def plot_correlation_hist(data):
""" Plot histogram of all correlations
"""
# gather data
corrs = []
for raw_res, enh_list in data:
_, raw_mat, _ = raw_res
if not raw_mat is None:
raw_vec = extract_sig_entries(raw_mat)
corrs.extend(raw_vec)
for enh_res in enh_list:
_, enh_mat, _ = enh_res
if not enh_mat is None:
enh_vec = extract_sig_entries(enh_mat)
corrs.extend(enh_vec)
# plot result
fig = plt.figure()
plot_histogram(corrs, plt.gca())
plt.xlabel('simulated correlations')
fig.savefig('images/all_sim_corrs.pdf')
def plot_correlation_hist(data):
""" Plot histogram of all correlations
"""
# gather data
corrs = []
for raw_res, enh_list in data:
_, raw_mat, _ = raw_res
if not raw_mat is None:
raw_vec = extract_sig_entries(raw_mat)
corrs.extend(raw_vec)
for enh_res in enh_list:
_, enh_mat, _ = enh_res
if not enh_mat is None:
enh_vec = extract_sig_entries(enh_mat)
corrs.extend(enh_vec)
# plot result
fig = plt.figure()
plot_histogram(corrs, plt.gca())
plt.xlabel('simulated correlations')
fig.savefig('images/all_sim_corrs.pdf')
def do(syst, ax):
# data
single_run_matrices = []
for _ in trange(reps):
sol = solve_system(syst)
sol_extract = sol.T[int(len(sol.T)*3/4):]
single_run_mat = compute_correlation_matrix(np.array([sol_extract]))
if single_run_mat.shape == (4, 4):
single_run_mat = single_run_mat[:-1,:-1]
assert single_run_mat.shape == (3, 3)
single_run_matrices.append(single_run_mat)
single_run_matrices = np.asarray(single_run_matrices)
# plotting
cols = cycle(['b', 'r', 'g', 'c', 'm', 'y', 'k'])
for i, row in enumerate(single_run_matrices.T):
for j, series in enumerate(row):
if i == j: break
plot_histogram(
series[series!=1], ax,
label=r'$c_{{{},{}}}$'.format(i,j),
facecolor=next(cols), alpha=0.5,
bins=100)
def plot_correlation_histogram(motifs, data):
""" Plot histogram of all observed intensity correlations
"""
colors = itertools.cycle(['b', 'r', 'g', 'c', 'm', 'y', 'k'])
plt.figure()
for (m, lbl) in motifs:
# compute correlations
corrs = []
for cs in tqdm(m):
for c1 in cs:
for c2 in cs:
if c1 == c2: break
# compute correlations
for i, int1 in enumerate(data[c1]['intensities']):
for j, int2 in enumerate(data[c2]['intensities']):
cc, _ = scis.pearsonr(int1, int2)
corrs.append(cc)
# plot
plotter.plot_histogram(
corrs, plt.gca(),
facecolor=next(colors), alpha=0.5,
label=lbl)
plt.title('Comparison of intensity correlation distributions')
plt.xlabel('intensity vector correlation')
plt.legend(loc='best')
plt.tight_layout()
plotter.save_figure('images/rl_corr_hist.pdf', bbox_inches='tight')
def plot_all_correlations(cs, data, ax):
""" Plot correlations between all intensity combinations
"""
col_list = ['red', 'blue', 'green']
tmp = collections.defaultdict(lambda: collections.defaultdict(list))
for c1 in cs:
for c2 in cs:
if c1 == c2: break
corrs = {}
# compute correlations
for i, int1 in enumerate(data[c1]['intensities']):
for j, int2 in enumerate(data[c2]['intensities']):
cc, _ = scis.pearsonr(int1, int2)
corrs[(i,j)] = cc
c1_idx, c2_idx = max(corrs.keys(), key=lambda k: corrs[k])
tmp[c1][c1_idx].append(corrs[(c1_idx, c2_idx)])
tmp[c2][c2_idx].append(corrs[(c1_idx, c2_idx)])
# plot histogram
if not ax is None:
plotter.plot_histogram(
list(corrs.values()), ax,
alpha=0.5, facecolor=col_list.pop())
# choose final selection
sel = {}
for c, maps in tmp.items():
choice = max(maps.keys(), key=lambda k: sum(maps[k]))
sel[c] = choice
return sel, corrs.values()
def parse_XML(doc):
cards = doc.getroot()
# Create general transition matrix (Model 3)
# (general_mtx,row_code,col_code) = model3.create_transition_matrix(extract_history(cards), 1)
# Updating global variable for progress measuring purposes
global total
total = len(cards)
global counter
counter = Value('i', 0)
# Executes card history evaluation in parallel
pool = Pool(processes=8, initializer = init, initargs = (counter, ))
rs = pool.map_async(exec_task, extract_history(cards), chunksize = 1)
while(True):
if (rs.ready()):
break
remaining = rs._number_left
sys.stdout.write("\tCurrent progress: %.2f %% of cards analyzed\r" % (100*(total-remaining)/total) )
sys.stdout.flush()
time.sleep(0.1)
rs.wait()
results = rs.get()
# Configurations for evaluation matrix
m_min_history = 0
m_max_history = 500
delta = 25
levels = int(math.floor((m_max_history-m_min_history)/delta))
mtx = np.zeros((levels,levels), dtype=np.float) - 1
counts = np.zeros((levels,levels), dtype=np.float)
min_range = range(m_min_history,delta*levels+m_min_history,delta)
max_range = range(m_min_history+delta,delta*levels+m_min_history+delta,delta)
###
mtx = update_evaluation_matrix(mtx,counts,results,min_range,min_range)
sys.stdout.write("\tCurrent progress: %.2f %% of cards analyzed\r\n" % (100.00) )
sys.stdout.flush()
precision = [float(i[0]) for i in results]
print "Average precision: " , sum(precision)/len(precision) , " ( min-history = " , min(1,m_min_history) , ", max-history = " , m_max_history, " )"
plotter.plot_matrix(mtx,max_range,min_range,"Model " + modelName + " precision","../results/model_"+modelName+"_matrix.png")
plotter.plot_histogram(np.asarray(precision),"Model " + modelName + " precision","../results/model_"+modelName+"_histogram.png")
print "Evaluation completed!"
def plot_result(motifs, data, fname_app='', sub_num=10):
""" Create result plot
"""
mpl.style.use('default')
sub_num = min(sub_num, len(motifs))
print(' > Plotting results ({})'.format(fname_app[1:]))
# overview plots
plt.figure(figsize=(30, 4 * sub_num))
gs = mpl.gridspec.GridSpec(sub_num, 3, width_ratios=[1, 2, 1])
idx = map(int,
np.linspace(0, len(motifs), num=sub_num, endpoint=False))
corrs = []
for ai, i in tqdm(enumerate(idx), total=sub_num):
c1, c2, c3 = motifs[i]
# plot all possible correlations and select optimal one
sel, all_corrs = plot_all_correlations(
(c1, c2, c3), data,
plt.subplot(gs[ai, 2]) if ai < sub_num else None)
# get intensities
sols = []
for foo in [c1, c2, c3]:
sols.append(data[foo]['intensities'][sel[foo]])
# compute correlation matrix
corr_mat = get_correlation_matrix(sols)
corrs.extend(all_corrs)
# plot rest
plotter.plot_corr_mat(corr_mat, plt.subplot(gs[ai, 0]))
series_ax = plt.subplot(gs[ai, 1])
plotter.plot_system_evolution(
sols, series_ax,
xlabel='sample')
series_ax.set_title('{}\n{}\n{}'.format(c1, c2, c3))
plt.tight_layout()
plotter.save_figure('images/rl_motifs{}.pdf'.format(fname_app), bbox_inches='tight')
# correlation histogram
plt.figure()
plotter.plot_histogram(corrs, plt.gca())
plt.tight_layout()
plotter.save_figure('images/rl_corr_hist{}.pdf'.format(fname_app), bbox_inches='tight')
def compute_overview_histogram(corrs):
""" Compute histogram of all possible correlations
"""
flat_corrs = list(itertools.chain.from_iterable(corrs))
plt.figure()
plot_histogram(flat_corrs, plt.gca())
plt.xlabel('molecule intensity correlation')
plt.title('Overview over all correlations')
plt.savefig('images/all_corrs_hist.pdf')
plt.close()
return flat_corrs
def movement_monte_carlo_plot_movenents():
start_time = time.perf_counter()
height = 20
width = 20
iterations = 10000
min_moves = 10
max_moves = 20
success = 0
wrong = 0
fail = 0
required_moves = np.zeros(max_moves + 1, dtype=int)
for i in range(iterations):
if i % 100 == 0:
log.debug("Running iteration", i)
m = mapgrid.generate_map(width=width, height=height)
v = vehicle.Vehicle(m, 10)
num_matches, x, y, possible_loc, movements = \
locator.locate_with_movement_and_error_fallback(m, v, max_moves, min_moves=min_moves)
if num_matches == 1:
correct_x, correct_y, direction = v.location()
if x == correct_x and y == correct_y:
success += 1
required_moves[movements] += 1
else:
wrong += 1
else:
fail += 1
#print("Success:", success)
#print("Wrong:", wrong)
#print("Fail", fail)
#print("Required moves", required_moves)
#print("Total time", time.perf_counter() - start_time)
x_axis = np.arange(1, 8)
y_axis = (required_moves / iterations)[10:17]
plotter.plot_histogram(x_axis,
y_axis,
xlabel="lisäaskelten lukumäärä",
ylim=None,
savename="lisa-askelten_lkm")
def do_hist(name1, ints1, name2, ints2):
corrs = []
for i1, i2 in itertools.product(ints1, ints2):
corr, p = scits.pearsonr(i1, i2)
corrs.append(corr)
if plot:
plt.figure()
plot_histogram(corrs, plt.gca())
plt.title('"{} ({})" vs "{} ({})"'.format(
name2, len(ints2), name1, len(ints1)))
plt.xlabel('correlation')
plt.savefig('corr_hists/corr_hist_{}_{}.pdf'.format(name1, name2))
plt.close()
return corrs
def compute_correlation_matrix(data, plot_hist=False, save_stdev=None):
""" Compute correlation matrix of given data points
"""
dim = data.shape[2]
mats = []
for rep_slice in data:
mat = np.empty((dim, dim))
for i in range(dim):
for j in range(dim):
xs, ys = extract(i, j, rep_slice)
cc = get_correlation(xs, ys)
mat[i, j] = cc
mats.append(mat)
mats = np.array(mats)
if plot_hist:
plt.figure(figsize=(6, 14))
gs = mpl.gridspec.GridSpec(int((dim**2-dim)/2), 1)
axc = 0
for i in range(dim):
for j in range(dim):
if i == j: break
ax = plt.subplot(gs[axc])
plotter.plot_histogram(mats[:,i,j], ax)
ax.set_title('Nodes {}, {}'.format(i, j))
ax.set_xlabel('correlation')
axc += 1
plt.tight_layout()
plt.savefig('images/simulated_corrs_hist.pdf')
plt.close()
if save_stdev is not None:
std = np.std(mats, axis=0)
np.save(save_stdev, std)
res_mat = np.mean(mats, axis=0)
return res_mat
def compute_correlation_matrix(data, plot_hist=False, save_stdev=None):
""" Compute correlation matrix of given data points
"""
dim = data.shape[2]
mats = []
for rep_slice in data:
mat = np.empty((dim, dim))
for i in range(dim):
for j in range(dim):
xs, ys = extract(i, j, rep_slice)
cc = get_correlation(xs, ys)
mat[i, j] = cc
mats.append(mat)
mats = np.array(mats)
if plot_hist:
plt.figure(figsize=(6, 14))
gs = mpl.gridspec.GridSpec(int((dim**2 - dim) / 2), 1)
axc = 0
for i in range(dim):
for j in range(dim):
if i == j: break
ax = plt.subplot(gs[axc])
plotter.plot_histogram(mats[:, i, j], ax)
ax.set_title('Nodes {}, {}'.format(i, j))
ax.set_xlabel('correlation')
axc += 1
plt.tight_layout()
plt.savefig('images/simulated_corrs_hist.pdf')
plt.close()
if save_stdev is not None:
std = np.std(mats, axis=0)
np.save(save_stdev, std)
res_mat = np.mean(mats, axis=0)
return res_mat
def main(target, cfg):
"""
Puts everything together.
"""
t0 = time.time()
#####################################################################
#
# READ SEGY
#
#####################################################################
if cfg['segy_library'].lower() == 'obspy':
s = Seismic.from_segy_with_obspy(target, params={'ndim': cfg['ndim']})
else:
s = Seismic.from_segy(target, params={'ndim': cfg['ndim']})
# Set the line and/or xline number.
try:
n, xl = cfg['number']
except:
n, xl = cfg['number'], 0.5
# Set the direction.
if (s.ndim) == 2:
direction = ['inline']
elif cfg['direction'].lower()[0] == 'i':
direction = ['inline']
elif cfg['direction'].lower()[0] in ['x', 'c']: # Allow 'crossline' too.
direction = ['xline']
elif cfg['direction'].lower()[0] == 't':
direction = ['tslice']
else:
direction = ['xline', 'inline']
# Get the data.
try:
ss = [
Seismic.from_seismic(s, n=n, direction=d)
for n, d in zip((n, xl), direction)
]
except IndexError:
# Perhaps misinterpreted 2D as 3D
s = Seismic.from_segy(target, params={'ndim': 2})
direction = ['inline']
ss = [
Seismic.from_seismic(s, n=n, direction=d)
for n, d in zip((n, xl), direction)
]
clip_val = np.percentile(s.data, cfg['percentile'])
if clip_val < 10:
fstr = '{:.3f}'
elif clip_val < 100:
fstr = '{:.2f}'
elif clip_val < 1000:
fstr = '{:.1f}'
else:
fstr = '{:.0f}'
# Notify user of parameters.
Notice.info("n_traces {}".format(s.ntraces))
Notice.info("n_samples {}".format(s.nsamples))
Notice.info("dt {}".format(s.dt))
Notice.info("t_start {}".format(s.tstart))
Notice.info("t_end {}".format(s.tend))
Notice.info("max_val " + fstr.format(np.amax(s.data)))
Notice.info("min_val " + fstr.format(np.amin(s.data)))
Notice.info("clip_val " + fstr.format(clip_val))
t1 = time.time()
Notice.ok("Read data in {:.1f} s".format(t1 - t0))
#####################################################################
#
# MAKE PLOT
#
#####################################################################
Notice.hr_header("Plotting")
# Plot size parameters.
wsl = 6 # Width of sidelabel, inches
mih = 11 # Minimum plot height, inches
fhh = 5 # File header box height, inches
m = 0.75 # basic unit of margins, inches
# Margins, CSS like: top, right, bottom, left.
mt, mr, mb, ml = m, 1.5 * m, m, 1.5 * m
mm = 2 * m # padded margin between seismic and label
# Determine plot dimensions. Kind of laborious and repetitive (see below).
if cfg['plot_width']:
seismic_width = cfg['plot_width'] - wsl - mm - ml - mr
tpi = max([s.ntraces for s in ss]) / seismic_width
else:
tpi = cfg['tpi']
if cfg['plot_height']:
seismic_height = max(
mih, cfg['plot_height']) - mb - 0.75 * (len(ss) - 1) - mt
seismic_height_raw = seismic_height / len(ss)
ips = seismic_height_raw / (s.tbasis[-1] - s.tbasis[0])
else:
ips = cfg['ips']
# Width is determined by seismic width, plus sidelabel, plus margins.
# Height is given by ips, but with a minimum of mih inches.
if 'tslice' in direction:
seismic_width = [s.ntraces / tpi for s in ss]
seismic_height_raw = max([s.nxlines for s in ss]) / tpi
else:
seismic_width = [s.ntraces / tpi for s in ss]
seismic_height_raw = ips * (s.tbasis[-1] - s.tbasis[0])
w = ml + max(seismic_width) + mm + wsl + mr # inches
seismic_height = len(ss) * seismic_height_raw
h_reqd = mb + seismic_height + 0.75 * (len(ss) - 1) + mt # inches
h = max(mih, h_reqd)
# Calculate where to start sidelabel and seismic data.
# Depends on whether sidelabel is on the left or right.
if cfg['sidelabel'] == 'right':
ssl = (ml + max(seismic_width) + mm) / w # Start of side label (ratio)
seismic_left = ml / w
else:
ssl = ml / w
seismic_left = (ml + wsl + mm) / w
adj = max(0, h - h_reqd) / 2
seismic_bottom = (mb / h) + adj / h
seismic_width_fraction = [sw / w for sw in seismic_width]
seismic_height_fraction = seismic_height_raw / h
# Publish some notices so user knows plot size.
Notice.info("plot width {:.2f} in".format(w))
Notice.info("plot height {:.2f} in".format(h))
# Make the figure.
fig = plt.figure(figsize=(w, h), facecolor='w')
# Set the tickformat.
tickfmt = mtick.FormatStrFormatter('%.0f')
# Could definitely do better for default fontsize than 10.
# Ideally would be adaptive to plot size.
cfg['fontsize'] = cfg['fontsize'] or 10
# Plot title.
if cfg['title']:
# Deal with Windows paths: \1 gets interpreted as a group by regex.
newt = re.sub(r'\\', '@@@@@', target)
temp = re.sub(r'_filename', newt, cfg['title'])
title = re.sub(r'@@@', r'\\', temp)
title_ax = fig.add_axes([ssl, 1 - mt / h, wsl / w, mt / h])
title_ax = plotter.plot_title(title_ax,
title,
fs=1.4 * cfg['fontsize'],
cfg=cfg)
# Plot title.
if cfg['subtitle']:
date = str(datetime.date.today())
subtitle = re.sub(r'_date', date, cfg['subtitle'])
subtitle_ax = fig.add_axes([ssl, 1 - mt / h, wsl / w, mt / h],
label='subtitle')
title_ax = plotter.plot_subtitle(subtitle_ax,
subtitle,
fs=0.75 * cfg['fontsize'],
cfg=cfg)
# Plot text header.
start = (h - 1.5 * mt - fhh) / h
head_ax = fig.add_axes([ssl, start, wsl / w, fhh / h])
head_ax = plotter.plot_header(head_ax,
s.header,
fs=9,
cfg=cfg,
version=__version__)
# Plot histogram.
# Params for histogram plot.
pady = 0.75 / h # 0.75 inch
padx = 0.75 / w # 0.75 inch
cstrip = 0.3 / h # color_strip height = 0.3 in
charth = 1.25 / h # height of charts = 1.25 in
chartw = wsl / w - mr / w - padx # or ml/w for left-hand sidelabel; same thing
chartx = (ssl + padx)
histy = 1.5 * mb / h + charth + pady
# Plot colourbar under histogram.
clrbar_ax = fig.add_axes([chartx, histy - cstrip, chartw, cstrip])
clrbar_ax = plotter.plot_colourbar(clrbar_ax, cmap=cfg['cmap'])
# Plot histogram itself.
hist_ax = fig.add_axes([chartx, histy, chartw, charth])
hist_ax = plotter.plot_histogram(hist_ax, s.data, tickfmt, cfg)
# Plot spectrum.
specy = 1.5 * mb / h
spec_ax = fig.add_axes([chartx, specy, chartw, charth])
try:
colour = utils.rgb_to_hex(cfg['highlight_colour'])
spec_ax = s.plot_spectrum(
ax=spec_ax,
tickfmt=tickfmt,
ntraces=20,
fontsize=cfg['fontsize'],
colour=colour,
)
except:
pass # No spectrum, oh well.
for i, line in enumerate(ss):
# Add the seismic axis.
ax = fig.add_axes([
seismic_left,
seismic_bottom + i * seismic_height_fraction + i * pady,
seismic_width_fraction[i], seismic_height_fraction
])
# Plot seismic data.
if cfg['display'].lower() in ['vd', 'varden', 'variable', 'both']:
im = ax.imshow(line.data.T,
cmap=cfg['cmap'],
clim=[-clip_val, clip_val],
extent=[
line.olineidx[0], line.olineidx[-1],
1000 * line.tbasis[-1], line.tbasis[0]
],
aspect='auto',
interpolation=cfg['interpolation'])
if np.argmin(seismic_width) == i:
cax = utils.add_subplot_axes(ax, [1.01, 0.02, 0.01, 0.2])
_ = plt.colorbar(im, cax=cax)
if cfg['display'].lower() in ['wiggle', 'both']:
ax = line.wiggle_plot(
cfg['number'],
direction,
ax=ax,
skip=cfg['skip'],
gain=cfg['gain'],
rgb=cfg['colour'],
alpha=cfg['opacity'],
lw=cfg['lineweight'],
)
valid = ['vd', 'varden', 'variable', 'wiggle', 'both']
if cfg['display'].lower() not in valid:
Notice.fail("You must specify the display: wiggle, vd, both.")
return
# Seismic axis annotations.
ax.set_ylabel(utils.LABELS[line.ylabel], fontsize=cfg['fontsize'])
ax.set_xlabel(utils.LABELS[line.xlabel],
fontsize=cfg['fontsize'],
ha='center')
ax.tick_params(axis='both', labelsize=cfg['fontsize'] - 2)
ax.xaxis.set_major_formatter(tickfmt)
ax.yaxis.set_major_formatter(tickfmt)
if ('tslice' not in direction):
ax.set_ylim(1000 * cfg['trange'][1] or 1000 * line.tbasis[-1],
1000 * cfg['trange'][0])
# Crossing point. Will only work for non-arb lines.
try:
ax.axvline(ss[i - 1].slineidx[0],
c=utils.rgb_to_hex(cfg['highlight_colour']),
alpha=0.5)
except IndexError:
pass # Nevermind.
# Grid, optional.
if cfg['grid_time'] or cfg['grid_traces']:
ax.grid()
for l in ax.get_xgridlines():
l.set_color(utils.rgb_to_hex(cfg['grid_colour']))
l.set_linestyle('-')
if cfg['grid_traces']:
l.set_linewidth(1)
else:
l.set_linewidth(0)
l.set_alpha(min(1, cfg['grid_alpha']))
for l in ax.get_ygridlines():
l.set_color(utils.rgb_to_hex(cfg['grid_colour']))
l.set_linestyle('-')
if cfg['grid_time']:
if 'tslice' in direction:
l.set_linewidth(1)
else:
l.set_linewidth(1.4)
else:
l.set_linewidth(0)
l.set_alpha(min(1, 2 * cfg['grid_alpha']))
# Watermark.
if cfg['watermark_text']:
ax = plotter.watermark_seismic(ax, cfg)
# Make parasitic (top) axis for labeling CDP number.
if (s.data.ndim > 2) and ('tslice' not in direction):
ylim = ax.get_ylim()
par1 = ax.twiny()
par1.spines["top"].set_position(("axes", 1.0))
par1.plot(line.slineidx, np.zeros_like(line.slineidx), alpha=0)
par1.set_xlabel(utils.LABELS[line.slabel],
fontsize=cfg['fontsize'])
par1.set_ylim(ylim)
# Adjust ticks
tx = par1.get_xticks()
newtx = [
line.slineidx[len(line.slineidx) * (i // len(tx))]
for i, _ in enumerate(tx)
]
par1.set_xticklabels(newtx, fontsize=cfg['fontsize'] - 2)
t2 = time.time()
Notice.ok("Built plot in {:.1f} s".format(t2 - t1))
#####################################################################
#
# SAVE FILE
#
#####################################################################
Notice.hr_header("Saving")
dname, fname, ext = utils.path_bits(target)
outfile = cfg['outfile'] or ''
if not os.path.splitext(outfile)[1]:
outfile = os.path.join(cfg['outfile'] or dname, fname + '.png')
fig.savefig(outfile)
t3 = time.time()
Notice.info("output file {}".format(outfile))
Notice.ok("Saved output in {:.1f} s".format(t3 - t2))
if cfg['stain_paper'] or cfg['coffee_rings'] or cfg['distort'] or cfg[
'scribble']:
fname = os.path.splitext(outfile)[0] + ".stupid.png"
fig.savefig(fname)
else:
return
#####################################################################
#
# SAVE STUPID FILE
#
#####################################################################
Notice.hr_header("Applying the stupidity")
stupid_image = Image.open(fname)
if cfg['stain_paper']:
utils.stain_paper(stupid_image)
utils.add_rings(stupid_image, cfg['coffee_rings'])
if cfg['scribble']:
utils.add_scribble(stupid_image)
# Trick to remove remaining semi-transparent pixels.
result = Image.new("RGB", stupid_image.size, (255, 255, 255))
result.paste(stupid_image)
result.save(fname)
t4 = time.time()
Notice.info("output file {}".format(fname))
Notice.ok("Saved stupidity in {:.1f} s".format(t4 - t3))
return
def check_ergodicity(reps=500):
""" Check whether simulated systems are ergodic
"""
def get_matrices(syst, entry_num=100):
""" Get correlation matrices for both cases
"""
# multiple entries from single run
single_run_matrices = []
for _ in range(entry_num):
sol = solve_system(syst)
extract = sol.T[-entry_num:]
single_run_mat = compute_correlation_matrix(np.array([extract]))
single_run_matrices.append(single_run_mat)
avg_single_mat = np.mean(single_run_matrices, axis=0)
# one entry from multiple runs
multiple_runs = []
for _ in range(entry_num):
sol = solve_system(syst)
extract = sol.T[-1].T
multiple_runs.append(extract)
multiple_mat = compute_correlation_matrix(np.array([multiple_runs]))
return avg_single_mat, multiple_mat
syst = generate_basic_system()
single_runs = []
multiple_runs = []
for _ in trange(reps):
sm, rm = get_matrices(syst)
single_runs.append(sm)
multiple_runs.append(rm)
single_runs = np.array(single_runs)
multiple_runs = np.array(multiple_runs)
# plot result
dim = syst.jacobian.shape[1]
plt.figure(figsize=(6, 14))
gs = mpl.gridspec.GridSpec(int((dim**2-dim)/2), 1)
axc = 0
for i in range(dim):
for j in range(dim):
if i == j: break
ax = plt.subplot(gs[axc])
plot_histogram(
single_runs[:,i,j], ax,
alpha=0.5,
label='Multiple entries from single run')
plot_histogram(multiple_runs[:,i,j], ax,
facecolor='mediumturquoise', alpha=0.5,
label='One entry from multiple runs')
ax.set_title('Nodes {}, {}'.format(i, j))
ax.set_xlabel('correlation')
ax.legend(loc='best')
axc += 1
plt.tight_layout()
plt.savefig('images/ergodicity_check.pdf')
def main(target, cfg):
"""
Puts everything together.
"""
t0 = time.time()
#####################################################################
#
# READ SEGY
#
#####################################################################
s = Seismic.from_segy(target, params={'ndim': cfg['ndim']})
# Set the line and/or xline number.
try:
n, xl = cfg['number']
except:
n, xl = cfg['number'], 0.5
# Set the direction.
if (s.ndim) == 2:
direction = ['inline']
elif cfg['direction'].lower()[0] == 'i':
direction = ['inline']
elif cfg['direction'].lower()[0] == 'x':
direction = ['xline']
elif cfg['direction'].lower()[0] == 't':
direction = ['tslice']
else:
direction = ['inline', 'xline']
# Get the data.
ss = [Seismic.from_seismic(s, n=n, direction=d) for n, d in zip((n, xl), direction)]
data = [s.data for s in ss]
clip_val = np.percentile(s.data, cfg['percentile'])
# Notify user of parameters.
Notice.info("n_traces {}".format(s.ntraces))
Notice.info("n_samples {}".format(s.nsamples))
Notice.info("dt {}".format(s.dt))
Notice.info("t_start {}".format(s.tstart))
Notice.info("t_end {}".format(s.tend))
Notice.info("max_val {:.3f}".format(np.amax(s.data)))
Notice.info("min_val {:.3f}".format(np.amin(s.data)))
Notice.info("clip_val {:.3f}".format(clip_val))
t1 = time.time()
Notice.ok("Read data in {:.1f} s".format(t1-t0))
#####################################################################
#
# MAKE PLOT
#
#####################################################################
Notice.hr_header("Plotting")
# Plot size parameters.
fs = cfg['fontsize']
wsl = 6 # Width of sidelabel, inches
mih = 12 # Minimum plot height, inches
fhh = 5 # File header box height, inches
m = 0.5 # basic unit of margins, inches
# Margins, CSS like: top, right, bottom, left.
mt, mr, mb, ml = m, 2 * m, m, 2 * m
mm = m # padded margin between seismic and label
# Width is determined by seismic width, plus sidelabel, plus margins.
# Height is given by ips, but with a minimum of mih inches.
if 'tslice' in direction:
print('doing tslice')
seismic_width = max([s.ninlines for s in ss]) / cfg['tpi']
seismic_height_raw = max([s.nxlines for s in ss]) / cfg['tpi']
print(seismic_width, seismic_height_raw)
else:
seismic_width = max([s.ntraces for s in ss]) / cfg['tpi']
seismic_height_raw = cfg['ips'] * (s.tbasis[-1] - s.tbasis[0])
w = ml + seismic_width + mm + wsl + mr # inches
seismic_height = len(ss) * seismic_height_raw
h_reqd = mb + seismic_height + 0.75*(len(ss)-1) + mt # inches
h = max(mih, h_reqd)
# Calculate where to start sidelabel and seismic data.
# Depends on whether sidelabel is on the left or right.
if cfg['sidelabel'] == 'right':
ssl = (ml + seismic_width + mm) / w # Start of side label (ratio)
seismic_left = ml / w
else:
ssl = ml / w
seismic_left = (ml + wsl + mm) / w
adj = max(0, h - h_reqd) / 2
seismic_bottom = (mb / h) + adj / h
seismic_width_fraction = seismic_width / w
seismic_height_fraction = seismic_height_raw / h
# Publish some notices so user knows plot size.
Notice.info("Width of plot {} in".format(w))
Notice.info("Height of plot {} in".format(h))
# Make the figure.
fig = plt.figure(figsize=(w, h), facecolor='w')
# Set the tickformat.
tickfmt = mtick.FormatStrFormatter('%.0f')
# Plot title.
if cfg['filename']:
title_ax = fig.add_axes([ssl, 1-mt/h, wsl/w, mt/h])
title_ax = plotter.plot_title(title_ax, target, fs=1.5*fs, cfg=cfg)
# Plot text header.
start = (h - 1.5*mt - fhh) / h
head_ax = fig.add_axes([ssl, start, wsl/w, fhh/h])
head_ax = plotter.plot_header(head_ax, s.header, fs=fs-1, cfg=cfg)
# Plot histogram.
# Params for histogram plot.
pady = 0.75 / h # 0.75 inch
padx = 0.75 / w # 0.75 inch
cstrip = 0.3/h # color_strip height = 0.3 in
charth = 1.5/h # height of charts = 1.5 in
chartw = wsl/w - mr/w - padx # or ml/w for left-hand sidelabel; same thing
chartx = (ssl + padx)
histy = 1.5 * mb/h + charth + pady
# Plot colourbar under histogram.
clrbar_ax = fig.add_axes([chartx, histy - cstrip, chartw, cstrip])
clrbar_ax = plotter.plot_colourbar(clrbar_ax, cmap=cfg['cmap'])
# Plot histogram itself.
hist_ax = fig.add_axes([chartx, histy, chartw, charth])
hist_ax = plotter.plot_histogram(hist_ax,
s.data,
tickfmt,
percentile=cfg['percentile'],
fs=fs)
# Plot spectrum.
specy = 1.5 * mb/h
spec_ax = fig.add_axes([chartx, specy, chartw, charth])
try:
spec_ax = s.plot_spectrum(ax=spec_ax,
tickfmt=tickfmt,
ntraces=20,
fontsize=fs)
except:
pass
for i, line in enumerate(ss):
# Add the seismic axis.
ax = fig.add_axes([seismic_left,
seismic_bottom + i*seismic_height_fraction + i*pady,
seismic_width_fraction,
seismic_height_fraction
])
# Plot seismic data.
if cfg['display'].lower() in ['vd', 'varden', 'variable', 'both']:
_ = ax.imshow(line.data.T,
cmap=cfg['cmap'],
clim=[-clip_val, clip_val],
extent=[0,
line.ntraces,
1000*line.tbasis[-1],
line.tbasis[0]],
aspect='auto'
)
# This does not work: should cut off line at cfg['tmax']
# ax.set_ylim(1000*cfg['tmax'] or 1000*line.tbasis[-1], line.tbasis[0])
if cfg['display'].lower() in ['wiggle', 'both']:
ax = line.wiggle_plot(cfg['number'], direction,
ax=ax,
skip=cfg['skip'],
gain=cfg['gain'],
rgb=cfg['colour'],
alpha=cfg['opacity'],
lw=cfg['lineweight'],
tmax=cfg['tmax'],
)
if cfg['display'].lower() not in ['vd', 'varden', 'variable', 'wiggle', 'both']:
Notice.fail("You must specify the type of display: wiggle, vd, both.")
return
# Seismic axis annotations.
fs = cfg['fontsize'] - 2
ax.set_xlim([0, line.data.shape[0]])
ax.set_ylabel(line.ylabel, fontsize=fs)
ax.set_xlabel(line.xlabel, fontsize=fs, horizontalalignment='center')
ax.set_xticklabels(ax.get_xticks(), fontsize=fs)
ax.set_yticklabels(ax.get_yticks(), fontsize=fs)
ax.xaxis.set_major_formatter(tickfmt)
ax.yaxis.set_major_formatter(tickfmt)
# Watermark.
if cfg['watermark_text']:
ax = plotter.watermark_seismic(ax, cfg)
# Make parasitic axes for labeling CDP number.
par1 = ax.twiny()
par1.spines["top"].set_position(("axes", 1.0))
par1.plot(s.xlines, np.zeros_like(s.xlines))
par1.set_xlabel(line.xlabel, fontsize=fs)
par1.set_xticklabels(par1.get_xticks(), fontsize=fs)
par1.xaxis.set_major_formatter(tickfmt)
t2 = time.time()
Notice.ok("Built plot in {:.1f} s".format(t2-t1))
#####################################################################
#
# SAVE FILE
#
#####################################################################
Notice.hr_header("Saving")
dname, fname, ext = utils.path_bits(target)
outfile = cfg['outfile'] or ''
if not os.path.splitext(outfile)[1]:
outfile = os.path.join(cfg['outfile'] or dname, fname + '.png')
fig.savefig(outfile)
t3 = time.time()
Notice.ok("Saved image file {} in {:.1f} s".format(outfile, t3-t2))
if cfg['stain_paper'] or cfg['coffee_rings'] or cfg['distort'] or cfg['scribble']:
fname = os.path.splitext(outfile)[0] + ".stupid.png"
fig.savefig(fname)
else:
return
#####################################################################
#
# SAVE STUPID FILE
#
#####################################################################
Notice.hr_header("Applying the stupidity")
stupid_image = Image.open(fname)
if cfg['stain_paper']: utils.stain_paper(stupid_image)
utils.add_rings(stupid_image, cfg['coffee_rings'])
if cfg['scribble']: utils.add_scribble(stupid_image)
stupid_image.save(fname)
t4 = time.time()
Notice.ok("Saved stupid file {} in {:.1f} s".format(fname, t4-t3))
return
def check_ergodicity(reps=500):
""" Check whether simulated systems are ergodic
"""
def get_matrices(syst, entry_num=100):
""" Get correlation matrices for both cases
"""
# multiple entries from single run
single_run_matrices = []
for _ in range(entry_num):
sol = solve_system(syst)
extract = sol.T[-entry_num:]
single_run_mat = compute_correlation_matrix(np.array([extract]))
single_run_matrices.append(single_run_mat)
avg_single_mat = np.mean(single_run_matrices, axis=0)
# one entry from multiple runs
multiple_runs = []
for _ in range(entry_num):
sol = solve_system(syst)
extract = sol.T[-1].T
multiple_runs.append(extract)
multiple_mat = compute_correlation_matrix(np.array([multiple_runs]))
return avg_single_mat, multiple_mat
syst = generate_basic_system()
single_runs = []
multiple_runs = []
for _ in trange(reps):
sm, rm = get_matrices(syst)
single_runs.append(sm)
multiple_runs.append(rm)
single_runs = np.array(single_runs)
multiple_runs = np.array(multiple_runs)
# plot result
dim = syst.jacobian.shape[1]
plt.figure(figsize=(6, 14))
gs = mpl.gridspec.GridSpec(int((dim**2-dim)/2), 1)
axc = 0
for i in range(dim):
for j in range(dim):
if i == j: break
ax = plt.subplot(gs[axc])
plot_histogram(
single_runs[:,i,j], ax,
alpha=0.5,
label='Multiple entries from single run')
plot_histogram(multiple_runs[:,i,j], ax,
facecolor='mediumturquoise', alpha=0.5,
label='One entry from multiple runs')
ax.set_title('Nodes {}, {}'.format(i, j))
ax.set_xlabel('correlation')
ax.legend(loc='best')
axc += 1
plt.tight_layout()
plt.savefig('images/ergodicity_check.pdf')
if progress%1000==0:
sys.stdout.write("\tCurrent progress: %d cards grouped\r" % (progress) )
sys.stdout.flush()
frequencies = np.asarray(frequencies)
amounts = np.asarray(amounts)
frequencies = frequencies[frequencies<=500]
amounts = amounts[amounts<=25000]
sys.stdout.write("\tCurrent progress: %d cards grouped\r\n" % (len(frequencies)) )
sys.stdout.flush()
plotter.plot_histogram(frequencies,"History length","../results/history_length_histogram.png",50)
plotter.plot_histogram(amounts,"Mean purchase amount","../results/mean_amounts_histogram.png",50)
print "Mean purchase amount: " , np.mean(amounts)
print "Median purchase amount: " , np.median(amounts)
print "Number of cards: " , len(frequencies)
############################################################
def main(target, cfg):
"""
Puts everything together.
"""
t0 = time.time()
# Read the file.
section = readSEGY(target, unpack_headers=True)
# Calculate some things.
# NB Getting nsamples and dt from the first trace assumes that all
# traces are the same length, which is not a safe assumption in SEGY v2.
ninlines = section.traces[-1].header.trace_sequence_number_within_line
last_tr = section.traces[-1].header.trace_sequence_number_within_segy_file
nxlines = last_tr / ninlines
nsamples = section.traces[0].header.number_of_samples_in_this_trace
dt = section.traces[0].header.sample_interval_in_ms_for_this_trace
ntraces = len(section.traces)
tbase = 0.001 * np.arange(0, nsamples * dt, dt)
tstart = 0
tend = np.amax(tbase)
# Make the data array.
data = np.vstack([t.data for t in section.traces]).T
threed = False
if nxlines > 1: # Then it's a 3D and `data` is an ensemble.
threed = True
cube = np.reshape(data.T, (ninlines, nxlines, nsamples))
l = cfg['number']
if cfg['direction'].lower()[0] == 'i':
direction = 'inline'
ntraces = nxlines
l *= ninlines if (l < 1) else 1
data = cube[l, :, :].T
else:
direction = 'xline'
ntraces = ninlines
l *= nxlines if (l < 1) else 1
data = cube[:, l, :].T
# Collect some other data. Use a for loop because there are several.
elev, esp, ens, tsq = [], [], [], []
for i, trace in enumerate(section.traces):
elev.append(trace.header.receiver_group_elevation)
esp.append(trace.header.energy_source_point_number)
tsq.append(trace.header.trace_sequence_number_within_line)
if threed:
trs = []
if direction == 'inline':
cdp_label_text = 'Crossline number'
trace_label_text = 'Trace number'
ens.append(trace.header.for_3d_poststack_data_this_field_is_for_cross_line_number)
trs.append(trace.header.for_3d_poststack_data_this_field_is_for_in_line_number)
else:
cdp_label_text = 'Inline number'
trace_label_text = 'Trace number'
ens.append(trace.header.for_3d_poststack_data_this_field_is_for_in_line_number)
trs.append(trace.header.for_3d_poststack_data_this_field_is_for_cross_line_number)
line_no = min(trs)
else:
cdp_label_text = 'CDP number'
trace_label_text = 'Trace number'
ens.append(trace.header.ensemble_number)
min_tr, max_tr = 0, ntraces
traces = (min_tr, max_tr)
clip_val = np.percentile(data, cfg['percentile'])
# Notify user of parameters
Notice.info("n_traces {}".format(ntraces))
Notice.info("n_samples {}".format(nsamples))
Notice.info("dt {}".format(dt))
Notice.info("t_start {}".format(tstart))
Notice.info("t_end {}".format(tend))
Notice.info("max_val {}".format(np.amax(data)))
Notice.info("min_val {}".format(np.amin(data)))
Notice.info("clip_val {}".format(clip_val))
t1 = time.time()
Notice.ok("Read data in {:.1f} s".format(t1-t0))
#####################################################################
#
# MAKE PLOT
#
#####################################################################
Notice.hr_header("Plotting")
##################################
# Plot size parameters
# Some constants
fs = cfg['fontsize']
wsl = 6 # Width of sidelabel, inches
mih = 12 # Minimum plot height, inches
fhh = 5 # File header box height, inches
m = 0.5 # basic unit of margins, inches
# Margins, CSS like: top, right, bottom, left.
mt, mr, mb, ml = m, 2 * m, m, 2 * m
mm = m # padded margin between seismic and label
# Width is determined by seismic width, plus sidelabel, plus margins.
seismic_width = ntraces / cfg['tpi']
w = ml + seismic_width + mm + wsl + mr # inches
# Height is given by ips, but with a minimum of mih inches
seismic_height = cfg['ips'] * (tbase[-1] - tbase[0]) / 1000
h_reqd = mb + seismic_height + mt # inches
h = max(mih, h_reqd)
# Calculate where to start sidelabel and seismic data.
# Depends on whether sidelabel is on the left or right.
if cfg['sidelabel'] == 'right':
ssl = (ml + seismic_width + mm) / w # Start of side label (ratio)
seismic_left = ml / w
else:
ssl = ml / w
seismic_left = (ml + wsl + mm) / w
adj = max(0, h - h_reqd) / 2
seismic_bottom = (mb / h) + adj / h
seismic_width_fraction = seismic_width / w
seismic_height_fraction = seismic_height / h
# Publish some notices so user knows plot size.
Notice.info("Width of plot {} in".format(w))
Notice.info("Height of plot {} in".format(h))
##################################
# Make the figure.
fig = plt.figure(figsize=(w, h), facecolor='w')
# Add the main seismic axis.
ax = fig.add_axes([seismic_left,
seismic_bottom,
seismic_width_fraction,
seismic_height_fraction
])
# make parasitic axes for labeling CDP number
par1 = ax.twiny()
par1.spines["top"].set_position(("axes", 1.0))
tickfmt = mtick.FormatStrFormatter('%.0f')
par1.plot(ens, np.zeros_like(ens))
par1.set_xlabel(cdp_label_text, fontsize=fs-2)
par1.set_xticklabels(par1.get_xticks(), fontsize=fs-2)
par1.xaxis.set_major_formatter(tickfmt)
# Plot title
title_ax = fig.add_axes([ssl, 1-mt/h, wsl/w, mt/(h)])
title_ax = plotter.plot_title(title_ax, target, fs=1.5*fs, cfg=cfg)
if threed:
title_ax.text(0.0, 0.0, '{} {}'.format(direction.title(), line_no))
# Plot text header.
s = section.textual_file_header.decode()
start = (h - 1.5*mt - fhh) / h
head_ax = fig.add_axes([ssl, start, wsl/w, fhh/h])
head_ax = plotter.plot_header(head_ax, s, fs=fs-1, cfg=cfg)
# Plot histogram.
# Params for histogram plot
pady = 0.75 / h # 0.75 inch
padx = 0.75 / w # 0.75 inch
cstrip = 0.3/h # color_strip height = 0.3 in
charth = 1.5/h # height of charts = 1.5 in
chartw = wsl/w - mr/w - padx # or ml/w for left-hand sidelabel; same thing
chartx = (ssl + padx)
histy = 1.5 * mb/h + charth + pady
# Plot colourbar under histogram
clrbar_ax = fig.add_axes([chartx, histy - cstrip, chartw, cstrip])
clrbar_ax = plotter.plot_colourbar(clrbar_ax, cmap=cfg['cmap'])
# Plot histogram itself
hist_ax = fig.add_axes([chartx, histy, chartw, charth])
hist_ax = plotter.plot_histogram(hist_ax,
data,
tickfmt,
percentile=cfg['percentile'],
fs=fs)
# Plot spectrum.
specy = 1.5 * mb/h
spec_ax = fig.add_axes([chartx, specy, chartw, charth])
try:
spec_ax = plotter.plot_spectrum(spec_ax,
data,
dt,
tickfmt,
ntraces=20,
fontsize=fs)
except:
pass
# Plot seismic data.
if cfg['display'].lower() in ['vd', 'varden', 'variable']:
_ = ax.imshow(data,
cmap=cfg['cmap'],
clim=[-clip_val, clip_val],
extent=[0, ntraces, tbase[-1], tbase[0]],
aspect='auto'
)
elif cfg['display'].lower() == 'wiggle':
ax = plotter.wiggle_plot(ax,
data,
tbase,
ntraces,
skip=cfg['skip'],
gain=cfg['gain'],
rgb=cfg['colour'],
alpha=cfg['opacity'],
lw=cfg['lineweight']
)
ax.set_ylim(ax.get_ylim()[::-1])
elif cfg['display'].lower() == 'both':
# variable density goes on first
_ = ax.imshow(data,
cmap=cfg['cmap'],
clim=[-clip_val, clip_val],
extent=[0, ntraces, tbase[-1], tbase[0]],
aspect='auto'
)
# wiggle plots go on top
ax = plotter.wiggle_plot(ax,
data,
tbase,
ntraces,
skip=cfg['skip'],
gain=cfg['gain'],
rgb=cfg['colour'],
alpha=cfg['opacity'],
lw=cfg['lineweight']
)
# ax.set_ylim(ax.get_ylim()[::-1])
else:
Notice.fail("You need to specify the type of display: wiggle or vd")
# Seismic axis annotations.
ax = plotter.decorate_seismic(ax, traces, trace_label_text, tickfmt, cfg)
# Watermark.
if cfg['watermark_text']:
Notice.info("Adding watermark")
ax = plotter.watermark_seismic(ax, cfg)
t2 = time.time()
Notice.ok("Built plot in {:.1f} s".format(t2-t1))
#####################################################################
#
# SAVE FILE
#
#####################################################################
Notice.hr_header("Saving")
if cfg['stain_paper'] or cfg['coffee_rings'] or cfg['distort'] or cfg['scribble']:
stupid = True
else:
stupid = False
s = "Saved image file {} in {:.1f} s"
if cfg['outfile']:
if os.path.isfile(cfg['outfile']):
outfile = cfg['outfile']
else: # is directory
stem, ext = os.path.splitext(os.path.split(target)[1])
outfile = os.path.join(cfg['outfile'], stem + '.png')
stem, _ = os.path.splitext(outfile) # Needed for stupidity.
fig.savefig(outfile)
t3 = time.time()
Notice.ok(s.format(outfile, t3-t2))
else: # Do the default: save a PNG in the same dir as the target.
stem, _ = os.path.splitext(target)
fig.savefig(stem)
t3 = time.time()
Notice.ok(s.format(stem+'.png', t3-t2))
if stupid:
fig.savefig(stem + ".stupid.png")
else:
return
#####################################################################
#
# SAVE STUPID FILE
#
#####################################################################
Notice.hr_header("Applying the stupidity")
stupid_image = Image.open(stem + ".stupid.png")
if cfg['stain_paper']:
utils.stain_paper(stupid_image)
utils.add_rings(stupid_image, cfg['coffee_rings'])
if cfg['scribble']:
utils.add_scribble(stupid_image)
stupid_image.save(stem + ".stupid.png")
s = "Saved stupid file stupid.png in {:.1f} s"
t4 = time.time()
Notice.ok(s.format(t4-t3))
return
