def plot_mean_firing_rates(self, trials, plt_title='Mean Firing Rates'):
if self.record_firing_rates:
mean_e_pop_rates=[]
std_e_pop_rates=[]
for i in range(self.network_params.num_groups):
pop_rate_mat=np.array(self.pop_rates['excitatory_rate_%d' % i])
mean_e_pop_rates.append(np.mean(pop_rate_mat[trials,:],axis=0))
std_e_pop_rates.append(np.std(pop_rate_mat[trials,:],axis=0)/np.sqrt(len(trials)))
mean_e_pop_rates=np.array(mean_e_pop_rates)
std_e_pop_rates=np.array(std_e_pop_rates)
pop_rate_mat=np.array(self.pop_rates['inhibitory_rate'])
mean_i_pop_rate=np.mean(pop_rate_mat[trials,:],axis=0)
std_i_pop_rate=np.std(pop_rate_mat[trials,:],axis=0)/np.sqrt(len(trials))
plot_network_firing_rates(np.array(mean_e_pop_rates), self.sim_params, self.network_params,
std_e_rates=std_e_pop_rates, i_rate=mean_i_pop_rate, std_i_rate=std_i_pop_rate, plt_title=plt_title)
def plot_sorted_mean_firing_rates(self, trials, plt_title='Mean Firing Rates'):
if self.record_firing_rates:
chosen_pop_rates=[]
unchosen_pop_rates=[]
for trial_idx in trials:
resp=self.trial_resp[0,trial_idx]
if resp>-1:
chosen_pop_rates.append(self.pop_rates['excitatory_rate_%d' % resp][trial_idx])
unchosen_pop_rates.append(self.pop_rates['excitatory_rate_%d' % (1-resp)][trial_idx])
if len(chosen_pop_rates)>1:
chosen_pop_rates=np.array(chosen_pop_rates)
unchosen_pop_rates=np.array(unchosen_pop_rates)
mean_e_pop_rates=np.array([np.mean(chosen_pop_rates,axis=0), np.mean(unchosen_pop_rates,axis=0)])
std_e_pop_rates=np.array([np.std(chosen_pop_rates,axis=0)/np.sqrt(len(trials)),
np.std(unchosen_pop_rates,axis=0)/np.sqrt(len(trials))])
plot_network_firing_rates(np.array(mean_e_pop_rates), self.sim_params, self.network_params,
std_e_rates=std_e_pop_rates, plt_title=plt_title, labels=['chosen','unchosen'])
def plot(self):
# Spike raster plots
if self.record_spikes:
num_plots=self.network_params.num_groups+1
figure()
for i in range(self.network_params.num_groups):
subplot(num_plots,1,i+1)
raster_plot(self.monitors['excitatory_spike_%d' % i],newfigure=False)
subplot(num_plots,1,num_plots)
raster_plot(self.monitors['inhibitory_spike'],newfigure=False)
# Network firing rate plots
if self.record_firing_rate:
e_rate_0=self.monitors['excitatory_rate_0'].smooth_rate(width=5*ms)/hertz
e_rate_1=self.monitors['excitatory_rate_1'].smooth_rate(width=5*ms)/hertz
i_rate=self.monitors['inhibitory_rate'].smooth_rate(width=5*ms)/hertz
plot_network_firing_rates(np.array([e_rate_0, e_rate_1]), i_rate, self.sim_params, self.network_params)
# Input firing rate plots
if self.record_inputs:
figure()
ax=subplot(111)
max_rate=0
task_rates=[]
for i in range(self.network_params.num_groups):
task_monitor=self.monitors['task_rate_%d' % i]
task_rate=task_monitor.smooth_rate(width=5*ms,filter='gaussian')/hertz
if np.max(task_rate)>max_rate:
max_rate=np.max(task_rate)
task_rates.append(task_rate)
rect=Rectangle((0,0),(self.sim_params.stim_end_time-self.sim_params.stim_start_time)/ms, max_rate+5,
alpha=0.25, facecolor='yellow', edgecolor='none')
ax.add_patch(rect)
# ax.plot(self.monitors['background_rate'].times/ms,
# self.monitors['background_rate'].smooth_rate(width=5*ms)/hertz)
for i in range(self.network_params.num_groups):
ax.plot((np.array(range(len(task_rates[i])))*self.sim_params.dt)/ms-self.sim_params.stim_start_time/ms, task_rates[i])
ylim(0,90)
ylabel('Firing rate (Hz)')
xlabel('Time (ms)')
# Network state plots
if self.record_neuron_state:
network_monitor=self.monitors['network']
max_conductances=[]
for neuron_idx in self.record_idx:
max_conductances.append(np.max(network_monitor['g_ampa_r'][neuron_idx]/nS))
max_conductances.append(np.max(network_monitor['g_ampa_x'][neuron_idx]/nS))
max_conductances.append(np.max(network_monitor['g_ampa_b'][neuron_idx]/nS))
max_conductances.append(np.max(network_monitor['g_nmda'][neuron_idx]/nS))
max_conductances.append(np.max(network_monitor['g_gaba_a'][neuron_idx]/nS))
#max_conductances.append(np.max(self.network_monitor['g_gaba_b'][neuron_idx]/nS))
max_conductance=np.max(max_conductances)
fig=figure()
for i in range(self.network_params.num_groups):
neuron_idx=self.record_idx[i]
ax=subplot(int('%d1%d' % (self.network_params.num_groups+1,i+1)))
title('e%d' % i)
ax.plot(network_monitor['g_ampa_r'].times/ms, network_monitor['g_ampa_r'][neuron_idx]/nS,
label='AMPA-recurrent')
ax.plot(network_monitor['g_ampa_x'].times/ms, network_monitor['g_ampa_x'][neuron_idx]/nS,
label='AMPA-task')
ax.plot(network_monitor['g_ampa_b'].times/ms, network_monitor['g_ampa_b'][neuron_idx]/nS,
label='AMPA-backgrnd')
ax.plot(network_monitor['g_nmda'].times/ms, network_monitor['g_nmda'][neuron_idx]/nS,
label='NMDA')
ax.plot(network_monitor['g_gaba_a'].times/ms, network_monitor['g_gaba_a'][neuron_idx]/nS,
label='GABA_A')
#ax.plot(network_monitor['g_gaba_b'].times/ms, network_monitor['g_gaba_b'][neuron_idx]/nS,
# label='GABA_B')
ylim(0,max_conductance)
xlabel('Time (ms)')
ylabel('Conductance (nS)')
legend()
neuron_idx=self.record_idx[self.network_params.num_groups]
ax=subplot('%d1%d' % (self.network_params.num_groups+1,self.network_params.num_groups+1))
title('i')
ax.plot(network_monitor['g_ampa_r'].times/ms, network_monitor['g_ampa_r'][neuron_idx]/nS,
label='AMPA-recurrent')
ax.plot(network_monitor['g_ampa_x'].times/ms, network_monitor['g_ampa_x'][neuron_idx]/nS,
label='AMPA-task')
ax.plot(network_monitor['g_ampa_b'].times/ms, network_monitor['g_ampa_b'][neuron_idx]/nS,
label='AMPA-backgrnd')
ax.plot(network_monitor['g_nmda'].times/ms, network_monitor['g_nmda'][neuron_idx]/nS,
label='NMDA')
ax.plot(network_monitor['g_gaba_a'].times/ms, network_monitor['g_gaba_a'][neuron_idx]/nS,
label='GABA_A')
#ax.plot(network_monitor['g_gaba_b'].times/ms, network_monitor['g_gaba_b'][neuron_idx]/nS,
# label='GABA_B')
ylim(0,max_conductance)
xlabel('Time (ms)')
ylabel('Conductance (nS)')
legend()
min_currents=[]
max_currents=[]
for neuron_idx in self.record_idx:
max_currents.append(np.max(network_monitor['I_ampa_r'][neuron_idx]/nS))
max_currents.append(np.max(network_monitor['I_ampa_x'][neuron_idx]/nS))
max_currents.append(np.max(network_monitor['I_ampa_b'][neuron_idx]/nS))
max_currents.append(np.max(network_monitor['I_nmda'][neuron_idx]/nS))
max_currents.append(np.max(network_monitor['I_gaba_a'][neuron_idx]/nS))
#max_currents.append(np.max(network_monitor['I_gaba_b'][neuron_idx]/nS))
min_currents.append(np.min(network_monitor['I_ampa_r'][neuron_idx]/nS))
min_currents.append(np.min(network_monitor['I_ampa_x'][neuron_idx]/nS))
min_currents.append(np.min(network_monitor['I_ampa_b'][neuron_idx]/nS))
min_currents.append(np.min(network_monitor['I_nmda'][neuron_idx]/nS))
min_currents.append(np.min(network_monitor['I_gaba_a'][neuron_idx]/nS))
#min_currents.append(np.min(network_monitor['I_gaba_b'][neuron_idx]/nS))
max_current=np.max(max_currents)
min_current=np.min(min_currents)
fig=figure()
for i in range(self.network_params.num_groups):
ax=subplot(int('%d1%d' % (self.network_params.num_groups+1,i+1)))
neuron_idx=self.record_idx[i]
title('e%d' % i)
ax.plot(network_monitor['I_ampa_r'].times/ms, network_monitor['I_ampa_r'][neuron_idx]/nA,
label='AMPA-recurrent')
ax.plot(network_monitor['I_ampa_x'].times/ms, network_monitor['I_ampa_x'][neuron_idx]/nA,
label='AMPA-task')
ax.plot(network_monitor['I_ampa_b'].times/ms, network_monitor['I_ampa_b'][neuron_idx]/nA,
label='AMPA-backgrnd')
ax.plot(network_monitor['I_nmda'].times/ms, network_monitor['I_nmda'][neuron_idx]/nA,
label='NMDA')
ax.plot(network_monitor['I_gaba_a'].times/ms, network_monitor['I_gaba_a'][neuron_idx]/nA,
label='GABA_A')
#ax.plot(network_monitor['I_gaba_b'].times/ms, network_monitor['I_gaba_b'][neuron_idx]/nA,
# label='GABA_B')
ylim(min_current,max_current)
xlabel('Time (ms)')
ylabel('Current (nA)')
legend()
ax=subplot(int('%d1%d' % (self.network_params.num_groups+1,self.network_params.num_groups+1)))
neuron_idx=self.record_idx[self.network_params.num_groups]
title('i')
ax.plot(network_monitor['I_ampa_r'].times/ms, network_monitor['I_ampa_r'][neuron_idx]/nA,
label='AMPA-recurrent')
ax.plot(network_monitor['I_ampa_x'].times/ms, network_monitor['I_ampa_x'][neuron_idx]/nA,
label='AMPA-task')
ax.plot(network_monitor['I_ampa_b'].times/ms, network_monitor['I_ampa_b'][neuron_idx]/nA,
label='AMPA-backgrnd')
ax.plot(network_monitor['I_nmda'].times/ms, network_monitor['I_nmda'][neuron_idx]/nA,
label='NMDA')
ax.plot(network_monitor['I_gaba_a'].times/ms, network_monitor['I_gaba_a'][neuron_idx]/nA,
label='GABA_A')
#ax.plot(network_monitor['I_gaba_b'].times/ms, network_monitor['I_gaba_b'][neuron_idx]/nA,
# label='GABA_B')
ylim(min_current,max_current)
xlabel('Time (ms)')
ylabel('Current (nA)')
legend()
# LFP plot
if self.record_lfp:
figure()
ax=subplot(111)
ax.plot(self.monitors['lfp'].times / ms, self.monitors['lfp'][0]/mA)
xlabel('Time (ms)')
ylabel('LFP (mA)')
# Voxel activity plots
if self.record_voxel:
voxel_monitor=self.monitors['voxel']
voxel_exc_monitor=None
if 'voxel_exc' in self.monitors:
voxel_exc_monitor=self.monitors['voxel_exc']
syn_max=np.max(voxel_monitor['G_total'][0] / nS)
y_max=np.max(voxel_monitor['y'][0])
y_min=np.min(voxel_monitor['y'][0])
figure()
if voxel_exc_monitor is None:
ax=subplot(211)
else:
ax=subplot(221)
syn_max=np.max([syn_max, np.max(voxel_exc_monitor['G_total'][0])])
y_max=np.max([y_max, np.max(voxel_exc_monitor['y'][0])])
y_min=np.min([y_min, np.min(voxel_exc_monitor['y'][0])])
ax.plot(voxel_monitor['G_total'].times / ms, voxel_monitor['G_total'][0] / nS)
xlabel('Time (ms)')
ylabel('Total Synaptic Activity (nS)')
ylim(0, syn_max)
if voxel_exc_monitor is None:
ax=subplot(212)
else:
ax=subplot(222)
ax.plot(voxel_monitor['y'].times / ms, voxel_monitor['y'][0])
xlabel('Time (ms)')
ylabel('BOLD')
ylim(y_min, y_max)
if voxel_exc_monitor is not None:
ax=subplot(223)
ax.plot(voxel_exc_monitor['G_total'].times / ms, voxel_exc_monitor['G_total'][0] / nS)
xlabel('Time (ms)')
ylabel('Total Synaptic Activity (nS)')
ylim(0, syn_max)
ax=subplot(224)
ax.plot(voxel_exc_monitor['y'].times / ms, voxel_exc_monitor['y'][0])
xlabel('Time (ms)')
ylabel('BOLD')
ylim(y_min, y_max)
if self.record_connections:
figure()
ax=subplot(111)
for mon in self.monitors:
if mon.startswith('connection_'):
conn_name=mon[11:]
conns=np.zeros((len(self.monitors[mon].values),1))
conn_times=[]
for idx, (time, conn_matrix) in enumerate(self.monitors[mon].values):
conn_diag=np.diagonal(conn_matrix.todense())
mean_w=np.mean(conn_diag)
conns[idx,0]=mean_w
conn_times.append(time)
ax.plot(np.array(conn_times) / ms, conns[:,0]/nS, label=conn_name)
legend(loc='best')
xlabel('Time (ms)')
ylabel('Connection Weight (nS)')