def construct_neural_populations(self, neuron_eqs, threshold_eqs,
reset_eqs):
self.G_TL2 = nc.generate_neuron_groups(self.N_TL2,
neuron_eqs,
threshold_eqs,
reset_eqs,
TL2_neuron_params,
name='TL2')
self.G_CL1 = nc.generate_neuron_groups(self.N_CL1,
neuron_eqs,
threshold_eqs,
reset_eqs,
CL1_neuron_params,
name='CL1')
self.G_TB1 = nc.generate_neuron_groups(self.N_TB1,
neuron_eqs,
threshold_eqs,
reset_eqs,
TB1_neuron_params,
name='TB1')
self.G_TN2 = nc.generate_neuron_groups(self.N_TN2,
neuron_eqs,
threshold_eqs,
reset_eqs,
TN2_neuron_params,
name='TN2')
self.G_CPU4 = nc.generate_neuron_groups(self.N_CPU4,
neuron_eqs,
threshold_eqs,
reset_eqs,
CPU4_neuron_params,
name='CPU4')
self.G_CPU1A = nc.generate_neuron_groups(self.N_CPU1A,
neuron_eqs,
threshold_eqs,
reset_eqs,
neuron_params,
name='CPU1A')
self.G_CPU1B = nc.generate_neuron_groups(self.N_CPU1B,
neuron_eqs,
threshold_eqs,
reset_eqs,
neuron_params,
name='CPU1B')
self.G_PONTINE = nc.generate_neuron_groups(self.N_PONTINE,
neuron_eqs,
threshold_eqs,
reset_eqs,
neuron_params,
name='PONTINE')
self.G_MOTOR = nc.generate_neuron_groups(self.N_MOTOR,
neuron_eqs,
threshold_eqs,
reset_eqs,
neuron_params,
name='MOTOR')
return [
self.G_TL2, self.G_CL1, self.G_TB1, self.G_TN2, self.G_CPU4,
self.G_CPU1A, self.G_CPU1B, self.G_PONTINE, self.G_MOTOR
]
start_scope()
time_step = 20 # ms
h_stimulus = TimedArray(headings * Hz, dt=1. * time_step * ms)
P_HEADING = PoissonGroup(N_TL2, rates='h_stimulus(t,i)')
SPM_H = SpikeMonitor(P_HEADING)
f_stimulus = TimedArray(flow * Hz, dt=1. * time_step * ms)
P_FLOW = PoissonGroup(N_TN2, rates='f_stimulus(t,i)')
SPM_FLOW = SpikeMonitor(P_FLOW)
# Neuron groups already optimised
G_TL2 = nc.generate_neuron_groups(N_TL2,
eqs,
threshold_eqs,
reset_eqs,
TL2_neuron_params,
name='TL2_source_network')
G_CL1 = nc.generate_neuron_groups(N_CL1,
eqs,
threshold_eqs,
reset_eqs,
CL1_neuron_params,
name='CL1_source_network')
G_TB1 = nc.generate_neuron_groups(N_TB1,
eqs,
threshold_eqs,
reset_eqs,
TB1_neuron_params,
name='TB1_source_network')
G_TN2 = nc.generate_neuron_groups(N_TN2,
v[:T_outbound]))
######################################
### SPIKE BASED CX
######################################
start_scope()
time_step = 20 # ms
h_stimulus = TimedArray(headings * Hz, dt=1. * time_step * ms)
P_HEADING = PoissonGroup(N_TL2, rates='h_stimulus(t,i)')
# Neuron group
G_TL2 = nc.generate_neuron_groups(N_TL2,
eqs,
threshold_eqs,
reset_eqs,
neuron_params,
name='TL2_source_network')
# Add monitors
#STM_TL2, SPM_TL2 = nc.add_monitors(G_TL2, name='TL2_source')
# Connect heading to TL2
S_P_HEADING_TL2 = nc.connect_synapses(P_HEADING,
G_TL2,
W_HEADING_TL2,
model=synapses_model,
params=synapses_params,
on_pre=synapses_eqs_ex)
#### Target
### SPIKE BASED CX
######################################
start_scope()
time_step = 60 # ms
h_stimulus = TimedArray(headings * Hz, dt=1. * time_step * ms)
P_HEADING = PoissonGroup(N_TL2, rates='h_stimulus(t,i)')
f_stimulus = TimedArray(flow * Hz, dt=1. * time_step * ms)
P_FLOW = PoissonGroup(N_TN2, rates='f_stimulus(t,i)')
# Neuron groups already optimised
G_TL2 = nc.generate_neuron_groups(N_TL2,
eqs,
threshold_eqs,
reset_eqs,
TL2_neuron_params,
name='TL2_source_network')
G_CL1 = nc.generate_neuron_groups(N_CL1,
eqs,
threshold_eqs,
reset_eqs,
CL1_neuron_params,
name='CL1_source_network')
G_TB1 = nc.generate_neuron_groups(N_TB1,
eqs,
threshold_eqs,
reset_eqs,
TB1_neuron_params,
name='TB1_source_network')
G_TN2 = nc.generate_neuron_groups(N_TN2,
f_stimulus = TimedArray(flow * Hz, dt=1. * time_step * ms)
P_FLOW = PoissonGroup(N_TN2, rates='f_stimulus(t,i)')
#global CPU4_memory_stimulus
CPU_MEMORY_starting_value = 50 #Hz
CPU4_memory_stimulus = CPU_MEMORY_starting_value * np.ones(
(T_outbound, N_CPU4)) * Hz
P_CPU4_MEMORY = PoissonGroup(N_CPU4,
rates=CPU4_memory_stimulus[0, :],
name='P_CPU4_MEMORY')
# Neuron groups already optimised
G_TL2 = nc.generate_neuron_groups(N_TL2,
eqs,
threshold_eqs,
reset_eqs,
TL2_neuron_params,
name='TL2')
G_CL1 = nc.generate_neuron_groups(N_CL1,
eqs,
threshold_eqs,
reset_eqs,
CL1_neuron_params,
name='CL1')
G_TB1 = nc.generate_neuron_groups(N_TB1,
eqs,
threshold_eqs,
reset_eqs,
TB1_neuron_params,
name='TB1')
G_TN2 = nc.generate_neuron_groups(N_TN2,
route=(h[:T_outbound], v[:T_outbound]))
######################################
### SPIKE BASED CX
######################################
start_scope()
time_step = 20 # ms
P_CL1 = PoissonGroup(N_CL1, rates=CL1_spike_rates*Hz)
# Neuron group
G_CL1 = nc.generate_neuron_groups(N_CL1, eqs, threshold_eqs, reset_eqs, neuron_params, name='CL1_source_network')
# Add monitors
#STM_TL2, SPM_TL2 = nc.add_monitors(G_TL2, name='TL2_source')
# Connect heading to TL2
S_P_CL1_CL1 = nc.connect_synapses(P_CL1, G_CL1, np.eye(N_CL1), model=synapses_model,
params=synapses_params, on_pre=synapses_eqs_ex)
#### Target
# Scale spike rates from rate-based CX in the right range
# transpose since log is neuron_index*time_step but we want the opposite
#TL2_stimulus = TimedArray(TL2_spike_rates*cx_log.tl2.T*Hz, dt=1.*time_step*ms)
P_CL1_TARGET = PoissonGroup(N_CL1, rates=CL1_spike_rates*Hz)
# SPM_TL2_IDEAL = SpikeMonitor(P_TL2, name='TL2_target')
