def plot_percentiles(self):
if not self.train_outputs and not self.test_outputs:
print("You need to fetch data first using fetch_range() or fetch_last().")
return
num_rows = len(self.blob_percentiles_keys) + len(self.weight_percentiles_keys)
colormap = cm.brg(np.linspace(0.5, 0.8, self.num_percentiles))
colormap = [colormap[4], colormap[2], colormap[0], colormap[2], colormap[4]]
iterations = self.cache_log_train[:, 0].astype(int)
fig, axes = plt.subplots(num_rows, 2)
fig.set_size_inches(15, 2*num_rows, forward=True)
fig.tight_layout()
row = 0
for layer in self.blob_percentiles_keys:
blob_percentiles_data = np.array([x[layer][0] for x in self.cache_log_train[:, 3]])
blob_percentiles_diff = np.array([x[layer][1] for x in self.cache_log_train[:, 3]])
self.make_subplot(axes, row, 0, layer, iterations, blob_percentiles_data, colormap)
self.make_subplot(axes, row, 1, layer, iterations, blob_percentiles_diff, colormap)
row += 1
if layer in self.weight_percentiles_keys:
weight_percentiles_data = np.array([x[layer][0] for x in self.cache_log_train[:, 4]])
weight_percentiles_diff = np.array([x[layer][1] for x in self.cache_log_train[:, 4]])
self.make_subplot(axes, row, 0, layer, iterations, weight_percentiles_data, colormap, 'weights')
self.make_subplot(axes, row, 1, layer, iterations, weight_percentiles_diff, colormap, 'weights')
row += 1
def draw_multigraph(G, plt):
colors = cm.brg(np.linspace(0, 1, len(G.nodes()) + 1))
for i, sg in enumerate(G.nodes()):
xs = list()
ys = list()
for _, _, data in sg.edges_iter(data=True):
draw_path(data["path"], color=colors[i],
linewidth=10)
for node in sg.nodes():
xs.append(node.x)
ys.append(node.y)
plt.scatter(xs, ys, color=colors[i], s=350,
edgecolors="k", linewidth=2, zorder=10)
from matplotlib import pyplot as plt
import matplotlib.cm as cm
import sys
import pandas as pd
import numpy as np
file = sys.argv[1]
data = pd.read_csv(file)
# separate by min resample limit
# Good RS nd PSC
bouts = list(set(data.number_of_bout))
colors = cm.brg(np.linspace(0, 1, len(bouts)))
legend_series_list = []
for bout in bouts:
green_data = data.loc[(data.met_resamp_min >= data.min_resampled) &
(data.mean_rs_per_mrm >= data.min_times_resampled) & (data.number_of_bout == bout)
]
# Good RS only
blue_data = data.loc[(data.met_resamp_min >= data.min_resampled) &
(data.mean_rs_per_mrm < data.min_times_resampled) & (data.number_of_bout == bout)
]
# Good RSC only
yellow_data = data.loc[(data.met_resamp_min < data.min_resampled) &
(data.mean_rs_per_mrm >= data.min_times_resampled) & (data.number_of_bout == bout)
]
# No criteria met
red_data = data.loc[(data.met_resamp_min < data.min_resampled) &
(data.mean_rs_per_mrm < data.min_times_resampled) & (data.number_of_bout == bout)
]
def showGraph(self, selected):
f = plt.figure()
a = f.add_subplot(111)
a2 = a.twinx()
f.canvas.mpl_connect('pick_event', self.onpick)
# a tk.DrawingArea
self.canvas = FigureCanvasTkAgg(f, self.master)
self.canvas.show()
self.canvas.get_tk_widget().pack(side=LEFT, fill=BOTH, expand=1)
averageTimes = [[], []]
[colorTop, top] = self.getLengthAndColors()
colors = cm.brg(np.linspace(0, 1, colorTop))
# add half the color value
colorAdd = (colorTop-1)/2
SE = None
for i in range(10, top):
data = self.readData(i)
if data[0]:
# set marker size, extra large if selected
s = 30
if i == selected:
s = 50
SE, = a.plot((i, i), (100, 1100), 'r--', zorder=-1)
# totalLengths errorbar
x = [i] * len(data[0])
y = data[0]
a.errorbar(x, y, zorder=-1, c='k')
# totalLengths errorbar
x = i
y = (sum(data[0])/float(len(data[0])))
if i == 10:
TL = a.scatter(x, y, c=colors[data[3][0] + colorAdd], s=s, zorder=3)
else:
a.scatter(x, y, c=colors[data[3][0] + colorAdd], s=s, zorder=0, picker=True)
# save times
averageTimes[0].append(i)
averageTimes[1].append((sum(data[1])/float(len(data[1]))))
# constraint relaxation scatter
x = i
y = (sum(data[2])/float(len(data[2])))
CR = a.scatter(x, y, c="blue", s=s, zorder=0, picker=True)
# plot the average time
Ti, = a2.plot(averageTimes[0], averageTimes[1], c="#cc9900", zorder=2, picker=True)
plt.legend((TL, CR, Ti, SE),
('Total Length (more red -> lower successrate)', 'Theoretical Minimum', 'Time', 'Selected Netlist'),
scatterpoints=1,
loc='upper left',
ncol=1,
fontsize=8)
plt.title("Results of Random Netlists on Print " + self.printvar.get())
a.set_xlabel('Amount of connections')
a.set_ylabel('Average Total Length')
a2.set_ylabel('Average Time(s)')
a2.set_ylim(bottom=0)
"outputs_gc_mb/bins.txt",
np.transpose([
0.5 * (z0bins + zfbins), 0.5 * (zfbins - z0bins),
sphz_red(0.5 * (z0bins + zfbins))
]))
ib_off = 7
if plot_stuff:
plt.figure()
for i in nbins - 1 - np.arange(nbins):
if i >= ib_off:
nz = nz_bins[i]
ran = np.where(nz > 1E-3 * np.amax(nz))[0]
plt.plot(zarr[ran],
nz[ran],
color=cm.brg((nbins - ib_off - i + ib_off - 0.5) /
(nbins - ib_off + 0.)),
lw=2)
plt.plot(zarr, nzarr, 'k-', lw=2)
plt.ylim([0, 1.05 * np.amax(nzarr)])
plt.xlim([0, 1.2 * zedge_hi])
plt.xlabel('$z$', fontsize=18)
plt.ylabel('$N(z)\\,\\,[{\\rm arcmin}^{-2}]$', fontsize=18)
plt.savefig('../Draft/Figs/nz_lsst_mb.pdf', bbox_inches='tight')
data_dd = cgf.read_cls_class("outputs_gc_mb/run_dens_denscl.dat")
data_mm = cgf.read_cls_class("outputs_gc_mb/run_mb_mbcl.dat")
data_dm = cgf.read_cls_class("outputs_gc_mb/run_dens_mbcl.dat")
larr = np.arange(2001)[2:1001]
cl_dd = data_dd['cl_dd'][2:1001, ib_off:, ib_off:]
cl_mm = data_mm['cl_dd'][2:1001, ib_off:, ib_off:]
def comparison_table(adata,
name_keys,
group=None,
color_thresholds=None,
n_genes=70):
## Very early: Take list of keys for which to search adata.uns.
## Then build a table of all included genes, see which w
## Trick: Map only the top 20/30 or so, color green/yellow/orange/red for top 40/ 60 / below
##TODO: Add functionality for group, color thresholds.
## Assume all annotations have the same length
name_list = {}
for i, j in enumerate(name_keys):
name_list[i] = adata.uns[j]
# We dont need rank list as we assume that name list is ordered.
length = len(name_list[0])
width = len(name_list)
# we create one large table (still no memory issue (realistic max. 10000*10--> 1 Mbyte approx). Truncuate later
rank_table = length * np.ones((length * width, width))
# Create full name list
full_name_list = list()
n = -1
for key in name_list:
for k, l in enumerate(name_list[key]):
# Only plot for each group the top 20 genes to avoid that the table becomes to large. Max_n should be
# a parameter in the future
# Problem: We should add everything, but only plot a certain truncu
if l not in full_name_list:
full_name_list.append(l)
n = n + 1
m = n
else:
m = full_name_list.index(l)
rank_table[m, key] = k
# Create table with all entries
if max_n < n:
n = max_n
trunc_table = rank_table[0:n + 1, :]
# Do the colorings:
colors = trunc_table.copy()
for i in range(n + 1):
# Here, for now we use the convention that the respective levels are true if the minimum rank is larger
# than the specified number
top20 = True
top50 = True
top100 = True
for j in range(width):
if colors[i, j] >= 100:
top20 = top50 = top100 = False
elif colors[i, j] >= 51:
top20 = top50 = False
elif colors[i, j] >= 21:
top20 = False
# Now depending on the boolean values, define colors.
if top100 is False:
colors[i, :] = 0
elif top50 is False:
colors[i, :] = 0.5
elif top20 is False:
colors[i, :] = 0.8
else:
colors[i, :] = 1
fig, ax = plt.subplots(1, 1)
ax.table(cellText=trunc_table,
rowLabels=full_name_list[0:n + 1],
colLabels=name_keys,
cellColours=cm.brg(colors))
plt.tight_layout()
plt.show()
path = 'X:\\Images\\150824_JVZ20_wrt2GFP_250x250x20\\'
worm = 'C10'
cells1 = ['b.','6.','c.','5.','4.','1.','2.']
cell2 = '3.'
cside = 'L'
fig = plt.figure(figsize = (5.8,5.8))
# plt.title(cell1+'-'+cell2)
ax = fig.add_subplot(111)
ax.set_xlim((-10000,10000))
ax.set_ylim((-10000,10000))
colors = np.array( [cm.brg(int(i)) for i in np.arange(len(cells1))*254./(len(cells1)-1)] )
colors = colors[::-1]
### find deviation from V3
dev = []
for idx, cell1 in enumerate( cells1 ):
c1Raw, c2Raw, c1Filt, c2Filt = loadData(path, worm, cell1, cell2, cside)
dev.append( getDeviation( c1Raw, c2Raw ) )
### sort the data according to deviation from V3
cells1sorted = [x for (y,x) in sorted(zip(dev,cells1))][::-1]
devsorted = [y for (y,x) in sorted(zip(dev,cells1))][::-1]
for idx, cell1 in enumerate(cells1sorted):
def showGraph(self, selected):
f = plt.figure()
# a tk.DrawingArea
self.canvas = FigureCanvasTkAgg(f, self.master)
self.canvas.show()
self.canvas.get_tk_widget().pack(side=LEFT, fill=BOTH, expand=1)
a = f.add_subplot(111)
a2 = a.twinx()
def OnClick(event):
self.cons = int(round(event.xdata, 0))
self.showInfoAndGraph()
cid = f.canvas.mpl_connect('button_press_event', OnClick)
averageTimes = [[], []]
[self.colorTop, top] = self.getLengthAndColors()
colors = cm.brg(np.linspace(0, 1, self.colorTop * 2 + 1))
# add half the color value
colorAdd = ((self.colorTop * 2 + 1) - 1) / 2
SE = None
for i in range(10, top):
data = self.readData(i)
if data[0]:
# set marker size, extra large if selected
s = 30
if i == selected:
s = 50
SE, = a.plot((i, i), (100, 1100), 'r--', zorder=-1)
# totalLengths errorbar
x = [i] * len(data[0])
y = data[0]
a.errorbar(x, y, zorder=1, c='k')
# totalLengths errorbar
x = i
y = (sum(data[0]) / float(len(data[0])))
if i == 10:
TL = a.scatter(x,
y,
c=colors[data[3][0] + colorAdd],
s=s,
zorder=2)
else:
a.scatter(x,
y,
c=colors[data[3][0] + colorAdd],
s=s,
zorder=2)
# save times
averageTimes[0].append(i)
averageTimes[1].append((sum(data[1]) / float(len(data[1]))))
# constraint relaxation scatter
x = i
y = (sum(data[2]) / float(len(data[2])))
CR = a.scatter(x, y, c="blue", s=s, zorder=2)
# plot the average time
Ti, = a2.plot(averageTimes[0], averageTimes[1], c="#cc9900", zorder=2)
plt.legend((TL, CR, Ti, SE),
('Total Length (more red -> lower successrate)',
'Theoretical Minimum', 'Time', 'Selected Netlist'),
scatterpoints=1,
loc='upper left',
ncol=1,
fontsize=8)
plt.title("Results of Random Netlists on Print " + self.printvar.get())
a.set_xlabel('Amount of connections')
a.set_ylabel('Average Total Length')
a2.set_ylabel('Average Time(s)')
a.set_xlim(left=9.8)
a.set_xlim(right=50.2)
a2.set_ylim(bottom=0)
plt.ylabel("log Peak Temperature Ratio")
plt.xticks([0.1, 1], [-1, 0])
plt.yticks([0.1, 1], [-1, 0])
plt.legend(loc="upper left")
# contour
H, xedges, yedges = np.histogram2d(p21,
r21,
bins=100,
range=([10**-1.2,
10**0.7], [10**-1.2, 10**0.7]))
extent = [xedges[0], xedges[-1], yedges[0], yedges[-1]]
plt.contour(H / H.max() * 100,
levels=[8, 16, 32, 64, 96],
extent=extent,
colors=[cm.brg(i / 2.5)],
zorder=1,
linewidths=2.5,
alpha=1.0,
label=galname.replace("ngc", "NGC "))
histo.extend(p21 / r21)
r21_all.extend(r21)
p21_all.extend(p21)
r21err_all.extend(r21err)
p21err_all.extend(p21err)
plt.legend(low="lower right")
p2r = np.array(p21_all) / np.array(r21_all)
r21err_all = np.array(r21err_all)
list_3627 = list_master[1]
list_4321 = list_master[2]
np.savetxt(dir_proj + "eps/ngc0628_noise.txt", list_0628)
np.savetxt(dir_proj + "eps/ngc3627_noise.txt", list_3627)
np.savetxt(dir_proj + "eps/ngc4321_noise.txt", list_4321)
fig = plt.figure(figsize=(10, 10))
ax1 = fig.add_subplot(111)
plt.rcParams["font.size"] = 22
plt.subplots_adjust(bottom=0.10, left=0.19, right=0.99, top=0.90)
ax1.plot(list_0628[:, 0],
np.log10(list_0628[:, 1] * 1.222e6 / list_0628[:, 0]**2 /
115.27120**2),
"-",
color=cm.brg(0 / 2.5),
markeredgewidth=0,
markersize=10,
alpha=0.5,
lw=7,
label="NGC 0628 CO(1-0)")
ax1.plot(list_0628[:, 0],
np.log10(list_0628[:, 2] * 1.222e6 / list_0628[:, 0]**2 /
230.53800**2),
"--",
color=cm.brg(0 / 2.5),
markeredgewidth=0,
markersize=10,
alpha=0.5,
lw=7,
label="NGC 0628 CO(2-1)")