def generate_connection_e(self, params):
"""
Generate the W_eu connection matrix. TODO: params should not be passed
as an argument again!
Arguments:
params -- bunch of simulation parameters from param.py
Returns:
W_eu -- FullSynapticMatrix of size (N_e, A), containing 1s and 0s
"""
# choose random, non-overlapping input neuron pools
W = np.zeros((self.N_e, self.A))
available = set(range(self.N_e))
for a in range(self.A):
temp = random.sample(available, self.N_u)
W[temp, a] = 1
available = set([n for n in available if n not in temp])
assert len(available) > 0,\
'Input alphabet too big for non-overlapping neurons'
# always use a full synaptic matrix
ans = synapses.FullSynapticMatrix(params, (self.N_e, self.A))
ans.W = W
return ans
def generate_connection_e(self, par):
"""Generate the W_eu connection matrix
Parameters:
par: Bunch
Main initial sorn parameters
"""
# use a full synaptic matrix
ans = synapses.FullSynapticMatrix(par, (par.N_e, self.N_u))
ans.W = np.zeros(par.N_e)
return ans
def generate_connection_e(self, par):
"""
Generate the W_eu connection matrix
Parameters:
N_e: number of excitatory neurons
"""
# choose random, overlapping input neuron pools
W = np.zeros((par.N_e, self.N_a))
available = set(range(par.N_e))
# TODO: be sure that the input pools are not equal - random.choice
for a in range(self.N_a):
temp = random.sample(available, self.N_u)
W[temp, a] = 1
# always use a full synaptic matrix
ans = synapses.FullSynapticMatrix(par, (par.N_e, self.N_a))
ans.W = W
return ans
def generate_connection_e(self, par):
"""Generate the W_eu connection matrix
Parameters:
par: Bunch
Main initial sorn parameters
"""
# choose random input neuron pools
W = np.zeros((par.N_e, self.N_a))
available = set(range(par.N_e))
for a in range(self.N_a):
temp = random.sample(available, self.N_u)
W[temp, a] = 1
if not self.overlap:
available -= set(temp)
# always use a full synaptic matrix
ans = synapses.FullSynapticMatrix(par, (par.N_e, self.N_a))
ans.W = W
return ans