def run(self):
#Get the columns we need, in reduced form
#since this is not MCMC
weight = self.weight_col()
weight = weight / weight.max()
#Sort the final 500 samples, since they
#are otherwise un-ordered
weight[-500:] = np.sort(weight[-500:])
#x-axis is just the iteration number
x = np.arange(weight.size)
#Choose a filename and make a figure for your plot.
figure, filename = self.figure("nest_weight")
#Do the plotting, and set the limits and labels
pylab.plot(x, weight / weight.max())
#pylab.xlim(-353, -341.5)
pylab.xlim(0, 8000)
pylab.xlabel("Iteration number $i$")
pylab.ylabel(r"$p_i$")
#Add some helpful text, on the title and the plot itself
pylab.title("Normalised posterior weight $p_i$", size='medium')
pylab.text(500,
0.4,
"Increasing likelihood,\n volume still large",
size="small")
pylab.text(5500, 0.65, "Volume\ndecreasing", size='small')
pylab.text(6500, 0.12, "Final $n_\mathrm{live}$\npoints", size='small')
#Return the list of filenames we made
return [filename]
def plot(self):
super(ShearCorrelationMinusPlot, self).plot()
nbin = 0
for i in range(1, 100):
filename = self.file_path("shear_xi_minus",
"bin_{0}_{0}".format(i))
if os.path.exists(filename):
nbin += 1
else:
break
if nbin == 0:
IOError("No data for plot: %s" % self.__class__.__name__[:-4])
theta = self.load_file("shear_xi_minus", "theta")
sz = 1.0 / (nbin + 2)
for i in range(1, nbin + 1):
for j in range(1, i + 1):
rect = (i * sz, j * sz, sz, sz)
self.figure.add_axes(rect)
#pylab.ploy()
#pylab.subplot(nbin, nbin, (nbin*nbin)-nbin*(j-1)+i)
xi = self.load_file("shear_xi_minus",
"bin_{0}_{1}".format(i, j))
pylab.loglog(theta, xi)
pylab.xlim(1e-4, 1e-1)
pylab.ylim(2e-7, 1e-3)
if i == 1 and j == 1:
pylab.xlabel("$\\theta$")
pylab.ylabel("$\\xi_+(\\theta)$")
else:
pylab.gca().xaxis.set_ticklabels([])
pylab.gca().yaxis.set_ticklabels([])
pylab.gca().tick_params(length=0.0, which='minor')
pylab.gca().tick_params(length=3.0, which='major')
pylab.gca().tick_params(labelsize=10)
pylab.text(1.5e-3,
1.8e-4,
"(%d,%d)" % (i, j),
fontsize=8,
color='red')
pylab.grid()
def make_2d_plot(self, name1, name2):
#Get the data
if 'planck' in self.source.name.lower():
def smooth_likelihood(obj, x, y):
n = obj.options.get("n_kde", 100)
factor = obj.options.get("factor_kde", 2.0)
kde = KDE([x, y], factor=factor)
x_range = (x.min(), x.max())
y_range = (y.min(), y.max())
(x_axis,
y_axis), like = kde.grid_evaluate(n, [x_range, y_range])
return n, x_axis, y_axis, like
x = self.reduced_col(name1)
y = self.reduced_col(name2)
else:
if name1[-2:] == 's8':
omm = self.reduced_col('cosmological_parameters--omega_m')
sigma_8 = self.reduced_col('cosmological_parameters--sigma_8')
x = sigma_8 * (omm / 0.3)**0.5
else:
x = self.find_or_pad(name1)
if name2[-2:] == 's8':
omm = self.reduced_col('cosmological_parameters--omega_m')
sigma_8 = self.reduced_col('cosmological_parameters--sigma_8')
y = sigma_8 * (omm / 0.3)**0.5
else:
y = self.find_or_pad(name2)
if x.max() - x.min() == 0 or y.max() - y.min() == 0:
return
print " (making %s vs %s)" % (name1, name2)
#import pdb ; pdb.set_trace()
#Interpolate using KDE
try:
n, x_axis, y_axis, like = self.smooth_likelihood(x, y)
except:
try:
n, x_axis, y_axis, like = self.smooth_likelihood_noweight(x, y)
except:
def smooth_likelihood(obj, x, y):
n = 100
factor = 2.0
kde = KDE([x, y], factor=factor)
x_range = (x.min(), x.max())
y_range = (y.min(), y.max())
(x_axis,
y_axis), like = kde.grid_evaluate(n, [x_range, y_range])
return n, x_axis, y_axis, like
n, x_axis, y_axis, like = smooth_likelihood(self, x, y)
figure, filename = self.figure("2D", name1, name2)
#Choose levels at which to plot contours
contour1 = 1 - 0.68
contour2 = 1 - 0.95
try:
level1, level2, total_mass = self._find_contours(
like, x, y, n, x_axis[0], x_axis[-1], y_axis[0], y_axis[-1],
contour1, contour2)
except:
def find_contours(like, x, y, n, xmin, xmax, ymin, ymax, contour1,
contour2):
N = len(x)
x_axis = np.linspace(xmin, xmax, n + 1)
y_axis = np.linspace(ymin, ymax, n + 1)
histogram, _, _ = np.histogram2d(x, y, bins=[x_axis, y_axis])
def objective(limit, target):
w = np.where(like >= limit)
count = histogram[w]
return count.sum() - target
target1 = histogram.sum() * (1 - contour1)
target2 = histogram.sum() * (1 - contour2)
level1 = scipy.optimize.bisect(objective,
like.min(),
like.max(),
args=(target1, ))
level2 = scipy.optimize.bisect(objective,
like.min(),
like.max(),
args=(target2, ))
return level1, level2, like.sum()
level1, level2, total_mass = find_contours(like, x, y, n,
x_axis[0], x_axis[-1],
y_axis[0], y_axis[-1],
contour1, contour2)
#import pdb ; pdb.set_trace()
level0 = 1.1
levels = [level2, level1, level0]
#Create the figure
fig, filename = self.figure("2D", name1, name2)
pylab.figure(fig.number)
#Make the plot
#pylab.figure(figure.number, figsize=(8,7))
keywords = self.keywords_2d()
fill = self.options.get("fill", True)
if self.fill_list != None:
fill = self.fill_list[self.source.index]
imshow = self.imshow[self.source.index]
#plot_points = self.options.get("plot_points", False)
color = self.colors[self.source.index]
linestyle = self.linestyles[self.source.index]
if self.linecolors[self.source.index] is not None:
linecolor = self.linecolors[self.source.index]
else:
linecolor = color
print self.source.index, linecolor
if self.opaque_list[self.source.index]:
pylab.contourf(x_axis,
y_axis,
like.T, [level2, level0],
colors=['white'])
print "fill", self.source.index, fill
if imshow:
pylab.imshow(like.T,
extent=(x_axis[0], x_axis[-1], y_axis[0], y_axis[-1]),
aspect='auto',
origin='lower',
cmap=cmap_white_to_color(color))
elif fill:
if self.fill_colors[self.source.index] is not None:
c = self.fill_colors[self.source.index]
pylab.contourf(x_axis,
y_axis,
like.T, [level2, level0],
colors=[c[0]])
pylab.contourf(x_axis,
y_axis,
like.T, [level1, level0],
colors=[c[1]])
else:
pylab.contourf(x_axis,
y_axis,
like.T, [level2, level0],
colors=[color],
alpha=self.alphas[self.source.index])
if self.opaque_centre[self.source.index]:
pylab.contourf(x_axis,
y_axis,
like.T, [level1, level0],
colors=[color])
else:
pylab.contourf(x_axis,
y_axis,
like.T, [level1, level0],
colors=[color],
alpha=self.alphas[self.source.index])
if (not fill) or self.line_list[self.source.index]:
print self.source.index, linecolor
pylab.contour(x_axis,
y_axis,
like.T, [level2, level1],
colors=[linecolor],
linestyles=linestyle,
linewidths=self.linewidths[self.source.index])
#if self.labels is not None:
# self.proxies.append(pylab.plot([],[], color=color, linestyle=linestyle, linewidth=self.linewidths[self.source.index])[0])
if self.labels is not None:
if self.fill_colors[self.source.index] is not None:
print self.fill_colors[self.source.index]
elif fill:
print self.source.index, self.line_list[self.source.index]
if self.line_list[self.source.index] is not None:
conv = matplotlib.colors.ColorConverter()
edgecolor = linecolor
self.proxies.append(
pylab.Line2D([0, 2.5], [0, 0],
linestyle=linestyle,
linewidth=3,
color=linecolor))
#facecolor=conv.to_rgba(color,alpha=1-(1-self.alphas[self.source.index])**2)
#self.proxies.append(pylab.Rectangle((0,0),1,1,facecolor=facecolor,edgecolor=edgecolor))#,alpha=(1-(1-self.alphas[self.source.index])**2)))
else:
self.proxies.append(
pylab.Rectangle(
(0, 0),
1,
1,
fc=color,
edgecolor=color,
alpha=(1 -
(1 - self.alphas[self.source.index])**2)))
else:
self.proxies.append(
pylab.plot(
[], [],
color=color,
linestyle=linestyle,
linewidth=self.linewidths[self.source.index])[0])
if self.plot_points is not None:
point_markers = ['x', '+', '^']
for p, m in zip(self.plot_points, point_markers):
print p, m
pylab.plot(p[0],
p[1],
m,
color='k',
markersize=10,
markerfacecolor='none')
#Do the labels
print self.proxies, self.labels
if self.labels is not None:
leg = pylab.legend(self.proxies,
self.labels,
loc="upper left",
fontsize=13)
leg.get_frame().set_alpha(
0.75) # this will make the box totally transparent
leg.get_frame().set_edgecolor(
'white') # this will make the edges of the
pylab.ylabel(r"$A_\mathrm{IA}$", fontsize=30, fontname='serif')
pylab.xlabel(r"$b_g$", fontsize=30, fontname='serif')
pylab.tight_layout()
for col, text in zip(self.colors, leg.get_texts()):
text.set_color(col)
if col == 'pink':
text.set_color('hotpink')
if blind:
pylab.xticks(visible=False)
pylab.xticks(visible=True, fontsize=25)
pylab.yticks(visible=True, fontsize=25)
pylab.xlim(self.axis[0], self.axis[1])
pylab.ylim(self.axis[2], self.axis[3])
pylab.title('$z=1.0$', fontsize=25)
# pylab.subplots_adjust(top=0.95)
x0 = [4.6]
dx = [0.5]
y0 = [1.77]
pylab.errorbar(x0,
y0,
xerr=dx,
yerr=[0.04],
markersize=9.0,
markerfacecolor='k',
color='k',
marker='*',
markeredgecolor='k',
linestyle='none')
# pylab.errorbar([4.35341365],[1.80416667], xerr=7.080459e-01, yerr=1.706998e-02,color='pink',marker='o')
#pylab.title('$\gamma \gamma + \delta_g \gamma + \delta_g\delta_g$')
#pylab.axhline(0,color="k",alpha=0.5,ls=':')
#pylab.axhline(5.0,color="darkviolet",lw=1, ls="--")
#print "Plotting bounds line..."
return filename
def plot(self):
super(BBPlot, self).plot()
pylab.xlim(0, 400)
def make_1d_plot(self, name1):
#Get the data
#import pdb ; pdb.set_trace()
filename = self.filename(name1)
col1 = self.source.get_col(name1)
like = self.source.get_col("post")
vals1 = np.unique(col1)
n1 = len(vals1)
like_sum = np.zeros(n1)
#normalizing like this is a big help
#numerically
like = like-like.max()
#marginalize
for k,v1 in enumerate(vals1):
w = np.where(col1==v1)
like_sum[k] = np.log(np.exp(like[w]).sum())
like = like_sum.flatten()
like -= like.max()
#linearly interpolate
n1_interp = n1*10
vals1_interp = np.linspace(vals1[0], vals1[-1], n1_interp)
like_interp = np.interp(vals1_interp, vals1, like)
if np.isfinite(like_interp).any():
vals1 = vals1_interp
like = like_interp
n1 = n1_interp
else:
print()
print("Parameter %s has a very wide range in likelihoods " % name1)
print("So I couldn't do a smooth likelihood interpolation for plotting")
print()
#normalize
like[np.isnan(like)] = -np.inf
like -= like.max()
#Determine the spacing in the different parameters
dx = vals1[1]-vals1[0]
#Set up the figure
fig,filename = self.figure(name1)
pylab.figure(fig.number)
#Plot the likelihood
print('----------------------- HITTA',self.source.index, self.colors[self.source.index])
pylab.plot(vals1, np.exp(like), linewidth=self.linewidths[self.source.index], label=self.labels[self.source.index], color=self.colors[self.source.index], linestyle='-')
self.proxies.append(pylab.Line2D([0,2.5],[0,0], color=self.colors[self.source.index], linewidth=self.linewidths[self.source.index]))
#Find the levels of the 68% and 95% contours
X, L = self.find_edges(np.exp(like), 0.68, 0.95, vals1)
#Plot black dotted lines from the y-axis at these contour levels
for (x, l) in zip(X,L):
if np.isnan(x[0]):
continue
pylab.plot([x[0],x[0]], [0, l[0]], ':', color='black')
pylab.plot([x[1],x[1]], [0, l[1]], ':', color='black')
#import pdb ; pdb.set_trace()
if self.labels is not None:
leg = pylab.legend(self.proxies,self.labels,loc="upper left", fontsize=16)
leg.get_frame().set_alpha(0.75) # this will make the box totally transparent
leg.get_frame().set_edgecolor('white') # this will make the edges of the
#Set the x and y limits
pylab.ylim(0,1.05)
pylab.yticks(visible=False)
pylab.xticks(visible=True, fontsize=16)
pylab.xlim(self.axis[0],self.axis[1])
#Add label
pylab.xlabel('$A_1$', fontsize=18)
return filename
def make_2d_plot(self, name1, name2):
#Get the data
x = self.reduced_col(name1)
y = self.reduced_col(name2)
if x.max()-x.min()==0 or y.max()-y.min()==0:
return
print " (making %s vs %s)" % (name1, name2)
#Interpolate using KDE
try:
n, x_axis, y_axis, like = self.smooth_likelihood(x, y)
except np.linalg.LinAlgError:
print " -- these two parameters have singular covariance - probably a linear relation"
print "Not making a 2D plot of them"
return []
figure,filename = self.figure("2D", name1, name2)
#Choose levels at which to plot contours
contour1=1-0.68
contour2=1-0.95
try:
level1, level2, total_mass = self._find_contours(like, x, y, n, x_axis[0], x_axis[-1], y_axis[0], y_axis[-1], contour1, contour2)
except:
import pdb ; pdb.set_trace()
level0 = 1.1
levels = [level2, level1, level0]
#Create the figure
fig,filename = self.figure("2D", name1, name2)
pylab.figure(fig.number)
#Make the plot
#pylab.figure(figure.number, figsize=(8,7))
keywords = self.keywords_2d()
fill = self.options.get("fill", True)
if self.fill_list!=None:
fill = self.fill_list[self.source.index]
imshow = self.imshow[self.source.index]
#plot_points = self.options.get("plot_points", False)
color = self.colors[self.source.index]
linestyle = self.linestyles[self.source.index]
if self.linecolors[self.source.index] is not None:
linecolor=self.linecolors[self.source.index]
else:
linecolor=color
print self.source.index,linecolor
if self.opaque_list[self.source.index]:
pylab.contourf(x_axis, y_axis, like.T, [level2,level0], colors=['white'])
print "fill",self.source.index, fill
if imshow:
pylab.imshow(like.T, extent=(x_axis[0], x_axis[-1], y_axis[0], y_axis[-1]), aspect='auto', origin='lower', cmap=cmap_white_to_color(color))
elif fill:
if self.fill_colors[self.source.index] is not None:
c=self.fill_colors[self.source.index]
pylab.contourf(x_axis, y_axis, like.T, [level2,level0], colors=[c[0]])
pylab.contourf(x_axis, y_axis, like.T, [level1,level0], colors=[c[1]])
else:
pylab.contourf(x_axis, y_axis, like.T, [level2,level0], colors=[color], alpha=self.alphas[self.source.index])
if self.opaque_centre[self.source.index]:
pylab.contourf(x_axis, y_axis, like.T, [level1,level0], colors=[color])
else:
pylab.contourf(x_axis, y_axis, like.T, [level1,level0], colors=[color], alpha=self.alphas[self.source.index])
if (not fill) or self.line_list[self.source.index]:
print self.source.index,linecolor
pylab.contour(x_axis, y_axis, like.T, [level2,level1], colors=[linecolor], linestyles=linestyle, linewidths=self.linewidths[self.source.index])
#if self.labels is not None:
# self.proxies.append(pylab.plot([],[], color=color, linestyle=linestyle, linewidth=self.linewidths[self.source.index])[0])
if self.labels is not None:
if self.fill_colors[self.source.index] is not None:
print self.fill_colors[self.source.index]
elif fill:
print self.source.index,self.line_list[self.source.index]
if self.line_list[self.source.index] is not None:
conv=matplotlib.colors.ColorConverter()
edgecolor=linecolor
self.proxies.append(pylab.Line2D([0,2.5], [0,0], linestyle=linestyle, linewidth=3, color=linecolor))
#facecolor=conv.to_rgba(color,alpha=1-(1-self.alphas[self.source.index])**2)
#self.proxies.append(pylab.Rectangle((0,0),1,1,facecolor=facecolor,edgecolor=edgecolor))#,alpha=(1-(1-self.alphas[self.source.index])**2)))
else:
self.proxies.append(pylab.Rectangle((0,0),1,1,fc=color,edgecolor=color,alpha=(1-(1-self.alphas[self.source.index])**2)))
else:
self.proxies.append(pylab.plot([],[], color=color, linestyle=linestyle, linewidth=self.linewidths[self.source.index])[0])
if self.plot_points is not None:
point_markers=['x','+','^']
for p,m in zip(self.plot_points,point_markers):
print p,m
pylab.plot(p[0], p[1], m, color='k', markersize=10, markerfacecolor='none')
#Do the labels
print self.proxies,self.labels
if self.labels is not None:
leg=pylab.legend(self.proxies,self.labels,loc="upper right")
leg.get_frame().set_alpha(0) # this will make the box totally transparent
leg.get_frame().set_edgecolor('white') # this will make the edges of the
pylab.xlabel(r"$S_8 \equiv \sigma_8 \left( \Omega_\mathrm{m} / 0.3 \right )^{\frac{1}{2}}$",fontsize=22)
pylab.ylabel("$A_\mathrm{IA}$", fontsize=22)
pylab.tight_layout()
if blind:
pylab.xticks(visible=False)
pylab.yticks(visible=False)
np.random.seed(2007)
dx = (np.random.rand()*2 - 1.0) * (0.4-1.2)*0.1
#pylab.xticks(np.array([0.4,0.5,0.6,0.7,0.8,0.9,1.0,1.1,1.2])+dx,["0.4","0.5","0.6","0.7",'0.8','0.9','1.0','1.1','1.2'])
#pylab.yticks(np.array([0.6,0.7,0.8,0.9,1.0])+d
pylab.xlim(self.axis[0], self.axis[1])
pylab.ylim(self.axis[2], self.axis[3])
pylab.xlim(self.axis[0], self.axis[1])
pylab.ylim(self.axis[2], self.axis[3])
#pylab.axhline(5.0,color="darkviolet",lw=1, ls="--")
#print "Plotting bounds line..."
return filename