ax1 = plt.subplot(2, 1, 1)
if data.shape[0] > 4:
minmean = data[3]
maxmean = data[4]
plt.fill_between(moves,
minmean,
maxmean,
edgecolor='none',
linewidth=0,
color=colors.color(method[1:]),
alpha=0.1,
zorder=-51)
my_S_error = errorinentropy[0:len(iterations)]
min_error = min(min_error, my_S_error[my_S_error > 0].min())
colors.loglog(moves, my_S_error, method=method[1:])
# make these tick labels invisible
plt.setp(ax1.get_xticklabels(), visible=False)
#plt.xlabel(r'$\textrm{Moves}$')
plt.ylabel(r'$\textrm{Average Entropy Error}$')
#plt.title('Average Entropy Error at Each Iteration, %s' %filebase)
if "default" in transcale:
plt.title(r'$N=%d$, $\eta = %g$ $\delta_0 = 0.05$' %
(int(N), ff))
elif "slow" in transcale:
plt.title(r'$N=%d$, $\eta = %g$ $\delta_0 = 0.005$' %
(int(N), ff))
elif "fast" in transcale:
plt.title(r'$N=%d$, $\eta = %g$ $\delta_0 = 0.5$' %
(int(N), ff))
wlmoves, wlfactor = np.loadtxt(wl, dtype = float, unpack = True)
data = np.loadtxt(wl)
moves = data[:, 0]
factor = data[:, 1]
if (data[0, 0] == 'wl_factor'): # using c++ data!
moves = np.zeros(len(wlmoves)*2+2)
factor = np.zeros_like(moves)
factor[0] = 1
moves[0] = 1
for i in range(len(wlmoves)):
moves[2*i+1] = wlmoves[i]
moves[2*i+2] = wlmoves[i]
factor[2*i+1] = wlfactor[i]*2
factor[2*i+2] = wlfactor[i]
colors.loglog(moves, factor,
'vanilla_wang_landau'
+ wl[len("data/gamma/%s/wl" % filename):-4])
plt.ylim(ymin=1e-10, ymax=1e1)
plt.xlim(xmin=1e0, xmax=2e12)
except:
pass
print((glob.glob("data/gamma/%s/sad*.dat" % filename)))
try:
for sad in glob.glob("data/gamma/%s/sad*.dat" % filename):
data = np.loadtxt(sad)
ts = data[:, 0]
avg_gamma = data[:, 1]
sadname = sad.split('/')[-1].split('.')[0]
colors.legend()
plt.savefig('figs/%s-round-trips-%g.pdf' % (tex_filebase, energy))
plt.figure('error-at-energy-round-trips')
colors.plot(Nrt_at_energy[Nrt_at_energy > 0],
erroratenergy[Nrt_at_energy > 0],
method=method[1:])
plt.title('Error at energy %g %s' % (energy, filebase))
plt.xlabel('Round Trips')
plt.ylabel('Error')
colors.legend()
plt.savefig('figs/%s-error-energy-Nrt-%g.pdf' %
(tex_filebase, energy))
plt.figure('maxerror')
colors.loglog(moves, maxerror, method=method[1:])
plt.xlabel(r'Moves')
plt.ylabel(r'Maximum Entropy Error')
#plt.title('Maximum Entropy Error vs Iterations, %s' %filebase)
colors.legend()
if type(iterations) is not np.float64:
plt.figure('errorinentropy')
if data.shape[0] > 4:
minmean = data[3]
maxmean = data[4]
plt.fill_between(moves,
minmean,
maxmean,
edgecolor='none',
linewidth=0,
plt.figure(figsize=(5,4))
try:
for wl in glob.glob("data/gamma/%s/wl*.txt" % filename):
wlmoves, wlfactor = np.loadtxt(wl, dtype = float, unpack = True)
moves = np.zeros(len(wlmoves)*2+2)
factor = np.zeros_like(moves)
factor[0] = 1
moves[0] = 1
for i in range(len(wlmoves)):
moves[2*i+1] = wlmoves[i]
moves[2*i+2] = wlmoves[i]
factor[2*i+1] = wlfactor[i]*2
factor[2*i+2] = wlfactor[i]
colors.loglog(moves, factor,
'vanilla_wang_landau'
+ wl[len("data/gamma/%s/wl" % filename):-4])
plt.ylim(ymin=1e-10, ymax=1e1)
plt.xlim(xmin=1e0, xmax=1e13)
except:
pass
for sad in glob.glob("data/gamma/%s/sad*.dat" % filename):
data = np.loadtxt(sad)
print data.shape
if data.shape[1] > 2: # i.e. we are not using parse-yaml-out.py
num_sad_states = data[:,0]
time = data[:,1]
ehi = data[:,2]
wlmoves, wlfactor = np.loadtxt(wl, dtype=float, unpack=True)
data = np.loadtxt(wl)
moves = data[:, 0]
factor = data[:, 1]
if (data[0, 0] == 'wl_factor'): # using c++ data!
moves = np.zeros(len(wlmoves) * 2 + 2)
factor = np.zeros_like(moves)
factor[0] = 1
moves[0] = 1
for i in range(len(wlmoves)):
moves[2 * i + 1] = wlmoves[i]
moves[2 * i + 2] = wlmoves[i]
factor[2 * i + 1] = wlfactor[i] * 2
factor[2 * i + 2] = wlfactor[i]
colors.loglog(
moves, factor,
'is' + wl[len("../ising/data/gamma/%s/wl" % filename):-4])
#plt.ylim(1e-10, 1e1)
#plt.xlim(1e3, 1e13)
except:
pass
try:
for sad in glob.glob("../ising/data/gamma/%s/*sad*.dat" % filename):
data = np.loadtxt(sad)
print(data.shape)
if data.shape[1] > 2: # i.e. we are not using parse-yaml-out.py
num_sad_states = data[:, 0]
time = data[:, 1]
ehi = data[:, 2]
longest_iters = data[:, 0]
best = np.zeros(most_iters) + 1e30
worst = np.zeros(most_iters)
for i in range(most_iters):
for dE in errors:
if i < len(errors[dE]):
best[i] = min(best[i], errors[dE][i])
worst[i] = max(worst[i], errors[dE][i])
if 'sad' in method:
# plt.fill_between(longest_iters, worst, best,
# edgecolor='none', linewidth=0,
# color=colors.color(method),
# alpha=0.1, zorder=-51)
# colors.loglog(iters[0.001], errors[0.001], method=r'SAD $\Delta E=0.001\epsilon$')
colors.loglog(iters[0.01],
errors[0.01],
method=r'SAD $\Delta E=0.01\epsilon$')
colors.loglog(iters[0.1],
errors[0.1],
method=r'SAD $\Delta E=0.1\epsilon$')
else:
colors.loglog(iters[0.01], errors[0.01], method=method)
for method in extra_files:
fname = extra_files[method]
data = np.loadtxt(datadir + fname + '-cv-error.txt')
if len(data) == 0:
continue
colors.loglog(data[:, 0], data[:, 2], method=method)
moves = np.array([1e6, 1e13])
plt.xlabel('# moves')
plt.ylabel('Round Trips')
colors.legend()
plt.savefig('figs/%s-round-trips-%g.pdf' % (tex_filebase, energy))
plt.figure('error-at-energy-round-trips')
colors.plot(Nrt_at_energy[Nrt_at_energy > 0], erroratenergy[Nrt_at_energy > 0],
method = method[1:])
plt.title('Error at energy %g %s' % (energy,filebase))
plt.xlabel('Round Trips')
plt.ylabel('Error')
colors.legend()
plt.savefig('figs/%s-error-energy-Nrt-%g.pdf' % (tex_filebase, energy))
plt.figure('maxerror')
colors.loglog(moves, maxerror, method = method[1:])
plt.xlabel(r'Moves')
plt.ylabel(r'Maximum Entropy Error')
#plt.title('Maximum Entropy Error vs Iterations, %s' %filebase)
colors.legend()
if type(iterations) is not np.float64:
plt.figure('errorinentropy')
my_S_error = errorinentropy[0:len(iterations)]
min_error = min(min_error, my_S_error[my_S_error > 0].min())
colors.loglog(moves, my_S_error, method = method[1:])
plt.xlabel(r'$\textrm{Moves}$')
plt.ylabel(r'$\textrm{Average Entropy Error}$')
#plt.title('Average Entropy Error at Each Iteration, %s' %filebase)
if "default" in transcale:
plt.title(r'$N=%d$, $\eta = %g$ $\delta_0 = 0.05$' % (int(N), ff))
plt.figure(figsize=(5, 4))
try:
for wl in glob.glob("data/gamma/%s/wl*.txt" % filename):
wlmoves, wlfactor = np.loadtxt(wl, dtype=float, unpack=True)
moves = np.zeros(len(wlmoves) * 2 + 2)
factor = np.zeros_like(moves)
factor[0] = 1
moves[0] = 1
for i in range(len(wlmoves)):
moves[2 * i + 1] = wlmoves[i]
moves[2 * i + 2] = wlmoves[i]
factor[2 * i + 1] = wlfactor[i] * 2
factor[2 * i + 2] = wlfactor[i]
colors.loglog(
moves, factor,
'vanilla_wang_landau' + wl[len("data/gamma/%s/wl" % filename):-4])
plt.ylim(ymin=1e-10, ymax=1e1)
plt.xlim(xmin=1e0, xmax=1e13)
except:
pass
for sad in glob.glob("data/gamma/%s/sad*.dat" % filename):
data = np.loadtxt(sad)
print data.shape
if data.shape[1] > 2: # i.e. we are not using parse-yaml-out.py
num_sad_states = data[:, 0]
time = data[:, 1]
ehi = data[:, 2]
elo = data[:, 3]
wlmoves, wlfactor = np.loadtxt(wl, dtype=float, unpack=True)
data = np.loadtxt(wl)
moves = data[:, 0]
factor = data[:, 1]
if (data[0, 0] == 'wl_factor'): # using c++ data!
moves = np.zeros(len(wlmoves) * 2 + 2)
factor = np.zeros_like(moves)
factor[0] = 1
moves[0] = 1
for i in range(len(wlmoves)):
moves[2 * i + 1] = wlmoves[i]
moves[2 * i + 2] = wlmoves[i]
factor[2 * i + 1] = wlfactor[i] * 2
factor[2 * i + 2] = wlfactor[i]
colors.loglog(
moves, factor,
'wl' + wl[len("../square-well/data/gamma/%s/wl" % filename):-4])
plt.ylim(ymin=1e-10, ymax=1e1)
plt.xlim(xmin=1e3, xmax=1e12)
except:
pass
try:
for sad in glob.glob("../square-well/data/gamma//%s/sad*.dat" % filename):
data = np.loadtxt(sad)
print((data.shape))
if data.shape[1] > 2: # i.e. we are not using parse-yaml-out.py
num_sad_states = data[:, 0]
time = data[:, 1]
ehi = data[:, 2]