def plot_correlation_demo2():
np.random.seed(0) # to reproduce the data later on
pylab.clf()
pylab.figure(num=None, figsize=(8, 8))
x = np.arange(-5, 5, 0.2)
pylab.subplot(221)
y = 0.5 * x**2 + norm.rvs(1, scale=.01, size=len(x))
_plot_correlation_func(x, y)
pylab.subplot(222)
y = 0.5 * x**2 + norm.rvs(1, scale=.1, size=len(x))
_plot_correlation_func(x, y)
pylab.subplot(223)
y = 0.5 * x**2 + norm.rvs(1, scale=1, size=len(x))
_plot_correlation_func(x, y)
pylab.subplot(224)
y = 0.5 * x**2 + norm.rvs(1, scale=10, size=len(x))
_plot_correlation_func(x, y)
pylab.autoscale(tight=True)
pylab.grid(True)
filename = "corr_demo_2.png"
def plot_p_leading_order(frame):
mat = scipy.io.loadmat('sound-speed_2D-wave.mat')
T = 5
nt = T / 0.5
pp = mat['U'][nt, :, :]
xx = mat['xx']
yy = mat['yy']
fig = pl.figure(figsize=(8, 3.5))
#pl.title("t= "+str(sol.state.t),fontsize=20)
pl.xticks(size=20)
pl.yticks(size=20)
pl.xlabel('x', fontsize=20)
pl.ylabel('y', fontsize=20)
#pl.pcolormesh(xx,yy,p_subxy,cmap=cm.OrRd)
pl.pcolormesh(xx, yy, pp, cmap='RdBu_r')
pl.autoscale(tight=True)
cb = pl.colorbar(ticks=[0.5, 1, 1.5, 2])
#pl.clim(ticks=[0.5,1,1.5,2])
imaxes = pl.gca()
pl.axes(cb.ax)
pl.yticks(fontsize=20)
pl.axes(imaxes)
#pl.xticks(fontsize=20); pl.axes(imaxes)
#pl.axis('equal')
pl.axis('tight')
fig.tight_layout()
pl.savefig('./_plots_to_paper/sound-speed_LO_t' + str(frame) +
'_pcolor.png')
pl.close()
def plot_mi_demo():
np.random.seed(0) # to reproduce the data later on
pylab.clf()
pylab.figure(num=None, figsize=(8, 8))
x = np.arange(0, 10, 0.2)
pylab.subplot(221)
y = 0.5 * x + norm.rvs(1, scale=.01, size=len(x))
_plot_mi_func(x, y)
# pylab.subplot(222)
# y = 0.5 * x + norm.rvs(1, scale=.1, size=len(x))
# _plot_mi_func(x, y)
# pylab.subplot(223)
# y = 0.5 * x + norm.rvs(1, scale=1, size=len(x))
# _plot_mi_func(x, y)
# pylab.subplot(224)
# y = norm.rvs(1, scale=10, size=len(x))
# _plot_mi_func(x, y)
pylab.autoscale(tight=True)
pylab.grid(True)
filename = "mi_demo_1.png"
pylab.savefig(os.path.join(CHART_DIR, filename), bbox_inches="tight")
def plot_mi_demo2():
np.random.seed(0)
pylab.clf()
pylab.figure(num=None, figsize=(8, 8))
x = np.arange(-5, 5, 0.2)
pylab.subplot(221)
y = 0.5 * x**2 + norm.rvs(1, scale=.01, size=len(x))
_plot_mi_func(x, y)
pylab.subplot(222)
y = 0.5 * x**2 + norm.rvs(1, scale=.1, size=len(x))
_plot_mi_func(x, y)
pylab.subplot(223)
y = 0.5 * x**2 + norm.rvs(1, scale=1, size=len(x))
_plot_mi_func(x, y)
pylab.subplot(224)
y = 0.5 * x**2 + norm.rvs(1, scale=10, size=len(x))
_plot_mi_func(x, y)
pylab.autoscale(tight=True)
pylab.grid(True)
filename = "mi_demo_2.png"
def plot_p(frame):
sol=Solution(frame,file_format='petsc',read_aux=False,path='./_output/_p/',file_prefix='claw_p')
x=sol.state.grid.x.centers; y=sol.state.grid.y.centers
mx=len(x); my=len(y)
mp=sol.state.num_eqn
yy,xx = np.meshgrid(y,x)
p=sol.state.q[0,:,:]
fig = pl.figure(figsize=(8, 3.5))
#pl.title("t= "+str(sol.state.t),fontsize=20)
pl.xticks(size=20); pl.yticks(size=20)
pl.xlabel('x',fontsize=20); pl.ylabel('y',fontsize=20)
#pl.pcolormesh(xx,yy,p_subxy,cmap=cm.OrRd)
pl.pcolormesh(xx,yy,p,cmap='RdBu_r')
pl.autoscale(tight=True)
cb = pl.colorbar(ticks=[0.5,1,1.5,2]);
#pl.clim(ticks=[0.5,1,1.5,2])
imaxes = pl.gca(); pl.axes(cb.ax)
pl.yticks(fontsize=20); pl.axes(imaxes)
#pl.xticks(fontsize=20); pl.axes(imaxes)
#pl.axis('equal')
pl.axis('tight')
fig.tight_layout()
pl.savefig('./_plots_to_paper/sound-speed_FV_t'+str(frame)+'_pcolor.png')
pl.close()
def plot_p_leading_order(frame):
mat = scipy.io.loadmat('sound-speed_2D-wave.mat')
T=5; nt=T/0.5
pp=mat['U'][nt,:,:]
xx=mat['xx']
yy=mat['yy']
fig=pl.figure(figsize=(8, 3.5))
#pl.title("t= "+str(sol.state.t),fontsize=20)
pl.xticks(size=20); pl.yticks(size=20)
pl.xlabel('x',fontsize=20); pl.ylabel('y',fontsize=20)
#pl.pcolormesh(xx,yy,p_subxy,cmap=cm.OrRd)
pl.pcolormesh(xx,yy,pp,cmap='RdBu_r')
pl.autoscale(tight=True)
cb = pl.colorbar(ticks=[0.5,1,1.5,2]);
#pl.clim(ticks=[0.5,1,1.5,2])
imaxes = pl.gca(); pl.axes(cb.ax)
pl.yticks(fontsize=20); pl.axes(imaxes)
#pl.xticks(fontsize=20); pl.axes(imaxes)
#pl.axis('equal')
pl.axis('tight')
fig.tight_layout()
pl.savefig('./_plots_to_paper/sound-speed_LO_t'+str(frame)+'_pcolor.png')
pl.close()
def plot_mi_demo1():
np.random.seed(0) # to reproduce the data later on
pylab.clf()
pylab.figure(num=None, figsize=(8, 8))
x = np.arange(0, 10, 0.2)
pylab.subplot(221)
y = 0.5 * x + norm.rvs(1, scale=.01, size=len(x))
_plot_mi_func(x, y)
pylab.subplot(222)
y = 0.5 * x + norm.rvs(1, scale=.1, size=len(x))
_plot_mi_func(x, y)
pylab.subplot(223)
y = 0.5 * x + norm.rvs(1, scale=1, size=len(x))
_plot_mi_func(x, y)
pylab.subplot(224)
y = norm.rvs(1, scale=10, size=len(x))
_plot_mi_func(x, y)
pylab.autoscale(tight=True)
pylab.grid(True)
filename = "mi_demo_1.png"
def plot_simple_demo_2():
pylab.clf()
fig = pylab.figure(num=None, figsize=(10, 4))
pylab.subplot(121)
title = "Original feature space"
pylab.title(title)
pylab.xlabel("$X_1$")
pylab.ylabel("$X_2$")
np.random.seed(3)
x1 = np.arange(0, 10, .2)
x2 = x1 + np.random.normal(loc=0, scale=1, size=len(x1))
good = x1 > x2
bad = ~good
x1g = x1[good]
x2g = x2[good]
pylab.scatter(x1g, x2g, edgecolor="blue", facecolor="blue")
x1b = x1[bad]
x2b = x2[bad]
pylab.scatter(x1b, x2b, edgecolor="red", facecolor="white")
pylab.grid(True)
pylab.subplot(122)
X = np.c_[(x1, x2)]
pca = decomposition.PCA(n_components=1)
Xtrans = pca.fit_transform(X)
Xg = Xtrans[good]
Xb = Xtrans[bad]
pylab.scatter(Xg[:, 0],
np.zeros(len(Xg)),
edgecolor="blue",
facecolor="blue")
pylab.scatter(Xb[:, 0],
np.zeros(len(Xb)),
edgecolor="red",
facecolor="white")
title = "Transformed feature space"
pylab.title(title)
pylab.xlabel("$X'$")
fig.axes[1].get_yaxis().set_visible(False)
print("======系统pca 数据2==========")
print("特征重要度:", pca.explained_variance_ratio_)
pylab.grid(True)
pylab.autoscale(tight=True)
filename = "pca_demo_2.png"
pylab.savefig(os.path.join(CHART_DIR, filename), bbox_inches="tight")
def plot_correlation_demo():
np.random.seed(0) # to reproduce the data later on
pylab.clf()
pylab.figure(num=None, figsize=(8, 8))
x = np.arange(0, 10, 0.2)
pylab.subplot(221)
y = 0.5 * x + norm.rvs(1, loc=0, scale=.01, size=len(x))
_plot_correlation_func(x, y)
pylab.subplot(222)
y = 0.5 * x + norm.rvs(1, loc=0, scale=.1, size=len(x))
_plot_correlation_func(x, y)
pylab.subplot(223)
y = 0.5 * x + norm.rvs(1, loc=0, scale=1, size=len(x))
_plot_correlation_func(x, y)
pylab.subplot(224)
y = norm.rvs(1, loc=0, scale=10, size=len(x))
_plot_correlation_func(x, y)
pylab.autoscale(tight=True)
pylab.grid(True)
filename = "corr_demo_1.png"
pylab.savefig(os.path.join(CHART_DIR, filename), bbox_inches="tight")
pylab.clf()
pylab.figure(num=None, figsize=(8, 8))
x = np.arange(-5, 5, 0.2)
pylab.subplot(221)
y = 0.5 * x ** 2 + norm.rvs(1, loc=0, scale=.01, size=len(x))
_plot_correlation_func(x, y)
pylab.subplot(222)
y = 0.5 * x ** 2 + norm.rvs(1, loc=0, scale=.1, size=len(x))
_plot_correlation_func(x, y)
pylab.subplot(223)
y = 0.5 * x ** 2 + norm.rvs(1, loc=0, scale=1, size=len(x))
_plot_correlation_func(x, y)
pylab.subplot(224)
y = 0.5 * x ** 2 + norm.rvs(1, loc=0, scale=10, size=len(x))
_plot_correlation_func(x, y)
pylab.autoscale(tight=True)
pylab.grid(True)
filename = "corr_demo_2.png"
pylab.savefig(os.path.join(CHART_DIR, filename), bbox_inches="tight")
def plot_simple_demo_lda():
#from sklearn import lda
# 与demo2不同的是,lda会利用(即有监督的标签信息,而pca并不会利用此信息)类别信息进行降维
from sklearn.discriminant_analysis import LinearDiscriminantAnalysis
pylab.clf()
fig = pylab.figure(num=None, figsize=(10, 4))
pylab.subplot(121)
title = "Original feature space"
pylab.title(title)
pylab.xlabel("$X_1$")
pylab.ylabel("$X_2$")
good = x1 > x2
bad = ~good
x1g = x1[good]
x2g = x2[good]
pylab.scatter(x1g, x2g, edgecolor="blue", facecolor="blue")
x1b = x1[bad]
x2b = x2[bad]
pylab.scatter(x1b, x2b, edgecolor="red", facecolor="white")
pylab.grid(True)
pylab.subplot(122)
X = np.c_[(x1, x2)]
#lda_inst = lda.LDA(n_components=1)
lda_inst = LinearDiscriminantAnalysis(n_components=1)
Xtrans = lda_inst.fit_transform(X, good)
Xg = Xtrans[good]
Xb = Xtrans[bad]
pylab.scatter(Xg[:, 0],
np.zeros(len(Xg)),
edgecolor="blue",
facecolor="blue")
pylab.scatter(Xb[:, 0],
np.zeros(len(Xb)),
edgecolor="red",
facecolor="white")
title = "Transformed feature space"
pylab.title(title)
pylab.xlabel("$X'$")
fig.axes[1].get_yaxis().set_visible(False)
pylab.grid(True)
pylab.autoscale(tight=True)
filename = "lda_demo.png"
pylab.savefig(os.path.join(CHART_DIR, filename), bbox_inches="tight")
def plot_correlation_demo():
np.random.seed(0)
pylab.clf()
pylab.figure(num=None, figsize=(8, 8))
x = np.arange(0, 10, 0.2)
pylab.subplot(221)
y = 0.5 * x + norm.rvs(1, scale=.01, size=len(x))
_plot_correlation_func(x, y)
pylab.subplot(222)
y = 0.5 * x + norm.rvs(1, scale=.1, size=len(x))
_plot_correlation_func(x, y)
pylab.subplot(223)
y = 0.5 * x + norm.rvs(1, scale=1, size=len(x))
_plot_correlation_func(x, y)
pylab.subplot(224)
y = norm.rvs(1, scale=10, size=len(x))
_plot_correlation_func(x, y)
pylab.autoscale(tight=True)
pylab.grid(True)
filename = "corr_demo_1.png"
pylab.savefig(os.path.join(CHART_DIR, filename), bbox_inches="tight")
pylab.clf()
pylab.figure(num=None, figsize=(8, 8))
x = np.arange(-5, 5, 0.2)
pylab.subplot(221)
y = 0.5 * x**2 + norm.rvs(1, scale=.01, size=len(x))
_plot_correlation_func(x, y)
pylab.subplot(222)
y = 0.5 * x**2 + norm.rvs(1, scale=.1, size=len(x))
_plot_correlation_func(x, y)
pylab.subplot(223)
y = 0.5 * x**2 + norm.rvs(1, scale=1, size=len(x))
_plot_correlation_func(x, y)
pylab.subplot(224)
y = 0.5 * x**2 + norm.rvs(1, scale=10, size=len(x))
_plot_correlation_func(x, y)
pylab.autoscale(tight=True)
pylab.grid(True)
filename = "corr_demo_2.png"
pylab.savefig(os.path.join(CHART_DIR, filename), bbox_inches="tight")
def plot_simple_demo_2():
pylab.clf()
fig = pylab.figure(num=None, figsize=(10, 4))
pylab.subplot(121)
title = "Original feature space"
pylab.title(title)
pylab.xlabel("$X_1$")
pylab.ylabel("$X_2$")
x1 = np.arange(0, 10, .2)
x2 = x1 + np.random.normal(loc=0, scale=1, size=len(x1))
good = x1 > x2
bad = ~good
x1g = x1[good]
x2g = x2[good]
pylab.scatter(x1g, x2g, edgecolor="blue", facecolor="blue")
x1b = x1[bad]
x2b = x2[bad]
pylab.scatter(x1b, x2b, edgecolor="red", facecolor="white")
pylab.grid(True)
pylab.subplot(122)
X = np.c_[(x1, x2)]
pca = decomposition.PCA(n_components=1)
Xtrans = pca.fit_transform(X)
Xg = Xtrans[good]
Xb = Xtrans[bad]
pylab.scatter(
Xg[:, 0], np.zeros(len(Xg)), edgecolor="blue", facecolor="blue")
pylab.scatter(
Xb[:, 0], np.zeros(len(Xb)), edgecolor="red", facecolor="white")
title = "Transformed feature space"
pylab.title(title)
pylab.xlabel("$X'$")
fig.axes[1].get_yaxis().set_visible(False)
print(pca.explained_variance_ratio_)
pylab.grid(True)
pylab.autoscale(tight=True)
filename = "pca_demo_2.png"
pylab.savefig(os.path.join(CHART_DIR, filename), bbox_inches="tight")
def Plot():
u = Calculator()
u = u[::-1]*4.47058
x = condition[:, 1]*1000000.
plt.plot(x, u)
plt.xlabel('micrometers depth')
plt.ylabel('argon concentration ppm')
plt.title('1D system state')
plt.grid(True)
#plt.savefig("F1D_result.png")
plt.autoscale(enable=True, axis='y')
plt.show()
return 'sucessfully plotted'
def plot_simple_demo_lda():
pylab.clf()
fig = pylab.figure(num=None, figsize=(10, 4))
pylab.subplot(121)
title = "Original feature space"
pylab.title(title)
pylab.xlabel("$X_1$")
pylab.ylabel("$X_2$")
good = x1 > x2
bad = ~good
x1g = x1[good]
x2g = x2[good]
pylab.scatter(x1g, x2g, edgecolor="blue", facecolor="blue")
x1b = x1[bad]
x2b = x2[bad]
pylab.scatter(x1b, x2b, edgecolor="red", facecolor="white")
pylab.grid(True)
pylab.subplot(122)
X = np.c_[(x1, x2)]
lda_inst = lda(n_components=1)
Xtrans = lda_inst.fit_transform(X, good)
Xg = Xtrans[good]
Xb = Xtrans[bad]
pylab.scatter(Xg[:, 0],
np.zeros(len(Xg)),
edgecolor="blue",
facecolor="blue")
pylab.scatter(Xb[:, 0],
np.zeros(len(Xb)),
edgecolor="red",
facecolor="white")
title = "Transformed feature space"
pylab.title(title)
pylab.xlabel("$X'$")
fig.axes[1].get_yaxis().set_visible(False)
pylab.grid(True)
pylab.autoscale(tight=True)
filename = "lda_demo.png"
pylab.savefig(os.path.join(CHART_DIR, filename), bbox_inches="tight")
def plot_clusters(x, y, title, mx=None, ymax=None, xmin=None, km=None):
pylab.figure(num=None, figsize=(8, 6))
if km:
pylab.scatter(x, y, s=50, c=km.predict(list(zip(x, y))))
else:
pylab.scatter(x, y, s=50)
pylab.title(title)
pylab.xlabel("Occurance of Word1")
pylab.ylabel("Occurance of Word2")
pylab.autoscale(tight=True)
pylab.ylim(ymin=0, ymax=1)
pylab.xlim(xmin=0, xmax=1)
pylab.grid(True, linestyle='-', color='0.75')
return pylab
def Plot():
u, u_sec = Calculator()
u = u[::-1]*4.47058
u_sec = u_sec[::-1]*2.17143
x = condition[:, 1]*1000000.
plt.plot(x, u)
plt.plot(x, u_sec)
plt.xlabel('micrometers depth')
plt.ylabel('argon concentration ppm')
plt.title('Zhang and measured D ranges')
plt.grid(True)
plt.savefig("F1D_result.png")
plt.autoscale(enable=True, axis='y')
plt.show()
def plot_simple_demo_lda():
pylab.clf()
fig = pylab.figure(num=None, figsize=(10, 4))
pylab.subplot(121)
title = "Original feature space"
pylab.title(title)
pylab.xlabel("$X_1$")
pylab.ylabel("$X_2$")
good = x1 > x2
bad = ~good
x1g = x1[good]
x2g = x2[good]
pylab.scatter(x1g, x2g, edgecolor="blue", facecolor="blue")
x1b = x1[bad]
x2b = x2[bad]
pylab.scatter(x1b, x2b, edgecolor="red", facecolor="white")
pylab.grid(True)
pylab.subplot(122)
X = np.c_[(x1, x2)]
lda_inst = lda.LDA(n_components=1)
Xtrans = lda_inst.fit_transform(X, good)
Xg = Xtrans[good]
Xb = Xtrans[bad]
pylab.scatter(
Xg[:, 0], np.zeros(len(Xg)), edgecolor="blue", facecolor="blue")
pylab.scatter(
Xb[:, 0], np.zeros(len(Xb)), edgecolor="red", facecolor="white")
title = "Transformed feature space"
pylab.title(title)
pylab.xlabel("$X'$")
fig.axes[1].get_yaxis().set_visible(False)
pylab.grid(True)
pylab.autoscale(tight=True)
filename = "lda_demo.png"
pylab.savefig(os.path.join(CHART_DIR, filename), bbox_inches="tight")
def plot_clustering(x, y, title, mx=None, ymax=None, xmin=None, km=None):
pylab.figure(num=None, figsize=(8, 6))
if km:
pylab.scatter(x, y, s=50, c=km.predict(list(zip(x, y))))
else:
pylab.scatter(x, y, s=50)
pylab.title(title)
pylab.xlabel("Occurrence word 1")
pylab.ylabel("Occurrence word 2")
pylab.autoscale(tight=True)
pylab.ylim(ymin=0, ymax=1)
pylab.xlim(xmin=0, xmax=1)
pylab.grid(True, linestyle='-', color='0.75')
return pylab
def plot_results(benchmark_series,
target_series,
target_balances,
n_assets,
columns,
name2plot = '',
path2save = './',
base_name_series = 'series'):
# N = len(np.array(benchmark_series).cumsum())
N = len(np.array([item for sublist in benchmark_series for item in sublist]).cumsum())
if not os.path.exists(path2save):
os.makedirs(path2save)
for i in range(0, len(target_balances)):
current_range = np.arange(0, N)
current_ts = np.zeros(N)
current_ts2 = np.zeros(N)
ts_benchmark = np.array([item for sublist in benchmark_series[:i+1] for item in sublist]).cumsum()
ts_target = np.array([item for sublist in target_series[:i+1] for item in sublist]).cumsum()
t = len(ts_benchmark)
current_ts[:t] = ts_benchmark
current_ts2[:t] = ts_target
current_ts[current_ts == 0] = ts_benchmark[-1]
current_ts2[current_ts2 == 0] = ts_target[-1]
plt.figure(figsize = (12, 10))
plt.subplot(2, 1, 1)
plt.bar(np.arange(n_assets), target_balances[i], color = 'grey')
plt.xticks(np.arange(n_assets), columns, rotation='vertical')
plt.subplot(2, 1, 2)
plt.colormaps = current_cmap
plt.plot(current_range[:t], current_ts[:t], color = 'black', label = 'Benchmark')
plt.plot(current_range[:t], current_ts2[:t], color = 'red', label = name2plot)
plt.plot(current_range[t:], current_ts[t:], ls = '--', lw = .1, color = 'black')
plt.autoscale(False)
plt.ylim([-1, 1])
plt.legend()
def plot_entropy():
pylab.clf()
pylab.figure(num=None, figsize=(5, 4))
title = "Entropy $H(X)$"
pylab.title(title)
pylab.xlabel("$P(X=$coin will show heads up$)$")
pylab.ylabel("$H(X)$")
pylab.xlim(xmin=0, xmax=1.1)
x = np.arange(0.001, 1, 0.001)
y = -x * np.log2(x) - (1-x) * np.log2(1-x)
pylab.plot(x, y)
pylab.autoscale(tight=True)
pylab.grid(True)
filename = "entropy_demo.png"
pylab.savefig(os.path.join(CHART_DIR, filename), bbox_inches="tight")
def plot_clustering(x, y, title, mx=None, ymax=None, xmin=None, km=None):
pylab.figure(num=None, figsize=(8, 6))
if km:
pylab.scatter(x, y, s=50, c=km.predict(list(zip(x, y))))
else:
pylab.scatter(x, y, s=50)
pylab.title(title)
pylab.xlabel("Occurrence word 1")
pylab.ylabel("Occurrence word 2")
# pylab.xticks([w*7*24 for w in range(10)], ['week %i'%w for w in range(10)])
pylab.autoscale(tight=True)
pylab.ylim(ymin=0, ymax=1)
pylab.xlim(xmin=0, xmax=1)
pylab.grid(True, linestyle='-', color='0.75')
return pylab
def show_image(img, messages=None, fig=None, ax=None, cmap='Gray'):
if not ax or not fig:
w, h = plt.figaspect(img.shape[0] / img.shape[1])
fig = plt.figure(figsize=(w, h))
ax = fig.add_axes([0.1, 0.1, 0.8, 0.8])
bbox_opts = dict(ec='k', fc='k')
ax.clear()
img_axes = ax.imshow(img, cmap=cmaps[cmap])
ax.axis('off')
ax.text(0.03, 0.97, messages['filename'], fontsize=8, color='w', ha='left', va='top', transform=ax.transAxes,
bbox=bbox_opts)
ax.text(0.03, 0.03, messages['text_bottom'], fontsize=8, color='w', ha='left', va='bottom',
transform=ax.transAxes, bbox=bbox_opts)
fig.canvas.draw()
plt.autoscale(tight=True)
fig.show()
return fig, ax, img_axes
def periodogram(pipe):
"""Create a plot of the periodogram with nice labels.
Highlights highest SNR point and also overplots harmonics
"""
pgram = pipe.pgram
# Nicely formatted periodogram ticks
xt = [
0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9,
1, 2, 3, 4, 5, 6, 7, 8, 9,
10, 20, 30, 40, 50, 60, 70, 80, 90,
100, 200, 300, 400, 500, 600, 700, 800, 900,
]
grid_P = pipe.header.ix['grid_P'].value
grid_s2n = pipe.header.ix['grid_s2n'].value
plt.semilogx()
peak = pgram.sort_values('s2n').iloc[-1]
# Plot periogram, peak, and harmonics
plt.plot(pgram.P, pgram.s2n)
plt.plot(grid_P, grid_s2n, 'ro')
harm = np.array([2,4]).astype(float)
harm = np.hstack([1,1/harm,harm])
harm *= grid_P
for h in harm:
plt.axvline(h,lw=6,alpha=0.2,color='r',zorder=0)
# Label plot
ax = plt.gca()
plt.autoscale(axis='x',tight=True)
xl = plt.xlim() # Save away the tight-fitting limits
plt.xticks(xt, xt) # Add the nicely formatted ticks
ax.xaxis.set_ticks_position('top')
ax.xaxis.set_label_position('top')
plt.xlim(*xl) # Reset the plot limits.
plt.xlabel('Period [days]')
plt.ylabel('SNR')
plt.draw()
def plot_entropy():
pylab.clf()
pylab.figure(num=None, figsize=(5, 4))
title = "Entropy $H(X)$"
pylab.title(title)
pylab.xlabel("$P(X=$coin will show heads up$)$")
pylab.ylabel("$H(X)$")
pylab.xlim(xmin=0, xmax=1.1)
x = np.arange(0.001, 1, 0.001)
y = -x * np.log2(x) - (1 - x) * np.log2(1 - x)
pylab.plot(x, y)
# pylab.xticks([w*7*24 for w in [0,1,2,3,4]], ['week %i'%(w+1) for w in
# [0,1,2,3,4]])
pylab.autoscale(tight=True)
pylab.grid(True)
filename = "entropy_demo.png"
pylab.savefig(os.path.join(CHART_DIR, filename), bbox_inches="tight")
def __init__(self,
x,
y,
title,
mx=None,
y_max=None,
x_min=None,
kmeans=None):
pylab.figure(num=None, figsize=(8, 6))
if kmeans:
pylab.scatter(x, y, s=50, c=kmeans.predict(list(zip(x, y))))
else:
pylab.scatter(x, y, s=50)
pylab.title(title)
pylab.xlabel("Occurrence of first word")
pylab.ylabel("Occurrence of second word")
pylab.autoscale(tight=True)
pylab.ylim(bottom=0, top=1)
pylab.xlim(left=0, right=1)
self._pylab = pylab
def plot_p(frame):
sol = Solution(frame,
file_format='petsc',
read_aux=False,
path='./_output/_p/',
file_prefix='claw_p')
x = sol.state.grid.x.centers
y = sol.state.grid.y.centers
mx = len(x)
my = len(y)
mp = sol.state.num_eqn
yy, xx = np.meshgrid(y, x)
p = sol.state.q[0, :, :]
fig = pl.figure(figsize=(8, 3.5))
#pl.title("t= "+str(sol.state.t),fontsize=20)
pl.xticks(size=20)
pl.yticks(size=20)
pl.xlabel('x', fontsize=20)
pl.ylabel('y', fontsize=20)
#pl.pcolormesh(xx,yy,p_subxy,cmap=cm.OrRd)
pl.pcolormesh(xx, yy, p, cmap='RdBu_r')
pl.autoscale(tight=True)
cb = pl.colorbar(ticks=[0.5, 1, 1.5, 2])
#pl.clim(ticks=[0.5,1,1.5,2])
imaxes = pl.gca()
pl.axes(cb.ax)
pl.yticks(fontsize=20)
pl.axes(imaxes)
#pl.xticks(fontsize=20); pl.axes(imaxes)
#pl.axis('equal')
pl.axis('tight')
fig.tight_layout()
pl.savefig('./_plots_to_paper/sound-speed_FV_t' + str(frame) +
'_pcolor.png')
pl.close()
def main():
#mat_dir = 'Annotations_Part/'
#origimage_dir = 'JPEGImages/'
mat_dir = 'Annotations_Part/'
origimage_dir = 'JPEGImages/'
label_dir = 'yolabel/'
personimage_dir = 'yoimage/'
os.mkdir(label_dir)
os.mkdir(personimage_dir)
list_mats = os.listdir(mat_dir)
for mat in list_mats:
imname = mat[:-4] + '.jpg'
image_size, part_mask = make_label(origimage_dir + imname, mat_dir + mat)
if image_size == 0:
continue
copy_cmd = 'cp ' + origimage_dir + imname + ' ' + personimage_dir
os.system(copy_cmd)
mpl.rcParams['savefig.pad_inches'] = 0
figsize = (image_size[1]*1.0/100, image_size[0]*1.0/100)
fig = plt.figure(figsize=figsize)
ax = plt.axes([0,0,1,1], frameon=False)
ax.get_xaxis().set_visible(False)
ax.get_yaxis().set_visible(False)
plt.autoscale(tight=True)
if np.max(part_mask) == 0:
ax.imshow(part_mask, cmap='gray')
else:
ax.imshow(part_mask, cmap=color_map(N=np.max(part_mask) + 1))
#plt.savefig(label_dir + imname)
cv2.imwrite(label_dir + imname, part_mask)
plt.close()
def ImageViewer(im, cmap=None, interpolation='nearest', **kwargs):
import matplotlib.pyplot as plt
import matplotlib.cm as cm
if type(im) is str: im = open_image(im, rgb2gray=False)
if cmap is None: cmap = cm.gray
fig = plt.figure()
ax = fig.add_subplot(111)
ax.imshow(im, interpolation=interpolation, cmap=cmap, **kwargs)
h = im.shape[0]
w = im.shape[1]
if im.ndim == 2:
def format_coord(x, y):
col = int(x + 0.5)
row = int(y + 0.5)
if col >= 0 and col < w and row >= 0 and row < h:
z = im[row, col]
return '%1.3f (y=%1.1f, x=%1.1f)' % (z, y, x)
else:
return 'x=%1.1f, y=%1.1f' % (x, y)
else:
def format_coord(x, y):
col = int(x + 0.5)
row = int(y + 0.5)
if col >= 0 and col < w and row >= 0 and row < h:
r = im[row, col, 0]
g = im[row, col, 1]
b = im[row, col, 2]
return 'y=%1.1f, x=%1.1f, R=%d, G=%d, B=%d' % (y, x, r, g, b)
else:
return 'x=%1.1f, y=%1.1f' % (x, y)
ax.format_coord = format_coord
plt.autoscale(tight=True)
return plt
def plot_entropy():
pylab.clf()
pylab.figure(num=None, figsize=(5, 4))
title = "Entropy $H(X)$"
pylab.title(title)
pylab.xlabel("$P(X=$coin will show heads up$)$")
pylab.ylabel("$H(X)$")
pylab.xlim(xmin=0, xmax=1.1)
x = np.arange(0.001, 1, 0.001)
# Claude Shannon's information entropy
# H(X) = - Sum_i( p(Xi) * log(p(Xi)))
y = -x * np.log2(x) - (1 - x) * np.log2(1 - x)
pylab.plot(x, y)
pylab.autoscale(tight=True)
pylab.grid(True)
filename = "b_entropy_coin.png"
pylab.savefig(os.path.join(CHART_DIR, filename), bbox_inches="tight")
def t_tms_from_Xc(M, savefig=None, plot_fig=None, ttms_true=None, Xc=None):
'''
Calculates the t(tms) for a given Xc and mass
Xc: float
M: float
'''
# ------------------------------------------------------------------------------
# Parameters from the models
mass = np.array([14.6, 12.5, 10.8, 9.6, 8.6, 7.7, 6.4, 5.5, 4.8,
4.2, 3.8, 3.4])
nm = len(mass)
str_mass = ['M14p60', 'M12p50', 'M10p80', 'M9p600', 'M8p600', 'M7p700',
'M6p400', 'M5p500', 'M4p800', 'M4p200', 'M3p800', 'M3p400']
st = ['B0.5', 'B1', 'B1.5', 'B2', 'B2.5', 'B3', 'B4', 'B5', 'B6', 'B7',
'B8', 'B9']
zsun = 'Z01400'
str_vel = ['V60000', 'V70000', 'V80000', 'V90000', 'V95000']
Hfracf = 0. # end of main sequence
# ****
folder_data = 'tables/models/models_bes/'
if plot_fig is True:
plt.xlabel(r'$t/t_{MS}$')
plt.ylabel(r'$X_c$')
plt.ylim([0.0, 0.8])
plt.xlim([0.0, 1.0])
# ------------------------------------------------------------------------------
# Loop (reading the models)
typ = (1, 3, 16, 21) # Age, Lum versus Teff versus Hfrac
arr_age = []
arr_Hfr = []
arr_t_tc = []
cor = phc.gradColor(np.arange(len(st)), cmapn='inferno')
iv = 2 # O que eh isto?
arr_Xc = []
for i in range(nm):
file_data = folder_data + str_mass[i] + zsun + str_vel[iv] + '.dat'
age, lum, Teff, Hfrac = np.loadtxt(file_data, usecols=typ,
unpack=True, skiprows=2)
arr_age.append(age)
arr_Hfr.append(Hfrac)
iMS = np.where(abs(Hfrac - Hfracf) == min(abs(Hfrac - Hfracf)))
X_c = Hfrac[0:iMS[0][0]]
arr_Xc.append(X_c)
t_tc = age[0:iMS[0][0]] / max(age[0:iMS[0][0]])
arr_t_tc.append(t_tc)
if plot_fig is True:
plt.plot(t_tc, X_c, color=cor[i], label=('%s' % st[i]))
# ------------------------------------------------------------------------------
# Interpolation
k = find_nearest(mass, M)[1]
if plot_fig is True:
plt.plot(ttms_true, Xc, 'o')
plt.autoscale()
plt.minorticks_on()
plt.legend(fontsize=10, ncol=2, fancybox=False, frameon=False)
# ------------------------------------------------------------------------------
if savefig is True:
pdfname = 'Xc_vs_Tsp.png'
plt.savefig(pdfname)
return k, arr_t_tc, arr_Xc
def setup(
subplt=None,
figsize=None,
ax=None,
xr=None,
xmin=None,
xmax=None,
yr=None,
ymin=None,
ymax=None,
xlog=False,
ylog=False,
xoffset=None,
yoffset=None,
xlabel=None,
ylabel=None,
xtickv=None,
xticknames=None,
xtickrotate=None,
ytickv=None,
yticknames=None,
ytickrotate=None,
halfxlog=False,
suptitle=None,
suptitle_prop=None,
subtitle=None,
subtitle_prop=None,
subtitleloc=1,
title=None,
xticks=None,
yticks=None,
autoticks=False,
embiggenx=None,
embiggeny=None,
secondx=False,
secondx_prop=None,
secondy=False,
secondy_prop=None,
grid=True,
tickmarks=True,
font=True,
adjust=True,
hspace=0.1,
wspace=0.1,
aspect=None,
rasterized=False,
):
'''Setup some nice defaults so that we are all fancy like
xr,yr -- xrange and yrange. by setting this it turns off the autoscale feature in that axis
xlog,ylog -- T/F I do love me my log log plots
xlabel,ylabel,title -- Set some nice text for the different axis
xticks, yticks -- set to False if you want to hide the tickmarks and labels
grid -- Turn on the grid in a nice way
tickmarks -- make me some nick minor and major ticks
you can pass in a gridspec
'''
# if notebook:
# # http://matplotlib.org/users/customizing.html
# # http://damon-is-a-geek.com/publication-ready-the-first-time-beautiful-reproducible-plots-with-matplotlib.html
# matplotlib.rcParams['savefig.dpi'] = 144
# matplotlib.rcParams.update({'font.size': 12})
# # matplotlib.rcParams['font.family'] = 'serif'
# # matplotlib.rcParams['font.serif'] = ['Computer Modern Roman']
# # matplotlib.rcParams['text.usetex'] = True
if figsize is not None:
fig = pylab.figure(figsize=figsize)
## Handle subplot being an int `223`, tuple `(2,2,3)` or gridspec
try:
if subplt is None:
if ax is None:
ax = pylab.gca()
elif isinstance(subplt,
(int, gridspec.GridSpec, gridspec.SubplotSpec)):
ax = pylab.subplot(subplt)
else:
ax = pylab.subplot(*subplt)
except Exception as e:
raise ValueError(
'Failed to setup subplt:[{}] -- probably indexing error'.format(
subplt))
# Ranges -- Setting either xr,yr stops the auto ranging
if xr is not None:
ax.set_xlim(xr)
pylab.autoscale(False, 'x', True)
if yr is not None:
ax.set_ylim(yr)
pylab.autoscale(False, 'y', True)
# Log stuff -- do this afterwards to ensure the minor tick marks are updated
# can set the specific ticks using subsx, subsy --
# ax.set_xscale('log', subsx=[2, 3, 4, 5, 6, 7, 8, 9])
# Look here: http://www.ianhuston.net/2011/02/minor-tick-labels-in-matplotlib
# clip ensures that lines that go off the edge of the plot are shown but clipped
if xlog:
ax.set_xscale('log', nonposx='clip')
if ylog:
ax.set_yscale('log', nonposy='clip')
# Labels
if xlabel is not None:
pylab.xlabel(xlabel)
if ylabel is not None:
pylab.ylabel(ylabel)
if title is not None:
pylab.title(title)
if suptitle is not None:
if suptitle_prop is None:
suptitle_prop = {}
pylab.suptitle(suptitle, **suptitle_prop)
if subtitle is not None:
prop = dict(transform=ax.transAxes)
if subtitleloc == 1:
prop.update({
'location': (0.95, 0.95),
'horizontalalignment': 'right',
'verticalalignment': 'top'
})
elif subtitleloc == 3:
prop.update({
'location': (0.05, 0.05),
'horizontalalignment': 'left',
'verticalalignment': 'bottom'
})
else:
raise NotImplementedError(
'Get to work adding the following subtitle location: %d' %
(subtitleloc))
if subtitle_prop is not None:
prop.update(subtitle_prop)
loc = prop.pop('location')
outline = prop.pop('outline', True)
outlinewidth = prop.pop('linewidth', 3.5)
txt = pylab.text(loc[0], loc[1], subtitle, **prop)
if outline:
# pylab.setp(txt,path_effects=[PathEffects.Stroke(linewidth=outlinewidth, foreground="w")]) # Broken
txt.set_path_effects([
PathEffects.Stroke(linewidth=outlinewidth, foreground="w"),
PathEffects.Normal()
])
# raise ValueError()
if xtickv is not None:
ax.xaxis.set_major_locator(matplotlib.ticker.FixedLocator(xtickv))
if xticknames is not None:
ax.xaxis.set_major_formatter(
matplotlib.ticker.FixedFormatter(xticknames))
if xtickrotate is not None:
if xtickrotate == 'vertical':
pylab.xticks(xtickv, xticknames, rotation='vertical')
else:
tmp = dict(rotation=30, ha='right')
if isinstance(xtickrotate, dict):
tmp.update(xtickrotate)
else:
tmp['rotation'] = xtickrotate
pylab.setp(pylab.xticks()[1], **tmp)
if ytickv is not None:
ax.yaxis.set_major_locator(matplotlib.ticker.FixedLocator(ytickv))
if yticknames is not None:
ax.yaxis.set_major_formatter(
matplotlib.ticker.FixedFormatter(yticknames))
if ytickrotate is not None:
pylab.setp(pylab.yticks()[1], rotation=ytickrotate, ha='right')
# Axis hiding
if autoticks:
if not (((isinstance(subplt, tuple) and len(subplt) == 3)) or
((isinstance(subplt, gridspec.SubplotSpec)))):
splog('Cannot setup auto ticks without a proper subplot')
else:
if isinstance(subplt, gridspec.SubplotSpec):
rows, cols, i, _ = subplt.get_geometry()
i += 1 # i is 0 indexed.
else:
rows, cols, i = subplt
if ((i % cols) != 1) and (not yticks):
yticks = False
# if ( i < (cols*(rows-1) + 1) ) and (not xticks):
# xticks = False
xticks = not ((i < (cols * (rows - 1) + 1)) and (not xticks))
# Tickmark hiding -- used by autoticks as well.
if xticks is False:
# ax.set_xticklabels([])
ax.set_xlabel('')
pylab.setp(ax.get_xticklabels(), visible=False)
if yticks is False:
# ax.set_yticklabels([])
ax.set_ylabel('')
pylab.setp(ax.get_yticklabels(), visible=False)
# some nice defaults
if grid is True:
pylab.grid(b=True,
which='major',
linestyle='solid',
color='0.3',
alpha=0.5)
ax.set_axisbelow(True)
if grid is False:
pylab.grid('off')
if tickmarks:
ax.tick_params('both', which='major', length=5, width=2)
ax.tick_params('both', which='minor', length=3, width=1)
ax.minorticks_on()
else:
ax.minorticks_off()
pylab.tick_params(axis='both',
which='both',
bottom='off',
top='off',
left='off',
right='off')
if adjust:
pylab.subplots_adjust(hspace=hspace, wspace=wspace)
if font:
# this in theory should work, but fails becuase of the stupidify known as `_`
# pylab.rc('font', **{'family':'sans-serif', 'sans-serif':['Helvetica']})
# prop = {'family' : 'normal', 'weight' : 'bold', 'size' : 22}
# pylab.rc('font', **{'family':'serif', 'serif':['Computer Modern Roman']})
# pylab.rc('text', usetex=True)
pass
if aspect is not None:
# 'auto', 'equal'
ax.set_aspect(aspect)
if embiggenx:
setup(xr=math.embiggen(ax.axis()[:2], embiggenx))
if embiggeny:
setup(yr=math.embiggen(ax.axis()[2:], embiggeny))
if (xticks) and (halfxlog):
xax = ax.xaxis
xax.set_minor_formatter(matplotlib.ticker.FormatStrFormatter('%g'))
tmp = (10.0**(np.arange(-1, 5, 1))) * 5.0
for x, label in zip(xax.get_minorticklocs(),
xax.get_minorticklabels()):
if x in tmp:
label.set_fontsize(8)
else:
label.set_fontsize(0)
pylab.setp(label, visible=False)
if secondx:
# second x axis
tmp = dict(
xlabel=None,
fcn=lambda arr: arr,
fmt='{}',
xr=ax.axis()[:2],
xlog=('log' in ax.get_xscale()),
# yr=ax.axis()[2:]
xtickv=ax.get_xticks(),
) # defaults
if secondx_prop:
tmp.update(secondx_prop)
# adjust some nice things
f = tmp.pop('fcn')
fcn = lambda arr: [f(x) for x in arr]
fmt = tmp.pop('fmt')
if isinstance(fmt, str): fmt = (fmt, fmt)
if 'xtickv' in tmp:
tmp['xtickv'] = np.array(tmp['xtickv'])
if 'xticknames' not in tmp:
arr = np.array(fcn(tmp['xtickv']))
tmp['xticknames'] = np.array([fmt[0].format(x) for x in arr])
if tmp['xlog'] and tmp['nicelog']:
ii = np.where((arr >= 1e1) & (arr < 1e4))[0]
tmp['xticknames'][ii] = [fmt[1].format(x) for x in arr[ii]]
ii = np.where(arr >= 1e4)
tmp['xticknames'][ii] = [
fmt[2].format(np.log10(x)) for x in arr[ii]
]
ax2 = ax.twiny() # I have never understood this
ax2.set_xlabel(tmp['xlabel'])
ax2.set_xlim(tmp['xr'])
if tmp['xlog']:
ax2.set_xscale('log', nonposx='clip')
if 'xtickv' in tmp:
ax2.xaxis.set_major_locator(
matplotlib.ticker.FixedLocator(tmp['xtickv']))
if 'xticknames' in tmp:
ax2.xaxis.set_major_formatter(
matplotlib.ticker.FixedFormatter(tmp['xticknames']))
ax.secondx = ax2
pylab.sca(ax)
if secondy:
# second x axis
tmp = dict(
ylabel=None,
fcn=lambda arr: arr,
fmt='{}',
yr=ax.axis()[2:],
ylog=('log' in ax.get_yscale()),
ytickv=ax.get_yticks(),
)
if secondy_prop:
tmp.update(secondy_prop)
# adjust some nice things
f = tmp.pop('fcn')
fcn = lambda arr: [f(x) for x in arr]
fmt = tmp.pop('fmt')
if isinstance(fmt, str): fmt = (fmt, fmt)
if 'ytickv' in tmp:
tmp['ytickv'] = np.array(tmp['ytickv'])
if 'yticknames' not in tmp:
arr = np.array(fcn(tmp['ytickv']))
tmp['yticknames'] = np.array([fmt[0].format(x) for x in arr])
if tmp['ylog'] and tmp['nicelog']:
ii = np.where((arr >= 1e1) & (arr < 1e4))[0]
tmp['yticknames'][ii] = [fmt[1].format(x) for x in arr[ii]]
ii = np.where(arr >= 1e4)
tmp['yticknames'][ii] = [
fmt[2].format(np.log10(x)) for x in arr[ii]
]
print(tmp)
ax2 = ax.twinx() # I have never understood this
ax2.set_ylabel(tmp['ylabel'])
ax2.set_ylim(tmp['yr'])
if tmp['ylog']:
ax2.set_yscale('log', nonposx='clip')
if 'ytickv' in tmp:
ax2.yaxis.set_major_locator(
matplotlib.ticker.FixedLocator(tmp['ytickv']))
if 'yticknames' in tmp:
ax2.yaxis.set_major_formatter(
matplotlib.ticker.FixedFormatter(tmp['yticknames']))
ax.secondy = ax2
pylab.sca(ax)
if xoffset is False:
tmp = matplotlib.ticker.ScalarFormatter(useOffset=False)
tmp.set_scientific(False)
ax.xaxis.set_major_formatter(tmp)
if yoffset is False:
tmp = matplotlib.ticker.ScalarFormatter(useOffset=False)
tmp.set_scientific(False)
ax.yaxis.set_major_formatter(tmp)
if rasterized:
ax.set_rasterized(True)
# temp
return ax
def plot_mi_demo():
# determinism!
np.random.seed(0)
# plot1
pylab.clf()
pylab.figure(num=None, figsize=(8, 8))
x = np.arange(0, 10, 0.2)
pylab.subplot(221)
y = 0.5 * x + norm.rvs(1, scale=.01, size=len(x))
plot_mi_func(x, y)
pylab.subplot(222)
y = 0.5 * x + norm.rvs(1, scale=.1, size=len(x))
plot_mi_func(x, y)
pylab.subplot(223)
y = 0.5 * x + norm.rvs(1, scale=1, size=len(x))
plot_mi_func(x, y)
pylab.subplot(224)
y = norm.rvs(1, scale=10, size=len(x))
plot_mi_func(x, y)
pylab.autoscale(tight=True)
pylab.grid(True)
filename = "c_mutual_information_1_ok.png"
pylab.savefig(os.path.join(CHART_DIR, filename), bbox_inches="tight")
##############
# plot 2
pylab.clf()
pylab.figure(num=None, figsize=(8, 8))
x = np.arange(-5, 5, 0.2)
pylab.subplot(221)
y = 0.5 * x ** 2 + norm.rvs(1, scale=.01, size=len(x))
plot_mi_func(x, y)
pylab.subplot(222)
y = 0.5 * x ** 2 + norm.rvs(1, scale=.1, size=len(x))
plot_mi_func(x, y)
pylab.subplot(223)
y = 0.5 * x ** 2 + norm.rvs(1, scale=1, size=len(x))
plot_mi_func(x, y)
pylab.subplot(224)
y = 0.5 * x ** 2 + norm.rvs(1, scale=10, size=len(x))
plot_mi_func(x, y)
pylab.autoscale(tight=True)
pylab.grid(True)
filename = "c_mutual_information_2_ok.png"
pylab.savefig(os.path.join(CHART_DIR, filename), bbox_inches="tight")
forecast = model_fit.forecast(forecast_len)
#forecast = np.exp(forecast) # for log correction
# eval metrics
rmse = np.sqrt(((forecast - df_test)**2).mean())
p_rmse = (rmse / df_test.mean()) * 100
# plot it again
plt.figure(figsize=(20, 10))
plt.plot(series, 'b')
plt.plot(forecast, 'r')
plt.title('RMSE: %.2f Percent Avg Error: %.2f' % (rmse, p_rmse))
plt.xlabel('Weeks')
plt.ylabel('Weekly Rates')
plt.autoscale(enable=True, axis='x', tight=True)
plt.axvline(x=series.index[len(df_train)], color='black')
# write out for testing
df_test.to_csv('../../data/processed/test_actuals.csv')
forecast.to_csv('../../data/processed/test_forecasted.csv')
##############
# prediction #
##############
# predict out for 52
model = SARIMAX(series,
order=(1, 1, 2),
seasonal_order=(1, 0, 1, 52),
enforce_stationarity=False,
def plot_MDS(x, y=None, data_set_ids=None):
if data_set_ids is None:
data_set_ids = np.ones(x.shape[0])
pl.close()
fig = pl.figure()
#axes = fig.add_subplot(111)
axes = pl.subplot(111)
to_use = false(x.shape[0])
max_to_use = 200
for i in np.unique(data_set_ids):
I = (data_set_ids == i).nonzero()[0]
if len(I) <= max_to_use:
to_use[I] = True
else:
choice = np.random.choice(I, max_to_use, replace=False)
to_use[choice] = True
x = x[to_use, :]
data_set_ids = data_set_ids[to_use]
if y is not None:
y = y[to_use]
x = try_toarray(x)
I_nonzero = x.sum(1).nonzero()[0]
x = x[I_nonzero, :]
data_set_ids = data_set_ids[I_nonzero]
y = y[I_nonzero]
'''
x = np.zeros(x.shape)
x[data_set_ids==0,0] = 1
x[data_set_ids==1,1] = 1
x[data_set_ids==2,2] = 1
'''
W = pairwise.pairwise_distances(x, x, 'cosine')
#W = pairwise.pairwise_distances(x,x,'euclidean')
W = make_rbf(x, sigma=10, metric='cosine')
W = 1 - W
#mds = sklearn.manifold.MDS(dissimilarity='precomputed',max_iter=1000,verbose=1,metric=False)
mds = sklearn.manifold.MDS(dissimilarity='precomputed',
max_iter=1000,
verbose=1,
n_init=4)
x_mds = mds.fit_transform(W)
#mds = sklearn.manifold.MDS()
#x_mds = mds.fit_transform(x)
colors = ['r', 'g', 'b']
labels = ['s', 'o']
for ind, i in enumerate(np.unique(data_set_ids)):
if ind == 0 and False:
continue
if ind == 2 and False:
continue
I = data_set_ids == i
alpha = 1 / float(I.sum())
alpha = alpha * 50
alpha = .3
if i == 0:
alpha = 1
y_curr = y[I]
x_curr = x_mds[I, :]
for ind2, j in enumerate(np.unique(y_curr)):
I2 = y_curr == j
axes.scatter(x_curr[I2, 0],
x_mds[I2, 1],
alpha=alpha,
c=colors[ind],
s=100,
marker=labels[ind2])
#axes.scatter(x_mds[I,0],x_mds[I,1], alpha=alpha,c=colors[ind],s=60)
move_fig(fig)
pl.autoscale()
pl.show(block=True)
pass
def plot_MDS(x, y=None, data_set_ids=None):
if data_set_ids is None:
data_set_ids = np.ones(x.shape[0])
pl.close()
fig = pl.figure()
#axes = fig.add_subplot(111)
axes = pl.subplot(111)
to_use = false(x.shape[0])
max_to_use = 200
for i in np.unique(data_set_ids):
I = (data_set_ids == i).nonzero()[0]
if len(I) <= max_to_use:
to_use[I] = True
else:
choice = np.random.choice(I, max_to_use, replace=False)
to_use[choice] = True
x = x[to_use,:]
data_set_ids = data_set_ids[to_use]
if y is not None:
y = y[to_use]
x = try_toarray(x)
I_nonzero = x.sum(1).nonzero()[0]
x = x[I_nonzero,:]
data_set_ids = data_set_ids[I_nonzero]
y = y[I_nonzero]
'''
x = np.zeros(x.shape)
x[data_set_ids==0,0] = 1
x[data_set_ids==1,1] = 1
x[data_set_ids==2,2] = 1
'''
W = pairwise.pairwise_distances(x,x,'cosine')
#W = pairwise.pairwise_distances(x,x,'euclidean')
W = make_rbf(x,sigma=10,metric='cosine')
W = 1 - W
#mds = sklearn.manifold.MDS(dissimilarity='precomputed',max_iter=1000,verbose=1,metric=False)
mds = sklearn.manifold.MDS(dissimilarity='precomputed',max_iter=1000,verbose=1,n_init=4)
x_mds = mds.fit_transform(W)
#mds = sklearn.manifold.MDS()
#x_mds = mds.fit_transform(x)
colors = ['r','g','b']
labels = ['s','o']
for ind,i in enumerate(np.unique(data_set_ids)):
if ind == 0 and False:
continue
if ind == 2 and False:
continue
I = data_set_ids == i
alpha = 1 / float(I.sum())
alpha = alpha*50
alpha = .3
if i == 0:
alpha = 1
y_curr = y[I]
x_curr = x_mds[I,:]
for ind2, j in enumerate(np.unique(y_curr)):
I2 = y_curr == j
axes.scatter(x_curr[I2,0],x_mds[I2,1], alpha=alpha,c=colors[ind],s=100,marker = labels[ind2])
#axes.scatter(x_mds[I,0],x_mds[I,1], alpha=alpha,c=colors[ind],s=60)
move_fig(fig)
pl.autoscale()
pl.show(block=True)
pass
def setup(subplt=None, figsize=None, ax=None,
xr=None, xmin=None, xmax=None,
yr=None, ymin=None, ymax=None,
xlog=False, ylog=False,
xoffset=None, yoffset=None,
xlabel=None, ylabel=None,
xtickv=None, xticknames=None, xtickrotate=None,
ytickv=None, yticknames=None, ytickrotate=None,
halfxlog=False,
suptitle=None, suptitle_prop=None,
subtitle=None, subtitle_prop=None, subtitleloc=1,
title=None,
xticks=None, yticks=None, autoticks=False,
embiggenx=None, embiggeny=None,
secondx=False, secondx_prop=None,
secondy=False, secondy_prop=None,
grid=True, tickmarks=True, font=True,
adjust=True, hspace=0.1, wspace=0.1, aspect=None,
rasterized=False,
):
'''Setup some nice defaults so that we are all fancy like
xr,yr -- xrange and yrange. by setting this it turns off the autoscale feature in that axis
xlog,ylog -- T/F I do love me my log log plots
xlabel,ylabel,title -- Set some nice text for the different axis
xticks, yticks -- set to False if you want to hide the tickmarks and labels
grid -- Turn on the grid in a nice way
tickmarks -- make me some nick minor and major ticks
you can pass in a gridspec
'''
# if notebook:
# # http://matplotlib.org/users/customizing.html
# # http://damon-is-a-geek.com/publication-ready-the-first-time-beautiful-reproducible-plots-with-matplotlib.html
# matplotlib.rcParams['savefig.dpi'] = 144
# matplotlib.rcParams.update({'font.size': 12})
# # matplotlib.rcParams['font.family'] = 'serif'
# # matplotlib.rcParams['font.serif'] = ['Computer Modern Roman']
# # matplotlib.rcParams['text.usetex'] = True
if figsize is not None:
fig = pylab.figure(figsize=figsize)
## Handle subplot being an int `223`, tuple `(2,2,3)` or gridspec
try:
if subplt is None:
if ax is None:
ax = pylab.gca()
elif isinstance(subplt, (int, gridspec.GridSpec, gridspec.SubplotSpec) ):
ax = pylab.subplot(subplt)
else:
ax = pylab.subplot(*subplt)
except Exception as e:
raise ValueError('Failed to setup subplt:[{}] -- probably indexing error'.format(subplt))
# Ranges -- Setting either xr,yr stops the auto ranging
if xr is not None:
ax.set_xlim(xr)
pylab.autoscale(False, 'x', True)
if yr is not None:
ax.set_ylim(yr)
pylab.autoscale(False, 'y', True)
# Log stuff -- do this afterwards to ensure the minor tick marks are updated
# can set the specific ticks using subsx, subsy --
# ax.set_xscale('log', subsx=[2, 3, 4, 5, 6, 7, 8, 9])
# Look here: http://www.ianhuston.net/2011/02/minor-tick-labels-in-matplotlib
# clip ensures that lines that go off the edge of the plot are shown but clipped
if xlog:
ax.set_xscale('log', nonposx='clip')
if ylog:
ax.set_yscale('log', nonposy='clip')
# Labels
if xlabel is not None:
pylab.xlabel(xlabel)
if ylabel is not None:
pylab.ylabel(ylabel)
if title is not None:
pylab.title(title)
if suptitle is not None:
if suptitle_prop is None:
suptitle_prop = {}
pylab.suptitle(suptitle, **suptitle_prop)
if subtitle is not None:
prop = dict(transform=ax.transAxes)
if subtitleloc == 1:
prop.update({'location':(0.95,0.95),
'horizontalalignment':'right',
'verticalalignment':'top'})
elif subtitleloc == 3:
prop.update({'location':(0.05,0.05),
'horizontalalignment':'left',
'verticalalignment':'bottom'})
else:
raise NotImplementedError('Get to work adding the following subtitle location: %d'%(subtitleloc))
if subtitle_prop is not None:
prop.update(subtitle_prop)
loc = prop.pop('location')
outline = prop.pop('outline',True)
outlinewidth = prop.pop('linewidth',3.5)
txt = pylab.text(loc[0], loc[1], subtitle, **prop)
if outline:
# pylab.setp(txt,path_effects=[PathEffects.Stroke(linewidth=outlinewidth, foreground="w")]) # Broken
txt.set_path_effects(
[PathEffects.Stroke(linewidth=outlinewidth, foreground="w"),
PathEffects.Normal()])
# raise ValueError()
if xtickv is not None:
ax.xaxis.set_major_locator(matplotlib.ticker.FixedLocator(xtickv))
if xticknames is not None:
ax.xaxis.set_major_formatter(matplotlib.ticker.FixedFormatter(xticknames))
if xtickrotate is not None:
if xtickrotate == 'vertical':
pylab.xticks(xtickv, xticknames, rotation='vertical')
else:
tmp = dict(rotation=30, ha='right')
if isinstance(xtickrotate, dict):
tmp.update(xtickrotate)
else:
tmp['rotation'] = xtickrotate
pylab.setp(pylab.xticks()[1], **tmp)
if ytickv is not None:
ax.yaxis.set_major_locator(matplotlib.ticker.FixedLocator(ytickv))
if yticknames is not None:
ax.yaxis.set_major_formatter(matplotlib.ticker.FixedFormatter(yticknames))
if ytickrotate is not None:
pylab.setp(pylab.yticks()[1], rotation=ytickrotate, ha='right')
# Axis hiding
if autoticks:
if not ( ( (isinstance(subplt, tuple) and len(subplt) == 3) ) or
( (isinstance(subplt, gridspec.SubplotSpec)) ) ):
splog('Cannot setup auto ticks without a proper subplot')
else:
if isinstance(subplt, gridspec.SubplotSpec):
rows,cols,i, _ = subplt.get_geometry()
i += 1 # i is 0 indexed.
else:
rows,cols,i = subplt
if ( (i%cols) != 1 ) and (not yticks):
yticks = False
# if ( i < (cols*(rows-1) + 1) ) and (not xticks):
# xticks = False
xticks = not ( ( i < (cols*(rows-1) + 1) ) and (not xticks) )
# Tickmark hiding -- used by autoticks as well.
if xticks is False:
# ax.set_xticklabels([])
ax.set_xlabel('')
pylab.setp(ax.get_xticklabels(), visible=False)
if yticks is False:
# ax.set_yticklabels([])
ax.set_ylabel('')
pylab.setp(ax.get_yticklabels(), visible=False)
# some nice defaults
if grid is True:
pylab.grid(b=True, which='major', linestyle='solid', color='0.3', alpha=0.5)
ax.set_axisbelow(True)
if grid is False:
pylab.grid('off')
if tickmarks:
ax.tick_params('both', which='major', length=5, width=2)
ax.tick_params('both', which='minor', length=3, width=1)
ax.minorticks_on()
else:
ax.minorticks_off()
pylab.tick_params(axis='both',which='both',
bottom='off', top='off',
left='off', right='off')
if adjust:
pylab.subplots_adjust(hspace=hspace, wspace=wspace)
if font:
# this in theory should work, but fails becuase of the stupidify known as `_`
# pylab.rc('font', **{'family':'sans-serif', 'sans-serif':['Helvetica']})
# prop = {'family' : 'normal', 'weight' : 'bold', 'size' : 22}
# pylab.rc('font', **{'family':'serif', 'serif':['Computer Modern Roman']})
# pylab.rc('text', usetex=True)
pass
if aspect is not None:
# 'auto', 'equal'
ax.set_aspect(aspect)
if embiggenx:
setup(xr=math.embiggen(ax.axis()[:2],embiggenx))
if embiggeny:
setup(yr=math.embiggen(ax.axis()[2:],embiggeny))
if (xticks) and (halfxlog):
xax = ax.xaxis
xax.set_minor_formatter(matplotlib.ticker.FormatStrFormatter('%g'))
tmp = (10.0**(np.arange(-1,5,1)))*5.0
for x,label in zip(xax.get_minorticklocs(), xax.get_minorticklabels()):
if x in tmp:
label.set_fontsize(8)
else:
label.set_fontsize(0)
pylab.setp(label, visible=False)
if secondx:
# second x axis
tmp = dict(xlabel=None,
fcn=lambda arr: arr,
fmt='{}',
xr=ax.axis()[:2],
xlog=('log' in ax.get_xscale()),
# yr=ax.axis()[2:]
xtickv=ax.get_xticks(),
) # defaults
if secondx_prop:
tmp.update(secondx_prop)
# adjust some nice things
f = tmp.pop('fcn')
fcn = lambda arr: [f(x) for x in arr]
fmt = tmp.pop('fmt')
if isinstance(fmt, str): fmt = (fmt,fmt)
if 'xtickv' in tmp:
tmp['xtickv'] = np.array(tmp['xtickv'])
if 'xticknames' not in tmp:
arr = np.array(fcn(tmp['xtickv']))
tmp['xticknames'] = np.array([fmt[0].format(x) for x in arr])
if tmp['xlog'] and tmp['nicelog']:
ii = np.where((arr >= 1e1)&
(arr < 1e4) )[0]
tmp['xticknames'][ii] = [fmt[1].format(x) for x in arr[ii]]
ii = np.where(arr >= 1e4)
tmp['xticknames'][ii] = [fmt[2].format(np.log10(x)) for x in arr[ii]]
ax2 = ax.twiny() # I have never understood this
ax2.set_xlabel(tmp['xlabel'])
ax2.set_xlim(tmp['xr'])
if tmp['xlog']:
ax2.set_xscale('log', nonposx='clip')
if 'xtickv' in tmp:
ax2.xaxis.set_major_locator(matplotlib.ticker.FixedLocator(tmp['xtickv']))
if 'xticknames' in tmp:
ax2.xaxis.set_major_formatter(matplotlib.ticker.FixedFormatter(tmp['xticknames']))
ax.secondx = ax2
pylab.sca(ax)
if secondy:
# second x axis
tmp = dict(ylabel=None,
fcn=lambda arr: arr,
fmt='{}',
yr=ax.axis()[2:],
ylog=('log' in ax.get_yscale()),
ytickv=ax.get_yticks(), )
if secondy_prop:
tmp.update(secondy_prop)
# adjust some nice things
f = tmp.pop('fcn')
fcn = lambda arr: [f(x) for x in arr]
fmt = tmp.pop('fmt')
if isinstance(fmt, str): fmt = (fmt,fmt)
if 'ytickv' in tmp:
tmp['ytickv'] = np.array(tmp['ytickv'])
if 'yticknames' not in tmp:
arr = np.array(fcn(tmp['ytickv']))
tmp['yticknames'] = np.array([fmt[0].format(x) for x in arr])
if tmp['ylog'] and tmp['nicelog']:
ii = np.where((arr >= 1e1)&
(arr < 1e4) )[0]
tmp['yticknames'][ii] = [fmt[1].format(x) for x in arr[ii]]
ii = np.where(arr >= 1e4)
tmp['yticknames'][ii] = [fmt[2].format(np.log10(x)) for x in arr[ii]]
print(tmp)
ax2 = ax.twinx() # I have never understood this
ax2.set_ylabel(tmp['ylabel'])
ax2.set_ylim(tmp['yr'])
if tmp['ylog']:
ax2.set_yscale('log', nonposx='clip')
if 'ytickv' in tmp:
ax2.yaxis.set_major_locator(matplotlib.ticker.FixedLocator(tmp['ytickv']))
if 'yticknames' in tmp:
ax2.yaxis.set_major_formatter(matplotlib.ticker.FixedFormatter(tmp['yticknames']))
ax.secondy = ax2
pylab.sca(ax)
if xoffset is False:
tmp = matplotlib.ticker.ScalarFormatter(useOffset=False)
tmp.set_scientific(False)
ax.xaxis.set_major_formatter(tmp)
if yoffset is False:
tmp = matplotlib.ticker.ScalarFormatter(useOffset=False)
tmp.set_scientific(False)
ax.yaxis.set_major_formatter(tmp)
if rasterized:
ax.set_rasterized(True)
# temp
return ax
def plot_simple_demo_1():
pylab.clf()
fig = pylab.figure(num=None, figsize=(10, 4))
pylab.subplot(131)
title = "Original feature space"
pylab.title(title)
pylab.xlabel("$X_1$")
pylab.ylabel("$X_2$")
np.random.seed(3)
x1 = np.arange(0, 10, .2)
x2 = x1 + np.random.normal(loc=0, scale=1, size=len(x1))
good = (x1 > 5) | (x2 > 5)
bad = ~good
x1g = x1[good]
x2g = x2[good]
pylab.scatter(x1g, x2g, edgecolor="blue", facecolor="blue")
x1b = x1[bad]
x2b = x2[bad]
pylab.scatter(x1b, x2b, edgecolor="red", facecolor="white")
pylab.grid(True)
pylab.subplot(132)
X = np.c_[(x1, x2)] # X:[N, 2]
pca = decomposition.PCA(n_components=2)
Xtrans = pca.fit_transform(X) # Xtrans:[N, 1], 只有一个主成份
#Xtrans *=-1
Xg = Xtrans[good]
Xb = Xtrans[bad]
"""
pylab.scatter( Xg[:, 0], np.zeros(len(Xg)), edgecolor="blue", facecolor="blue")
pylab.scatter( Xb[:, 0], np.zeros(len(Xb)), edgecolor="red", facecolor="white")
"""
pylab.scatter(Xg[:, 0], Xg[:, 1], edgecolor="blue", facecolor="blue")
pylab.scatter(Xb[:, 0], Xb[:, 1], edgecolor="red", facecolor="white")
title = "Transformed feature space"
pylab.title(title)
pylab.xlabel("$X'$")
fig.axes[1].get_yaxis().set_visible(False)
title = "Transformed feature space(1-d)"
pylab.subplot(133)
pylab.scatter(Xg[:, 0],
np.zeros(len(Xg)),
edgecolor="blue",
facecolor="blue")
pylab.scatter(Xb[:, 0],
np.zeros(len(Xb)),
edgecolor="red",
facecolor="white")
pylab.title(title)
pylab.xlabel("$X'$")
fig.axes[2].get_yaxis().set_visible(False)
print("======系统pca 数据1==========")
print("特征值:", pca.explained_variance_)
print("特征重要度比率:", pca.explained_variance_ratio_)
# 可以看出, 手工pca与系统pca数据有个符号相反,并不影响分解,
# 比如 X=U*S*(V^T) = (-U) S*(-V^T)
# 对于二维数据, 其物理意义对应是按照顺时针旋转还是逆时针旋转,并不影响保持最大分量这一物理意义
print("系统Xg:", Xg)
pylab.grid(True)
pylab.autoscale(tight=True)
filename = "pca_demo_1.png"
pylab.savefig(os.path.join(CHART_DIR, filename), bbox_inches="tight")
callback = (odl.solvers.CallbackPrintIteration())
# Solve with initial guess x = 0.
# Step size parameters are selected to ensure convergence.
# See douglas_rachford_pd doc for more information.
x = ray_trafo.domain.zero()
odl.solvers.douglas_rachford_pd(x,
f,
g,
lin_ops,
tau=0.1,
sigma=[0.1, 0.1, 0.02],
lam=1.5,
niter=100,
callback=callback)
name = 'reco_l2TV_datafunc__lam_' + lam_str + '__l2reg_' + l2_reg_param_str
np.savetxt(path_load_result + name, x)
fig = x.show(clim=[-1, 1])
plt.axis('off')
fig.delaxes(fig.axes[1])
plt.autoscale(False)
typefig = 'png'
plt.savefig(path_load_result + name + '.' + typefig,
transparent=True,
bbox_inches='tight',
pad_inches=0,
format=typefig)
plt.close('all')
def plot_correlation():
# rerun
# plot 1
np.random.seed(66)
pylab.clf()
pylab.figure(num=None, figsize=(8, 8))
x = np.arange(0, 10, 0.2)
# print x
pylab.subplot(221)
y = 0.5 * x + norm.rvs(1, scale=0.01, size=len(x))
# print y
plot_correlation_func(x, y)
pylab.subplot(222)
y = 0.5 * x + norm.rvs(1, scale=.1, size=len(x))
plot_correlation_func(x, y)
pylab.subplot(223)
y = 0.5 * x + norm.rvs(1, scale=1, size=len(x))
plot_correlation_func(x, y)
pylab.subplot(224)
y = norm.rvs(1, scale=10, size=len(x))
plot_correlation_func(x, y)
# pylab.autoscale(tight=True)
pylab.grid(True)
filename = "a_correlation_pearsonr_1_linear_ok.png"
CHART_DIR = os.path.join(os.path.dirname(os.path.realpath(__file__)), "charts")
if not os.path.exists(CHART_DIR):
os.mkdir(CHART_DIR)
pylab.savefig(os.path.join(CHART_DIR, filename), bbox_inches="tight")
#########
# plot 2
pylab.clf()
pylab.figure(num=None, figsize=(8, 8))
x = np.arange(-5, 5, 0.2)
pylab.subplot(221)
y = 0.5 * x ** 2 + norm.rvs(1, scale=.01, size=len(x))
plot_correlation_func(x, y)
pylab.subplot(222)
y = 0.5 * x ** 2 + norm.rvs(1, scale=.1, size=len(x))
plot_correlation_func(x, y)
pylab.subplot(223)
y = 0.5 * x ** 2 + norm.rvs(1, scale=1, size=len(x))
plot_correlation_func(x, y)
pylab.subplot(224)
y = 0.5 * x ** 2 + norm.rvs(1, scale=10, size=len(x))
plot_correlation_func(x, y)
pylab.autoscale(tight=True)
pylab.grid(True)
filename = "a_correlation_pearsonr_2_quad_bad.png"
pylab.savefig(os.path.join(CHART_DIR, filename), bbox_inches="tight")
def plot_correlation():
# rerun
# plot 1
np.random.seed(66)
pylab.clf()
pylab.figure(num=None, figsize=(8, 8))
x = np.arange(0, 10, 0.2)
# print x
pylab.subplot(221)
y = 0.5 * x + norm.rvs(1, scale=0.01, size=len(x))
# print y
plot_correlation_func(x, y)
pylab.subplot(222)
y = 0.5 * x + norm.rvs(1, scale=.1, size=len(x))
plot_correlation_func(x, y)
pylab.subplot(223)
y = 0.5 * x + norm.rvs(1, scale=1, size=len(x))
plot_correlation_func(x, y)
pylab.subplot(224)
y = norm.rvs(1, scale=10, size=len(x))
plot_correlation_func(x, y)
# pylab.autoscale(tight=True)
pylab.grid(True)
filename = "a_correlation_pearsonr_1_linear_ok.png"
CHART_DIR = os.path.join(os.path.dirname(os.path.realpath(__file__)),
"charts")
if not os.path.exists(CHART_DIR):
os.mkdir(CHART_DIR)
pylab.savefig(os.path.join(CHART_DIR, filename), bbox_inches="tight")
#########
# plot 2
pylab.clf()
pylab.figure(num=None, figsize=(8, 8))
x = np.arange(-5, 5, 0.2)
pylab.subplot(221)
y = 0.5 * x**2 + norm.rvs(1, scale=.01, size=len(x))
plot_correlation_func(x, y)
pylab.subplot(222)
y = 0.5 * x**2 + norm.rvs(1, scale=.1, size=len(x))
plot_correlation_func(x, y)
pylab.subplot(223)
y = 0.5 * x**2 + norm.rvs(1, scale=1, size=len(x))
plot_correlation_func(x, y)
pylab.subplot(224)
y = 0.5 * x**2 + norm.rvs(1, scale=10, size=len(x))
plot_correlation_func(x, y)
pylab.autoscale(tight=True)
pylab.grid(True)
filename = "a_correlation_pearsonr_2_quad_bad.png"
pylab.savefig(os.path.join(CHART_DIR, filename), bbox_inches="tight")
#pylab.grid(True)
#pylab.autoscale(tight=True)
##case 2 LDA
good = x1 > x2
bad = ~good
lda_inst = lda.LDA(n_components=1)
Xtrans = lda_inst.fit_transform(X, good)
pylab.clf()
fig = pylab.figure(num=None, figsize=(10, 4))
pylab.subplot(121)
pylab.xlabel("$X_1$")
pylab.ylabel("$X_2$")
pylab.scatter(x1[good], x2[good], edgecolors="blue", facecolor="blue")
pylab.scatter(x1[bad], x2[bad], edgecolors="red", facecolor="white")
pylab.grid(True)
pylab.subplot(122)
pylab.xlabel("$X'$")
pylab.scatter(Xtrans[good],
np.zeros(len(Xtrans[good])),
edgecolors="blue",
facecolor="blue")
pylab.scatter(Xtrans[bad],
np.zeros(len(Xtrans[bad])),
edgecolors="red",
facecolor="white")
pylab.grid(True)
pylab.autoscale(tight=True)
def density_along_orbit(self, ax=None, annotate=True, min_fp_local_length=0, bg_color='dimgrey', cmap=None, vmin=1., vmax=3.):
if cmap is None:
cmap = plt.cm.autumn
cmap.set_bad('dimgrey',0.)
cmap.set_under('dimgrey',0.)
if ax is None:
ax = plt.gca()
fp_local_list = [d for d in self.digitization_list if np.isfinite(d.fp_local)]
plt.sca(ax)
mex.plot_planet(lw=3.)
mex.plot_bs(lw=1., ls='dashed', color='k')
mex.plot_mpb(lw=1., ls='dotted', color='k')
ax.set_aspect('equal', 'box')
plt.xlim(2,-2)
plt.autoscale(False,tight=True)
plt.ylim(0,1.9999)
if annotate:
plt.annotate('%d' % self.orbit, (0.05, 0.85), xycoords='axes fraction', va='top')
def f_x(pos):
return pos[0] / mex.mars_mean_radius_km
def f_y(pos):
return np.sqrt(pos[1]**2. + pos[2]**2.) / mex.mars_mean_radius_km
# def f_y(pos):
# return pos[2] / mex.mars_mean_radius_km
plt.plot( f_x(self.mso_pos), f_y(self.mso_pos),
color=bg_color, lw=1., zorder=-10)
inx = np.interp(np.array([d.time for d in self.ionogram_list]), self.t, np.arange(self.t.shape[0]))
inx = inx.astype(int)
plt.plot( f_x(self.mso_pos[:,inx]), f_y(self.mso_pos[:,inx]),
color=bg_color, ls='None',marker='o', ms=8.,mew=0., mec=bg_color, zorder=-9)
if fp_local_list:
val = np.empty_like(self.t) + np.nan
# for t, v in [(float(f.time), np.log10(ais_code.fp_to_ne(f.maximum_fp_local))) for f in fp_local_list]:
# val[np.abs(self.t - t) < ais_code.ais_spacing_seconds] = v
for f in fp_local_list:
t = float(f.time)
# if (f.fp_local_error / f.fp_local) > 0.3:
# v = np.log10(20.)
# else:
v = np.log10(ais_code.fp_to_ne(f.fp_local))
# print t, ais_code.fp_to_ne(f.fp_local), f.fp_local_error/f.fp_local > 0.3
val[np.abs(self.t - t) < ais_code.ais_spacing_seconds] = v
points = np.array([f_x(self.mso_pos), f_y(self.mso_pos)]).T.reshape(-1, 1, 2)
segments = np.concatenate([points[:-1], points[1:]], axis=1)
lc = LineCollection(segments, cmap=cmap, norm=plt.Normalize(vmin=vmin, vmax=vmax, clip=True))
lc.set_array(val)
lc.set_linewidth(5)
plt.gca().add_collection(lc)
else:
lc = None
plt.ylabel(r'$\rho / R_M$')
# plt.ylabel(r'$z / R_M$')
plt.xlabel(r'$x / R_M$')
if lc:
ticks = [i for i in range(10) if ((float(i)+0.1) > vmin) & ((float(i)-0.1) < vmax)]
old_ax = plt.gca()
plt.colorbar(lc, cax = celsius.make_colorbar_cax(offset=0.001, height=0.8),
ticks=ticks).set_label(r'$log_{10}\;n_e / cm^{-3}$')
plt.sca(old_ax)
def modb_along_orbit(self, ax=None, annotate=True, bg_color='dimgrey', cmap=None, vmin=0., vmax=20.):
if ax is None:
ax = plt.gca()
if cmap is None:
cmap = plt.cm.autumn
cmap.set_bad('dimgrey',0.)
cmap.set_under('dimgrey',0.)
td_cyclotron_list = [d for d in self.digitization_list if np.isfinite(d.td_cyclotron)]
plt.sca(ax)
mex.plot_planet(lw=3.)
mex.plot_bs(lw=1., ls='dashed', color='k')
mex.plot_mpb(lw=1., ls='dotted', color='k')
ax.set_aspect('equal','box')
plt.xlim(2,-2)
plt.autoscale(False,tight=True)
plt.ylim(0., 1.999)
if annotate:
plt.annotate('%d' % self.orbit, (0.05, 0.85),
xycoords='axes fraction', va='top')
def f_x(pos):
return pos[0] / mex.mars_mean_radius_km
def f_y(pos):
return np.sqrt(pos[1]**2. + pos[2]**2.) / mex.mars_mean_radius_km
# def f_y(pos):
# return pos[2] / mex.mars_mean_radius_km
plt.plot( f_x(self.mso_pos), f_y(self.mso_pos),
color=bg_color, lw=2., zorder=-10)
inx = np.interp(np.array([d.time for d in self.ionogram_list]),
self.t, np.arange(self.t.shape[0]))
inx = inx.astype(int)
plt.plot(f_x(self.mso_pos[:,inx]), f_y(self.mso_pos[:,inx]),
color=bg_color, ls='None',marker='o', ms=8., mew=0., mec=bg_color, zorder=-9)
if td_cyclotron_list:
val = np.empty_like(self.t) + np.nan
for t, v in [(float(f.time), 1E9 * ais_code.td_to_modb(f.td_cyclotron))
for f in td_cyclotron_list]:
val[np.abs(self.t - t) < ais_code.ais_spacing_seconds] = v
points = np.array([f_x(self.mso_pos), f_y(self.mso_pos)]).T.reshape(-1, 1, 2)
segments = np.concatenate([points[:-1], points[1:]], axis=1)
lc = LineCollection(segments, cmap=cmap,
norm=plt.Normalize(vmin=vmin, vmax=vmax, clip=True))
lc.set_array(val)
lc.set_linewidth(5)
plt.gca().add_collection(lc)
else:
lc = None
plt.ylabel(r'$\rho / R_M$')
# plt.ylabel(r'$z / R_M$')
plt.xlabel(r'$x / R_M$')
if lc:
old_ax = plt.gca()
plt.colorbar(lc, cax = celsius.make_colorbar_cax(offset=0.001, height=0.8)
).set_label(r'$|B| / nT$')
plt.sca(old_ax)
def plot_simple_demo_my_pca():
pylab.clf()
fig = pylab.figure(num=None, figsize=(25, 4))
pylab.subplot(151)
title = "Original feature space"
pylab.title(title)
pylab.xlabel("$X_1$")
pylab.ylabel("$X_2$")
np.random.seed(3)
x1 = np.arange(0, 10, .2)
x2 = x1 + np.random.normal(loc=0, scale=1, size=len(x1))
# 两者中,有一个>5就是good
good = (x1 > 5) | (x2 > 5)
bad = ~good
x1g = x1[good]
x2g = x2[good]
pylab.scatter(x1g, x2g, edgecolor="blue", facecolor="blue")
x1b = x1[bad]
x2b = x2[bad]
pylab.scatter(x1b, x2b, edgecolor="red", facecolor="white")
pylab.grid(True)
pylab.subplot(152)
X = np.c_[(x1, x2)] # X:[N, 2]
Xtrans, eigen_values, V, x_mean = my_pca(X, k_components=2, whiten=False)
Xtrans_whiten, _, _, _ = my_pca(np.copy(X), k_components=2, whiten=True)
Xg = Xtrans[good]
Xb = Xtrans[bad]
print("======手工pca==========")
print("原始特征值 eigen_values:", eigen_values)
print("手工Xg:", Xg)
print("手工Xg(whiten):", Xtrans_whiten[good])
pylab.scatter(Xg[:, 0], Xg[:, 1], edgecolor="blue", facecolor="blue")
pylab.scatter(Xb[:, 0], Xb[:, 1], edgecolor="red", facecolor="white")
#Xb[:, 0], np.zeros(len(Xb)), edgecolor="red", facecolor="white")
title = "Transformed feature space"
pylab.title(title)
pylab.xlabel("$X'$")
fig.axes[1].get_yaxis().set_visible(True)
fig.axes[1].set_xlim(-6, 6)
fig.axes[1].set_ylim(-6, 6)
#fig.axes.set_ylim(-6,6)
# 白化
Xg = Xtrans_whiten[good]
Xb = Xtrans_whiten[bad]
title = "Transformed feature space(whiten)"
pylab.subplot(153)
pylab.scatter(Xg[:, 0], Xg[:, 1], edgecolor="blue", facecolor="blue")
pylab.scatter(Xb[:, 0], Xb[:, 1], edgecolor="red", facecolor="white")
pylab.title(title)
pylab.xlabel("$X'$")
fig.axes[2].get_yaxis().set_visible(True)
fig.axes[2].set_xlim(-6, 6)
fig.axes[2].set_ylim(-6, 6)
# X_new_1d = X*V[:,0:1]
title = "Transformed feature space(1-d)"
pylab.subplot(154)
pylab.scatter(Xg[:, 0],
np.zeros(len(Xg)),
edgecolor="blue",
facecolor="blue")
pylab.scatter(Xb[:, 0],
np.zeros(len(Xb)),
edgecolor="red",
facecolor="white")
pylab.title(title)
pylab.xlabel("$X'$")
fig.axes[3].get_yaxis().set_visible(True)
fig.axes[3].set_xlim(-6, 6)
fig.axes[3].set_ylim(-6, 6)
# X_new_1d = X*V[:,0:1]
# 重构信号
X_reconstruct = Xtrans[:, 0:1] * (V[:, 0:1].T) + x_mean
Xg = X_reconstruct[good]
Xb = X_reconstruct[bad]
title = "reconstruct from transformed feature space(1-d)"
pylab.xlabel("$X'$")
pylab.subplot(155)
pylab.scatter(Xg[:, 0], Xg[:, 1], edgecolor="blue", facecolor="blue")
pylab.scatter(Xb[:, 0], Xb[:, 1], edgecolor="red", facecolor="white")
pylab.title(title)
pylab.xlabel("$X'$")
fig.axes[4].get_yaxis().set_visible(True)
fig.axes[4].set_xlim(-6, 10)
fig.axes[4].set_ylim(-6, 10)
#pylab.grid(True)
pylab.autoscale(tight=True)
filename = "my_pca_demo.png"
pylab.savefig(os.path.join(CHART_DIR, filename), bbox_inches="tight")
def plot_mi_demo():
# determinism!
np.random.seed(0)
# plot1
pylab.clf()
pylab.figure(num=None, figsize=(8, 8))
x = np.arange(0, 10, 0.2)
pylab.subplot(221)
y = 0.5 * x + norm.rvs(1, scale=.01, size=len(x))
plot_mi_func(x, y)
pylab.subplot(222)
y = 0.5 * x + norm.rvs(1, scale=.1, size=len(x))
plot_mi_func(x, y)
pylab.subplot(223)
y = 0.5 * x + norm.rvs(1, scale=1, size=len(x))
plot_mi_func(x, y)
pylab.subplot(224)
y = norm.rvs(1, scale=10, size=len(x))
plot_mi_func(x, y)
pylab.autoscale(tight=True)
pylab.grid(True)
filename = "c_mutual_information_1_ok.png"
pylab.savefig(os.path.join(CHART_DIR, filename), bbox_inches="tight")
##############
# plot 2
pylab.clf()
pylab.figure(num=None, figsize=(8, 8))
x = np.arange(-5, 5, 0.2)
pylab.subplot(221)
y = 0.5 * x**2 + norm.rvs(1, scale=.01, size=len(x))
plot_mi_func(x, y)
pylab.subplot(222)
y = 0.5 * x**2 + norm.rvs(1, scale=.1, size=len(x))
plot_mi_func(x, y)
pylab.subplot(223)
y = 0.5 * x**2 + norm.rvs(1, scale=1, size=len(x))
plot_mi_func(x, y)
pylab.subplot(224)
y = 0.5 * x**2 + norm.rvs(1, scale=10, size=len(x))
plot_mi_func(x, y)
pylab.autoscale(tight=True)
pylab.grid(True)
filename = "c_mutual_information_2_ok.png"
pylab.savefig(os.path.join(CHART_DIR, filename), bbox_inches="tight")
def diagnostic(pipe):
"""
Print a 1-page diagnostic plot of a given pipeline object
Args:
pipe (terra.pipeline.Pipeline) : pipeline object
"""
fig = plt.figure(figsize=(20,12))
gs_kw = dict(left=0.04,right=0.99)
# Provision the upper plots
gs1 = GridSpec(3,10)
gs1.update(hspace=0.2, wspace=0.6,bottom=0.65,top=0.95, **gs_kw)
axPeriodogram = fig.add_subplot(gs1[0,0:8])
axACF = fig.add_subplot(gs1[0,8])
# These 4 plots all share an xaxis
ax_transit = fig.add_subplot(gs1[1,0:2])
ax_transit_zoom = fig.add_subplot(gs1[2,0:2],sharex=ax_transit)
ax_phasefold_180 = fig.add_subplot(gs1[1,2:4],sharex=ax_transit)
ax_phasefold_secondary = fig.add_subplot(gs1[2,2:4],sharex=ax_transit)
ax_phasefold = fig.add_subplot(gs1[1:,4:8])
axTransitTimes = fig.add_subplot(gs1[1,8])
axTransitRp = fig.add_subplot(gs1[2,8])
# Provision the lower plots
gs2 = GridSpec(1,5)
gs2.update(hspace=0.05,wspace=0.2,bottom=0.05,top=0.6,**gs_kw)
axSES = fig.add_subplot(gs2[0 ,0:4])
axTransitStack = fig.add_subplot(gs2[0 ,4])
# Periodogram
plt.sca(axPeriodogram)
periodogram(pipe)
ax = plt.gca()
at = AnchoredText('Periodogram',prop=tprop, frameon=True,loc=2)
ax.add_artist(at)
# Auto correlation plot, and header text, defined to the right
plt.sca(axACF)
ax = plt.gca()
autocorr(pipe)
AddAnchored("ACF",prop=tprop,frameon=True,loc=2)
# Transit times, holding depth constant
plt.sca(axTransitTimes)
transits(pipe,mode='transit-times')
# Transit radius ratio
plt.sca(axTransitRp)
transits(pipe,mode='transit-rp')
# Phasefolded photometry
plt.sca(ax_transit)
phasefold_transit(pipe,mode='transit')
plt.ylabel('Flux')
# Zoomed phase folded photometry
plt.sca(ax_transit_zoom)
phasefold_transit(pipe,mode='transit-zoom')
# Light curve 180 degrees out of phase
plt.sca(ax_phasefold_180)
phasefold_shifted(pipe, 0.5)
# Phase-folded light curve at the most significant secondary eclipse
plt.sca(ax_phasefold_secondary)
phasefold_shifted(pipe, pipe.se_phase)
# Phase-folded all phases
plt.sca(ax_phasefold)
lc, lcbin = phasefold_shifted(pipe, 0, xdata='phase')
yl = [lcbin.fmed.min(),lcbin.fmed.max()]
span = lcbin.fmed.ptp()
yl[0] -= 0.25 * span
yl[1] += 0.25 * span
plt.ylim(*yl)
plt.autoscale(axis='x',tight=True)
plt.xlabel('Phase')
plt.ylabel('Flux')
# Single Event Statistic Stack
plt.sca(axSES)
mad = pipe.lc.f.abs().median()
ystep = 3 * 1.6 * mad
ystep = max(pipe.fit_rp**2 * 1.0, ystep)
single_event_statistic_stack(pipe, ystep=ystep)
pipe.se_phase
plt.axvline(0, alpha=.1,lw=10,color='m',zorder=1)
# Single Event Statistic Stack
plt.sca(axTransitStack)
transit_stack(pipe,ystep=ystep)
axL = [
ax_transit, ax_transit_zoom, ax_phasefold_180, ax_phasefold_secondary,
ax_phasefold, axTransitTimes, axTransitRp
]
for ax in axL:
ax.yaxis.set_major_locator(MaxNLocator(4))
axL = [
ax_transit, ax_transit_zoom, ax_phasefold_180, ax_phasefold,
axTransitTimes, axTransitRp, axSES, axTransitStack,
]
for ax in fig.get_axes():
ax.grid()
plt.sca(axACF)
text = header_text(pipe)
plt.text(
1.1, 1.0, text, family='monospace', size='large',
transform=axACF.transAxes, va='top',
)
pylab.clf()
pylab.figure(num=None, figsize=(8, 8))
x = np.arange(0, 10, 0.2)
y = 0.5 * x + norm.rvs(1, scale=.01, size=len(x))
obj.set_data_set(x, y)
obj.plot_correlation_func(221)
y = 0.5 * x + norm.rvs(1, scale=.1, size=len(x))
obj.set_data_set(x, y)
obj.plot_correlation_func(222)
y = 0.5 * x + norm.rvs(1, scale=1, size=len(x))
obj.set_data_set(x, y)
obj.plot_correlation_func(223)
y = norm.rvs(1, scale=10, size=len(x))
obj.set_data_set(x, y)
obj.plot_correlation_func(224)
pylab.autoscale(tight=True)
pylab.grid(True)
filename = "corr_demo_1.png"
pylab.savefig(os.path.join(CHART_DIR, filename), bbox_inches="tight")
pylab.clf()
pylab.figure(num=None, figsize=(8, 8))
x = np.arange(-5, 5, 0.2)
y = 0.5 * x ** 2 + norm.rvs(1, scale=.01, size=len(x))
obj.set_data_set(x, y)
obj.plot_correlation_func(221)
y = 0.5 * x ** 2 + norm.rvs(1, scale=.1, size=len(x))
obj.set_data_set(x, y)
obj.plot_correlation_func(222)
y = 0.5 * x ** 2 + norm.rvs(1, scale=1, size=len(x))
