def iter_partial_rectification(trajectories, eff_targets, G, metric='centrality', N_per_sector=1, **kwargs_rectificate):
"""
Used to iterate a partial_rectification without recomputing the best nodes each time.
Working but problematic from a theoretical point of view.
"""
trajectories_copy = deepcopy(trajectories)
G_copy = deepcopy(G)
# Make groups
#n_best = select_interesting_navpoints(G, OD=OD(trajectories), N_per_sector=N_per_sector, metric=metric) # Selecting points with highest betweenness centrality within each sector
n_best = select_interesting_navpoints_per_trajectory(trajectories_copy, G_copy, OD=OD(trajectories_copy), N_per_sec_per_traj=N_per_sector, metric=metric) # Selecting points with highest betweenness centrality within each sector
n_best = [n for sec, points in n_best.items() for n in points]
print 'n_best', n_best
groups = {"C":[], "N":[]} # C for "critical", N for "normal"
for n in G_copy.G_nav.nodes():
if n in n_best:
groups["C"].append(n)
else:
groups["N"].append(n)
probabilities = {"C":0., "N":1.} # Fix nodes with best score (critical points).
final_trajs_list, final_eff_list, final_G_list, final_groups_list = [], [], [], []
for eff_target in eff_targets:
# print "Trajectories:"
# for traj in trajectories_copy:
# print traj
final_trajs, final_eff, final_G, final_groups = rectificate_trajectories_network(trajectories_copy, eff_target, G_copy.G_nav, groups=groups, probabilities=probabilities,\
remove_nodes=True, **kwargs_rectificate)
for new_el, listt in [(final_trajs, final_trajs_list), (final_eff, final_eff_list), (final_G, final_G_list), (final_groups, final_groups_list)]:
listt.append(deepcopy(new_el))
print
return final_trajs_list, final_eff_list, final_G_list, final_groups_list, n_best
def partial_rectification(trajectories, eff_target, G, metric='centrality', N_per_sector=1, **kwargs_rectificate):
"""
High level function for rectification. Fix completely N_per_sector points with
highest metric value per sector.
Working but problematic from a theoretical point of view.
"""
# Make groups
#n_best = select_interesting_navpoints(G, OD=OD(trajectories), N_per_sector=N_per_sector, metric=metric) # Selecting points with highest betweenness centrality within each sector
n_best = select_interesting_navpoints_per_trajectory(trajectories, G, OD=OD(trajectories), N_per_sec_per_traj=N_per_sector, metric=metric) # Selecting points with highest betweenness centrality within each sector
n_best = [n for sec, points in n_best.items() for n in points]
groups = {"C":[], "N":[]} # C for "critical", N for "normal"
for n in G.G_nav.nodes():
if n in n_best:
groups["C"].append(n)
else:
groups["N"].append(n)
probabilities = {"C":0., "N":1.} # Fix nodes with best score (critical points).
final_trajs, final_eff, final_G, final_groups = rectificate_trajectories_network(trajectories, eff_target, G.G_nav, remove_nodes=True,
groups=groups, probabilities=probabilities, **kwargs_rectificate)
return final_trajs, final_eff, final_G, final_groups
