"M_mil_hi" : M_lsm_hi,
"R_mil_mu" : R_lsm_mu,
"R_mil_lo" : R_lsm_lo,
"R_mil_hi" : R_lsm_hi
}
)
df.to_csv("../../data/an_mil_c{0:}.csv".format(MYCASE))
##################################################
# Report
##################################################
print("Execution time: {} sec".format(t1-t0))
Colors = linspecer(5)
### Effective margin
plt.figure()
## Data
# Basis Value
plt.plot(N_ALL,100*M_bv_lo,
color = Colors[0,:], linewidth = 1.0, linestyle = ":")
plt.plot(N_ALL,100*M_bv_mu,
color = Colors[0,:], linewidth = 2.0, label='AN + BV')
for i in range(len(N_ALL)):
plt.plot([N_ALL[i]*0.95, N_ALL[i]*0.95],
[100*M_bv_mu[i], 100*M_bv_lo[i]],
color = Colors[0,:],
linewidth = 0.5)
# matplotlib import block
import matplotlib
# choose the backend to handle window geometry
matplotlib.use("Qt4Agg")
# Import pyplot
import matplotlib.pyplot as plt
# Plot settings
offset = [(0,500),(700,500),(1400,500)] # window locations
# Example details
import numpy as np
import pyutil.plotting as ut
# Example plots
C = ut.linspecer(len(offset))
for i in range(len(offset)):
# Generate some fake data
X = np.linspace(-1,1)
Y = np.random.random(len(X))
# Plot
plt.figure()
plt.plot(X,Y,color=C[i])
plt.xlabel('X')
plt.ylabel('Y')
plt.title('Data Set {}'.format(i+1))
# Set plot location on screen
manager = plt.get_current_fig_manager()
# Grab window dimensions
x,y,dx,dy = manager.window.geometry().getRect()
df["R_pri{0:d}_mu".format(p)] = R_pri_mc_mu[:, lnd]
df["R_pri{0:d}_lo".format(p)] = R_pri_mc_lo[:, lnd]
df["R_pri{0:d}_hi".format(p)] = R_pri_mc_hi[:, lnd]
df.to_csv("../../data/mc_mip_c{0:}{1:}.csv".format(MYCASE, suffix))
##################################################
# Report
##################################################
print("Execution time: {} sec".format(t1 - t0))
### Label formatting
Labels = ['L={0:1.0e}'.format(l) for l in L_ALL]
nplt = len(pIdx)
Colors = linspecer(len(L_ALL) * nplt)
### Effective margin --------------------------------------------------
plt.figure()
## Data
for lnd in range(len(L_ALL)):
# MC mip
if 0 in pIdx:
plt.plot(N_ALL,
100 * M_mip_mc_lo[:, lnd],
color=Colors[0 + nplt * lnd, :],
linewidth=1.0,
linestyle=":")
plt.plot(N_ALL,
100 * M_mip_mc_mu[:, lnd],
color=Colors[0 + nplt * lnd, :],
"""
# matplotlib import block
import matplotlib
# choose the backend to handle window geometry
matplotlib.use("Qt4Agg")
# Import pyplot
import matplotlib.pyplot as plt
# Plot settings
offset = [(0, 500), (700, 500), (1400, 500)] # window locations
# Example details
import numpy as np
import pyutil.plotting as ut
# Example plots
C = ut.linspecer(len(offset))
for i in range(len(offset)):
# Generate some fake data
X = np.linspace(-1, 1)
Y = np.random.random(len(X))
# Plot
plt.figure()
plt.plot(X, Y, color=C[i])
plt.xlabel('X')
plt.ylabel('Y')
plt.title('Data Set {}'.format(i + 1))
# Set plot location on screen
manager = plt.get_current_fig_manager()
# Grab window dimensions
x, y, dx, dy = manager.window.geometry().getRect()
if dist != None:
print "Subspace Dist = \n{}".format(dist)
print "Leading Vectors:"
for i in range(m_des):
print "W[:," + str(i) + "] = \n{}".format(W[:, i])
##################################################
# Plotting
##################################################
if plotting:
### Residual sequences
length = len(Res_full)
label_len = min(length, 10)
longest = 0
colors = linspecer(length)
sty = '-'
mkr = 'o'
fig = plt.figure()
for i in range(length):
plt.plot(Res_full[i], color=colors[i], linestyle=sty, marker=mkr)
longest = max(longest, len(Res_full[i]))
plt.yscale('log')
plt.xlim([-0.5, longest + 0.5])
# Annotation
plt.title("Residual Sequences: " + title_case)
plt.xlabel("Iteration")
plt.ylabel("Residual")
plt.legend(['Stage ' + str(i + 1) for i in range(label_len)])
W_all.append(W_r)
L_all.append(L_r)
D_all.append(d_r)
# Form numpy arrays
L_all = np.array(L_all)
D_all = np.array(D_all)
# Bootstrap analysis
L_mean = np.array([L_all[:,i].mean() for i in range(L_all.shape[1])])
L_std = np.array([L_all[:,i].std() for i in range(L_all.shape[1])])
D_mean = np.array([D_all[:,i].mean() for i in range(D_all.shape[1])])
D_std = np.array([D_all[:,i].std() for i in range(D_all.shape[1])])
# Plot results
# --------------------------------------------------
colors = linspecer(2)
sig = 2. # Number of sigma to plot as interval
# Subspace Distance
ind = range(1,m)
D_lo= D_mean-D_std*sig; D_lo = [max(d,1e-25) for d in D_lo]
D_hi= D_mean+D_std*sig
plt.figure()
plt.plot(ind,D_mean,color=colors[0],marker='*',linestyle='-')
plt.fill_between(ind,D_lo,D_hi,alpha=0.5,color=colors[0])
plt.title('Subspace Distance')
plt.yscale('log')
plt.ylim((1e-15,max(D_mean)*10))
plt.xlim((0,m))
# Set plot location on screen