temporal_patterns = temporal_patterns_full.copy()
for idx in range(run + args.n_cells[4],
temporal_patterns.shape[0]):
temporal_patterns[idx] = np.array([])
for idx in range(0, run):
temporal_patterns[idx] = np.array([])
nw = net_tunedrevexpdrives.TunedNetwork(
seed=args.seed + run,
n_gcs=args.n_cells[0],
n_mcs=args.n_cells[1],
n_bcs=args.n_cells[2],
n_hcs=args.n_cells[3],
W_pp_gc=args.W_pp_gc,
W_pp_bc=ff_weight,
n_pp_gc=args.n_pp_gc,
n_pp_bc=args.n_pp_bc,
W_gc_bc=fb_weight,
W_gc_hc=fb_weight,
W_bc_gc=args.W_bc_gc,
W_hc_gc=args.W_hc_gc,
ff_t_offset=args.t_pp_to_bc_offset,
temporal_patterns=temporal_patterns,
rec_cond=bool(args.rec_cond))
# Run the model
"""Initialization for -2000 to -100"""
h.cvode.active(0)
dt = 0.1
h.steps_per_ms = 1.0 / dt
h.finitialize(-60)
h.t = -2000
# Generate the PP -> BC mapping as above
innervation_pattern_bc = np.array(
[np.random.choice(400, 20, replace=False) for x in range(24)])
innervation_pattern_bc = innervation_pattern_bc.swapaxes(0, 1)
PP_to_BCs = []
for x in range(0, 400):
PP_to_BCs.append(np.argwhere(innervation_pattern_bc == x)[:, 1])
PP_to_BCs = np.array(PP_to_BCs)
all_targets = np.array([y for x in PP_to_GCs for y in x])
# Start the runs of the model
for run in runs:
nw = net_tunedrevexpdrives.TunedNetwork(
10000, temporal_patterns[0 + run:24 + run],
PP_to_GCs[0 + run:24 + run], PP_to_BCs[0 + run:24 + run])
# Attach voltage recordings to all cells
nw.populations[0].voltage_recording(range(2000))
nw.populations[1].voltage_recording(range(60))
nw.populations[2].voltage_recording(range(24))
nw.populations[3].voltage_recording(range(24))
# Run the model
"""Initialization for -2000 to -100"""
h.cvode.active(0)
dt = 0.1
h.steps_per_ms = 1.0 / dt
h.finitialize(-60)
h.t = -2000
h.secondorder = 0