def make_plot_distance(args):
room = Roof(args.L,
args.B,
args.H,
objective_type=args.obj_function,
obj_weight=args.obj_weight)
base_algo = args.algorithm
scipy_algo = scipy_algo_name[base_algo]
our_save_string = get_save_string(args)
args.algorithm = scipy_algo
scipy_save_string = get_save_string(args)
x_range = list(range(args.seed_min, args.seed_max))
with open(our_save_string, 'rb') as f:
our_results = pkl.load(f)
with open(scipy_save_string, 'rb') as f:
scipy_results = pkl.load(f)
metric = 'final_x'
val_ours = onp.around(our_results[metric], 3)
val_scipy = onp.around(scipy_results[metric], 3)
distance_vec = []
for set1, set2 in zip(val_ours, val_scipy):
set1 = onp.array(room.to_pos(set1))
set2 = onp.array(room.to_pos(set2))
set1 = set1.reshape(5, 2)
set2 = set2.reshape(5, 2)
# print(set1, set2)
# dist = conv_distance(set1, set2)
# print(dist)
# distance_vec.append(dist[0])
min_dist = np.inf
indices = np.arange(5, dtype=int)
for x in multiset_permutations(indices):
set2perm = set2[x]
print(set1, '\n', set2perm)
dist = np.sqrt(np.linalg.norm(set1 - set2perm))
min_dist = min(min_dist, dist)
print(dist, min_dist)
distance_vec.append(min_dist)
save_plot_name = our_save_string.replace('.pkl',
'_%s_distance.png' % metric)
print(distance_vec)
save_plot(
distance_vec, x_range, '%s for %s on %s problem.' %
(metric, args.algorithm, args.environment), 'Seeds', metric,
save_plot_name)
print('Metric %s:' % 'Distance', distance_vec)
def on_save(self, _event):
dlg = wx.FileDialog(self, "Save a scan", self.dirname,
self.filename, File.RFS,
wx.SAVE | wx.OVERWRITE_PROMPT)
if dlg.ShowModal() == wx.ID_OK:
self.status.set_general("Saving")
self.filename = dlg.GetFilename()
self.dirname = dlg.GetDirectory()
save_plot(self.dirname, self.filename, self.scanInfo,
self.spectrum)
self.isSaved = True
self.status.set_general("Finished")
dlg.Destroy()
def make_plot_one_seed_range(args):
save_string = get_save_string(args)
with open(save_string, 'rb') as f:
results = pkl.load(f)
metrics = ['final_obj', 'n_iter', 'runtime']
x_range = list(range(args.seed_min, args.seed_max))
for metric in metrics:
save_plot_name = save_string.replace('.pkl', '_%s.png' % metric)
save_plot(
onp.around(results[metric],
3), x_range, '%s for %s on %s problem.' %
(metric, args.algorithm, args.environment), 'Seeds', metric,
save_plot_name)
print('Metric %s:' % metric, results[metric])
print('Metric %s:' % 'final_x', results['final_x'])
def plot(self,
euler_angles=True,
nutation_angle=True,
longitudinal_axis=True,
show_plot=False,
save_plot=True,
filepath=os.path.join(os.path.expanduser('~'), 'SpinFigures'),
filenames=['Euler.png', 'Nut.png', 'Axis.png']):
"""Plots figures and saves them to file by default. All arguments are optional and can be changed if necessary.
"""
if euler_angles:
plot.plot_euler_angles(self.t_span, self.psi, self.theta, self.phi)
if save_plot:
if not os.path.exists(filepath):
os.makedirs(filepath)
plot.save_plot(os.path.join(filepath, filenames[0]))
if nutation_angle:
plot.plot_nutation_angle(self.t_span, self.nutation_angle)
if save_plot:
if not os.path.exists(filepath):
os.makedirs(filepath)
plot.save_plot(os.path.join(filepath, filenames[1]))
if longitudinal_axis:
plot.plot_longitudinal_axis(self.t_span, self.nutation_angle,
self.precession_angle)
if save_plot:
if not os.path.exists(filepath):
os.makedirs(filepath)
plot.save_plot(os.path.join(filepath, filenames[2]))
if show_plot:
plot.show_plot()
def accumulated_swe_plots(nc_dir=UT_NC_DIR,
domain='d02',
domain_name="UT2.6km",
labels=[],
extent=None,
central_longitude=-110):
for nc_file in domain_netcdf_files(path=nc_dir, domain=domain):
ds = Dataset(nc_dir + '/' + nc_file)
swe_in = ds.variables['SNOWNC'][0] * MM_TO_IN
lons = ds.variables['XLONG'][0]
lats = ds.variables['XLAT'][0]
init_time = parser.parse(ds.START_DATE.replace('_', ' '))
cycle = str(init_time.hour).zfill(2)
fhour = int(ds.variables['XTIME'][0] / 60)
fhour_str = str(fhour).zfill(2)
valid_time = init_time + timedelta(hours=fhour)
title = plot_title(init_time, valid_time, fhour, 'swe', 'danwrf', 'in')
print('saving swe', domain_name, cycle, fhour)
plot = SurfacePlot(lons,
lats,
swe_in,
extent=extent,
colormap=PRECIP_CMAP_DATA,
color_levels=PRECIP_CLEVS,
central_longitude=central_longitude,
labels=labels,
display_counties=True,
title=title)
plot.save_plot(
f'wrf_prod/images/{cycle}z/{domain_name}-{cycle}z-swe-{fhour_str}.png'
)
def update(self, position, wpm, volume):
if fps_on:
global frame_count
frame_count+=1
optimal_wpm = (MAXWPM + MINWPM)/2
wpm = realtime_interpreter.get_wpm() #grab wpm value
wpm = clamp(MAXWPM - wpm, MINWPM, MAXWPM) #clamp it
# print("Updating...")
# print("Current wpm (clamped): {}".format(wpm))
# print("Running average WPM: {}".format(float(sum(self.past_wpm[-6:-1]))/len(self.past_wpm[-6:-1])))
if enable_graphing:
self.past_wpm.append(wpm)
# self.past_volume.append(volume)
if fps_on:
self.pace_value['text'] = frame_count#'{} WPM'.format(int(wpm))
if enable_graphing:
self.wpm_average_history.append(float(sum(self.past_wpm[-3:-1]))/len(self.past_wpm[-3:-1]))
plot.save_plot(self.wpm_average_history[-10:-1], 'pace_graph.png', MAXWPM/ 2.0, MINWPM, MAXWPM)
# self.scrollb.set(.1, 0.8) #SET SCROLL
wpm_settings = plot.get_wpm_settings_outside(MAXWPM / 2.0, MINWPM, MAXWPM)
#print(wpm_settings)
if wpm > wpm_settings [0] and wpm / 2.0 <= wpm_settings [1]:
self.speed_tip['text'] = "Speak faster"
self.speed_tip.config(fg='#ff0000')
elif wpm >= wpm_settings [1] and wpm / 2.0 <= wpm_settings [2]:
self.speed_tip['text'] = "Speak a little slower"
self.speed_tip.config(fg='#c9cf00')
elif wpm >= wpm_settings [2] and wpm / 2.0 <= wpm_settings [3]:
self.speed_tip['text'] = "Good speed"
self.speed_tip.config(fg='#009800')
elif wpm >= wpm_settings [3] and wpm / 2.0 <= wpm_settings [4]:
self.speed_tip['text'] = "Speak a little faster"
self.speed_tip.config(fg='#c9cf00')
elif wpm >= wpm_settings [4] and wpm / 2.0 <= wpm_settings [5]:
self.speed_tip['text'] = "Speak a lot faster"
self.speed_tip.config(fg='#ff0000')
else:
self.speed_tip['text'] = "Please speak at a consistent pace"
self.speed_tip.config(fg='black')
# self.text.see(1)
# print(self.scrollb.get())
# os.system("python3 plot.py 0 150 300 {} volume_graph.png 140".format(str.join(',',past_wpm_str)))
# print("Done computing...")
img2 = ImageTk.PhotoImage(Image.open("pace_graph.png"))
self.Artwork.configure(image=img2)
self.Artwork.image = img2
num_words = realtime_interpreter.get_word_number()
# Update by word number
#print("hi____"+str(num_words))
self.bold_by_word_number(num_words)
variations = realtime_interpreter.get_variations()
for variation in variations:
self.highlight_by_word_number(variation.expected_index)
def update(self, position, wpm, volume):
# return None
# print("Current wpm: {}".format(wpm))
optimal_wpm = (MAXWPM + MINWPM)/2
wpm = realtime_interpreter.get_wpm() #grab wpm value
#process wpm value
wpm -= optimal_wpm
wpm *= 0.2
if wpm > 0:
wpm = abs(wpm ** 1.5)
else:
wpm = -abs(wpm ** 1.5)
wpm += optimal_wpm
wpm = clamp(MAXWPM - wpm, MINWPM, MAXWPM) #clamp it
# print("Updating...")
# print("Current wpm (clamped): {}".format(wpm))
# print("Running average WPM: {}".format(float(sum(self.past_wpm[-6:-1]))/len(self.past_wpm[-6:-1])))
self.past_wpm.append(wpm)
# self.past_volume.append(volume)
self.pace_value['text'] = ""#'{} WPM'.format(int(wpm))
self.wpm_average_history.append(float(sum(self.past_wpm[-3:-1]))/len(self.past_wpm[-3:-1]))
plot.save_plot(self.wpm_average_history[-10:-1], 'rect1.png', MAXWPM/ 2.0, MINWPM, MAXWPM)
# plot.save_plot(self.past_wpm, 'rect2.png', 5, 0, 10)
# self.scrollb.set(.1, 0.8)
wpm_settings = plot.get_wpm_settings_outside(MAXWPM / 2.0, MINWPM, MAXWPM)
#print(wpm_settings)
if wpm > wpm_settings [0] and wpm / 2.0 <= wpm_settings [1]:
self.speed_tip['text'] = "Speak faster"
self.speed_tip.config(fg='#ff0000')
elif wpm >= wpm_settings [1] and wpm / 2.0 <= wpm_settings [2]:
self.speed_tip['text'] = "Speak a little slower"
self.speed_tip.config(fg='#c9cf00')
elif wpm >= wpm_settings [2] and wpm / 2.0 <= wpm_settings [3]:
self.speed_tip['text'] = "Good speed"
self.speed_tip.config(fg='#009800')
elif wpm >= wpm_settings [3] and wpm / 2.0 <= wpm_settings [4]:
self.speed_tip['text'] = "Speak a little faster"
self.speed_tip.config(fg='#c9cf00')
elif wpm >= wpm_settings [4] and wpm / 2.0 <= wpm_settings [5]:
self.speed_tip['text'] = "Speak a lot faster"
self.speed_tip.config(fg='#ff0000')
else:
self.speed_tip['text'] = "Please speak at a consistent pace"
self.speed_tip.config(fg='black')
# self.text.see(1)
# print(self.scrollb.get())
# os.system("python3 plot.py 0 150 300 {} rect2.png 140".format(str.join(',',past_wpm_str)))
#
# print("Done computing...")
img2 = ImageTk.PhotoImage(Image.open("rect1.png"))
self.Artwork.configure(image=img2)
self.Artwork.image = img2
num_words = realtime_interpreter.get_word_number()
filename = sys.argv[1]
print "Got data file", filename
with open(filename, 'rb') as f:
r = pickle.load(f)
os.system('mkdir -p ' + r.output)
plot.plot_init()
for i in range(len(r.names)):
f, ax = plot.run_plot(r.X, r.Y, r.Zs[i], r.names[i])
ax.set_xlabel(r.xlabel)
ax.set_ylabel(r.ylabel)
if r.xticks != []:
ax.set_xticks(r.xticks)
if r.yticks != []:
ax.set_yticks(r.yticks)
if r.xticklabels != []:
ax.set_xticklabels(r.xticklabels, rotation=90, ha='right')
if r.yticklabels != []:
ax.set_yticklabels(r.yticklabels)
#xticklabels = ax.get_xticklabels()
#xticklabels[0].set_visible(False)
plot.save_plot(f, r.output + r.names[i])
def __init__(self, pool, args):
start = args.start
end = args.end
gain = args.gain
dwell = args.dwell
nfft = args.fft
lo = args.lo
index = args.index
remote = args.remote
directory, filename = os.path.split(args.file)
_null, ext = os.path.splitext(args.file)
self.stepsTotal = 0
self.steps = 0
self.spectrum = {}
self.settings = Settings(load=False)
error = None
if end <= start:
error = "Start should be lower than end"
elif dwell <= 0:
error = "Dwell should be positive"
elif nfft <= 0:
error = "FFT bins should be positive"
elif ext != ".rfs" and ext != ".csv":
error = "File extension should be .rfs or .csv"
else:
device = Device()
if remote is None:
self.settings.devices = get_devices()
count = len(self.settings.devices)
if index > count - 1:
error = "Device not found ({0} devices in total):\n".format(count)
for device in self.settings.devices:
error += "\t{0}: {1}\n".format(device.index, device.name)
else:
device.isDevice = False
url = urlparse('//' + remote)
if url.hostname is not None:
device.server = url.hostname
else:
error = "Invalid hostname"
if url.port is not None:
device.port = url.port
else:
device.port = 1234
self.settings.devices.append(device)
index = len(self.settings.devices) - 1
if error is not None:
print "Error: {0}".format(error)
exit(1)
if end - 1 < start:
end = start + 1
if remote is None:
gain = nearest(gain, self.settings.devices[index].gains)
self.settings.start = start
self.settings.stop = end
self.settings.dwell = calc_real_dwell(dwell)
self.settings.nfft = nfft
self.settings.devices[index].gain = gain
self.settings.devices[index].lo = lo
print "{0} - {1}MHz".format(start, end)
print "{0}dB Gain".format(gain)
print "{0}s Dwell".format(self.settings.dwell)
print "{0} FFT points".format(nfft)
print "{0}MHz LO".format(lo)
if remote is not None:
print remote
else:
print self.settings.devices[index].name
self.scan(self.settings, index, pool)
if ext == ".rfs":
scanInfo = ScanInfo()
scanInfo.setFromSettings(self.settings)
save_plot(directory, filename, scanInfo, self.spectrum)
else:
export_plot(directory, filename, self.spectrum)
print "Done"
comb = (lr, drop, reg, size)
histories[comb] = model.fit_generator(train_generator,
steps_per_epoch=TRAINING_STEPS,
epochs=EPOCHS,
validation_data=val_generator,
validation_steps=VALIDATION_STEPS,
callbacks=[es_val_loss])
# update best combination of hyperparameters based on validation loss
if minimum > min(histories[comb].history['val_loss']):
minimum = min(histories[comb].history['val_loss'])
best_comb = comb
# save plots
save_path = './plots/' + '_'.join(str(comb[i]) for i in range(len(comb))) + '.pdf'
save_plot(save_path, histories[comb].history)
# storying histories for later usages
with open('./histories/'+'_'.join(str(label) for label in comb), 'wb') as fd:
pickle.dump(histories[comb].history, fd)
print('Trained from scratch with: lr={}, dropout={}, regularization term={}, dense size={}'.
format(lr, drop, reg, size))
print('The best combination of hyperparameters is found to be: {}'.format(best_comb))
print('The corresponding metrics are: val_loss={} and val_categorical_accuracy={}'.
format(minimum, histories[best_comb].history['val_categorical_accuracy'][np.argmin(histories[best_comb].history['val_loss'])]))
lr, drop, reg, size = best_comb
### Train the best model ###
import plot as plt
import scipy.optimize as optimization
from numpy import *
k = 13
path = "h_125.dat"
data = array(plt.parse(path))
f = lambda x, a, b, c, d: a*exp(-b*x**c) + d
fig, ax = plt.new_plot(title="K-means, %d Clusters" % k,
xaxis="Time [ms]",
yaxis="K-means distortion")
p0 = [2.7, 1., 1.1, .25]
p, e = optimization.curve_fit(f, data[:,0], data[:,1], p0)
print p
X = linspace(0, max(data[:,0]), 100)
plt.dot(data[:,0]*.005, data[:,1], "H = 50 mT")
plt.add(X*.005, f(X, p[0], p[1], p[2], p[3]), "Exponential fit")
plt.save_plot(fig, ax, path[:-4]+".png")
import plot as plt
from numpy import *
k = 13
path = "h1000.dat"
data = array(plt.parse(path))
blur = 15
smoothed = array(data[:,1])
for i in range(blur, len(smoothed)-blur):
for j in range(-blur, blur):
smoothed[i] += data[i+j,1]
smoothed[i] /= (2*blur)
fig, ax = plt.new_plot(title="K-means, %d Clusters" % k,
xaxis="Time [ms]",
yaxis="K-means distortion")
plt.add(arange(len(smoothed))*.001, smoothed,
"K-Means Cluster Distortion Smoothed")
plt.dot(arange(len(data))*.001, data[:,1],
"K-Means Cluster Distortion")
plt.save_plot(fig, ax, "plt1.png")