Npost_1=1
group_dict = {}
# run(run_time)
P.simulation['trials']=10
for trial in range(P.simulation['trials']):
net.restore('initial')
P.init_synapses['1']['w_ampa'] = param._weight_matrix_randn(Npre=Npre_1, Npost=Npost_1, w_mean=1, w_std=0.5)
synapses['1'].w_ampa = P.init_synapses['1']['w_ampa']
print synapses['1'].w_ampa
net.store('randomized')
synapses['1'].w_ampa = P.init_synapses['1']['w_ampa']
P.simulation['trial_id'] = str(uuid.uuid4())
for field_i, field in enumerate(P.simulation['field_mags']):
P.simulation['field_mag'] = field
P.neurons['1']['I_field'] = field
net.restore('randomized')
neurons['1'].I_field= field
net.run(P.simulation['run_time'])
data_df = analysis._rec2df(rec=rec, P=P)
group_df = group_df.append(data_df, ignore_index=True)
data_dict = analysis._rec2dict(rec=rec, P=P)
group_dict = analysis._add_to_group_data(group_data=group_dict, data_dict=data_dict)
group_df = analysis._dict2frame(data_dict=group_dict)
# for group_key, group in neurons.iter
# brian_objects = collect(level=0)
# net = Network(nrn, self.input_nrn, self.input_syn, self.rec)
net.run(P.simulation['run_time'])
print 'first run finished'
# get trained weights
trained_weights = {}
weight_keys = ['w_ampa', 'w_nmda', 'w_gaba', 'w_clopath', 'w_vogels']
for syn_group, syn in synapses.iteritems():
trained_weights[syn_group] = {}
for weight_key in weight_keys:
if hasattr(syn, weight_key):
trained_weights[syn_group][weight_key] = getattr(
syn, weight_key)[-1]
# training data
train_df = analysis._rec2df(rec=rec, P=P, include_P=False)
# Test
#==================================================================
# restore randomized network
net.restore('randomized')
# set ampa weights to be fixed
# synapses['EE'].update_ampa_online = 0
P.synapses['EE']['update_ampa_online'] = 0
P.synapses['FE_train']['update_ampa_online'] = 0
P.synapses['FE_test']['update_ampa_online'] = 0
print synapses['EE'].namespace
# initialize weights to trained values
for syn_group, syn in synapses.iteritems():