val, val], [val, val]]
# Set up the experiment with a constant input current
ex.const_current(AL, num_layers, neuron_inds, current_vals)
#set up the lab
f, initial_conditions, neuron_inds = lm.set_up_lab(AL)
#run for specified time with dt
time_len = 1000.0 #Run for 600 ms
dt = 0.02 # Integration time step 0.02 ms
time_sampled_range = np.arange(0., time_len, dt)
# Run the lab and get output
data = lm.run_lab(f,
initial_conditions,
time_sampled_range,
integrator='dopri5',
compile=True)
#save PN data for analysis
V_vec = []
net, base_rate, i_max=i_max, num_sniffs=num_sniffs,
time_per_sniff=time_per_sniff)
# # for more than two pre-synaptic neurons, use the following instead
# ex.feed_gaussian_rate_poisson_spikes_DL(
# net, base_rate, i_max=i_max, num_sniffs=num_sniffs,
# time_per_sniff=time_per_sniff)
# periodic
# total_time = 1000
# i_max=100.
# base_rate = 0.01
# ratio = 0.95
# ex.feed_periodic_spikes(net, base_rate, ratio, total_time, i_max)
# step 3: ask our lab manager to set up the lab for the experiment.
f, initial_conditions, _ = lm.set_up_lab(net)
# step 4: run the lab and gather data
time_sampled_range = np.arange(0., total_time, 0.1)
data = lm.run_lab(f, initial_conditions, time_sampled_range)
# Time to witness some magic
for layer_idx in range(len(net.layers)):
lm.show_all_neuron_in_layer(
time_sampled_range, data, net, layer_idx)
for layer_idx in range(len(net.layers)):
lm.show_all_dendrite_onto_layer(
time_sampled_range, data, net, layer_idx, delta_time=None)
nm.PlasticNMDASynapseWithCaJL, neuron_nums)
networks.draw_layered_digraph(net)
# step 2: design an experiment. (Fixing input currents really)
#experiments.delay_pulses_on_layer_0_and_1(net)
i_max = 50.
num_sniffs = 5
time_per_sniff = 200.
total_time = num_sniffs * time_per_sniff
base_rate = 0.05
experiments.feed_gaussian_rate_poisson_spikes(net,
base_rate,
i_max=i_max,
num_sniffs=num_sniffs,
time_per_sniff=time_per_sniff)
# step 3: ask our lab manager to set up the lab for the experiment.
f, initial_conditions, _ = lab_manager.set_up_lab(net)
# step 4: run the lab and gather data
time_sampled_range = np.arange(0., total_time, 0.1)
data = lab_manager.run_lab(f, initial_conditions, time_sampled_range)
# Time to witness some magic
#lab_manager.sample_plot(time_sampled_range, data, net)
for layer_idx in range(len(net.layers)):
lab_manager.show_all_neuron_in_layer(time_sampled_range, data, net,
layer_idx)
lab_manager.show_all_synaspe_onto_layer(time_sampled_range, data, net,
layer_idx)