def run_GS(W, b1, b2, n):
steps = n
from neusa.RBMSampler import RBMGibbsSampler
g = RBMGibbsSampler(W, b1, b2)
i = np.random.randint(0,2, size=N_h)
lv, lh = g.manysteps(i,steps)
a=np.hstack([lv,lh])
states = kl_tools.spikes2states(a)
return states
def run_GS(W, b1, b2, n):
steps = n
from neusa.RBMSampler import RBMGibbsSampler
g = RBMGibbsSampler(W, b1, b2)
i = np.random.randint(0, 2, size=N_h)
lv, lh = g.manysteps(i, steps)
a = np.hstack([lv, lh])
states = kl_tools.spikes2states(a)
return states
def states_NS(Mv, Mh, t_conv=1.0):
Sv = monitor_to_spikelist(Mv).time_slice(10*t_ref, t_sim*1000)
Sh = monitor_to_spikelist(Mh).time_slice(10*t_ref, t_sim*1000)
Sv.time_offset(-10*t_ref)
Sh.time_offset(-10*t_ref)
Sv.t_start = 0
Sv.t_stop = t_sim*1000 - 10*t_ref
Sh.t_start = 0
Sh.t_stop = t_sim*1000 - 10*t_ref
Sv.complete(np.arange(N_v))
Sh.complete(np.arange(N_h))
sv = Sv.spike_histogram(time_bin=1.).astype('bool').astype('int')
sh = Sh.spike_histogram(time_bin=1.).astype('bool').astype('int')
s = np.row_stack([sv, sh])
s_conv = []
for row in s:
s_conv.append(np.convolve(row, np.ones([t_conv*t_ref/ms])))
s_conv = np.array(s_conv).T.astype('int')
s_conv[s_conv>1]=1
states = kl_tools.spikes2states(s_conv)
return states
def states_NS(Mv, Mh, t_conv=1.0):
Sv = monitor_to_spikelist(Mv).time_slice(10 * t_ref, t_sim * 1000)
Sh = monitor_to_spikelist(Mh).time_slice(10 * t_ref, t_sim * 1000)
Sv.time_offset(-10 * t_ref)
Sh.time_offset(-10 * t_ref)
Sv.t_start = 0
Sv.t_stop = t_sim * 1000 - 10 * t_ref
Sh.t_start = 0
Sh.t_stop = t_sim * 1000 - 10 * t_ref
Sv.complete(np.arange(N_v))
Sh.complete(np.arange(N_h))
sv = Sv.spike_histogram(time_bin=1.).astype('bool').astype('int')
sh = Sh.spike_histogram(time_bin=1.).astype('bool').astype('int')
s = np.row_stack([sv, sh])
s_conv = []
for row in s:
s_conv.append(np.convolve(row, np.ones([t_conv * t_ref / ms])))
s_conv = np.array(s_conv).T.astype('int')
s_conv[s_conv > 1] = 1
states = kl_tools.spikes2states(s_conv)
return states
