def plot_features_train_val(paradigm, models, train, val):
"""
plot RF scores as function of the given key name
:param paradigm: the current paradigm name that is being tested (for the title)
:param models: all the models. list of tuples where the second index of the tuple is the model's name
:param train: list of train scores (same order as the models)
:param val: list of validation scores (same order as the models)
"""
title = "%s_train_validation_results" % paradigm
models_names = [model[1] for model in models] # gets the models name
# gets the data frame for the current plot
train = special_df(models_names, train, "train", FEATURES_NUMS, "features")
if len(train) <= 1:
# no data to plot (no 2 different values for x axis)
return
test = special_df(models_names, val, "val", FEATURES_NUMS, "features")
fig = plt.figure()
ax = fig.add_subplot(111)
# set the plot's colors so it will have different color for each model
ax.set_prop_cycle(color=COLORS[:test.shape[1]])
test.plot(ax=ax)
train.plot(ax=ax, style="--")
set_plt_params(ax, title.replace("_", " "), "Number of features",
"Accuracy")
# saves figure legend as a different file
fig_legend = pylab.figure(figsize=(4, 10))
pylab.figlegend(*ax.get_legend_handles_labels(), loc='center')
fig_legend.savefig(path.join(FIGS_DIR, title + '_legend.png'))
ax.get_legend().remove()
fig.savefig(path.join(FIGS_DIR, title + '.png'))
plt.close() # close the figure
def piechart(self, domain_type):
pie_file = self.output[0]
pylab.figure(figsize=(5, 5))
probs, labels = ([], [])
for type in DOMAIN_TYPES:
mean_coverage = mean(
domain_type[type]) if len(domain_type[type]) > 0 else 0.0
debug(self.family_name, type, mean_coverage)
probs.append(mean_coverage)
labels.append(type) #+":%0.2f" % mean_coverage)
#ax = pylab.axes([0.6, 0.6, 0.4, 0.4])
explode = [0.05 for i in xrange(len(DOMAIN_TYPES))]
patches, texts = pylab.pie(probs,
explode=None,
labels=None,
shadow=False,
colors=DOMAIN_COLORS)
pylab.figlegend(patches,
labels,
"lower left",
fancybox=True,
markerscale=0.2)
pylab.title(self.family_name)
pylab.savefig(pie_file)
def plot(x, y, yci, methods, title, xlabel, ylabel, filename):
fig = pylab.figure()
ax = pylab.gca()
for method in methods:
print method, y(method), yci(method)
lines = ax.errorbar(x(method),
y(method),
yci(method),
fmt=markers[method],
label=legends[method],
mfc='none',
markersize=15,
capsize=5)
# easiest way to plot brute force
"""
method = 'brute'
lines = ax.errorbar(range(1,4), [0,] * 3, [0,] * 3, fmt=markers[method], label=legends[method], mfc='none', markersize=15, capsize=5)
"""
ax.xaxis.set_major_formatter(FormatStrFormatter('%.0f'))
pylab.title(title)
pylab.xlabel(xlabel)
pylab.ylabel(ylabel)
#pylab.ylim([0, 0.2])
pylab.gcf().subplots_adjust(bottom=0.2, left=0.2)
fig.savefig(filename + ".pdf", dpi=300, format="pdf")
figLegend = pylab.figure(figsize=(6, 4))
pylab.figlegend(*ax.get_legend_handles_labels(), loc='upper left')
figLegend.savefig("legend.pdf", dpi=300, format="pdf")
pylab.close()
def plot(x, y, yci, methods, title, xlabel, ylabel, filename):
fig = pylab.figure()
ax = pylab.gca()
for method in methods:
print method, y(method), yci(method)
lines = ax.errorbar(x(method), y(method), yci(method), fmt=markers[method], label=legends[method], mfc='none', markersize=15, capsize=5)
# easiest way to plot brute force
"""
method = 'brute'
lines = ax.errorbar(range(1,4), [0,] * 3, [0,] * 3, fmt=markers[method], label=legends[method], mfc='none', markersize=15, capsize=5)
"""
ax.xaxis.set_major_formatter(FormatStrFormatter('%.0f'))
pylab.title(title)
pylab.xlabel(xlabel)
pylab.ylabel(ylabel)
#pylab.ylim([0, 0.2])
pylab.gcf().subplots_adjust(bottom=0.2, left=0.2)
fig.savefig(filename + ".pdf", dpi=300, format="pdf")
figLegend = pylab.figure(figsize = (6, 4))
pylab.figlegend(*ax.get_legend_handles_labels(), loc = 'upper left')
figLegend.savefig("legend.pdf", dpi=300, format="pdf")
pylab.close()
def findBasicLineData(data, coordPairs, bPlot = True) :
xdata =pylab.arange(0,len(data),1)
basicData = []
for lineRange in coordPairs :
#print 'findBasicLineData> lineRange', lineRange
xmean = (data[lineRange[0]:lineRange[1]]*xdata[lineRange[0]:lineRange[1]]).sum()/(data[lineRange[0]:lineRange[1]]).sum()
ymax = data[lineRange[0]:lineRange[1]].min()
#amplitude set as the minimum value between the defined coords
xwidth = lineRange[1]-lineRange[0]
basicData.append([xmean,ymax,xwidth])
if bPlot:
pylab.figure(20)
pylab.clf()
pylab.plot(data)
i=1
pylab.xlabel('Pixels')
pylab.ylabel('no. Photoelectrons')
pylab.title('Guess Parameters')
for d in basicData :
a=pylab.axvline(d[0],color='r',ls='--')
pylab.text(d[0],data.max(),i,fontsize=10)
b=pylab.axhline(d[1],color='k',ls='--')
i+=1
pylab.figlegend((a,b),('$\mu$','Amplitude'),'best',prop={'size':10})
return pylab.array(basicData)
def plot(self,yl='',yr='',x='s',idx=slice(None),clist='k r b g c m',lattice=True,newfig=True):
self._clist=clist.split()
if newfig:
f=_p.figure()
else:
f=_p.gcf()
sp=_p.subplot(111)
_p.subplots_adjust(right=0.72)
_p.setp( sp.yaxis, visible=False)
xd=getattr(self,x)[idx]
out=qdplot(figure=f,subplot=sp,plots=[],lines=[],legends=[],xaxis=xd)
if lattice:
self._lattice(out,idx,['kn0l','angle'],"#a0ffa0",'Bend h')
self._lattice(out,idx,['ks0l'],"#ffa0a0",'Bend v')
self._lattice(out,idx,['kn1l','k1l'],"#a0a0ff",'Quad')
self._lattice(out,idx,['hkick'],"#e0a0e0",'Kick h')
self._lattice(out,idx,['vkick'],"#a0e0e0",'Kick v')
for i in yl.split():
self._column(out,idx,i,'left')
for i in yr.split():
self._column(out,idx,i,'right')
_p.xlabel(_mylbl(axlabel,x))
_p.xlim(min(xd),max(xd))
_p.figlegend(out.lines,out.legends,'upper right')
_p.grid()
_p.draw()
del self._clist
if hasattr(self,'_colAxleft'): delattr(self,'_colAxleft')
if hasattr(self,'_colAxright'): delattr(self,'_colAxright')
self.currplot=out
return out
def update(self):
if callable(self.pre):
self.pre()
_p.ioff()
self.lines=[]
self.legends=[]
# self.figure.lines=[]
# self.figure.patches=[]
# self.figure.texts=[]
# self.figure.images = []
self.figure.legends = []
if self.lattice:
self.lattice.patches=[]
self._lattice(['k0l','kn0l','angle'],"#a0ffa0",'Bend h')
self._lattice(['ks0l'],"#ffa0a0",'Bend v')
self._lattice(['kn1l','k1l'],"#a0a0ff",'Quad')
self._lattice(['hkick'],"#e0a0e0",'Kick h')
self._lattice(['vkick'],"#a0e0e0",'Kick v')
if self.left:
self.left.lines=[]
for i in self.yl:
self._column(i,self.left,self.color[i])
if self.right:
self.right.lines=[]
for i in self.yr:
self._column(i,self.right,self.color[i])
_p.xlabel(_mylbl(axlabel,self.x))
self.figure.gca().set_xlim(min(self.xaxis[self.idx]),max(self.xaxis[self.idx]))
_p.figlegend(self.lines,self.legends,'upper right')
_p.grid(True)
# self.figure.canvas.mpl_connect('button_release_event',self.button_press)
# self.figure.canvas.mpl_connect('pick_event',self.pick)
_p.ion()
_p.draw()
def plot_all_shapelets(result):
f, rows = plt.subplots(5, 2, sharex=True)
axs = list(rows[:, 0]) + list(rows[:, 1])
lines = dict()
classifiers = dict()
for i, (label, (classifier, _)) in enumerate(result.items()):
classifiers[label] = classifier
order = [("wipe", "wipe_start"), ("force_inc", "force_inc"),
("slide", "slide_left_start"), ("slide_r", "slide_right_start"),
("push", "movable_box"), ("wipe_end", "wipe_end"),
("force_dec", "force_dec"), ("slide_end", "slide_left_end"),
("slide_r_end", "slide_right_end"), ("screw", "fixed_screw")]
j = 0
for i, (label, title) in enumerate(order):
if classifiers.has_key(label):
classifier = classifiers[label]
axs[j].set_title(title)
lines.update(
plot_shapelet(axs[j], classifier.shapelet, classifier.dim_s))
axs[j].set_xlim(0, 50)
plt.setp(axs[j].get_yticklabels(), visible=False)
j += 1
classifiers = sorted(lines.items(), key=lambda x: x[0])
plt.figlegend([x[1] for x in classifiers], [x[0] for x in classifiers],
loc='lower center',
ncol=5,
labelspacing=0.)
plt.subplots_adjust(left=.05,
bottom=.15,
right=.95,
top=.93,
wspace=.12,
hspace=.51)
plt.show()
def plot_train_test(general_results, general_train, mode):
for feature_set in FEATURES_FILES:
name = feature_set[0][:-4]
title = "%strain_test_results_%s" % (mode, name)
test = prepare_train_test_df(general_results, name, "test")
train = prepare_train_test_df(general_train, name, "train")
fig = plt.figure()
ax = fig.add_subplot(111)
# setting the amount of colors in the plot to be the amount of models
# cm = plt.rcParams['axes.prop_cycle'].by_key()['color']
# cm = list(mcolors.XKCD_COLORS.values())
# ax.set_color_cycle(cm[::len(cm) // (test.shape[1] - 1)])
ax.set_color_cycle(COLORS[:test.shape[1]])
test.plot(ax=ax)
train.plot(ax=ax, style="--")
set_plt_params(ax,
title.replace("_", " "),
"Number of features",
"Accuracy score",
width=WIDE_WIN)
if DEBUG:
plt.show()
else:
# saves figure legend as a different file
fig_legend = pylab.figure(figsize=(4, 10))
pylab.figlegend(*ax.get_legend_handles_labels(), loc='center')
fig_legend.savefig(
path.join(OUTPUT_PATH, FIGS_DIR, title + '_legend.png'))
ax.get_legend().remove()
fig.savefig(path.join(OUTPUT_PATH, FIGS_DIR, title + '.png'))
plt.close() # close the figure
def plot_results(t_hr, I_mA, avg_t_hr, lifetime_hr, lifetime_error = None,
show_plot = False):
""""""
# Set show_plot to True to stop the code here for examining the plot
if show_plot:
fig = plt.figure()
ax1 = fig.add_subplot(111)
h1 = ax1.plot(t_hr, I_mA, '.-', label='Data', linewidth=0.5)
ax1.set_xlabel('Time [hr]')
ax1.set_ylabel('Beam Current [mA]')
ax2 = ax1.twinx()
if lifetime_error is not None :
h2 = ax2.errorbar(avg_t_hr, lifetime_hr,
yerr = lifetime_error,
fmt = '-', ecolor = 'r',
color = 'r', linestyle = '-', marker = '.',
label = 'Lifetime')
else :
h2 = ax2.plot(avg_t_hr, lifetime_hr,
color = 'r', linestyle = '-', marker = '.',
label = 'Lifetime')
ax2.set_ylabel('Lifetime [hr]')
plt.figlegend( (h1[0], h2[0]),
(h1[0].get_label(), h2[0].get_label()),
loc='upper right')
plt.show()
def plotMatches(imgsources, refsources, matches, wcs, W, H, prefix,
saveplot=True, format='png'):
plt.clf()
# Image sources
ix = np.array([s.getXAstrom() for s in imgsources])
iy = np.array([s.getYAstrom() for s in imgsources])
iflux = np.array([s.getPsfFlux() for s in imgsources])
I = np.argsort(-iflux)
# First 200: red dots
II = I[:200]
p1 = plt.plot(ix[II], iy[II], 'r.', zorder=10)
# Rest: tiny dots
II = I[200:]
p2 = plt.plot(ix[II], iy[II], 'r.', markersize=1, zorder=9)
# Ref sources:
# Only getRa() (not getRaAstrom(), getRaObject()) is non-zero.
rx,ry = [],[]
for r in refsources:
xy = wcs.skyToPixel(r.getRaDec())
rx.append(xy[0])
ry.append(xy[1])
rx = np.array(rx)
ry = np.array(ry)
p3 = plt.plot(rx, ry, 'bo', mec='b', mfc='none', markersize=6, zorder=20)
x,y = [],[]
dx,dy = [],[]
for m in matches:
x0,x1 = m.first.getXAstrom(), m.second.getXAstrom()
y0,y1 = m.first.getYAstrom(), m.second.getYAstrom()
#plt.plot([x0, x1], [y0, y1], 'g.-')
x.append(x0)
y.append(y0)
dx.append(x1-x0)
dy.append(y1-y0)
#plt.plot(x, y, 's', mec='g', mfc='none', markersize=5)
p4 = plt.plot(x, y, 'o', mec='g', mfc='g', alpha=0.5, markersize=8, zorder=5)
p5 = plt.quiver(x, y, dx, dy, angles='xy', scale=30., zorder=30)
plt.axis('scaled')
plt.axis([0, W, 0, H])
#print p1, p2, p3, p4, p5
plt.figlegend((p1, p2, p3, p4), #, p5),
('Image sources (brightest 200)',
'Image sources (rest)',
'Reference sources',
'Matches',),
'center right',
numpoints=1,
prop=FontProperties(size='small'))
fn = prefix + '-matches.' + format
return _output(fn, format, saveplot)
def plot_circles(optimal_choices, labels, num_col, title, dir_path, file_name, color_set,
b_nach=(0.5, -0.05), legend_loc=9, print_title=True, print_legend=True):
legend_labels = []
legend_markers = []
for i in range(0, len(labels)):
legend_labels.append(labels[i])
legend_markers.append(patches.Circle((0.5, 0.5), 0.25, facecolor=color_set[i],
edgecolor=color_set[i], linewidth=1))
fig = plt.figure()
ax = fig.add_subplot(111, aspect='equal')
if print_title is True:
# LaTeX rendering case. You may use the next line if you have LaTeX installed.
# plt.title(r'Classification Model Recommendation vs.\ \textsc{' + title.title() + '} Data Stream Over Time'
# + '\n' + r'(\textit{from top-left corner to bottom-right corner})', fontsize=8, loc='center')
plt.title('Classification Model Recommendation vs. ' + title.title() + ' Data Stream Over Time'
+ '\n' + 'from top-left corner to bottom-right corner', fontsize=8, loc='center')
ax.set_xlabel(r'$\rightarrow$')
ax.set_ylabel(r'$\downarrow$')
ax.set_xticklabels([])
ax.set_yticklabels([])
x = 0.0125
y = 0.9875
l = 0.0125
for i in range(0, len(optimal_choices)):
ax.add_patch(patches.Circle((x, y), 0.011, color=color_set[optimal_choices[i][0]]))
if x > 1 - 2 * l:
x = 0.0125
y -= 2 * l
y = round(y, 4)
else:
x += 2 * l
x = round(x, 4)
file_path = (dir_path + file_name + "_" + title).lower()
if print_legend is True:
ax.legend(legend_markers, legend_labels, bbox_to_anchor=b_nach, loc=legend_loc,
fontsize=8, ncol=num_col, frameon=True, framealpha=1,
handler_map={patches.Circle: HandlerCircle()})
fig.savefig(file_path + '_circle.png', dpi=150, bbox_inches='tight')
fig.savefig(file_path + '_circle.pdf', dpi=150, bbox_inches='tight')
else:
fig_leg = pylab.figure(figsize=(13.5, 3.5), dpi=150)
pylab.figlegend(legend_markers, legend_labels, loc='center', fontsize=14, ncol=num_col, framealpha=1,
handler_map={patches.Circle: HandlerCircle()})
fig_leg.savefig(file_path + "_circle_legend.pdf", dpi=150)
fig.savefig(file_path + "_circle.pdf", dpi=150, bbox_inches='tight')
fig.savefig(file_path + "_circle.png", dpi=150, bbox_inches='tight')
def plot(cluster, filename="plot.png", func=lambda a: a.id,
plot_title='', cmap='Paired', filter=lambda a: True,
draw_legend=False, radius='2.25', sym=None):
ac = rc.load(cluster)
clusters = rc.load_clusters(cluster)
p = get_population()
pops = [0 for i in range(30000)]
cluster_pops = []
cluster_pop_max = []
for clust in range(len(clusters)):
cluster_pops.append(pops[:])
cluster_pop_max.append([0,0,-1,0])
for clust,agents in enumerate(clusters):
for agent in agents:
a = Agent(agent)
for i in range(a.birth, a.death):
cluster_pops[clust][i] += 1
if cluster_pop_max[clust][2] == -1:
cluster_pop_max[clust][2] = i
if i > cluster_pop_max[clust][3]:
cluster_pop_max[clust][3] = i
if cluster_pops[clust][i] > cluster_pop_max[clust][0]:
cluster_pop_max[clust][0] = cluster_pops[clust][i]
cluster_pop_max[clust][1] = i
lines=[]
for i,clust in enumerate(cluster_pops):
lines.append(pylab.plot(range(30000),clust,
label=("%d: k=%d" % (i, len(clusters[i]))),
color=pylab.cm.Paired(float(i)/len(clusters))))
if draw_legend:
pylab.figlegend(lines, ["%d: k=%d" % (i, len(clust))
for i,clust in enumerate(clusters)],
'center right',
ncol=((len(clusters)/35)+1), prop=dict(size=6))
else:
print "should not draw!!!"
title = r"Cluster Population ($\epsilon$ = %s, %d clusters)" % (radius, len(clusters))
pylab.title(title, weight='black')
pylab.xlabel("Time", weight='bold')
pylab.ylabel("Population Size", weight='bold')
if sym is not None:
pylab.figtext(0,.954, '(%s)' % sym, size=6, weight='black')
pylab.savefig(filename, dpi=300)
print 'cluster, totalPop, start, peak, stop, maxPop'
for clust,agents in enumerate(clusters):
print clust, len(agents), cluster_pop_max[clust][2], cluster_pop_max[clust][1], cluster_pop_max[clust][3]+1, cluster_pop_max[clust][0]
def figure2 ( ):
w,h = 20,7
fig = pl.figure ( figsize=(fullwidth,h*fullwidth/w) )
a,b = place_axes ( fig, 1.1,1, [7,7,5], [6,6,0], [True,False],
[.5,.5,1], (w,h), twiny=True )
kl = convenience.kernelplot ( data, results, plotinfo, a, b, None )
textfile.write ( "Figure 2\n" )
M = results['model_w_hist']
hr = data.gethistorykernel ( M.w[data.hf0:data.hf0+data.hlen], M.w[1] )
hz = data.gethistorykernel ( M.w[data.hf0+data.hlen:], M.w[1] )
bootstrap = results['bootstrap']
kernellen = (bootstrap.shape[1]-1)/2
ci_stim,ci_resp,ci_corr,ci_inco = statistics.history_kernel_ci (
bootstrap[:,kernellen:-2], bootstrap[:,:kernellen],
hz, hr )
textfile.write ( " stim[-1] = %g in (%g,%g)\n" % (hz[0], ci_stim[0,0], ci_stim[1,0]) )
textfile.write ( " resp[-1] = %g in (%g,%g)\n" % (hr[0], ci_resp[0,0], ci_resp[1,0]) )
textfile.write ( " correct[-1] = %g in (%g,%g)\n" % (hz[0]+hr[0], ci_corr[0,0], ci_corr[1,0]) )
textfile.write ( " incorrect[-1] = %g in (%g,%g)\n" % (-hz[0]+hr[0], ci_inco[0,0], ci_inco[1,0]) )
a.text ( .05, laby, r"{\bf a}", transform=a.transAxes )
b.text ( .05, laby, r"{\bf b}", transform=b.transAxes )
pl.setp ( (a,b), title="", xlabel="Lag" )
pl.setp ( b, yticks=(), ylabel="" )
pl.setp ( a, ylabel="weight" )
a.xaxis.set_major_formatter ( myformatter )
b.xaxis.set_major_formatter ( myformatter )
a.yaxis.set_major_formatter ( myformatter )
handles = []
annotations = []
for l in kl:
handles.append ( l )
annotations.append ( l.get_label() )
assert len(handles)==4
pl.figlegend ( handles, annotations,
'lower left', bbox_to_anchor=(15./w,2./h),
numpoints=1,
title='Weights of\nprevious...')
# mx = abs ( pl.array([ci_inco, ci_stim, ci_resp, ci_corr]) ).max( )
# mx = 0.5*10**pl.ceil(pl.log10(2*mx))
# pl.setp ( (a,b), ylim = (-mx,mx) )
a.yaxis.set_major_locator ( tckr ( density=0.5, figure=fig, which=1 ) )
b.yaxis.set_visible ( False )
if observer in ['sim_KP','sim_KP_nh']:
pl.setp ( (a,b), ylim = (-1,1) )
pl.savefig ( "figures/%s2.pdf" % (figname,) )
pl.savefig ( "figures/%s2.eps" % (figname,) )
def plot(cluster_file,
filename="plot.png",
func=lambda a: a.id,
plot_title='',
cmap='Paired',
filter=lambda a: True,
draw_legend=False,
radius='2.25',
sym=None,
run_dir='../run/'):
"""
Creates a line plot showing population per cluster
"""
# retrieve cluster data from cluster file
clusters = rc.load_clusters(cluster_file)
# grab cluster population
p = get_population(run_dir=run_dir)
lines = []
for cluster, agents in enumerate(clusters):
pop_by_time = list(repeat(0, 30000))
for agent in agents:
a = Agent(agent)
for i in range(a.birth, a.death):
pop_by_time[i] += 1
lines.append(
pylab.plot(range(30000),
pop_by_time,
label=("%d: k=%d" % (i, len(clusters[cluster]))),
color=pylab.cm.Paired(float(cluster) / len(clusters))))
if draw_legend:
pylab.figlegend(lines, [
"%d: k=%d" % (i, len(cluster))
for i, cluster in enumerate(clusters)
],
'center right',
ncol=((len(clusters) / 35) + 1),
prop=dict(size=6))
title = r"Cluster Population ($\epsilon$ = %s, %d clusters)" % (
radius, len(clusters))
pylab.title(title, weight='black')
pylab.xlabel("Time", weight='bold')
pylab.ylabel("Population Size", weight='bold')
if sym is not None:
pylab.figtext(0, .954, '(%s)' % sym, size=6, weight='black')
pylab.savefig(filename, dpi=300)
def plot_graph(processes, series_filter, save_file_path=None):
"""
Create a graph. Either display the graph or save the
graph. This function will only create ONE graph.
processes -- a list of Process instance.
save_file_path -- If not None, save the graph to the file path
specified by this parameter (a string).
"""
g.figure(figsize=(12, 7))
handles = []
labels = []
sizer = Sizer(len(series_filter.series))
for p in processes:
time_series = make_time_series(p.time_series)
colour = COLOURS.next()
print p.id, colour
for count, series in enumerate(series_filter.series):
ax = g.axes(sizer.coordinates(count))
try:
h = g.plot(time_series,
p.get(series.name),
series.marker,
color=colour,
label=p.id)
except:
print "Error in generating graph for %s of %s!" \
% (p.id, series.name)
raise
g.ylabel("%s %s" % (series.name, series.unit))
g.grid(True)
g.setp(ax.get_xticklabels(), visible=False)
handles.append(h)
labels.append(p.id)
g.setp(ax.get_xticklabels(), visible=True)
g.xlabel("Elapsed Time (min)")
legend = g.figlegend(handles,
labels,
'lower right',
pad=0.1,
labelsep=0.0025,
handlelen=0.025,
handletextsep=0.01,
axespad=0.08)
for t in legend.get_texts():
t.set_fontsize(8)
if save_file_path != None:
g.savefig(save_file_path, orientation='landscape')
g.close("all")
else:
g.show()
def piechart(self, domain_type):
pie_file = self.output[0]
pylab.figure(figsize=(5,5))
probs,labels = ([],[])
for type in DOMAIN_TYPES:
mean_coverage = mean(domain_type[type]) if len(domain_type[type])>0 else 0.0
debug(self.family_name,type,mean_coverage)
probs.append(mean_coverage)
labels.append(type)#+":%0.2f" % mean_coverage)
#ax = pylab.axes([0.6, 0.6, 0.4, 0.4])
explode = [0.05 for i in xrange(len(DOMAIN_TYPES))]
patches, texts = pylab.pie(probs,explode=None,labels=None,shadow=False,colors=DOMAIN_COLORS)
pylab.figlegend(patches, labels, "lower left", fancybox=True, markerscale=0.2)
pylab.title(self.family_name)
pylab.savefig(pie_file)
def plot_categories(categories, category_names=None, distribution_names=None, numrows=None, numcols=None):
"Plot the categories as sub-plots."
import pylab as P
assert 2 == len(categories.shape)
# determine number of rows and columns
num_dists = categories.shape[0]
num_categories = categories.shape[1]
if not numrows and not numcols:
numrows = int(numpy.sqrt(num_categories))
if not numrows:
numrows = num_categories / numcols
if num_categories % numcols:
numrows += 1
elif not numcols:
numcols = num_categories / numrows
if num_categories % numrows:
numcols += 1
# plot each category
max, min = categories.max(), categories.min()
ind = numpy.arange(num_dists) # the x locations for the groups
# the width of the bars: can also be len(x) sequence
width = 1.
for c in range(num_categories):
# overlaps, subplot(111) is killed
splt = P.subplot(numrows, numcols, c + 1)
handles = [
P.bar((i,), (x,), width, color=color)
for color, i, x
in zip(pylab_utils.simple_colours, ind, categories[:, c])
]
splt.set_ylim(min, max)
if category_names:
P.title(category_names[c])
if c % numcols:
splt.set_yticklabels(())
if distribution_names:
P.xticks(ind + width / 2., distribution_names)
else:
splt.set_xticklabels(())
splt.set_xticks(tuple())
if distribution_names:
P.figlegend(handles, distribution_names, 'upper right')
def plot(self,
yl='',
yr='',
x='s',
idx=slice(None),
clist='k r b g c m',
lattice=True,
newfig=True):
self._clist = clist.split()
if newfig:
f = _p.figure()
else:
f = _p.gcf()
sp = _p.subplot(111)
_p.subplots_adjust(right=0.72)
_p.setp(sp.yaxis, visible=False)
xd = getattr(self, x)[idx]
out = qdplot(figure=f,
subplot=sp,
plots=[],
lines=[],
legends=[],
xaxis=xd)
if lattice:
self._lattice(out, idx, ['kn0l', 'angle'], "#a0ffa0", 'Bend h')
self._lattice(out, idx, ['ks0l'], "#ffa0a0", 'Bend v')
self._lattice(out, idx, ['kn1l', 'k1l'], "#a0a0ff", 'Quad')
self._lattice(out, idx, ['hkick'], "#e0a0e0", 'Kick h')
self._lattice(out, idx, ['vkick'], "#a0e0e0", 'Kick v')
for i in yl.split():
self._column(out, idx, i, 'left')
for i in yr.split():
self._column(out, idx, i, 'right')
_p.xlabel(_mylbl(axlabel, x))
_p.xlim(min(xd), max(xd))
_p.figlegend(out.lines, out.legends, 'upper right')
_p.grid()
_p.draw()
del self._clist
if hasattr(self, '_colAxleft'): delattr(self, '_colAxleft')
if hasattr(self, '_colAxright'): delattr(self, '_colAxright')
self.currplot = out
return out
def draw_stacked_bar_w_err_bar(c, means_d, stderr_d, category_labels, outpfile, f_type):
N = len(category_labels)
# Get colors from ColorBrewer (all colorbrewer scales: http://bl.ocks.org/mbostock/5577023)
custom_color_list = list()
custom_color_list.extend(brewer2mpl.get_map('Paired', 'qualitative', 10).mpl_colors)
custom_color_list.extend(brewer2mpl.get_map('Set2', 'qualitative', 6).mpl_colors)
custom_color_list.extend(brewer2mpl.get_map('Set1', 'qualitative', 7).mpl_colors)
custom_color_list.extend(brewer2mpl.get_map('Set3', 'qualitative', 10).mpl_colors)
# Set the random seed for consistency
numpy.random.seed(12)
ind = numpy.arange(N) # the x locations for the groups
width = 0.35 # the width of the bars
bottom_value = [0 for m in category_labels]
legend_bar = list()
legend_text = list()
counter = 0
#taxons_sorted_list = sorted(means_d) # this sorts the taxons (keys of the means_d dictionary) and saves it as a list
# print in bar graph (bottom to top), the taxons in decreasing order of average abundance across groups.
#http://www.quora.com/How-do-I-sum-the-values-of-each-key-in-a-Python-dictionary-and-then-display-the-result-as-key-summed-value
d_total_list_values = dict((k, sum([float(i) for i in v])/float(len(v))) for k,v in means_d.items()) # convert values in list to float, avg them, assign total as value to key
taxons_sorted_list = sorted(d_total_list_values, key=d_total_list_values.get, reverse=True)# this sorts the taxons (keys) as per values and saves it as a list
#print d_total_list_values, taxons_sorted_list
for t in taxons_sorted_list:
if counter == len(custom_color_list): # cycle through the list of available colors if we have more taxons
counter = 0
mean = [float(i) for i in means_d[t]]
stderr = [float(i) for i in stderr_d[t]]
p = plt.bar(ind, mean, width, color=custom_color_list[counter], yerr=stderr, bottom=bottom_value)
bottom_value = [x+y for x,y in zip(bottom_value, mean)]
legend_bar.append(p[0])
legend_text.append(t)
counter += 1
plt.ylabel('Cumulative relative abundance')
plt.xlabel(c)
plt.title('Taxonomic summary')
plt.xticks(ind+width/2., tuple(category_labels) )
#plt.legend( tuple(legend_bar), tuple(legend_text) )
figname = outpfile + '_' + c + '.' + f_type
plt.savefig(figname, close=True)
# Klaus se at stackexchange
#http://stackoverflow.com/questions/4534480/get-legend-as-a-seperate-picture-in-matplotlib
legend_fig = pylab.figure()
# print in legend (top to bottom), the taxons in decreasing order of average abundance across groups.
#legend = pylab.figlegend( tuple(legend_bar), tuple(legend_text), loc = 'center' )
#print in legend (bottom to top), the taxons in decreasing order of average abundance across groups.
rev_legend_bar = list(reversed(legend_bar))
rev_legend_text = list(reversed(legend_text))
legend = pylab.figlegend( tuple(rev_legend_bar), tuple(rev_legend_text), loc = 'center' )
legend_fig.canvas.draw()
legend_figname = 'legend_' + figname
legend_fig.savefig(legend_figname, bbox_inches=legend.get_window_extent().transformed(legend_fig.dpi_scale_trans.inverted()))
pylab.close()
def plot_results(t_hr,
I_mA,
avg_t_hr,
lifetime_hr,
lifetime_error=None,
show_plot=False):
""""""
# Set show_plot to True to stop the code here for examining the plot
if show_plot:
fig = plt.figure()
ax1 = fig.add_subplot(111)
h1 = ax1.plot(t_hr, I_mA, '.-', label='Data', linewidth=0.5)
ax1.set_xlabel('Time [hr]')
ax1.set_ylabel('Beam Current [mA]')
ax2 = ax1.twinx()
if lifetime_error is not None:
h2 = ax2.errorbar(avg_t_hr,
lifetime_hr,
yerr=lifetime_error,
fmt='-',
ecolor='r',
color='r',
linestyle='-',
marker='.',
label='Lifetime')
else:
h2 = ax2.plot(avg_t_hr,
lifetime_hr,
color='r',
linestyle='-',
marker='.',
label='Lifetime')
ax2.set_ylabel('Lifetime [hr]')
plt.figlegend((h1[0], h2[0]), (h1[0].get_label(), h2[0].get_label()),
loc='upper right')
plt.show()
def plot_multiple(pairs_names, num_instances, performances_array, y_title,
project_name, dir_path, file_name, y_lim, b_anch, legend_loc, col_num, zip_size,
color_set, z_orders, print_legend=True):
x = []
y = []
for i in range(0, len(pairs_names)):
y.append([])
for i in range(0, num_instances):
if i % zip_size == 0 or i == num_instances - 1:
x.append((i / num_instances) * 100)
for j in range(0, len(pairs_names)):
y[j].append(performances_array[j][i])
fig = plt.figure()
ax = fig.add_subplot(111)
# LaTeX rendering case. You may use the next line if you have LaTeX installed.
# ax.set_title(r'\textsc{' + project_name.title() + '}', fontsize=14)
ax.set_title(project_name.title(), fontsize=14)
ax.set_xlim(0, 100)
if y_lim is not None:
ax.set_ylim(y_lim[0], y_lim[1])
ax.set_ylabel(y_title, fontsize=14)
ax.set_xlabel("Percentage of Instances", fontsize=14)
ax.grid()
for i in range(0, len(pairs_names)):
ax.plot(x, y[i], label=pairs_names[i], color=color_set[i], linewidth=1.2, zorder=z_orders[i])
# LaTeX rendering case. You may use the next line if you have LaTeX installed.
# ax.xaxis.set_major_formatter(FuncFormatter(lambda ix, _: '%1.0f' % ix + '\%'))
ax.xaxis.set_major_formatter(FuncFormatter(lambda ix, _: '%1.0f' % ix + '%'))
file_path = (dir_path + file_name + "_" + y_title).lower()
if print_legend is True:
leg = ax.legend(bbox_to_anchor=b_anch, loc=legend_loc, ncol=col_num, fontsize=8, framealpha=1)
for leg_obj in leg.legendHandles:
leg_obj.set_linewidth(1)
fig.savefig(file_path + ".pdf", dpi=150, bbox_inches='tight')
fig.savefig(file_path + ".png", dpi=150, bbox_inches='tight')
else:
fig_leg = pylab.figure(figsize=(13.5, 3.5), dpi=150)
leg = pylab.figlegend(*ax.get_legend_handles_labels(), loc='center', ncol=col_num, fontsize=14
, framealpha=1)
for leg_obj in leg.legendHandles:
leg_obj.set_linewidth(3.0)
fig_leg.savefig(file_path + "_legend.pdf", dpi=150)
fig_leg.savefig(file_path + "_legend.png", dpi=150)
fig.savefig(file_path + ".pdf", dpi=150, bbox_inches='tight')
fig.savefig(file_path + ".png", dpi=150, bbox_inches='tight')
def plot(cluster_file, filename="plot.png", func=lambda a: a.id,
plot_title='', cmap='Paired', filter=lambda a: True,
draw_legend=False, radius='2.25', sym=None, run_dir='../run/'):
"""
Creates a line plot showing population per cluster
"""
# retrieve cluster data from cluster file
clusters = rc.load_clusters(cluster_file)
# grab cluster population
p = get_population(run_dir=run_dir)
lines=[]
for cluster, agents in enumerate(clusters):
pop_by_time = list(repeat(0, 30000))
for agent in agents:
a = Agent(agent)
for i in range(a.birth, a.death):
pop_by_time[i] += 1
lines.append(
pylab.plot(range(30000), pop_by_time,
label=("%d: k=%d" % (i, len(clusters[cluster]))),
color=pylab.cm.Paired(float(cluster)/len(clusters))))
if draw_legend:
pylab.figlegend(lines, ["%d: k=%d" % (i, len(cluster))
for i,cluster in enumerate(clusters)],
'center right',
ncol=((len(clusters)/35)+1), prop=dict(size=6))
title = r"Cluster Population ($\epsilon$ = %s, %d clusters)" % (radius, len(clusters))
pylab.title(title, weight='black')
pylab.xlabel("Time", weight='bold')
pylab.ylabel("Population Size", weight='bold')
if sym is not None:
pylab.figtext(0,.954, '(%s)' % sym, size=6, weight='black')
pylab.savefig(filename, dpi=300)
def plot_all_shapelets(result):
f, rows = plt.subplots(2, 2, sharex=True)
axs = list(rows[:, 0]) + list(rows[:, 1])
lines = dict()
classifiers = dict()
for i, (label, (classifier, _)) in enumerate(result.items()):
classifiers[label] = classifier
order = [('59d638667bfe0b5f22bd6420: Motek - Erebus Unmount',
"Erebus Unmount"),
('59d638667bfe0b5f22bd6443: Motek - White Part Mount Tilted',
"Motek - White Part Mount Tilted"),
('59d638667bfe0b5f22bd6446: Pitasc-Sub - White Part Mount Tilted',
"Pitasc-Sub - White Part Mount Tilted"),
('59d638667bfe0b5f22bd6449: Pitasc - Insert Upright',
"Pitasc - Insert Upright")]
j = 0
for i, (label, title) in enumerate(order):
if classifiers.has_key(label):
classifier = classifiers[label]
axs[j].set_title(title)
lines.update(
plot_shapelet(axs[j], classifier.shapelet, classifier.dim_s))
axs[j].set_xlim(0, 50)
plt.setp(axs[j].get_yticklabels(), visible=False)
j += 1
classifiers = sorted(lines.items(), key=lambda x: x[0])
plt.figlegend([x[1] for x in classifiers], [x[0] for x in classifiers],
loc='lower center',
ncol=5,
labelspacing=0.)
plt.subplots_adjust(left=.05,
bottom=.15,
right=.95,
top=.93,
wspace=.12,
hspace=.51)
plt.show()
def fig_pca(directory, outfn, coords, model, title, sample_population=None):
import pylab
if sample_population is None:
sample_population = df_samples.population.values
populations = list(sample_population.unique())
colordict = randColor(len(populations), populations)
# plot coords for PCs 1 vs 2, 3 vs 4
figData = pylab.figure(figsize=(10, 6))
#fig = pylab.figure()
#figlegend = pylab.figure(figsize=(3,2))
#ax = fig.add_subplot(121)
#lines = ax.plot(range(10), pylab.randn(10), range(10), pylab.randn(10))
ax = figData.add_subplot(1, 2, 1)
plot_pca_coords(coords, model, 0, 1, ax, sample_population, populations,
colordict)
ax = figData.add_subplot(1, 2, 2)
plot_pca_coords(coords, model, 2, 3, ax, sample_population, populations,
colordict)
# create a second figure for the legend
figLegend = pylab.figure(figsize=(5, 10))
# produce a legend for the objects in the other figure
pylab.figlegend(*ax.get_legend_handles_labels(), loc='upper left')
# save the two figures to files
#figData.savefig("plot.png")
figLegend.savefig(directory + "graphics/" + outfn + "_pca_legend.jpg")
#ax.legend(bbox_to_anchor=(1, 1), loc='center')
figData.suptitle(title, y=1.02)
#fig.tight_layout()
figData.savefig(directory + "graphics/" + outfn + "_pca.jpg")
def do_plot(data, legend, fn_png):
pylab.clf()
pylab.figure(0, (6,8))
#pylab.axes([left, bottom, width, height])
# in fractional coordinates
pylab.axes([0.20, 0.070, 0.76, 0.84])
handles = []
labels = []
for key, symbol, label in legend:
x = []
y = []
for index, lot in enumerate(lots_list):
av = data.get((key, lot.key))
if av is not None:
x.append(av)
y.append(-index)
handle = pylab.plot(x, y, symbol)
handles.append(handle)
labels.append(label)
pos = -numpy.arange(len(lots_list))
ylabels = [lot.key for lot in lots_list]
pylab.yticks(pos, ylabels, size="small")
pylab.xticks(size="small")
#pylab.xlabel(r"Geometric mean of $k_{\theory}/k_{\exp}$", size="small")
pylab.xlabel(r"$\exp(RMSD(\ln(k_{\theory}) - \ln($k_{\exp})))$", size="small")
pylab.semilogx()
pylab.axvline(0.1, color="k", zorder=-5)
pylab.axvline(1.0, color="k", zorder=-5)
pylab.axvline(10.0, color="k", zorder=-5)
for y in 0, -5, -10, -15, -20, -25, -30, -35, -40:
pylab.axhline(y, color="k", alpha=0.2, lw=10, zorder=-5)
if len(legend) > 1:
legend = pylab.figlegend(handles, labels, (0.22,0.915), numpoints=1, handlelength=0, prop={"size":"small"})
frame = legend.get_frame()
frame.set_lw(0.0)
#pylab.xlim(1e-8, 1e5)
pylab.xlim(1e0, 1e5)
pylab.ylim(pos[-1]-0.5, 0.5)
pylab.savefig(fn_png)
def wise_psf(bands = [1,2,3,4], fftcentral=None, noplots=False, bw=False):
colors = ['b', 'g', 'r', 'm']
dbs = []
slices = []
centrals = []
for band in bands:
if band in [1,2,3]:
# im = fitsio.read('wise-psf-w%i-507.5-507.5.fits' % band)
im = fitsio.read('wise-psf-w%i-507.5-507.5-bright.fits' % band)
else:
# im = fitsio.read('wise-psf-w%i-253.5-253.5.fits' % band)
im = fitsio.read('wise-psf-w%i-253.5-253.5-bright.fits' % band)
im /= im.max()
print 'Image', im.shape, im.dtype
H,W = im.shape
assert(H == W)
print 'center / max:', im[H/2, W/2]
assert(im[H/2, W/2] == 1.0)
c = W/2
# plt.clf()
# plt.imshow(np.log10(im), interpolation='nearest', origin='lower',
# vmin=-8, vmax=0)
# ps.savefig()
nclip = (W - 251) / 2
slc = slice(nclip, -nclip)
central = im[slc,slc]
centrals.append(central)
ch,cw = central.shape
cc = ch/2
slices.append((central[cc,:], central[:,cc]))
if fftcentral is not None:
nclip = (W - fftcentral) / 2
im = im[nclip:-nclip, nclip:-nclip]
H,W = im.shape
Y = im[H/2, :]
X = im[:, W/2]
Fy = np.fft.fft(Y)
Fx = np.fft.fft(X)
Sy = np.fft.fftshift(Fy)
Sx = np.fft.fftshift(Fx)
Px = Sx.real**2 + Sx.imag**2
Py = Sy.real**2 + Sy.imag**2
freqs1 = np.fft.fftfreq(len(Y))
sfreqs1 = np.fft.fftshift(freqs1)
s0 = np.flatnonzero(sfreqs1 == 0)
s0 = s0[0]
Px = Px[s0:]
Py = Py[s0:]
sf = sfreqs1[s0:]
Sx = Sx[s0:]
Sy = Sy[s0:]
dbs.append((Sx, Sy, Px, Py, sf))
# plt.clf()
# plt.plot(sf, 10.*np.log10(Px / Px[0]), 'g-')
# plt.plot(sf, 10.*np.log10(Py / Py[0]), 'b-')
# plt.ylabel('dB')
# plt.ylim(-40, 3)
# plt.title('PSF models: W%i' % band)
# ps.savefig()
if noplots:
return dbs
cmap = 'jet'
ps.basefn = 'wisepsf'
if bw:
ps.pattern = 'wisepsf-%s-bw.%s'
#cmap = 'gray'
cmap = antigray
colors = ['k']*4
plt.figure(figsize=(4,4))
if True:
vmin=-7
vmax=0
# bot = 0.05
# top = 0.02
# left = 0.1
# right = 0.02
bot = 0.12
top = 0.03
left = 0.12
right = 0.03
# colorbar cbar
plt.figure(figsize=(0.75,4))
plt.clf()
ax = plt.gca()
print 'ax pos', ax.get_position()
ax.set_position([0.01, bot, 0.35, 1.-bot-top])
#cax,kw = matplotlib.colorbar.make_axes(ax, fraction=0.9)
#fraction 0.15; fraction of original axes to use for colorbar
#pad 0.05 if vertical, 0.15 if horizontal; fraction of original axes between colorbar and new image axes
#shrink 1.0; fraction by which to shrink the colorbar
#aspect 20; ratio of long to short dimensions
cb = matplotlib.colorbar.ColorbarBase(ax, cmap=cmap,
norm=matplotlib.colors.Normalize(vmin=vmin, vmax=vmax))
tv = np.arange(vmin, vmax+1)
cb.set_ticks(tv, update_ticks=False)
tt = ['$10^{%i}$' % t for t in tv]
tt[-1] = '$1$'
cb.set_ticklabels(tt, update_ticks=False)
cb.update_ticks()
ps.savefig()
plt.figure(figsize=(4,4))
plt.subplots_adjust(left=left, right=1.-right, bottom=bot, top=1.-top)
for central in centrals:
ch,cw = central.shape
cc = ch/2
plt.clf()
plt.imshow(np.log10(np.maximum(10.**(vmin-1), central)),
interpolation='nearest', origin='lower',
vmin=vmin, vmax=vmax, cmap=cmap,
extent=[-cc,cc,-cc,cc])
ps.savefig()
if True:
left = 0.17
right = 0.03
bot = 0.12
top = 0.03
plt.subplots_adjust(left=left, right=1.-right, bottom=bot, top=1.-top)
for i,(sx,sy) in enumerate(slices):
band = bands[i]
plt.clf()
S = len(sx)
mid = S/2
#vmin,vmax = -6,0
#vmin,vmax = -3,0
x = np.arange(len(sx)) - mid
pa = plt.semilogy(x, sx, '-', color=colors[i])
pb = plt.plot(x, sy, '-', color=colors[i], lw=3, alpha=0.5)
xlo,xhi = -10, 10
xx = np.linspace(xlo, xhi, 500)
scale = np.pi / 2.
yy = (2. * scipy.special.j1(xx * scale) / (xx * scale))**2
yy[xx == 0] = 1.
pc = plt.plot(xx, yy, '--', color='k')
plt.figlegend((pb[0], pa[0], pc[0]),
('W%i(y)' % band, 'W%i(x)' % band, 'Airy*'),
loc='upper right', fontsize=fontsize)
plt.xlim(xlo, xhi)
plt.ylim(1e-3/1.1, 1.1)
plt.xlabel('Pixel')
plt.ylabel('PSF profile')
ps.savefig()
if True:
left = 0.17
right = 0.03
bot = 0.12
top = 0.03
plt.subplots_adjust(left=left, right=1.-right, bottom=bot, top=1.-top)
# How much to compress the Airy profile so that it is just
# well-sampled (with unit pixels)
scale = np.pi / 2.
# Its half-width at half-max
hwhm = (1./scale * 1.61633)
# Gaussian with same HWHM
sigma = 2. * hwhm / 2.35
print 'sigma', sigma
x = np.linspace(-10, 10, 1000)
dx = x[1]-x[0]
g = np.exp(-0.5 * x**2 / sigma**2)
airy = (2. * scipy.special.j1(x * scale) / (x * scale))**2
Fg = np.fft.fft(g)
Sg = np.fft.fftshift(Fg)
Pg = Sg.real**2 + Sg.imag**2
freqs1 = np.fft.fftfreq(len(g))
sfg = np.fft.fftshift(freqs1)
s0 = np.flatnonzero(sfg == 0)
Fa = np.fft.fft(airy)
Sa = np.fft.fftshift(Fa)
Pa = Sa.real**2 + Sa.imag**2
Pg = Pg[s0:]
Pa = Pa[s0:]
sfg = sfg[s0:]
print 'dx', dx
sfg /= dx
for i,(nil,nil,Px,Py,sf) in enumerate(dbs):
band = bands[i]
plt.clf()
pa = plt.plot(sf, 10.*np.log10(Px / Px[0]), '-', color=colors[i])
pb = plt.plot(sf, 10.*np.log10(Py / Py[0]), '-', color=colors[i],
lw=3, alpha=0.5)
pc = plt.plot(sfg, 10.*np.log10(Pg / Pg[0]), 'k--')
pd = plt.plot(sfg, 10.*np.log10(Pa / Pa[0]), 'k:')
lp = (pb[0], pa[0], pc[0], pd[0])
lt = ('W%i (y)' % band, 'W%i (x)' % band,
'Gaussian*', 'Airy*')
if band == 1:
pitch = 0.001
lan3 = lanczos_test(noplots=True, pitch=pitch)
(Slan,Plan,sflan) = lan3[0]
sflan /= pitch
pe = plt.plot(sflan, 10.*np.log10(Plan / Plan[0]), 'k-.')
lp = (pe[0],) + lp
lt = ('Lanczos-3',) + lt
plt.ylabel('Power in Fourier component (dB)')
plt.ylim(-40, 3)
plt.xlabel('Frequency (cycles/pixel)')
plt.legend(lp, lt, loc='lower left', fontsize=fontsize)
plt.xlim(0, 0.5)
ps.savefig()
def plot_icd_vs_mass():
galaxies = pickle.load(open('galaxies.pickle', 'rb'))
arrow_up = [[0, 0], [-1, -1], [0, 0], [0, -2], [0, 0], [1, -1], [0, 0]]
# Make figure
f1 = pyl.figure(1, figsize=(4, 6))
f1s1 = f1.add_subplot(211)
f1s2 = f1.add_subplot(212)
for galaxy in galaxies:
if galaxy.ston_I > 30. and galaxy.ICD_IH != None:
f1s1.scatter(galaxy.z,
galaxy.ICD_IH * 100,
c='0.8',
marker='o',
s=25,
edgecolor='0.8')
if galaxy.ston_V > 30. and galaxy.ICD_VJ != None:
f1s2.scatter(galaxy.z,
galaxy.ICD_VJ * 100,
c='0.8',
marker='o',
s=25,
edgecolor='0.8')
# Add the box and whiskers
galaxies2 = filter(lambda galaxy: galaxy.ston_I > 30., galaxies)
galaxies2 = pyl.asarray(galaxies2)
x = [galaxy.z for galaxy in galaxies2]
ll = 1.5
ul = 4
bins_x = pyl.arange(ll, ul, 0.5)
grid = []
for i in range(bins_x.size - 1):
xmin = bins_x[i]
xmax = bins_x[i + 1]
cond = [cond1 and cond2 for cond1, cond2 in zip(x >= xmin, x < xmax)]
grid.append(galaxies2.compress(cond))
icd = []
for i in range(len(grid)):
icd.append([galaxy.ICD_IH * 100 for galaxy in grid[i]])
from boxplot_percentile import percentile_box_plot as pbp
pbp(f1s1,
icd,
indexer=list(pyl.delete(bins_x, -1) + 0.25),
whisker_top=None,
whisker_bottom=None)
# Add the box and whiskers
galaxies = pickle.load(open('galaxies.pickle', 'rb'))
galaxies2 = filter(lambda galaxy: galaxy.ston_V > 30., galaxies)
galaxies2 = pyl.asarray(galaxies2)
x = [galaxy.z for galaxy in galaxies2]
ll = 1.5
ul = 4
bins_x = pyl.arange(ll, ul, 0.5)
grid = []
for i in range(bins_x.size - 1):
xmin = bins_x[i]
xmax = bins_x[i + 1]
cond = [cond1 and cond2 for cond1, cond2 in zip(x >= xmin, x < xmax)]
grid.append(galaxies2.compress(cond))
icd = []
for i in range(len(grid)):
icd.append([galaxy.ICD_VJ * 100 for galaxy in grid[i]])
pbp(f1s2,
icd,
indexer=list(pyl.delete(bins_x, -1) + 0.25),
whisker_top=None,
whisker_bottom=None)
f1s1.set_xlim(1.5, 3.5)
f1s2.set_xlim(1.5, 3.5)
f1s1.set_ylim(-1, 10)
f1s2.set_ylim(-1, 10)
f1s1.set_xticklabels([])
f1s2.set_xticks([1.5, 2, 2.5, 3, 3.5])
f1s1.set_ylabel(r"$\xi[i_{775},H_{160}]$ (%)")
f1s2.set_xlabel("Redshift")
f1s2.set_ylabel(r"$\xi[V_{606},J_{125}]$ (%)")
import matplotlib.font_manager
line1 = pyl.Line2D([], [],
marker='o',
mfc='0.8',
mec='0.8',
markersize=8,
linewidth=0)
line2 = pyl.Line2D([], [],
marker='s',
mec='#348ABD',
mfc='None',
markersize=10,
linewidth=0,
markeredgewidth=2)
line3 = pyl.Line2D([], [], color='#A60628', linewidth=2)
prop = matplotlib.font_manager.FontProperties(size='small')
pyl.figlegend((line1, line2, line3), ('Data', 'Quartiles', 'Medians'),
'lower center',
prop=prop,
ncol=3)
pyl.show()
ax.grid()
if I > 1:
ax.set_xscale('log')
ax.set_yscale('log')
if I == 0:
# x = We ** aWe / Ca ** aCa
ax.set_xlabel(r'$We^{{{0:0.3f}}} Ca^{{{1:0.3f}}}$'.format(aWe, -aCa))
if I == 1:
# x = We ** aWe * Re ** aRe
ax.set_xlabel(r'$We^{{{0:0.3f}}} We^{{{1:0.3f}}}$'.format(aWe, aRe))
elif I == 2:
# x = St ** aSt * (Ca ** aCa * c1 + c2)
ax.set_xlabel(r'$St^{{{0:0.3f}}} Ca^{{{1:0.3f}}}$'.format(aSt, aCa))
elif I == 3:
# x = Re * St ** aSt / We ** aWe
ax.set_xlabel(r'$St^{{{0:0.3f}}} We^{{{1:0.3f}}} Re$'.format(aSt, -aWe))
fig.suptitle('C' + repr(I + 1))
plt.subplots_adjust(top=0.8)
ax.plot(x, y, color='black')
plt.savefig('validationData/plots/C' + repr(I + 1) + '.pgf',
bbox_inches='tight')
figLegend = pylab.figure(figsize=(1.8, 0.3))
pylab.figlegend(*ax.get_legend_handles_labels(), loc='upper left')
plt.savefig('validationData/plots/legend.pgf', bbox_inches='tight')
plt.show()
pylab.semilogy()
pylab.fill(
numpy.concatenate((invtemps, invtemps[::-1])),
numpy.concatenate((experimental_k*10, experimental_k[::-1]/10)),
"k", alpha=0.2, lw=0,
)
pylab.xticks(
1/numpy.array([300, 350, 400, 450, 500, 550, 600], float),
["300", "350", "400", "450", "500", "550", "600"],
)
pylab.title(title)
pylab.xlabel("T [K]")
pylab.ylabel("k [(m**3/mol)/s]")
pylab.ylim(1e-8,1e7)
legend = pylab.figlegend(
lines, labels, (0.07,0.06), ncol=3, prop={"size":11},
handlelength=3, labelspacing=0.1, columnspacing=1
)
#legend.get_frame().set_linewidth(0)
legend.get_frame().set_fill(True)
legend.get_frame().set_alpha(0.5)
pylab.savefig(fn_img % "rates")
pylab.clf()
lines = []
labels = []
line = pylab.plot([experimental_Ea], [experimental_A], color="k", marker="o", ms=11, mew=2, lw=0, ls=" ")
lines.append(line)
labels.append("experiment")
for lot in lots_list:
if lot.spin == "ROS":
label = "ro" + lot.label
P.close()
# do AUC bar-chart
#P.rcParams['xtick.direction'] = 'out'
def x(a, b, m):
return b*(len(methods)+1) + m
P.figure(figsize=(14,6))
for a, auc in enumerate(('AUC', 'AUC50')):
ax = P.subplot(1, 2, a+1)
xlocs = []
xlabels = []
for b, bg in enumerate(harness.options.backgrounds):
rects = P.bar(
[x(a, b, m) for m in xrange(len(methods))],
[aucs[bg][auc][method] for method in methods],
color=[colors[method] for method in methods],
width=1.
)
xlocs.append(x(a, b, len(methods)/2.))
xlabels.append(bg)
P.xticks(xlocs, xlabels)
P.ylim(0,1)
P.title(auc)
for tl in ax.get_xticklines():
tl.set_visible(False)
#P.text(.5, .9, '%s' % fragment, horizontalalignment='center')
P.figlegend(rects, methods, loc='upper right')
P.savefig(os.path.join(options.results_dir, 'AUC-%s.png' % fragment))
P.savefig(os.path.join(options.results_dir, 'AUC-%s.eps' % fragment))
P.close()
import pylab as plt
A = fits_table('Almanac_2016-03-31.fits')
A.cut(A.ra_offset < 99)
T = fits_table('db.fits')
T.cut(np.array([e in A.expnum for e in T.expnum]))
print(len(A), len(T))
from camera_mosaic import dradec_to_ref_chip
refdra, refddec = dradec_to_ref_chip(T)
plt.clf()
plt.subplot(2, 1, 1)
p1 = plt.plot(A.expnum, A.ra_offset, 'b.')
p2 = plt.plot(T.expnum, refdra, 'r.')
plt.figlegend([p1[0], p2[0]], ('Mosstat', 'Copilot'), 'upper right')
plt.ylabel('dRA (arcsec)')
plt.subplot(2, 1, 2)
plt.plot(A.expnum, A.dec_offset, 'b.')
plt.plot(T.expnum, refddec, 'r.')
plt.ylabel('dDec (arcsec)')
plt.xlabel('Expnum')
plt.suptitle('Copilot and Mosstat, 2016-03-31')
plt.savefig('astrom.png')
plt.clf()
p1 = plt.plot(A.expnum, A.ra_offset - refdra, 'b.')
p2 = plt.plot(A.expnum, A.dec_offset - refddec, 'r.')
plt.figlegend([p1[0], p2[0]], ('RA', 'Dec'), 'upper right')
plt.ylabel('Mosstat - Copilot (arcsec)')
plt.xlabel('Expnum')
def plot_icd_vs_mass():
galaxies = pickle.load(open('galaxies.pickle','rb'))
# Make figure
f1 = pyl.figure(1, figsize=(6,4))
f1s1 = f1.add_subplot(221)
f1s2 = f1.add_subplot(222)
f1s3 = f1.add_subplot(223)
f1s4 = f1.add_subplot(224)
#Upper and Lower limit arrow verts
arrowup_verts = [[0.,0.], [-1., -1], [0.,0.],
[0.,-2.], [0.,0.], [1,-1], [0,0]]
#arrowdown_verts = [[0.,0.], [-1., 1], [0.,0.],
# [0.,2.], [0.,0.], [1, 1]]
for galaxy in galaxies:
if galaxy.ston_I > 30. and galaxy.ICD_IH != None:
# Add arrows first
if galaxy.ICD_IH > 0.5:
f1s1.scatter(galaxy.Mass, 0.5*100, s=100, marker=None,
verts=arrowup_verts)
else:
f1s1.scatter(galaxy.Mass, galaxy.ICD_IH * 100, c='0.8',
marker='o', s=25, edgecolor='0.8')
f1s2.scatter(galaxy.Mass, galaxy.ICD_IH * 100, c='0.8',
marker='o', s=25, edgecolor='0.8')
if galaxy.ston_J > 30. and galaxy.ICD_JH != None:
# Add arrows first
if galaxy.ICD_JH > 0.12:
f1s3.scatter(galaxy.Mass, 12, s=100, marker=None,
verts=arrowup_verts)
else:
f1s3.scatter(galaxy.Mass, galaxy.ICD_JH * 100, c='0.8',
marker='o', s=25, edgecolor='0.8')
f1s4.scatter(galaxy.Mass, galaxy.ICD_JH * 100, c='0.8',
marker='o', s=25, edgecolor='0.8')
# Add the box and whiskers
galaxies2 = filter(lambda galaxy: galaxy.ston_I > 30., galaxies)
galaxies2 = pyl.asarray(galaxies2)
x = [galaxy.Mass for galaxy in galaxies2]
ll = 8.5
ul= 12
#bins_x =pyl.arange(8.5, 12.5, 0.5)
bins_x =pyl.array([8.5, 9., 9.5, 10., 10.5, 11., 12.])
grid = []
for i in range(bins_x.size-1):
xmin = bins_x[i]
xmax = bins_x[i+1]
cond=[cond1 and cond2 for cond1, cond2 in zip(x>=xmin, x<xmax)]
grid.append(galaxies2.compress(cond))
icd = []
for i in range(len(grid)):
icd.append([galaxy.ICD_IH*100 for galaxy in grid[i]])
from boxplot_percentile import percentile_box_plot as pbp
#bp1 = f1s1.boxplot(icd, positions=pyl.delete(bins_x,-1)+0.25, sym='')
pbp(f1s1, icd, indexer=list(pyl.delete(bins_x,-1)+0.25))
pbp(f1s2, icd, indexer=list(pyl.delete(bins_x,-1)+0.25))
# Add the box and whiskers
galaxies2 = filter(lambda galaxy: galaxy.ston_J > 30., galaxies)
galaxies2 = pyl.asarray(galaxies2)
x = [galaxy.Mass for galaxy in galaxies2]
ll = 8.5
ul= 12
#bins_x =pyl.linspace(ll, ul, 7)
#bins_x =pyl.arange(8.5, 12.5, 0.5)
bins_x =pyl.array([8.5, 9., 9.5, 10., 10.5, 11., 12.])
grid = []
for i in range(bins_x.size-1):
xmin = bins_x[i]
xmax = bins_x[i+1]
cond=[cond1 and cond2 for cond1, cond2 in zip(x>=xmin, x<xmax)]
grid.append(galaxies2.compress(cond))
icd = []
for i in range(len(grid)):
icd.append([galaxy.ICD_JH*100 for galaxy in grid[i]])
#bp2 = f1s2.boxplot(icd, positions=pyl.delete(bins_x,-1)+0.25, sym='')
pbp(f1s3, icd, indexer=list(pyl.delete(bins_x,-1)+0.25))
pbp(f1s4, icd, indexer=list(pyl.delete(bins_x,-1)+0.25))
# Finish Plot
# Tweak colors on the boxplot
#pyl.setp(bp1['boxes'], lw=2)
#pyl.setp(bp1['whiskers'], lw=2)
#pyl.setp(bp1['medians'], lw=2)
#pyl.setp(bp2['boxes'], lw=2)
#pyl.setp(bp2['whiskers'], lw=2)
#pyl.setp(bp2['medians'], lw=2)
#pyl.setp(bp['fliers'], color='#8CFF6F', marker='+')
#f1s1.axvspan(7.477, 9, facecolor='#FFFDD0', ec='None', zorder=0)
#f1s1.axvspan(11, 12, facecolor='#FFFDD0', ec='None', zorder=0)
#f1s2.axvspan(7.477, 9, facecolor='#FFFDD0', ec='None', zorder=0)
#f1s2.axvspan(11, 12, facecolor='#FFFDD0', ec='None', zorder=0)
f1s1.set_xlim(8,12)
f1s2.set_xlim(8,12)
f1s3.set_xlim(8,12)
f1s4.set_xlim(8,12)
f1s1.set_ylim(-10,50)
f1s2.set_ylim(0,15)
f1s3.set_ylim(-5,12)
f1s4.set_ylim(-1,3)
f1s1.set_xticks([8,9,10,11,12])
f1s1.set_xticklabels([])
f1s2.set_xticks([8,9,10,11,12])
f1s2.set_xticklabels([])
f1s3.set_xticks([8,9,10,11,12])
f1s4.set_xticks([8,9,10,11,12])
f1s4.set_yticks([-1, 0, 1, 2, 3])
f1s3.set_xlabel(r"Log Mass ($M_{\odot})$")
f1s1.set_ylabel(r"$\xi[i_{775},H_{160}]$ (%)")
f1s4.set_xlabel(r"Log Mass ($M_{\odot})$")
f1s3.set_ylabel(r"$\xi[J_{125},H_{160}]$ (%)")
import matplotlib.font_manager
line1 = pyl.Line2D([], [], marker='o', mfc='0.8', mec='0.8', markersize=8,
linewidth=0)
line2 = pyl.Line2D([], [], marker='s', mec='blue', mfc='None',
markersize=10, linewidth=0, markeredgewidth=2)
line3 = pyl.Line2D([], [], color='r', linewidth=2)
prop = matplotlib.font_manager.FontProperties(size='small')
pyl.figlegend((line1, line2, line3), ('Data', 'Quartiles',
'Medians'), 'lower center', prop=prop, ncol=3)
from matplotlib.patches import ConnectionPatch
xy = (12, 15)
xy2 = (8, 15)
con = ConnectionPatch(xyA=xy, xyB=xy2, coordsA='data', coordsB='data',
axesA=f1s1, axesB=f1s2)
xy = (12, 0)
xy2 = (8, 0)
con2 = ConnectionPatch(xyA=xy, xyB=xy2, coordsA='data', coordsB='data',
axesA=f1s1, axesB=f1s2)
f1s1.add_artist(con)
f1s1.add_artist(con2)
xy = (12, 3)
xy2 = (8, 3)
con = ConnectionPatch(xyA=xy, xyB=xy2, coordsA='data', coordsB='data',
axesA=f1s3, axesB=f1s4)
xy = (12, -1)
xy2 = (8, -1)
con2 = ConnectionPatch(xyA=xy, xyB=xy2, coordsA='data', coordsB='data',
axesA=f1s3, axesB=f1s4)
f1s3.add_artist(con)
f1s3.add_artist(con2)
pyl.draw()
pyl.show()
def main():
parser = E.OptionParser(
version="%prog version: $Id: plot_histogram.py 2782 2009-09-10 11:40:29Z andreas $", usage=globals()["__doc__"])
parser.add_option("-l", "--legend", dest="legend", type="string",
help="legend for plot [default=%default].")
parser.add_option("-t", "--title", dest="title", type="string",
help="title for plot [default=%default].")
parser.add_option("-p", "--hardcopy", dest="hardcopy", type="string",
help="filename for hardcopy of plot. The extension defines the format. Known extensions are: 'emf, eps, jpeg, jpg, pdf, png, ps, raw, rgba, svg, svgz' [default=%default].", metavar="FILE")
parser.add_option("", "--xrange", dest="xrange", type="string",
help="x viewing range of plot [default=%default].")
parser.add_option("", "--yrange", dest="yrange", type="string",
help="y viewing range of plot[default=%default].")
parser.add_option("-o", "--logscale", dest="logscale", type="string",
help="use logscale on x, y or xy [default=%default]")
parser.add_option("-x", "--xtitle", dest="xtitle", type="string",
help="title for x axis [default=%default]")
parser.add_option("-y", "--ytitle", dest="ytitle", type="string",
help="title for y axis [default=%default]")
parser.add_option("-d", "--dpi", dest="dpi", type="int",
help="dpi of images [default=%default]")
parser.add_option("-n", "--normalize", dest="normalize", action="store_true",
help="normalize histograms [default=%default]")
parser.add_option("--cumulate", dest="cumulate", action="store_true",
help="calculate cumulative histogram [default=%default].")
parser.add_option("--reverse-cumulate", dest="reverse_cumulate", action="store_true",
help="calculate cumulative histogram in reverse order [default=%default].")
parser.add_option("--legend-location", dest="legend_location", type="choice",
choices=("upper left", "upper right", "lower left",
"lower right", "center", "center right", "center left", "none"),
help="location of legend [default=%default]")
parser.add_option("--backend", dest="backend", type="string",
help="backend to use [Agg|SVG|PS] [default=%default]")
parser.add_option("--symbols", dest="symbols", type="string",
help="symbols to use for each histogram [steps|...] [default=%default].")
parser.add_option("--dump", dest="dump", action="store_true",
help="dump data for debug purposes [default=%default].")
parser.add_option("-c", "--columns", dest="columns", type="string",
help="columns to use for plotting [default=%default].")
parser.add_option("--truncate", dest="truncate", action="store_true",
help="truncate date within x range. If not set, xrange is simply a viewing range [default=%default].")
parser.add_option("--as-lines", dest="as_lines", action="store_true",
help="plot only lines, no symbols [default=%default].")
parser.add_option("--noheaders", dest="headers", action="store_false",
help="do not take first input line as header [default=%default].")
parser.add_option("--stacked", dest="stacked", action="store_true",
help="do a stacked plot [default=%default].")
parser.add_option("--add-function", dest="function", type="string",
help="add a function to the plot [default=%default].")
parser.add_option("--add-error-bars", dest="error_bars", type="choice",
choices=("interleaved", "blocked"),
help="add error bars. The input format is 'interleaved' or 'blocked'. In the interleaved format the error follows each column. I the blocked format first the data, then the errors in the same order [default=%default].")
parser.set_defaults(
legend=None,
title=None,
hardcopy=None,
logscale=None,
xtitle=None,
ytitle=None,
xrange=None,
yrange=None,
normalize=None,
columns="all",
headers=True,
legend_location="upper right",
backend="cairo",
symbols="g-D,b-h,r-+,c-+,m-+,y-+,k-o,g-^,b-<,r->,c-D,m-h",
dump=False,
truncate=False,
cumulate=False,
reverse_cumulate=False,
function=None,
add_error_bars=None,
as_lines=False,
stacked=False,
dpi=80,
)
(options, args) = E.Start(parser)
# import matplotlib/pylab. Has to be done here
# for batch scripts without GUI.
import matplotlib
if options.hardcopy:
matplotlib.use("cairo")
import pylab
# put this method here (because it requires pylab)
def doStackedPlot(data, legend):
colors = ["red",
"blue",
"green",
"cyan",
"magenta",
"yellow",
"brown",
"silver",
"purple",
"lightyellow",
"black",
"ivory",
"pink",
"orange",
"gray",
"teal"]
ax = data[:, 0]
xvals = numpy.concatenate((ax, ax[::-1]))
y_top = numpy.zeros(len(ax))
min_y = min(data[:, 1:].flat)
max_y = min_y
new_legend, dummy_lines = [], []
for i in range(1, len(legend)):
new_y_top = y_top + data[:, i]
yvals = numpy.concatenate((new_y_top, y_top[::-1]))
p = pylab.fill(xvals,
yvals,
colors[i % len(colors)])
y_top = new_y_top
max_y = max(y_top)
dummy_lines.append(pylab.plot(xvals,
yvals,
colors[i % len(colors)]))
new_legend.append(legend[i])
if not options.xrange:
options.xrange = min(data[:, 0]), max(data[:, 0])
if not options.yrange:
options.yrange = 0, max_y
return dummy_lines, new_legend
if options.as_lines:
options.symbols = []
for y in ("-", ":", "--"):
for x in "gbrcmyk":
options.symbols.append(y + x)
else:
options.symbols = options.symbols.split(",")
if options.xrange:
options.xrange = map(float, options.xrange.split(","))
if options.yrange:
options.yrange = map(float, options.yrange.split(","))
# Added support for (inclusive) range format: "1,3,5,7-100" (Gerton
# 13/12/06)
if options.columns != "all":
cols = []
for d in options.columns.split(','):
colopts = d.split('-')
if len(colopts) == 2:
cols += range(int(colopts[0]), int(colopts[1]) + 1)
else:
cols += [int(d) - 1]
options.columns = cols
if args:
if args[0] == "-":
infile = sys.stdin
else:
infile = open(args[0], "r")
else:
infile = sys.stdin
if options.truncate:
xr = options.xrange
else:
xr = None
data, legend = IOTools.readTable(infile,
numeric_type=numpy.float,
take=options.columns,
headers=options.headers,
truncate=xr)
if infile != sys.stdin:
infile.close()
if len(data) == 0: # or data is None:
E.info("empty table: no plot")
E.Stop()
return
nrows, ncols = data.shape
# note: because of MA, iteration makes copy of slices
# Solution: inplace edits.
if options.cumulate:
if options.add_error_bars:
raise "can not add error bars to cumulative histogram."
if data.mask.any():
# cumsum does not work with masked arrays, so do it manually
for y in range(1, ncols):
c = 0
for x in range(0, nrows):
if not data.mask[x, y]:
data[x, y] += c
c = data[x, y]
else:
for x in range(1, ncols):
data[:, x] = data[:, x].cumsum()
elif options.reverse_cumulate:
if options.add_error_bars:
raise "can not add error bars to cumulative histogram."
if data.mask.any():
l = [0] * ncols
for x in range(nrows - 1, -1, -1):
for y in range(1, ncols):
if not data.mask[x, y]:
data[x, y] += l[y]
l[y] = data[x, y]
else:
l = [0] * ncols
for x in range(nrows - 1, -1, -1):
for y in range(1, ncols):
data[x, y] += l[y]
l[y] = data[x, y]
if options.normalize:
if options.add_error_bars:
raise "can not add error bars to normalized histogram."
if data.mask.any():
m = [0] * ncols
for x in range(nrows):
for y in range(1, ncols):
if not data.mask[x, y]:
m[y] = max(m[y], float(data[x, y]))
for y in range(1, ncols):
if m[y] == 0:
m[y] = 1.0
for x in range(nrows):
for y in range(1, ncols):
data[x, y] = data[x, y] / m[y]
else:
for x in range(1, ncols):
m = float(data[:, x].max())
data[:, x] /= m
if options.legend:
legend = options.legend.split(",")
if options.dump:
for d in data:
print d
if options.title:
pylab.title(options.title)
if options.xtitle:
pylab.xlabel(options.xtitle)
else:
pylab.xlabel(legend[0])
if options.ytitle:
pylab.ylabel(options.ytitle)
lines = []
# use dummy_lines to workaround a bug in errorbars that
# causes the line styles to be set incorrectly.
dummy_lines = []
new_legend = []
if options.error_bars:
if options.error_bars == "interleaved":
step_size = 2
max_size = len(legend)
elif options.error_bars == "blocked":
step_size = 1
max_size = (len(legend) - 1) / 2
else:
step_size = 1
max_size = len(legend)
if options.stacked:
dummy_lines, new_legend = doStackedPlot(data, legend)
else:
nplotted = 0
nskipped = 0
for x in range(1, max_size, step_size):
s = options.symbols[nplotted % len(options.symbols)]
yvals = data[:, x]
xvals = numpy.ma.masked_array(data[:, 0], numpy.ma.getmask(yvals))
xvals = xvals.compressed()
yvals = yvals.compressed()
if len(xvals) == 0:
E.warn("skipped empty column %i: %s" % (x, legend[x]))
if options.error_bars == "interleaved":
yerr = data[:, x + 1]
yerr = yerr.compressed()
else:
yerr = None
lines.append(pylab.errorbar(xvals,
yvals,
yerr=yerr,
fmt=s))
dummy_lines.append(pylab.plot(xvals,
yvals,
s))
new_legend.append(legend[x])
nplotted += 1
E.info("nplotted=%i, nskipped=%i" % (nplotted, nskipped))
if len(lines) == 0:
E.Stop()
return
if options.legend_location != "none":
pylab.figlegend(dummy_lines,
new_legend,
options.legend_location)
if options.logscale:
if "x" in options.logscale:
pylab.gca().set_xscale('log')
if "y" in options.logscale:
pylab.gca().set_yscale('log')
if options.xrange:
pylab.xlim(options.xrange)
if options.yrange:
pylab.ylim(options.yrange)
if options.function:
xstart, xend = pylab.gca().get_xlim()
increment = (xend - xstart) / 100.0
exec("f = lambda x: %s" % options.function) in locals()
xvals, yvals = [], []
for x in range(0, 100):
xvals.append(xstart)
yvals.append(f(xstart))
xstart += increment
xvals.append(xstart)
yvals.append(f(xstart))
pylab.plot(xvals, yvals)
if options.hardcopy:
pylab.savefig(os.path.expanduser(options.hardcopy), dpi=options.dpi)
else:
pylab.show()
E.Stop()
# calculate the nets output for train and the test data
trnSequenceOutput = net.extrapolate(trnSequenceWashout, len(trnSequenceTarget))
tstSequenceOutput = net.extrapolate(tstSequenceWashout, len(tstSequenceTarget))
# plot training data
sp = subplot(211) # switch to the first subplot
cla() # clear the subplot
title("Training Set") # set the subplot's title
sp.set_autoscale_on(True) # enable autoscaling
targetline = plot(trnSequenceTarget, "r-") # plot the targets
sp.set_autoscale_on(False) # disable autoscaling
outputline = plot(trnSequenceOutput, "b-") # plot the actual output
# plot test data
sp = subplot(212)
cla()
title("Test Set")
sp.set_autoscale_on(True)
plot(tstSequenceTarget, "r-")
sp.set_autoscale_on(False)
plot(tstSequenceOutput, "b-")
# create a legend
figlegend((targetline, outputline), ("target", "output"), ("upper right"))
# draw everything
draw()
show()
def interaction_plot(df, val, xaxis,
seplines=None, sepxplots=None, sepyplots=None,
xmin='AUTO', xmax='AUTO', ymin='AUTO', ymax='AUTO',
where=None, fname=None, output_dir='',
quality='low', yerr=None):
"""
makes an interaction plot
args:
df:
a pyvttbl.DataFrame object
val:
the label of the dependent variable
xaxis:
the label of the variable to place on the xaxis of each subplot
kwds:
seplines:
label specifying seperate lines in each subplot
sepxplots:
label specifying seperate horizontal subplots
sepyplots:
label specifying separate vertical subplots
xmin:
('AUTO' by default) minimum xaxis value across subplots
xmax:
('AUTO' by default) maximum xaxis value across subplots
ymin:
('AUTO' by default) minimum yaxis value across subplots
ymax:
('AUTO' by default) maximum yaxis value across subplots
where:
a string, list of strings, or list of tuples
applied to the DataFrame before plotting
fname:
output file name
quality:
{'low' | 'medium' | 'high'} specifies image file dpi
yerr:
{float, 'ci', 'stdev', 'sem'} designates errorbars across
datapoints in all subplots
"""
##############################################################
# interaction_plot programmatic flow #
##############################################################
# 1. Check to make sure a plot can be generated with the #
# specified arguments and parameter #
# 2. Set yerr aggregate #
# 3. Figure out ymin and ymax if 'AUTO' is specified #
# 4. Figure out how many subplots we need to make and the #
# levels of those subplots #
# 5. Initialize pylab.figure and set plot parameters #
# 6. Build and set main title #
# 7. loop through the the rlevels and clevels and make #
# subplots #
# 7.1 Create new axes for the subplot #
# 7.2 Add subplot title #
# 7.3 Format the subplot #
# 7.4 Iterate plotnum counter #
# 8. Place yerr text in bottom right corner #
# 9. Save the figure #
# 10. return the test dictionary #
##############################################################
# 1. Check to make sure a plot can be generated with the
# specified arguments and parameter
##############################################################
# pylab doesn't like not being closed. To avoid starting
# a plot without finishing it, we do some extensive checking
# up front
if where == None:
where = []
# check for data
if df == {}:
raise Exception('Table must have data to plot marginals')
# check to see if data columns have equal lengths
if not df._are_col_lengths_equal():
raise Exception('columns have unequal lengths')
# check to make sure arguments are column labels
if val not in df.keys():
raise KeyError(val)
if xaxis not in df.keys():
raise KeyError(xaxis)
if seplines not in df.keys() and seplines != None:
raise KeyError(seplines)
if sepxplots not in df.keys() and sepxplots != None:
raise KeyError(sepxplots)
if sepyplots not in df.keys() and sepyplots != None:
raise KeyError(sepyplots)
# check for duplicate names
dup = Counter([val, xaxis, seplines, sepxplots, sepyplots])
del dup[None]
if not all([count == 1 for count in dup.values()]):
raise Exception('duplicate labels specified as plot parameters')
# check fname
if not isinstance(fname, _strobj) and fname != None:
raise TypeError('fname must be None or string')
if isinstance(fname, _strobj):
if not (fname.lower().endswith('.png') or \
fname.lower().endswith('.svg')):
raise Exception('fname must end with .png or .svg')
# check cell counts
cols = [f for f in [seplines, sepxplots, sepyplots] if f in df.keys()]
counts = df.pivot(val, rows=[xaxis], cols=cols,
where=where, aggregate='count')
# unpack conditions DictSet from counts PyvtTbl
# we need to do it this way instead of using df.conditions
# because the counts PyvtTbl reflects the where exclusions
conditions = counts.conditions
# flatten counts to MaskedArray
counts = counts.flatten()
for count in counts:
if count < 1:
raise Exception('cell count too low to calculate mean')
# 2. Initialize test dictionary
##############################################################
# To test the plotting a dict with various plot parameters
# is build and returned to the testing module. In this
# scenario our primary concern is that the values represent
# what we think they represent. Whether they match the plot
# should be fairly obvious to the user.
test = {}
# 3. Set yerr aggregate so sqlite knows how to calculate yerr
##############################################################
# check yerr
aggregate = None
if yerr == 'sem':
aggregate = 'sem'
elif yerr == 'stdev':
aggregate = 'stdev'
elif yerr == 'ci':
aggregate = 'ci'
for count in counts:
if aggregate != None and count < 2:
raise Exception('cell count too low to calculate %s'%yerr)
test['yerr'] = yerr
test['aggregate'] = aggregate
# 4. Figure out ymin and ymax if 'AUTO' is specified
##############################################################
desc = df.descriptives(val)
if ymin == 'AUTO':
# when plotting postive data always have the y-axis go to 0
if desc['min'] >= 0.:
ymin = 0.
else:
ymin = desc['mean'] - 3.*desc['stdev']
if ymax == 'AUTO':
ymax = desc['mean'] + 3.*desc['stdev']
if any([math.isnan(ymin), math.isinf(ymin), math.isnan(ymax), math.isinf(ymax)]):
raise Exception('calculated plot bounds nonsensical')
test['ymin'] = ymin
test['ymax'] = ymax
# 5. Figure out how many subplots we need to make and the
# levels of those subplots
##############################################################
numrows = 1
rlevels = [1]
if sepyplots != None:
rlevels = copy(conditions[sepyplots]) # a set
numrows = len(rlevels) # a int
rlevels = sorted(rlevels) # set -> sorted list
numcols = 1
clevels = [1]
if sepxplots != None:
clevels = copy(conditions[sepxplots])
numcols = len(clevels)
clevels = sorted(clevels) # set -> sorted list
test['numrows'] = numrows
test['rlevels'] = rlevels
test['numcols'] = numcols
test['clevels'] = clevels
# 6. Initialize pylab.figure and set plot parameters
##############################################################
fig = pylab.figure(figsize=(6*numcols, 4*numrows+1))
fig.subplots_adjust(wspace=.05, hspace=0.2)
# 7. Build and set main title
##############################################################
maintitle = '%s by %s'%(val,xaxis)
if seplines:
maintitle += ' * %s'%seplines
if sepxplots:
maintitle += ' * %s'%sepxplots
if sepyplots:
maintitle += ' * %s'%sepyplots
fig.text(0.5, 0.95, maintitle,
horizontalalignment='center',
verticalalignment='top')
test['maintitle'] = maintitle
# 8. loop through the the rlevels and clevels and make
# subplots
##############################################################
test['y'] = []
test['yerr'] = []
test['subplot_titles'] = []
test['xmins'] = []
test['xmaxs'] = []
plotnum = 1 # subplot counter
axs = []
for r, rlevel in enumerate(rlevels):
for c, clevel in enumerate(clevels):
where_extension = []
if sepxplots!=None:
where_extension.append((sepxplots, '=', [clevel]))
if sepyplots!=None:
where_extension.append((sepyplots, '=', [rlevel]))
# 8.1 Create new axes for the subplot
######################################################
axs.append(pylab.subplot(numrows, numcols, plotnum))
######## If separate lines are not specified #########
if seplines == None:
y = df.pivot(val, cols=[xaxis],
where=where+where_extension,
aggregate='avg')
cnames = y.cnames
y = y.flatten()
if aggregate != None:
yerr = df.pivot(val, cols=[xaxis],
where=where+where_extension,
aggregate=aggregate).flatten()
x = [name for [(label, name)] in cnames]
if _isfloat(yerr):
yerr = np.array([yerr for a in x])
if all([_isfloat(a) for a in x]):
axs[-1].errorbar(x, y, yerr)
if xmin == 'AUTO' and xmax == 'AUTO':
xmin, xmax = axs[-1].get_xlim()
xran = xmax - xmin
xmin = xmin - 0.05*xran
xmax = xmax + 0.05*xran
axs[-1].plot([xmin, xmax], [0., 0.], 'k:')
else : # categorical x axis
axs[-1].errorbar(_xrange(len(x)), y, yerr)
pylab.xticks(_xrange(len(x)), x)
xmin = - 0.5
xmax = len(x) - 0.5
axs[-1].plot([xmin, xmax], [0., 0.], 'k:')
########## If separate lines are specified ###########
else:
y = df.pivot(val, rows=[seplines], cols=[xaxis],
where=where+where_extension,
aggregate='avg')
if aggregate != None:
yerrs = df.pivot(val,
rows=[seplines],
cols=[xaxis],
where=where+where_extension,
aggregate=aggregate)
x = [name for [(label, name)] in y.cnames]
if _isfloat(yerr):
yerr = np.array([yerr for a in x])
plots = []
labels = []
for i, name in enumerate(y.rnames):
if aggregate != None:
yerr = yerrs[i].flatten()
labels.append(name[0][1])
if all([_isfloat(a) for a in x]):
plots.append(
axs[-1].errorbar(x, y[i].flatten(), yerr)[0])
if xmin == 'AUTO' and xmax == 'AUTO':
xmin , xmax = axs[-1].get_xlim()
xran = xmax - xmin
xmin = xmin - .05*xran
xmax = xmax + .05*xran
axs[-1].plot([xmin, xmax], [0.,0.], 'k:')
else : # categorical x axis
plots.append(
axs[-1].errorbar(
_xrange(len(x)), y[i].flatten(), yerr)[0])
pylab.xticks(_xrange(len(x)), x)
xmin = - 0.5
xmax = len(x) - 0.5
axs[-1].plot([xmin, xmax], [0., 0.], 'k:')
pylab.figlegend(plots, labels, loc=1,
labelsep=.005,
handlelen=.01,
handletextsep=.005)
test['y'].append(y)
if yerr == None:
test['yerr'].append([])
else:
test['yerr'].append(yerr)
test['xmins'].append(xmin)
test['xmaxs'].append(xmax)
# 8.2 Add subplot title
######################################################
if rlevels == [1] and clevels == [1]:
title = ''
elif rlevels == [1]:
title = _str(clevel)
elif clevels == [1]:
title = _str(rlevel)
else:
title = '%s = %s, %s = %s' \
% (sepyplots, _str(rlevel),
sepxplots, _str(rlevel))
pylab.title(title, fontsize='medium')
test['subplot_titles'].append(title)
# 8.3 Format the subplot
######################################################
pylab.xlim(xmin, xmax)
pylab.ylim(ymin, ymax)
# supress tick labels unless subplot is on the bottom
# row or the far left column
if r != (len(rlevels) - 1):
pylab.setp(axs[-1].get_xticklabels(), visible=False)
if c != 0:
pylab.setp(axs[-1].get_yticklabels(), visible=False)
# Set the aspect ratio for the subplot
Dx = abs(axs[-1].get_xlim()[0] - axs[-1].get_xlim()[1])
Dy = abs(axs[-1].get_ylim()[0] - axs[-1].get_ylim()[1])
axs[-1].set_aspect(.75*Dx/Dy)
# 8.4 Iterate plotnum counter
######################################################
plotnum += 1
# 9. Place yerr text in bottom right corner
##############################################################
if aggregate != None:
if aggregate == 'ci':
aggregate = '95% ci'
pylab.xlabel('\n\n '
'*Error bars reflect %s'\
%aggregate.upper())
# 10. Save the figure
##############################################################
if fname == None:
fname = 'interaction_plot(%s'%val
factors = [xaxis,seplines,sepxplots,sepyplots]
fname += '~' + '_X_'.join([str(f) for f in factors if f != None])
if aggregate != None:
fname += ',yerr=' + aggregate
elif yerr != None:
fname += ',yerr=' + str(yerr[0])
fname += ').png'
fname = os.path.join(output_dir, fname)
if quality == 'low' or fname.endswith('.svg'):
pylab.savefig(fname)
elif quality == 'medium':
pylab.savefig(fname, dpi=200)
elif quality == 'high':
pylab.savefig(fname, dpi=300)
else:
pylab.savefig(fname)
pylab.close()
test['fname'] = fname
# 11. return the test dictionary
##############################################################
if df.TESTMODE:
return test
sp = subplot(211) # switch to the first subplot
cla() # clear the subplot
title("Training Set") # set the subplot's title
sp.set_autoscale_on( True ) # enable autoscaling
targetline = plot(trnSequenceTarget,"r-") # plot the targets
sp.set_autoscale_on( False ) # disable autoscaling
outputline = plot(trnSequenceOutput,"b-") # plot the actual output
# plot test data
sp = subplot(212)
cla()
title("Test Set")
sp.set_autoscale_on( True )
plot(tstSequenceTarget,"r-")
sp.set_autoscale_on( False )
plot(tstSequenceOutput,"b-")
# create a legend
figlegend((targetline, outputline),('target','output'),('upper right'))
# draw everything
draw()
show()
def _plot_psf_models(fn, ps):
plt.clf()
plt.subplots_adjust(top=0.92, bottom=0.1, left=0.1, right=0.9,
hspace=0.25)
lp,lt = [],[]
for band in [1,2,3,4]:
print
print 'Band W%i' % band
#plt.clf()
plt.subplot(2,2, band)
plt.title('W%i' % band)
pix = fitsio.read('wise-psf-avg-pix-bright.fits', ext=band-1)
h,w = pix.shape
scale = 1.
vscale = 1
if band == 4:
scale = 0.5
vscale = 2.
# Slices along axes
xx = (np.arange(w)-w/2) / scale
p1 = plt.plot(xx, pix[h/2, :] * vscale, 'b-', lw=2, alpha=0.5)
xx = (np.arange(h)-h/2) / scale
p2 = plt.plot(xx, pix[:, w/2] * vscale, 'b--', lw=2, alpha=0.5)
lp.append(p1[0])
lt.append('Meisner+')
yy = pix[:,w/2]
print 'Meisner max:', pix.max(), 'sum', pix.sum()
pix = reduce(np.add,
[fitsio.read('wise-w%i-psf-wpro-09x09-%02ix%02i.fits' %
(band, 1+(i/9), 1+(i%9)))
for i in range(9*9)])
pix /= 81.
h,w = pix.shape
scale = 8.
vscale = scale**2
if band == 4:
scale = 4.
xx = (np.arange(w)-w/2) / scale
p3 = plt.plot(xx, pix[h/2, :] * vscale, 'r-')
xx = (np.arange(h) - h/2) / scale
p4 = plt.plot(xx, pix[:, w/2] * vscale, 'r--')
lp.append(p3[0])
lt.append('AllWISE')
print 'AllWISE max:', pix.max(), 'sum', pix.sum()
T = fits_table(fn, hdu=band)
# T.about()
print 'Amp sum:', np.sum(T.amp)
print 'Amp', T.amp
# print 'Mean', T.mean
print 'Var', T.var[:,0,0]
vscale = 1
if band == 4:
vscale = 4.
#xlo,xhi = -22, 22
xlo,xhi = -16, 16
xx = np.linspace(xlo, xhi, 201)
yy = reduce(np.add,
[amp * 1./(2.*np.pi*var) * np.exp(-0.5 * xx**2 / var)
for amp,var in zip(T.amp, T.var[:,0,0])])
p5 = plt.plot(xx, yy * vscale, 'k-')
lp.append(p5[0])
lt.append('Model')
# for amp,var in zip(T.amp, T.var[:,0,0]):
# yy = amp * 1./(2.*np.pi*var) * np.exp(-0.5 * xx**2 / var)
# if amp < 0:
# plt.plot(xx, np.abs(yy) * vscale, 'k--', alpha=0.5)
# else:
# plt.plot(xx, yy * vscale, 'k-', alpha=0.5)
print 'Mid:', yy[len(yy)/2]
print 'Model max:', yy.max()
xx,yy = np.meshgrid(np.arange(w)-w/2, np.arange(h)-h/2)
zz = reduce(np.add,
[amp * 1./(2.*np.pi*var) * np.exp(-0.5 * (xx**2 + yy**2) / var)
for amp,var in zip(T.amp, T.var[:,0,0])])
print 'Model sum:', zz.sum()
#plt.yscale('symlog', linthreshy=1e-6)
plt.yscale('log')
plt.ylim(2.5e-6, 0.25)
#plt.xscale('symlog', linthreshx=10, linscalex=3)
#plt.xlim(-100, 100)
plt.xticks([-20,-10,0,10,20])
plt.xlim(xlo, xhi)
if band == 1:
plt.figlegend(lp, lt, 'upper right')
if band in [1,3]:
plt.ylabel('PSF profile')
if band in [3,4]:
plt.xlabel('Pixel')
ps.savefig()
def twin_fig_legend(axes=None, loc="best", **kw):
"""Skapa figur legend for alla linjer i en plot som skapats med
dubbla y-axlar (twinx)
"""
(lines, labels) = build_figlegend(axes)
pylab.figlegend(lines, labels, loc=loc, **kw)
def plot_fingerprint(data, alpha=0.7, \
show_legend=True, graph_name='fingerprint.png', has_mean=True,
which_blocks='quartets', multiple=False, graph_grid='t', prob_axes=True, \
edge_colors='k', **kwargs):
"""Outputs a bubble plot of four-codon amino acid blocks
labeled with the colors from Sueoka 2002.
takes: data: array-elements in the col order x, y, r of
each of the four codon Amino Acids in the row order:
ALA, ARG4, GLY, LEU4, PRO, SER, THR, VAL
(for traditional fingerprint), or:
UU -> GG (for 16-block fingerprint).
last row is the mean (if has_mean is set True)
**kwargs passed on to init_graph_display (these include
graph_shape, graph_grid, x_label, y_label, dark, with_parens).
title: will be printed on graph (default: 'Unknown Species')
num_genes (number of genes contributing to graph: default None)
NOTE: will not print if None.)
size: of graph in inches (default = 8.0)
alpha: transparency of bubbles
(ranges from 0, transparent, to 1, opaque; default 0.7)
show_legend: bool, default True, whether to print legend
graph_name: name of file to write (default 'fingerprint.png')
has_mean: whether the data contain the mean (default: True)
which_blocks: which codon blocks to print (default is 'quartets'
for the 4-codon amino acid blocks, but can also use 'all' for all
quartets or 'split' for just the split quartets.)
multiple: if False (the default), assumes it got a single block
of data. Otherwise, assumes multiple blocks of data in a list or
array.
edge_colors: if multiple is True (ignored otherwise), uses this
sequence of edge color strings to hand out edge colors to successive
series. Will iterate over this, so can be a string of 1-letter
color codes or a list of color names.
note: that the data are always expected to be in the range (0,1)
since we're plotting frequencies. axes, gid, etc. are hard-coded
to these values.
"""
#figure out which type of fingerprint plot we're doing, and get the
#right colors
if which_blocks == 'quartets':
blocks = CodonUsage.SingleAABlocks
elif which_blocks == 'split':
blocks = CodonUsage.SplitBlocks
else:
blocks = CodonUsage.Blocks
colors = [doublets_to_colors[i] for i in blocks]
#formatting the labels in latex
x_label="$G_3/(G_3+C_3)$"
y_label="$A_3/(A_3+T_3)$"
#initializing components of the graph
font,label_font_size=init_graph_display(graph_shape='sqr', \
graph_grid=graph_grid, x_label=x_label, \
y_label=y_label, prob_axes=prob_axes, **kwargs)
if not multiple:
data = [data]
alpha = broadcast(alpha, len(data))
edge_colors = broadcast(edge_colors, len(data))
for al, d, edge_color in zip(alpha, data, edge_colors):
#skip this series if no data
if d is None or not d.any():
continue
for i, color in enumerate(colors):
j = i+1
#note: doing these as slices because scatter_classic needs the
#extra level of nesting
patches = scatter_classic(d[i:j,0], d[i:j,1],
s=(d[i:j,2]/2), c=color)
#set alpha for the patches manually
for p in patches:
p.set_alpha(al)
p.set_edgecolor(edge_color)
#plot mean as its own point -- can't do cross with scatter
if has_mean:
mean_index = len(blocks) #next index after the blocks
plot([d[mean_index,0]], [d[mean_index,1]],
'-k+',markersize=label_font_size, alpha=al)
abbrev = CodonUsage.BlockAbbreviations
a = gca()
#if show_legend is True prints a legend in the right center area
if show_legend:
legend_key = [abbrev[b] for b in blocks]
#copy legend font properties from the x axis tick labels
legend_font_props = \
a.xaxis.get_label().get_fontproperties().copy()
legend_font_scale_factor = 0.7
curr_size = legend_font_props.get_size()
legend_font_props.set_size(curr_size*legend_font_scale_factor)
l = figlegend(a.patches[:len(blocks)],
legend_key,
prop=legend_font_props,
loc='center right',borderpad=0.1,labelspacing=0.5,
handlelength=1.0,handletextpad=0.5, borderaxespad=0.0)
#fix transparency of patches
for p in l.get_patches():
p.set_alpha(1)
#initialize the ticks
set_axis_to_probs()
init_ticks(a, label_font_size)
a.set_xticks([0, 0.5, 1])
a.set_yticks([0,0.5,1])
#output the figure
if graph_name is not None:
savefig(graph_name)
pylab.errorbar(0.5, 0.5, xerr=0.1, yerr=0.1, \
color='None', markersize=7, markeredgecolor='darkorange', fmt='o', capsize=0, elinewidth=0.4, \
label=r'$\left< \mathrm{West\ Bridge} \right>$')
pylab.errorbar(np.arange(10), np.arange(10), xerr=np.arange(10), yerr=np.arange(10), \
color='darkgreen', markersize=3.5, markeredgecolor='None', fmt='o', capsize=0, elinewidth=0.2, \
label='Taffy-N')
pylab.errorbar(np.arange(10), np.arange(10), xerr=np.arange(10), yerr=np.arange(10), \
color='darkgreen', markersize=6, markeredgecolor='darkgreen', fmt='+', zorder=5, capsize=0, elinewidth=0.2, \
label='West Taffy-N')
pylab.errorbar(np.arange(10), np.arange(10), xerr=np.arange(10), yerr=np.arange(10), \
color='midnightblue', markersize=3.5, markeredgecolor='None', fmt='o', capsize=0, elinewidth=0.2, \
label='Taffy-S')
pylab.errorbar(np.arange(10), np.arange(10), xerr=np.arange(10), yerr=np.arange(10), \
color='midnightblue', markersize=4.5, markeredgecolor='midnightblue', fmt='d', zorder=5, capsize=0, elinewidth=0.2, \
label='Taffy-S nuclear region')
#pylab.errorbar(np.arange(10), np.arange(10), xerr=np.arange(10), yerr=np.arange(10), \
# color='None', markersize=5, markeredgecolor='limegreen', fmt='o', zorder=5, capsize=0, elinewidth=0.2, \
# label='Taffy-S nuclear minor axis')
# create a second figure for the legend
figLegend = pylab.figure()
# produce a legend for the objects in the other figure
pylab.figlegend(*ax.get_legend_handles_labels(), loc='center', frameon=False)
# save the two figures to files
figLegend.savefig(taffy_extdir + "emissionline_diagnostic_legend.png", dpi=300)
sys.exit(0)
for k in tests:
print k, objective(results[k]['sol'])
import pylab
lines = []
colors = ('b', 'g', 'r', 'c', 'm', 'y', 'k')
styles = ['-', ':']
count = 0
for label in tests.keys():
count += 1
color = colors[count % len(colors)]
style = styles[count % len(styles)]
res_pri = results[label]['res_pri']
res_dual = results[label]['res_dual']
#errs = results[label]['errs']
pylab.subplot(2, 1, 1)
line = pylab.semilogy(range(runs), res_pri, style, color=color)
lines.append(line[0])
pylab.ylabel('primal residual')
pylab.subplot(2, 1, 2)
pylab.semilogy(range(runs), res_dual, style, color=color)
pylab.xlabel('iteration')
pylab.ylabel('dual residual')
pylab.figlegend(lines, tests.keys(), loc='upper center', shadow=True,
fancybox=True, ncol=3)
pylab.show()
#pylab.ylabel('Correctly classified loci [%]')
pylab.yticks(range(50,101,10), [])
pylab.text(0.07, 1.02, 'C', transform = ax.transAxes, horizontalalignment='right', fontsize=12)
######### HGDP WinSize ################
ax=pylab.subplot(1,4,4)
with open(OUTPUT_TWO_POP_SVM_WIN,'r') as fp: g=cPickle.load(fp)
success=np.asarray(g.success)
lines=[]; labels=[]
for i in range(7):
l=pylab.semilogx(g.fst[:8], success[i*8:(i+1)*8,0], '-o', linewidth=.5, markersize=3)
lines.append(l)
labels.append('-'.join([l.capitalize() for l in g.files[i*8]]))
pylab.axis([8.5, 5500, 50, 100.5])
pylab.xticks(np.take(g.fst, [0,1,2,3,4,6,7]), np.take(g.fst, [0,1,2,3,4,6]), fontsize=6); pylab.yticks(range(50,101,10), [])
pylab.xlabel('Window size [# SNPs]')
pylab.figlegend(lines, labels, "lower right", ncol=4, numpoints=1, borderaxespad=1)
pylab.subplots_adjust(left=.078, bottom=.3, right=.98, top=.9, hspace=.19, wspace=.05)
pylab.text(0.07, 1.02, 'D', transform = ax.transAxes, horizontalalignment='right', fontsize=12)
pylab.savefig('fig1.'+FILETYPE,format=FILETYPE)
################################
# Supplemental Figure 1 - Structure+all chroms
################################
pylab.figure(figsize=(8.8, 10.8))
####### SupportMix Plots ############x##
startPos=.95
hSpace=0.001
for i in range(1,23):
CHR='chr%i' %i
chrIdx=qatarPositions[:,0]==CHR
height=.77*sum(chrIdx)/nQatarWins
def barplot_neighbors(Nrange=2 ** np.arange(1, 11),
Drange=2 ** np.arange(7),
krange=2 ** np.arange(10),
N=1000,
D=64,
k=5,
leaf_size=30,
dataset='digits'):
algorithms = ('kd_tree', 'brute', 'ball_tree')
fiducial_values = {'N': N,
'D': D,
'k': k}
#------------------------------------------------------------
# varying N
N_results_build = dict([(alg, np.zeros(len(Nrange)))
for alg in algorithms])
N_results_query = dict([(alg, np.zeros(len(Nrange)))
for alg in algorithms])
for i, NN in enumerate(Nrange):
print("N = %i (%i out of %i)" % (NN, i + 1, len(Nrange)))
X = get_data(NN, D, dataset)
for algorithm in algorithms:
nbrs = neighbors.NearestNeighbors(n_neighbors=min(NN, k),
algorithm=algorithm,
leaf_size=leaf_size)
t0 = time()
nbrs.fit(X)
t1 = time()
nbrs.kneighbors(X)
t2 = time()
N_results_build[algorithm][i] = (t1 - t0)
N_results_query[algorithm][i] = (t2 - t1)
#------------------------------------------------------------
# varying D
D_results_build = dict([(alg, np.zeros(len(Drange)))
for alg in algorithms])
D_results_query = dict([(alg, np.zeros(len(Drange)))
for alg in algorithms])
for i, DD in enumerate(Drange):
print("D = %i (%i out of %i)" % (DD, i + 1, len(Drange)))
X = get_data(N, DD, dataset)
for algorithm in algorithms:
nbrs = neighbors.NearestNeighbors(n_neighbors=k,
algorithm=algorithm,
leaf_size=leaf_size)
t0 = time()
nbrs.fit(X)
t1 = time()
nbrs.kneighbors(X)
t2 = time()
D_results_build[algorithm][i] = (t1 - t0)
D_results_query[algorithm][i] = (t2 - t1)
#------------------------------------------------------------
# varying k
k_results_build = dict([(alg, np.zeros(len(krange)))
for alg in algorithms])
k_results_query = dict([(alg, np.zeros(len(krange)))
for alg in algorithms])
X = get_data(N, DD, dataset)
for i, kk in enumerate(krange):
print("k = %i (%i out of %i)" % (kk, i + 1, len(krange)))
for algorithm in algorithms:
nbrs = neighbors.NearestNeighbors(n_neighbors=kk,
algorithm=algorithm,
leaf_size=leaf_size)
t0 = time()
nbrs.fit(X)
t1 = time()
nbrs.kneighbors(X)
t2 = time()
k_results_build[algorithm][i] = (t1 - t0)
k_results_query[algorithm][i] = (t2 - t1)
pl.figure(figsize=(8, 11))
for (sbplt, vals, quantity,
build_time, query_time) in [(311, Nrange, 'N',
N_results_build,
N_results_query),
(312, Drange, 'D',
D_results_build,
D_results_query),
(313, krange, 'k',
k_results_build,
k_results_query)]:
ax = pl.subplot(sbplt, yscale='log')
pl.grid(True)
tick_vals = []
tick_labels = []
bottom = 10 ** np.min([min(np.floor(np.log10(build_time[alg])))
for alg in algorithms])
for i, alg in enumerate(algorithms):
xvals = 0.1 + i * (1 + len(vals)) + np.arange(len(vals))
width = 0.8
c_bar = pl.bar(xvals, build_time[alg] - bottom,
width, bottom, color='r')
q_bar = pl.bar(xvals, query_time[alg],
width, build_time[alg], color='b')
tick_vals += list(xvals + 0.5 * width)
tick_labels += ['%i' % val for val in vals]
pl.text((i + 0.02) / len(algorithms), 0.98, alg,
transform=ax.transAxes,
ha='left',
va='top',
bbox=dict(facecolor='w', edgecolor='w', alpha=0.5))
pl.ylabel('Time (s)')
ax.xaxis.set_major_locator(ticker.FixedLocator(tick_vals))
ax.xaxis.set_major_formatter(ticker.FixedFormatter(tick_labels))
for label in ax.get_xticklabels():
label.set_rotation(-90)
label.set_fontsize(10)
title_string = 'Varying %s' % quantity
descr_string = ''
for s in 'NDk':
if s == quantity:
pass
else:
descr_string += '%s = %i, ' % (s, fiducial_values[s])
descr_string = descr_string[:-2]
pl.text(1.01, 0.5, title_string,
transform=ax.transAxes, rotation=-90,
ha='left', va='center', fontsize=20)
pl.text(0.99, 0.5, descr_string,
transform=ax.transAxes, rotation=-90,
ha='right', va='center')
pl.gcf().suptitle("%s data set" % dataset.capitalize(), fontsize=16)
pl.figlegend((c_bar, q_bar), ('construction', 'N-point query'),
'upper right')
def stage3(cat=None,
variances=None,
T=None,
bands=None,
ps=None,
targetwcs=None,
T2=None,
M=None,
**kwargs):
ccmap = dict(g='g', r='r', i='m')
for band in bands:
mag = M.get('mag_%s' % band)
# sdss
smag = M.get('smag_%s' % band)
ok = np.flatnonzero(mag != smag)
mag = mag[ok]
smag = smag[ok]
cc = ccmap[band]
plt.clf()
plt.subplot(2, 1, 1)
plt.plot(smag, mag, '.', color=cc, alpha=0.5)
lo, hi = 16, 24
plt.plot([lo, hi], [lo, hi], 'k-')
plt.xlabel('SDSS mag')
plt.ylabel('CFHT mag')
plt.axis([lo, hi, lo, hi])
plt.subplot(2, 1, 2)
plt.plot(smag, mag - smag, '.', color=cc, alpha=0.5)
lo, hi = 16, 24
plt.plot([lo, hi], [0, 0], 'k-')
plt.xlabel('SDSS mag')
plt.ylabel('CFHT - SDSS mag')
plt.axis([lo, hi, -1, 1])
plt.suptitle('%s band' % band)
ps.savefig()
plt.clf()
lp, lt = [], []
for band in bands:
sn = (T2.get('decam_%s_nanomaggies' % band) *
np.sqrt(T2.get('decam_%s_nanomaggies_invvar' % band)))
mag = T2.get('decam_%s_mag' % band)
cc = ccmap[band]
p = plt.semilogy(mag, sn, '.', color=cc, alpha=0.5)
lp.append(p[0])
lt.append('%s band' % band)
plt.xlabel('mag')
plt.ylabel('Flux Signal-to-Noise')
#tt = [1,2,3,4,5,10,20,30,40,50]
tt = [
1, 3, 5, 10, 30, 50, 100, 300, 500, 1000, 3000, 5000, 10000, 30000,
50000
]
plt.yticks(tt, ['%i' % t for t in tt])
plt.axhline(5., color='k')
#plt.axis([21, 26, 1, 20])
plt.legend(lp, lt, loc='upper right')
plt.title('CFHT depth')
lo, hi = 16, 27
plt.xlim(lo, hi)
ps.savefig()
# zps = {}
# plt.clf()
# lp,lt = [],[]
# for band in bands:
# sn = (T2.get('decam_%s_nanomaggies' % band) *
# np.sqrt(T2.get('decam_%s_nanomaggies_invvar' % band)))
# mag = T2.get('decam_%s_mag' % band)
# cc = ccmap[band]
# I = (np.isfinite(mag) * (T2.type == 'S') * np.isfinite(sn))
# Ti = T2[I]
# Ti.mag = mag[I]
# Ti.sn = sn[I]
# zps[band] = np.median(np.log10(Ti.sn) + 0.4*Ti.mag)
# plt.plot(Ti.mag, np.log10(Ti.sn) + 0.4*Ti.mag, '.', color=cc)
# ps.savefig()
metadata = dict(
g=(1080946, 634),
r=(1617879, 445),
i=(1080306, 411),
)
zps = {}
lo, hi = 16, 27
plt.clf()
lp, lt = [], []
for band in bands:
sn = (T2.get('decam_%s_nanomaggies' % band) *
np.sqrt(T2.get('decam_%s_nanomaggies_invvar' % band)))
mag = T2.get('decam_%s_mag' % band)
cc = ccmap[band]
I = (np.isfinite(mag) * (T2.type == 'S') * np.isfinite(sn))
Ti = T2[I]
Ti.mag = mag[I]
Ti.sn = sn[I]
zp = np.median(np.log10(Ti.sn) + 0.4 * Ti.mag)
zps[band] = zp
print('zp', zp)
xx = np.array([lo, hi])
plt.plot(xx, 10.**(zp - 0.4 * xx), '-', alpha=0.4, color=cc)
plt.plot(xx, 10.**(zp - 0.4 * xx), 'k-', alpha=0.4)
p = plt.semilogy(Ti.mag, Ti.sn, '.', color=cc, alpha=0.5)
(expnum, exptime) = metadata[band]
depth = (zp - np.log10(5.)) / 0.4
lp.append(p[0])
lt.append('%s band: exptime %i, depth %.2f (exp %i)' %
(band, exptime, depth, expnum))
plt.plot([depth, depth], [1, 5], color=cc, alpha=0.5)
plt.xlabel('mag')
plt.ylabel('Flux Signal-to-Noise')
#tt = [1,2,3,4,5,10,20,30,40,50]
tt = [
1, 3, 5, 10, 30, 50, 100, 300, 500, 1000, 3000, 5000, 10000, 30000,
50000
]
plt.yticks(tt, ['%i' % t for t in tt])
plt.axhline(5., color='k')
#plt.axis([21, 26, 1, 20])
plt.xlim(lo, hi)
#ylo,yhi = plt.ylim()
#plt.ylim(1, yhi)
plt.ylim(1, 1e5)
plt.legend(lp, lt, loc='upper right')
plt.title('CFHT depth (point sources)')
ps.savefig()
plt.clf()
lp, lt = [], []
for band in bands:
sn = (T2.get('decam_%s_nanomaggies' % band) *
np.sqrt(T2.get('decam_%s_nanomaggies_invvar' % band)))
mag = T2.get('decam_%s_mag' % band)
cc = ccmap[band]
I = (np.isfinite(mag) * (T2.type != 'S') * np.isfinite(sn))
Ti = T2[I]
Ti.mag = mag[I]
Ti.sn = sn[I]
xx = np.array([lo, hi])
zp = zps[band]
plt.plot(xx, 10.**(zp - 0.4 * xx), '-', alpha=0.4, color=cc)
plt.plot(xx, 10.**(zp - 0.4 * xx), 'k-', alpha=0.4)
p = plt.semilogy(Ti.mag, Ti.sn, '.', color=cc, alpha=0.5)
lp.append(p[0])
lt.append('%s band' % band)
plt.xlabel('mag')
plt.ylabel('Flux Signal-to-Noise')
tt = [
1, 3, 5, 10, 30, 50, 100, 300, 500, 1000, 3000, 5000, 10000, 30000,
50000
]
plt.yticks(tt, ['%i' % t for t in tt])
plt.axhline(5., color='k')
plt.xlim(lo, hi)
# ylo,yhi = plt.ylim()
# plt.ylim(1, yhi)
plt.ylim(1, 1e5)
plt.legend(lp, lt, loc='upper right')
plt.title('CFHT depth (extended sources)')
ps.savefig()
plt.subplots_adjust(hspace=0.)
plt.clf()
lp, lt = [], []
for i, band in enumerate(bands):
plt.subplot(3, 1, i + 1)
sn = (T2.get('decam_%s_nanomaggies' % band) *
np.sqrt(T2.get('decam_%s_nanomaggies_invvar' % band)))
mag = T2.get('decam_%s_mag' % band)
cc = ccmap[band]
I = (np.isfinite(mag) * (T2.type != 'S') * np.isfinite(sn))
Ti = T2[I]
Ti.mag = mag[I]
Ti.sn = sn[I]
xx = np.array([lo, hi])
zp = zps[band]
plt.plot(xx, [1., 1.], '-', alpha=0.4, color=cc)
plt.plot(xx, [1., 1.], 'k-', alpha=0.4)
plt.plot(xx, [0.4, 0.4], 'k-', alpha=0.1)
plt.plot(xx, [0.16, 0.16], 'k-', alpha=0.1)
p = plt.semilogy(Ti.mag,
Ti.sn / 10.**(zp - 0.4 * Ti.mag),
'.',
color=cc,
alpha=0.5)
lp.append(p[0])
lt.append('%s band' % band)
if i == 1:
plt.ylabel('Flux Signal-to-Noise vs Point Source')
if i != 2:
plt.xticks([])
plt.ylim(0.08, 1.2)
plt.xlim(lo, hi)
plt.xlabel('mag')
#tt = [1,3,5,10,30,50,100,300,500,1000,3000,5000,10000,30000,50000]
#plt.yticks(tt, ['%i' % t for t in tt])
plt.figlegend(lp, lt, loc='upper right')
plt.suptitle('CFHT depth (extended sources)')
ps.savefig()
plt.clf()
lp, lt = [], []
for i, band in enumerate(bands):
sn = (T2.get('decam_%s_nanomaggies' % band) *
np.sqrt(T2.get('decam_%s_nanomaggies_invvar' % band)))
mag = T2.get('decam_%s_mag' % band)
cc = ccmap[band]
I = (np.isfinite(mag) * (T2.type != 'S') * np.isfinite(sn))
Ti = T2[I]
Ti.mag = mag[I]
Ti.sn = sn[I]
zp = zps[band]
snloss = Ti.sn / 10.**(zp - 0.4 * Ti.mag)
for J, name, style, deV in [
(np.flatnonzero(Ti.type == 'D'), 'deV', 'x', True),
(np.flatnonzero(Ti.type == 'E'), 'exp', 'o', False),
(np.flatnonzero(
(Ti.type == 'C') * (Ti.fracDev > 0.5)), 'comp/deV', 's', True),
(np.flatnonzero((Ti.type == 'C') * (Ti.fracDev <= 0.5)),
'comp/exp', '^', False),
]:
print(len(J), name)
if deV:
size = Ti.shapeDev[:, 0]
else:
size = Ti.shapeExp[:, 0]
ylo = 5e-2
p = plt.loglog(np.clip(size[J], 1e-2, 1e3),
np.clip(snloss[J], ylo, 10.),
style,
color=cc,
mfc='none',
mec=cc,
alpha=0.5)
if i == 0:
lp.append(p[0])
lt.append(name)
plt.axhline(0.4**0, color='k', alpha=0.1)
plt.axhline(0.4**1, color='k', alpha=0.1)
plt.axhline(0.4**2, color='k', alpha=0.1)
plt.axhline(0.4**3, color='k', alpha=0.1)
plt.xlim(0.9 * 1e-2, 1.1 * 1e3)
plt.ylim(0.9 * ylo, 1.2)
plt.xlabel('Galaxy effective radius (arcsec)')
plt.ylabel('S/N loss vs point source')
plt.legend(lp, lt, loc='upper right')
plt.title('CFHT extended sources: S/N vs size')
ps.savefig()
def plot_multiple(pairs_names,
num_instances,
performances_array,
y_title,
project_name,
dir_path,
file_name,
y_lim,
b_anch,
legend_loc,
col_num,
zip_size,
color_set,
z_orders,
step=1,
print_legend=True,
datasetName=None,
dataName=None):
# x = []
# y = []
# for i in range(0, len(pairs_names)):
# y.append([])
# for i in range(0, num_instances):
# if i % zip_size == 0 or i == num_instances - 1:
# x.append((i / num_instances) * 100)
# for j in range(0, len(pairs_names)):
# y[j].append(performances_array[j][i])
fig = plt.figure()
ax = fig.add_subplot(111)
# LaTeX rendering case. You may use the next line if you have LaTeX installed.
# ax.set_title(r'\textsc{' + project_name.title() + '}', fontsize=14)
ax.set_title(project_name.title(), fontsize=14)
# ax.set_xlim(0, 100)
# if y_lim is not None:
# ax.set_ylim(y_lim[0], y_lim[1])
# else:
# y1_y2 = performances_array[0] + performances_array[1]
# y_lim = [min(y1_y2)-1.0, 1.05]
# print(y_lim)
# ax.set_ylim(y_lim[0], y_lim[1])
if y_lim is not None:
ax.set_ylim(y_lim[0], y_lim[1])
else:
interval = 0.1
top = 1.05
bottom = -0.05
if datasetName == 'prsa_data':
interval = 0.01
top = 1.01
bottom = 0.79
elif datasetName == 'movie_data':
if dataName == 'movie_lens':
interval = 0.1
top = 1.05
bottom = 0.29
elif dataName == 'netflix':
interval = 0.1
top = 1.05
bottom = 0.45
elif dataName == 'rotten_tomatoes':
interval = 0.05
top = 1.04
bottom = 0.66
elif dataName == 'twitter':
interval = 0.05
top = 1.04
bottom = 0.56
y_major_locator = MultipleLocator(interval)
ax.yaxis.set_major_locator(y_major_locator)
ax.set_ylim(bottom, top)
ax.set_ylabel(y_title, fontsize=14)
ax.set_xlabel("Step = {}".format(step), fontsize=14)
ax.grid()
for i in range(0, len(pairs_names)):
ax.plot([i for i in range(len(performances_array[i]))],
performances_array[i],
label=pairs_names[i],
color=color_set[i],
linewidth=1,
zorder=z_orders[i])
# LaTeX rendering case. You may use the next line if you have LaTeX installed.
# ax.xaxis.set_major_formatter(FuncFormatter(lambda ix, _: '%1.0f' % ix + '\%'))
# ax.xaxis.set_major_formatter(FuncFormatter(lambda ix, _: '%1.0f' % ix + '%'))
file_path = (dir_path + file_name + "_" + y_title).lower()
if print_legend is True:
leg = ax.legend(bbox_to_anchor=b_anch,
loc=legend_loc,
ncol=col_num,
fontsize=8,
framealpha=1)
for leg_obj in leg.legendHandles:
leg_obj.set_linewidth(1)
fig.savefig(file_path + ".pdf", dpi=150, bbox_inches='tight')
fig.savefig(file_path + ".png", dpi=150, bbox_inches='tight')
else:
fig_leg = pylab.figure(figsize=(13.5, 3.5), dpi=150)
leg = pylab.figlegend(*ax.get_legend_handles_labels(),
loc='center',
ncol=col_num,
fontsize=14,
framealpha=1)
for leg_obj in leg.legendHandles:
leg_obj.set_linewidth(3.0)
fig_leg.savefig(file_path + "_legend.pdf", dpi=150)
fig_leg.savefig(file_path + "_legend.png", dpi=150)
fig.savefig(file_path + ".pdf", dpi=150, bbox_inches='tight')
fig.savefig(file_path + ".png", dpi=150, bbox_inches='tight')
def plot_icd_vs_mass():
galaxies = pickle.load(open('galaxies.pickle','rb'))
arrow_up = [[0,0], [-1,-1], [0,0], [0,-2], [0,0], [1,-1], [0,0]]
# Make figure
f1 = pyl.figure(1, figsize=(6,4))
f1s1 = f1.add_subplot(221)
f1s2 = f1.add_subplot(222)
f1s3 = f1.add_subplot(223)
f1s4 = f1.add_subplot(224)
for galaxy in galaxies:
if galaxy.ston_I > 30. and galaxy.ICD_IH != None:
if galaxy.ICD_IH*100 < 50:
f1s1.scatter(galaxy.z, galaxy.ICD_IH * 100, c='0.8',
marker='o', s=25, edgecolor='0.8')
else:
f1s1.scatter(galaxy.z, 50, s=100, marker=None, verts=arrow_up)
f1s2.scatter(galaxy.z, galaxy.ICD_IH * 100, c='0.8',
marker='o', s=25, edgecolor='0.8')
if galaxy.ston_V > 30. and galaxy.ICD_VJ != None:
f1s3.scatter(galaxy.z, galaxy.ICD_VJ * 100, c='0.8',
marker='o', s=25, edgecolor='0.8')
f1s4.scatter(galaxy.z, galaxy.ICD_VJ * 100, c='0.8',
marker='o', s=25, edgecolor='0.8')
# Add the box and whiskers
galaxies2 = filter(lambda galaxy: galaxy.ston_I > 30., galaxies)
galaxies2 = pyl.asarray(galaxies2)
x = [galaxy.z for galaxy in galaxies2]
ll = 1.5
ul= 4
bins_x =pyl.arange(ll, ul, 0.5)
grid = []
for i in range(bins_x.size-1):
xmin = bins_x[i]
xmax = bins_x[i+1]
cond=[cond1 and cond2 for cond1, cond2 in zip(x>=xmin, x<xmax)]
grid.append(galaxies2.compress(cond))
icd = []
for i in range(len(grid)):
icd.append([galaxy.ICD_IH*100 for galaxy in grid[i]])
from boxplot_percentile import percentile_box_plot as pbp
pbp(f1s1, icd, indexer=list(pyl.delete(bins_x,-1)+0.25))
pbp(f1s2, icd, indexer=list(pyl.delete(bins_x,-1)+0.25))
# Add the box and whiskers
galaxies = pickle.load(open('galaxies.pickle','rb'))
galaxies2 = filter(lambda galaxy: galaxy.ston_V > 30., galaxies)
galaxies2 = pyl.asarray(galaxies2)
x = [galaxy.z for galaxy in galaxies2]
ll = 1.5
ul= 4
bins_x =pyl.arange(ll, ul, 0.5)
grid = []
for i in range(bins_x.size-1):
xmin = bins_x[i]
xmax = bins_x[i+1]
cond=[cond1 and cond2 for cond1, cond2 in zip(x>=xmin, x<xmax)]
grid.append(galaxies2.compress(cond))
icd = []
for i in range(len(grid)):
icd.append([galaxy.ICD_VJ*100 for galaxy in grid[i]])
pbp(f1s3, icd, indexer=list(pyl.delete(bins_x,-1)+0.25))
pbp(f1s4, icd, indexer=list(pyl.delete(bins_x,-1)+0.25))
f1s1.set_xlim(1.5, 3.5)
f1s2.set_xlim(1.5, 3.5)
f1s3.set_xlim(1.5, 3.5)
f1s4.set_xlim(1.5, 3.5)
f1s1.set_ylim(-5, 50)
f1s2.set_ylim(-1, 10)
f1s3.set_ylim(-5, 50)
f1s4.set_ylim(-1, 10)
f1s1.set_xticks([1.5,2,2.5,3,3.5])
f1s2.set_xticks([1.5,2,2.5,3,3.5])
f1s1.set_xticklabels([])
f1s2.set_xticklabels([])
f1s3.set_xticks([1.5,2,2.5,3,3.5])
f1s4.set_xticks([1.5,2,2.5,3,3.5])
f1s1.axhline(0.0, lw=2, c='b', zorder=0)
f1s2.axhline(0.0, lw=2, c='b', zorder=0)
f1s3.axhline(0.0, lw=2, c='b', zorder=0)
f1s4.axhline(0.0, lw=2, c='b', zorder=0)
f1s3.set_xlabel("Redshift")
f1s1.set_ylabel(r"$\xi[i_{775},H_{160}]$ (%)")
f1s4.set_xlabel("Redshift")
f1s3.set_ylabel(r"$\xi[V_{606},J_{125}]$ (%)")
import matplotlib.font_manager
line1 = pyl.Line2D([], [], marker='o', mfc='0.8', mec='0.8', markersize=8,
linewidth=0)
line2 = pyl.Line2D([], [], marker='s', mec='blue', mfc='None',
markersize=10, linewidth=0, markeredgewidth=2)
line3 = pyl.Line2D([], [], color='r', linewidth=2)
prop = matplotlib.font_manager.FontProperties(size='small')
pyl.figlegend((line1, line2, line3), ('Data', 'Quartiles',
'Medians'), 'lower center', prop=prop, ncol=3)
from matplotlib.patches import ConnectionPatch
xy = (3.5, 10)
xy2 = (1.5, 10)
con = ConnectionPatch(xyA=xy, xyB=xy2, coordsA='data', coordsB='data',
axesA=f1s1, axesB=f1s2)
xy = (3.5, -1)
xy2 = (1.5, -1)
con2 = ConnectionPatch(xyA=xy, xyB=xy2, coordsA='data', coordsB='data',
axesA=f1s1, axesB=f1s2)
f1s1.add_artist(con)
f1s1.add_artist(con2)
xy = (3.5, 10)
xy2 = (1.5, 10)
con = ConnectionPatch(xyA=xy, xyB=xy2, coordsA='data', coordsB='data',
axesA=f1s3, axesB=f1s4)
xy = (3.5, -1)
xy2 = (1.5, -1)
con2 = ConnectionPatch(xyA=xy, xyB=xy2, coordsA='data', coordsB='data',
axesA=f1s3, axesB=f1s4)
f1s3.add_artist(con)
f1s3.add_artist(con2)
pyl.draw()
pyl.show()
def plot_fingerprint(data, alpha=0.7, \
show_legend=True, graph_name='fingerprint.png', has_mean=True,
which_blocks='quartets', multiple=False, graph_grid='t', prob_axes=True, \
edge_colors='k', **kwargs):
"""Outputs a bubble plot of four-codon amino acid blocks
labeled with the colors from Sueoka 2002.
takes: data: array-elements in the col order x, y, r of
each of the four codon Amino Acids in the row order:
ALA, ARG4, GLY, LEU4, PRO, SER, THR, VAL
(for traditional fingerprint), or:
UU -> GG (for 16-block fingerprint).
last row is the mean (if has_mean is set True)
**kwargs passed on to init_graph_display (these include
graph_shape, graph_grid, x_label, y_label, dark, with_parens).
title: will be printed on graph (default: 'Unknown Species')
num_genes (number of genes contributing to graph: default None)
NOTE: will not print if None.)
size: of graph in inches (default = 8.0)
alpha: transparency of bubbles
(ranges from 0, transparent, to 1, opaque; default 0.7)
show_legend: bool, default True, whether to print legend
graph_name: name of file to write (default 'fingerprint.png')
has_mean: whether the data contain the mean (default: True)
which_blocks: which codon blocks to print (default is 'quartets'
for the 4-codon amino acid blocks, but can also use 'all' for all
quartets or 'split' for just the split quartets.)
multiple: if False (the default), assumes it got a single block
of data. Otherwise, assumes multiple blocks of data in a list or
array.
edge_colors: if multiple is True (ignored otherwise), uses this
sequence of edge color strings to hand out edge colors to successive
series. Will iterate over this, so can be a string of 1-letter
color codes or a list of color names.
note: that the data are always expected to be in the range (0,1)
since we're plotting frequencies. axes, gid, etc. are hard-coded
to these values.
"""
#figure out which type of fingerprint plot we're doing, and get the
#right colors
if which_blocks == 'quartets':
blocks = CodonUsage.SingleAABlocks
elif which_blocks == 'split':
blocks = CodonUsage.SplitBlocks
else:
blocks = CodonUsage.Blocks
colors = [doublets_to_colors[i] for i in blocks]
#formatting the labels in latex
x_label="$G_3/(G_3+C_3)$"
y_label="$A_3/(A_3+T_3)$"
#initializing components of the graph
font,label_font_size=init_graph_display(graph_shape='sqr', \
graph_grid=graph_grid, x_label=x_label, \
y_label=y_label, prob_axes=prob_axes, **kwargs)
if not multiple:
data = [data]
alpha = broadcast(alpha, len(data))
edge_colors = broadcast(edge_colors, len(data))
for al, d, edge_color in zip(alpha, data, edge_colors):
#skip this series if no data
if d is None or not d.any():
continue
for i, color in enumerate(colors):
j = i+1
#note: doing these as slices because scatter_classic needs the
#extra level of nesting
patches = scatter_classic(d[i:j,0], d[i:j,1],
s=(d[i:j,2]/2), c=color)
#set alpha for the patches manually
for p in patches:
p.set_alpha(al)
p.set_edgecolor(edge_color)
#plot mean as its own point -- can't do cross with scatter
if has_mean:
mean_index = len(blocks) #next index after the blocks
plot([d[mean_index,0]], [d[mean_index,1]],
'-k+',markersize=label_font_size, alpha=al)
abbrev = CodonUsage.BlockAbbreviations
a = gca()
#if show_legend is True prints a legend in the right center area
if show_legend:
legend_key = [abbrev[b] for b in blocks]
#copy legend font properties from the x axis tick labels
legend_font_props = \
a.xaxis.get_label().get_fontproperties().copy()
legend_font_scale_factor = 0.7
curr_size = legend_font_props.get_size()
legend_font_props.set_size(curr_size*legend_font_scale_factor)
l = figlegend(a.patches[:len(blocks)],
legend_key,
prop=legend_font_props,
loc='center right',borderpad=0.1,labelspacing=0.5,
handlelength=1.0,handletextpad=0.5, borderaxespad=0.0)
#fix transparency of patches
for p in l.get_patches():
p.set_alpha(1)
#initialize the ticks
set_axis_to_probs()
init_ticks(a, label_font_size)
a.set_xticks([0, 0.5, 1])
a.set_yticks([0,0.5,1])
#output the figure
if graph_name is not None:
savefig(graph_name)
trnSequenceOutput = net.extrapolate(trnSequenceWashout,
len(trnSequenceTarget))
tstSequenceOutput = net.extrapolate(tstSequenceWashout,
len(tstSequenceTarget))
# plot training data
sp = subplot(211) # switch to the first subplot
cla() # clear the subplot
title("Training Set") # set the subplot's title
sp.set_autoscale_on(True) # enable autoscaling
targetline = plot(trnSequenceTarget, "r-") # plot the targets
sp.set_autoscale_on(False) # disable autoscaling
outputline = plot(trnSequenceOutput, "b-") # plot the actual output
# plot test data
sp = subplot(212)
cla()
title("Test Set")
sp.set_autoscale_on(True)
plot(tstSequenceTarget, "r-")
sp.set_autoscale_on(False)
plot(tstSequenceOutput, "b-")
# create a legend
figlegend((targetline, outputline), ('target', 'output'), ('upper right'))
# draw everything
draw()
show()
fake_vms_end_line = lines.Line2D([jvmBootDays,0], [jvmBootDays,0], label='VMS cache refresh end', color='c')
ax.add_line(fake_vms_start_line)
ax.add_line(fake_vms_end_line)
try_to_draw_vms_cache_refresh_lines()
# various chart options
smallEnvDict = getSmallEnvDict()
ax.set_title('gc events over time for %s %s %s %s %s %s' % (smallEnvDict['appname'], smallEnvDict['env'], smallEnvDict['ec2PublicHostname'], smallEnvDict['instanceID'], smallEnvDict['instanceType'], smallEnvDict['asg']))
ax.grid(True)
ax.set_xlabel('gc event start timestamp')
ax.set_ylabel('gc event duration (seconds)')
fig.autofmt_xdate()
plt.ylim([0,maxSTWGCEventDuration])
fig.set_size_inches(21,12)
pylab.figlegend(*ax.get_legend_handles_labels(), loc='lower center', ncol=5)
# BUG: add sar charts, including but not limited to cpu, network
# BUG: plot secs since jvm start
# BUG: visualize the facet data collected in vms-cache-refresh-facet-info.csv. Maybe leave this up to visualize-instance.
savedImageBaseFileName = vmsGCReportDirectory + os.path.sep + smallEnvDict['appname'] + '-gc-events-' + now
savedIamgeFileName = savedImageBaseFileName + '.png'
plt.savefig(savedIamgeFileName)
print "output on %s" % (savedIamgeFileName,)
savedIamgeFileName = savedImageBaseFileName + '.pdf'
plt.savefig(savedIamgeFileName)
print "output on %s" % (savedIamgeFileName,)
# Potential BUG: do we want to issue the plt.show()
plt.show()
pylab.xlabel(xlabel(tex(r'[112]') + ' tilt angle ' + tex(r'\theta')),
labelpad=xlabelpad)
pylab.ylabel(ylabel("Energy " + tex("(J/m^2)")), labelpad=ylabelpad)
offset = 0.9
if args.annotate:
pylab.annotate(annotate(tex(r'(1\bar{1}0)(1\bar{1}0)')),
xycoords=('data', 'axes fraction'),
xy=(0 + offset, 1.05),
va="bottom",
ha="center",
rotation=90)
handles, labels = ax.get_legend_handles_labels()
if not args.labels_off:
if unrelaxed:
lgd = pylab.figlegend(handles, labels, loc=legend_loc, ncol=2)
else:
lgd = pylab.figlegend(handles, labels, loc=legend_loc, ncol=3)
pylab.tight_layout()
if args.annotate:
pylab.subplots_adjust(bottom=0.175,
left=0.075,
right=0.98,
top=0.875,
wspace=0.2,
hspace=0.75)
else:
pylab.subplots_adjust(bottom=0.2,
left=0.075,
def plot_icd_vs_mass():
galaxies = pickle.load(open('galaxies.pickle','rb'))
arrow_up = [[0,0], [-1,-1], [0,0], [0,-2], [0,0], [1,-1], [0,0]]
# Make figure
f1 = pyl.figure(1, figsize=(4,6))
f1s1 = f1.add_subplot(211)
f1s2 = f1.add_subplot(212)
for galaxy in galaxies:
if galaxy.ston_I > 30. and galaxy.ICD_IH != None:
f1s1.scatter(galaxy.z, galaxy.ICD_IH * 100, c='0.8',
marker='o', s=25, edgecolor='0.8')
if galaxy.ston_V > 30. and galaxy.ICD_VJ != None:
f1s2.scatter(galaxy.z, galaxy.ICD_VJ * 100, c='0.8',
marker='o', s=25, edgecolor='0.8')
# Add the box and whiskers
galaxies2 = filter(lambda galaxy: galaxy.ston_I > 30., galaxies)
galaxies2 = pyl.asarray(galaxies2)
x = [galaxy.z for galaxy in galaxies2]
ll = 1.5
ul= 4
bins_x =pyl.arange(ll, ul, 0.5)
grid = []
for i in range(bins_x.size-1):
xmin = bins_x[i]
xmax = bins_x[i+1]
cond=[cond1 and cond2 for cond1, cond2 in zip(x>=xmin, x<xmax)]
grid.append(galaxies2.compress(cond))
icd = []
for i in range(len(grid)):
icd.append([galaxy.ICD_IH*100 for galaxy in grid[i]])
from boxplot_percentile import percentile_box_plot as pbp
pbp(f1s1, icd, indexer=list(pyl.delete(bins_x,-1)+0.25), whisker_top=None,
whisker_bottom=None)
# Add the box and whiskers
galaxies = pickle.load(open('galaxies.pickle','rb'))
galaxies2 = filter(lambda galaxy: galaxy.ston_V > 30., galaxies)
galaxies2 = pyl.asarray(galaxies2)
x = [galaxy.z for galaxy in galaxies2]
ll = 1.5
ul= 4
bins_x =pyl.arange(ll, ul, 0.5)
grid = []
for i in range(bins_x.size-1):
xmin = bins_x[i]
xmax = bins_x[i+1]
cond=[cond1 and cond2 for cond1, cond2 in zip(x>=xmin, x<xmax)]
grid.append(galaxies2.compress(cond))
icd = []
for i in range(len(grid)):
icd.append([galaxy.ICD_VJ*100 for galaxy in grid[i]])
pbp(f1s2, icd, indexer=list(pyl.delete(bins_x,-1)+0.25), whisker_top=None,
whisker_bottom=None)
f1s1.set_xlim(1.5, 3.5)
f1s2.set_xlim(1.5, 3.5)
f1s1.set_ylim(-1, 10)
f1s2.set_ylim(-1, 10)
f1s1.set_xticklabels([])
f1s2.set_xticks([1.5,2,2.5,3,3.5])
f1s1.set_ylabel(r"$\xi[i_{775},H_{160}]$ (%)")
f1s2.set_xlabel("Redshift")
f1s2.set_ylabel(r"$\xi[V_{606},J_{125}]$ (%)")
import matplotlib.font_manager
line1 = pyl.Line2D([], [], marker='o', mfc='0.8', mec='0.8', markersize=8,
linewidth=0)
line2 = pyl.Line2D([], [], marker='s', mec='#348ABD', mfc='None',
markersize=10, linewidth=0, markeredgewidth=2)
line3 = pyl.Line2D([], [], color='#A60628', linewidth=2)
prop = matplotlib.font_manager.FontProperties(size='small')
pyl.figlegend((line1, line2, line3), ('Data', 'Quartiles',
'Medians'), 'lower center', prop=prop, ncol=3)
pyl.show()
qdo.Task.SUCCEEDED: 'g',
qdo.Task.FAILED: 'r',
}
for state in [qdo.Task.WAITING, qdo.Task.SUCCEEDED, qdo.Task.PENDING,
qdo.Task.RUNNING, qdo.Task.FAILED]:
if not state in state_radec:
continue
ra,dec = state_radec[state]
p = plt.plot(ra, dec, '.', color=cmap.get(state, 'y'))
lp.append(p[0])
lt.append(state)
plt.xlim([0, 360])
plt.figlegend(lp, lt, 'upper left')
plt.savefig('status.png')
sys.exit(0)
allra = []
alldec = []
allstate = []
alltasks = []
# allra.append(ra)
# alldec.append(dec)
# allstate.append([state] * len(ra))
# alltasks.append(tasks)
def __doPlot(self):
if len(self.groupedPlots) + len(self.plots) == 0:
print("PlotLayout.plot(): No data to plot!")
return
if self.rcParams is not None:
pylab.rcParams.update(self.rcParams)
groupedPlotLengths = [
len(plots) for plots in self.groupedPlots.values()
]
if len(groupedPlotLengths) == 0:
maxRowLength = self.width
else:
maxRowLength = max(groupedPlotLengths)
numPlots = len(self.plots)
if numPlots == 0:
numExcessRows = 0
elif numPlots <= maxRowLength:
numExcessRows = 1
elif numPlots % maxRowLength == 0:
numExcessRows = numPlots / maxRowLength
else:
numExcessRows = (numPlots / maxRowLength) + 1
numRows = len(self.groupedPlots.keys()) + numExcessRows
if self.groupOrder is not None:
keyList = self.groupOrder
else:
keyList = self.groupedPlots.keys()
currentRow = 0
if self.figdimensions is not None:
fig = pyplot.figure(figsize=(self.figdimensions[0],
self.figdimensions[1]))
elif self.dimensions is not None:
fig = pyplot.figure(figsize=(self.dimensions[0] * maxRowLength,
self.dimensions[1] * numRows))
else:
(figWidth, figHeight) = getGoldenRatioDimensions(8.0)
figWidth *= maxRowLength
figHeight *= numRows
fig = pyplot.figure(figsize=(figWidth, figHeight))
# figWidth = fig.get_figwidth()
# print figWidth
# print fig.get_figheight()
# (goldenWidth, goldenHeight) = getGoldenRatioDimensions(figWidth)
# fig.set_figheight(goldenHeight)
# print fig.get_figheight()
plotHandles = []
plotLabels = []
for grouping in keyList:
currentColumn = 1
plots = self.groupedPlots[grouping]
numPlots = len(plots)
for plot in plots:
myRows = numRows
myCols = numPlots
myPos = (currentRow * numPlots) + currentColumn
(currPlotHandles,
currPlotLabels) = plot.subplot(myRows, myCols, myPos)
for i in range(len(currPlotHandles)):
if currPlotLabels[i] in plotLabels:
continue
if isinstance(currPlotHandles[i], list):
plotHandles.append(currPlotHandles[i][0])
else:
plotHandles.append(currPlotHandles[i])
plotLabels.append(currPlotLabels[i])
# plotHandles.extend(currPlotHandles)
# plotLabels.extend(currPlotLabels)
currentColumn += 1
currentRow += 1
ungroupedPlotsRemaining = len(self.plots)
if ungroupedPlotsRemaining > 0:
currentColumn = 1
for plot in self.plots:
if currentColumn == 1:
numColumns = min(maxRowLength, ungroupedPlotsRemaining)
myRows = numRows
myCols = numColumns
myPos = (currentRow * numColumns) + currentColumn
(currPlotHandles,
currPlotLabels) = plot.subplot(myRows, myCols, myPos)
for i in range(len(currPlotHandles)):
if currPlotLabels[i] in plotLabels:
continue
if isinstance(currPlotHandles[i], list):
plotHandles.append(currPlotHandles[i][0])
else:
plotHandles.append(currPlotHandles[i])
plotLabels.append(currPlotLabels[i])
currentColumn += 1
if currentColumn > numColumns:
currentRow += 1
currentColumn = 1
ungroupedPlotsRemaining -= 1
if self.figLegendLoc is not None:
figLegendKeywords = {}
if self.figLegendCols is not None:
versionPieces = [
int(x) for x in matplotlib.__version__.split('.')
]
(superMajor, major, minor) = versionPieces[0:3]
if superMajor == 0 and major < 98:
print >> sys.stderr, "Number of columns support not available in versions of matplotlib prior to 0.98"
else:
figLegendKeywords["ncol"] = self.figLegendCols
pylab.figlegend(plotHandles, plotLabels, self.figLegendLoc,
**figLegendKeywords)
if self.plotParams is not None:
pylab.subplots_adjust(left=self.plotParams["left"],
bottom=self.plotParams["bottom"],
right=self.plotParams["right"],
top=self.plotParams["top"],
wspace=self.plotParams["wspace"],
hspace=self.plotParams["hspace"])
def main():
parser = argparse.ArgumentParser()
parser.add_argument('log_dir', nargs='*')
parser.add_argument('--n_cols', type=int, default=4)
parser.add_argument('--smoothing', type=float, default=0.6)
args = parser.parse_args()
if not args.log_dir:
raise Exception("Specify at least one log directory")
for d in args.log_dir:
if not os.path.exists(d):
raise Exception(f"Directory {d} does not exist")
print("Reading events...")
events_by_log_dir = read_all_events(args.log_dir)
i = len(os.path.commonprefix(args.log_dir))
len_longest_suffix = len(os.path.commonprefix([s[::-1] for s in args.log_dir]))
if len_longest_suffix > 0:
j = -len_longest_suffix
else:
j = None
print("Plotting...")
lines, labels = [], []
plot_n = 1
axes = {}
n_graphs = len(set([key for events in events_by_log_dir.values() for key in events.keys()])) + 1
subplot_dims = (int(ceil(n_graphs / args.n_cols)), args.n_cols)
figure(figsize=(6 * subplot_dims[1], 4 * subplot_dims[0]))
for log_dir_n, (log_dir, events) in enumerate(events_by_log_dir.items()):
print(log_dir)
color = f"C{log_dir_n % 10}"
label = log_dir[i:j]
for key, events in events.items():
if key not in axes:
axes[key] = subplot(*(subplot_dims + (plot_n,)))
plot_n += 1
ax = axes[key]
timestamps, values = list(zip(*events))
relative_timestamps = [t - timestamps[0] for t in timestamps]
relative_timestamps_hours = [t / 3600 for t in relative_timestamps]
line = plot_values(ax, values, relative_timestamps_hours, 'Value', 'Time (hours)', key,
label=label, color=color, smoothing=args.smoothing)
lines.append(line)
labels.append(label)
unique_lines = []
unique_labels = []
for line, label in zip(lines, labels):
if label not in unique_labels:
unique_labels.append(label)
unique_lines.append(line)
figlegend(unique_lines, unique_labels, loc='upper center')
tight_layout(rect=[0, 0, 1, 0.90])
print("Saving figure...")
# Removing trailing slashes
normed_log_dirs = [os.path.normpath(d) for d in args.log_dir]
if len(args.log_dir) == 1:
out_filename = os.path.basename(normed_log_dirs[0])
else:
longest_common_prefix = os.path.commonprefix(normed_log_dirs)
longest_common_suffix = os.path.commonprefix([d[::-1] for d in normed_log_dirs])[::-1]
out_filename = longest_common_prefix + '*' + longest_common_suffix + '.png'
savefig(out_filename)
rr,dd = wcs.pixelxy2radec(np.array([1,W,W,1,1]),
np.array([1,1,H,H,1]))
print 'included:', np.sum(T.included)
T.use = (T.use == 1)
T.included = (T.included == 1)
print 'included:', len(np.flatnonzero(T.included))
plt.clf()
plt.plot(T.ra, T.dec, 'r.')
I = (T.qual_frame == 0)
p1 = plt.plot(T.ra[I], T.dec[I], 'mx')
p2 = plt.plot(T.ra[T.use], T.dec[T.use], 'b.')
I = (T.npixrchi > (T.npixoverlap * 0.01))
p3 = plt.plot(T.ra[I], T.dec[I], 'r+')
I = T.included
p4 = plt.plot(T.ra[I], T.dec[I], 'go')
plt.plot(rr, dd, 'k-')
plt.figlegend((p[0] for p in (p1,p2,p3,p4)),
('Bad qual_frame', 'Used', 'Bad rchi', 'Included'),
loc='upper right')
ps.savefig()