def cluster_estimation_study(folder_experiment, cluster_dir):
for dirName, subdirList, fileList in os.walk(folder_experiment):
biggest_clusters_avg = np.array([])
n_clusters_avg = np.array([])
num_robots = "-1"
rho = -1.0
alpha = -1.0
elements = dirName.split("_")
for e in elements:
if e.startswith("robots"):
num_robots = e.split("#")[-1]
if e.startswith("rho"):
rho = float(e.split("#")[-1])
if e.startswith("alpha"):
alpha = float(e.split("#")[-1])
# print(num_robots+' '+str(rho)+' '+str(alpha))
if num_robots == "-1" or rho == -1.0 or alpha == -1:
continue
else:
print(dirName)
runs = len([f for f in fileList if f.endswith('position.tsv')])
# print(runs)
[_, df_experiment] = utils.load_pd_positions(dirName, "experiment")
positions_concatenated = df_experiment.values[:, 1:] # [robots, times]
[num_robot, num_times] = positions_concatenated.shape
positions_concatenated = np.array(
[x.split(',') for x in positions_concatenated.ravel()],
dtype=float)
positions_concatenated = positions_concatenated.reshape(
num_robot, num_times, 2)
position_concatenated_split = np.split(positions_concatenated, runs)
for single_run in position_concatenated_split:
# print('single run processing')
biggest_clusters_per_time, n_clusters = get_connected_components(
single_run)
biggest_clusters_avg = np.vstack([
biggest_clusters_avg, biggest_clusters_per_time
]) if biggest_clusters_avg.size else biggest_clusters_per_time
n_clusters_avg = np.vstack([
n_clusters_avg, n_clusters
]) if n_clusters_avg.size else n_clusters
biggest_clusters_avg = np.mean(biggest_clusters_avg, axis=0)
n_clusters_avg = np.mean(n_clusters_avg, axis=0)
print('Plotting')
plot_info_clusters(biggest_clusters_avg, n_clusters_avg, str(rho),
str(alpha), num_robots, cluster_dir)
def distance_from_origin_distribution(main_folder, folder_experiments,
powerlaw_dir):
Ncolors = 200
colormap = plt.cm.viridis # LinearSegmentedColormap
# print("colormap.N", colormap.N)
Ncolors = min(colormap.N, Ncolors)
mapcolors = [
colormap(int(x * colormap.N / Ncolors)) for x in range(Ncolors)
]
# print(len(mapcolors))
for dirName, subdirList, fileList in os.walk(main_folder + '/' +
folder_experiments):
num_robots = "-1"
rho = -1.0
alpha = -1.0
elements = dirName.split("_")
for e in elements:
if e.startswith("robots"):
num_robots = e.split("#")[-1]
if (e.startswith("rho")):
rho = float(e.split("#")[-1])
if (e.startswith("alpha")):
alpha = float(e.split("#")[-1])
if (num_robots == "-1" or rho == -1.0 or alpha == -1):
continue
rho_str = str(rho)
alpha_str = str(alpha)
# print("rho", rho_str)
# print("alpha", alpha_str)
# print(dirName)
df_experiment = pd.DataFrame()
[_, df_experiment] = utils.load_pd_positions(dirName, "experiment")
# print(number_of_experiments)
positions_concatenated = df_experiment.values[:, 1:]
[num_robot, num_times] = positions_concatenated.shape
positions_concatenated = np.array(
[x.split(',') for x in positions_concatenated.ravel()],
dtype=float)
positions_concatenated = positions_concatenated.reshape(
num_robot, num_times, 2)
# print("positions_concatenated.shape", positions_concatenated.shape)
distances = utils.distance_from_the_origin(positions_concatenated)
# print("distances.shape", distances.shape)
fig = plt.figure(figsize=(10, 5), dpi=160)
plt.xlim((0.001, 100))
plt.ylim((0.0001, 100))
for i, d in enumerate(distances):
fit = powerlaw.Fit(d, xmin=0.00001)
# print(i)
fit.plot_pdf(linewidth=2, color=mapcolors[i])
plt.ylabel('p(x)')
plt.xlabel('distance from origin')
plt.title(
"origin distance distribution with %s robots, alpha=%s, rho=%s" %
(num_robots, alpha, rho))
file_name = "powerlaw_%s_rho_%s_alpha_%s.png" % (num_robots, rho,
alpha)
plt.savefig(powerlaw_dir + '/' + file_name)
plt.close(fig)
def evaluate_history_WMSD_and_time_diffusion(main_folder, folder_experiments, baseline_dir, msd_type, b_edges,
result_time_dir, distance_heatmap_dir):
print("MSD Evaluation and/or Density Maps Generation")
for dirName, subdirList, fileList in os.walk(main_folder + '/' + folder_experiments):
# print(dirName)
num_robots = "-1"
rho = -1.0
alpha = -1.0
elements = dirName.split("_")
for e in elements:
if e.startswith("robots"):
num_robots = e.split("#")[-1]
if e.startswith("rho"):
rho = float(e.split("#")[-1])
if e.startswith("alpha"):
alpha = float(e.split("#")[-1])
# print(num_robots+' '+str(rho)+' '+str(alpha))
if num_robots == "-1" or rho == -1.0 or alpha == -1:
continue
# print("dirName: ", dirName)
runs = len([f for f in fileList if
(os.path.isfile(os.path.join(dirName, f)) and f.endswith('position.tsv'))])
# print("runs: ", runs)
rho_str = str(rho)
alpha_str = str(alpha)
# print("rho", rho_str)
# print("alpha", alpha_str)
total_experiment_wmsd = []
baseline_experiment_wmsd = []
# print(alpha_str)
folder_baseline = baseline_dir + "alpha#%s_rho#%s_baseline_1800" % (alpha_str, rho_str)
# if not os.path.isdir(main_folder + '/' + folder_baseline):
# print("folder_baseline is not an existing directory")
# exit(-1)
number_of_experiments = 0
df_experiment = pd.DataFrame()
df_baseline = pd.DataFrame()
# print("W_size=", window_size)
[number_of_experiments, df_experiment] = utils.load_pd_positions(dirName, "experiment")
[_, df_baseline] = utils.load_pd_positions(folder_baseline, "baseline")
# print(number_of_experiments)
positions_concatenated = df_experiment.values[:, 1:]
[num_robot, num_times] = positions_concatenated.shape
positions_concatenated = np.array([x.split(',') for x in positions_concatenated.ravel()], dtype=float)
positions_concatenated = positions_concatenated.reshape(num_robot, num_times, 2)
# print(positions_concatenated.shape)
if config.comparison_plots_flag:
baseline_concatenated = df_baseline.values[:, 1:]
[num_robot, num_times] = baseline_concatenated.shape
baseline_concatenated = np.array([x.split(',') for x in baseline_concatenated.ravel()], dtype=float)
baseline_concatenated = baseline_concatenated.reshape(num_robot, num_times, 2)
for window_size in range(1, 10):
w_displacement_array = np.array([])
base_w_displacement_array = np.array([])
if msd_type == 'windowed':
base_win_disp = utils.window_displacement(baseline_concatenated, window_size)
win_disp = utils.window_displacement(positions_concatenated, window_size)
else:
# win_disp = utils.fixed_window_displacement(positions_concatenated, window_size)
# base_win_disp = utils.fixed_window_displacement(baseline_concatenated, window_size)
win_disp = utils.time_mean_square_displacement(positions_concatenated)
base_win_disp = utils.time_mean_square_displacement(baseline_concatenated)
w_displacement_array = np.vstack(
[w_displacement_array, win_disp]) if w_displacement_array.size else win_disp
base_w_displacement_array = np.vstack(
[base_w_displacement_array, base_win_disp]) if base_w_displacement_array.size else base_win_disp
# mean_wmsd = win_disp.mean()
total_experiment_wmsd.append(w_displacement_array)
baseline_experiment_wmsd.append(base_w_displacement_array)
if msd_type == 'time_msd':
break
# utils.plot_both_wmsd(windowed, baseline_experiment_wmsd, total_experiment_wmsd, alpha_str, rho_str, num_robots,
# title, result_time_dir)
utils.plot_both_wmsd(baseline_experiment_wmsd, total_experiment_wmsd, alpha_str, rho_str, num_robots,
result_time_dir, msd_type)
if config.density_maps_flag:
distances = utils.distance_from_the_origin(positions_concatenated)
occurrences = utils.get_occurrences(distances, b_edges, runs)
if not config.open_space_flag and config.density_maps_flag:
utils.time_plot_histogram(occurrences.T, b_edges[1:], alpha_str, rho_str, num_robots,
distance_heatmap_dir)
def avg_connection_degree_heatmap(folder_experiment,
avg_connection_degree_dir):
for dirName, subdirList, fileList in os.walk(folder_experiment):
num_robots = "-1"
rho = -1.0
alpha = -1.0
elements = dirName.split("_")
for e in elements:
if e.startswith("robots"):
num_robots = e.split("#")[-1]
if (e.startswith("rho")):
rho = float(e.split("#")[-1])
if (e.startswith("alpha")):
alpha = float(e.split("#")[-1])
# print(num_robots+' '+str(rho)+' '+str(alpha))
if (num_robots == "-1" or rho == -1.0 or alpha == -1):
continue
else:
# print(dirName)
runs = len([f for f in fileList if f.endswith('position.tsv')])
# print(runs)
[_, df_experiment] = utils.load_pd_positions(dirName, "experiment")
positions_concatenated = df_experiment.values[:, 1:]
[num_robot, num_times] = positions_concatenated.shape
positions_concatenated = np.array(
[x.split(',') for x in positions_concatenated.ravel()],
dtype=float)
positions_concatenated = positions_concatenated.reshape(
num_robot, num_times, 2)
position_concatenated_split = np.split(positions_concatenated, runs)
# print("positions_concatenated.shape: ", positions_concatenated.shape)
# CONNECTIONS
connection_number_history = np.array([])
for single_run in position_concatenated_split:
connection_number_history = np.vstack(
(connection_number_history, get_connections(single_run)
)) if connection_number_history.size else get_connections(
single_run)
# print(connection_number_history.shape)
# ORIGIN DISTANCE
origin_distance = utils.distance_from_the_origin(
positions_concatenated).T
connection_in_time = np.ones(
(config.bin_edges.size - 1, origin_distance.shape[1])) * -1
for idx, distance_t in enumerate(origin_distance.T):
# print(distance_t)
for edge_idx in range(config.bin_edges.size - 1):
# print(bin_edges[edge_idx],bin_edges[edge_idx+1])
# print("\t",edge_idx)
if edge_idx < config.bin_edges.size - 1 or not config.open_space_flag:
where_index = np.where(
np.logical_and(
distance_t >= config.bin_edges[edge_idx],
distance_t < config.bin_edges[edge_idx + 1]))
connection_in_time[edge_idx, idx] = np.mean(
connection_number_history[where_index])
else:
where_index = np.where(
distance_t >= config.bin_edges[edge_idx])
connection_in_time[edge_idx, idx] = np.mean(
connection_number_history[where_index])
connection_heatmap(config.bin_edges, connection_in_time, str(alpha),
str(rho), num_robots, avg_connection_degree_dir)
def avg_connection_plot_different_population_sizes(folder_experiment,
avg_connection_degree_dir):
for a in config.alpha_array:
for r in config.rho_array:
# print("a",a,"r",r)
robot_arr = []
fig = plt.figure(figsize=(20, 10), dpi=80)
# for dirName, subdirList, fileList in os.walk(folder_experiment):
for dirName in natsorted(os.listdir(folder_experiment)):
dirPath = os.path.join(folder_experiment, dirName)
num_robots = "-1"
elements = dirName.split("_")
for e in elements:
if e.startswith("robots"):
num_robots = e.split("#")[-1]
if e.startswith("rho"):
rho = float(e.split("#")[-1])
if e.startswith("alpha"):
alpha = float(e.split("#")[-1])
# print(num_robots+' '+str(rho)+' '+str(alpha))
# print(alpha, rho)
if num_robots == "-1" or rho != r or alpha != a:
continue
else:
# print(dirName)
robot_arr += [int(num_robots)]
# print(num_robots, int(num_robots))
runs = len([
f for f in os.listdir(dirPath)
if (os.path.isfile(os.path.join(dirPath, f))
and f.endswith('position.tsv'))
])
# print(runs)
rho_str = str(rho)
alpha_str = str(alpha)
[_, df_experiment
] = utils.load_pd_positions(dirPath, "experiment")
positions_concatenated = df_experiment.values[:, 1:]
[num_robot, num_times] = positions_concatenated.shape
positions_concatenated = np.array(
[x.split(',') for x in positions_concatenated.ravel()],
dtype=float)
positions_concatenated = positions_concatenated.reshape(
num_robot, num_times, 2)
print("runs", runs)
print("positions_concatenated.shape:",
positions_concatenated.shape)
position_concatenated_split = np.split(positions_concatenated,
runs)
connection_number_history = np.array([])
for single_run in position_concatenated_split:
connection_number_history = np.vstack(
(connection_number_history,
get_connections(single_run))
) if connection_number_history.size else get_connections(
single_run)
connection_number_history_mean = np.mean(
connection_number_history, axis=0)
# print(num_times)
# print(connection_number_history_mean.shape)
# print(len(robot_arr))
plt.plot(np.arange(num_times),
connection_number_history_mean,
linewidth=2,
label=num_robots,
color=mapcolors[len(robot_arr)])
plt.title("Average connection with " + r" $\bf{\rho}:$" + rho_str +
" and " + r"$\bf{\alpha}:$" + alpha_str)
plt.ylabel('mean connection link')
plt.xlabel('time')
# plt.legend(loc='lower right')
if not config.open_space_flag:
plt.yticks(np.arange(0, 9, 0.5))
else:
plt.yticks(np.arange(0, 20, 0.5))
# plt.grid(which='minor')
plt.grid()
plt.legend(loc=2) # , bbox_to_anchor=(0.95, 0.5))
file_name = "average_connection_plot_rho_%s_alpha_%s.png" % (
rho_str, alpha_str)
plt.savefig(avg_connection_degree_dir + '/' + file_name)
plt.close(fig)
def evaluate_WMSD_heatmap(main_folder, folder_experiments, baseline_dir,
msd_type, heatmap_dir):
for window_size in range(1, 10):
total_dict = dict()
number_dict = dict()
for dirName, subdirList, fileList in os.walk(main_folder + '/' +
folder_experiments):
num_robots = "-1"
rho = -1.0
alpha = -1.0
elements = dirName.split("_")
for e in elements:
if e.startswith("robots"):
num_robots = e.split("#")[-1]
if num_robots not in total_dict:
total_dict[num_robots] = dict()
number_dict[num_robots] = dict()
if (e.startswith("rho")):
rho = float(e.split("#")[-1])
if (e.startswith("alpha")):
alpha = float(e.split("#")[-1])
# print(str(count) + " : " + dirName)
if num_robots == "-1" or rho == -1.0 or alpha == -1:
continue
rho_str = str(rho)
alpha_str = str(alpha)
# print("rho", rho_str)
# print("alpha", alpha_str)
# print(dirName)
if rho_str not in total_dict[num_robots]:
total_dict[num_robots][rho_str] = dict()
number_dict[num_robots][rho_str] = dict()
# print(total_dict)
total_experiment_wmsd = []
baseline_experiment_wmsd = []
# folder_baseline = "baseline_2020-02-14/2020-02-14_robots#1_alpha#%s_rho#%s_baseline_1800" %(alpha_str, rho_str)
folder_baseline = baseline_dir + "alpha#%s_rho#%s_baseline_1800" % (
alpha_str, rho_str)
number_of_experiments = 0
df_experiment = pd.DataFrame()
df_baseline = pd.DataFrame()
# print("W_size=", window_size)
[number_of_experiments,
df_experiment] = utils.load_pd_positions(dirName, "experiment")
[_, df_baseline] = utils.load_pd_positions(folder_baseline,
"baseline")
# print(number_of_experiments)
positions_concatenated = df_experiment.values[:, 1:]
[num_robot, num_times] = positions_concatenated.shape
positions_concatenated = np.array(
[x.split(',') for x in positions_concatenated.ravel()],
dtype=float)
positions_concatenated = positions_concatenated.reshape(
num_robot, num_times, 2)
baseline_concatenated = df_baseline.values[:, 1:]
[num_robot, num_times] = baseline_concatenated.shape
baseline_concatenated = np.array(
[x.split(',') for x in baseline_concatenated.ravel()],
dtype=float)
baseline_concatenated = baseline_concatenated.reshape(
num_robot, num_times, 2)
w_displacement_array = np.array([])
base_w_displacement_array = np.array([])
if msd_type == 'windowed':
base_msd = utils.window_displacement(baseline_concatenated,
window_size)
msd = utils.window_displacement(positions_concatenated,
window_size)
elif msd_type == 'fixed':
msd = utils.fixed_window_displacement(positions_concatenated,
window_size)
base_msd = utils.fixed_window_displacement(
baseline_concatenated, window_size)
elif msd_type == 'time_msd':
msd = utils.time_mean_square_displacement(
positions_concatenated)
base_msd = utils.time_mean_square_displacement(
baseline_concatenated)
w_displacement_array = np.vstack([
w_displacement_array, msd
]) if w_displacement_array.size else msd
base_w_displacement_array = np.vstack([
base_w_displacement_array, base_msd
]) if base_w_displacement_array.size else base_msd
mean_wmsd = msd.mean()
total_dict[num_robots][rho_str][alpha_str] = mean_wmsd
number_dict[num_robots][rho_str][alpha_str] = number_of_experiments
total_experiment_wmsd.append(w_displacement_array)
baseline_experiment_wmsd.append(base_w_displacement_array)
# print(heatmap_dir)
# print(total_dict)
total_dict = utils.sort_nested_dict(total_dict)
utils.plot_heatmap(total_dict, msd_type, window_size, heatmap_dir)
if msd_type == 'time_msd':
break